From 2a7e2990340a1c42321d79dc99755683ea5a1026 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 21 Sep 2005 18:50:51 -0700 Subject: [SPARC64]: Move kernel TLB miss handling into a seperate file. Signed-off-by: David S. Miller --- arch/sparc64/kernel/dtlb_base.S | 8 +- arch/sparc64/kernel/entry.S | 153 ----------------------------------- arch/sparc64/kernel/head.S | 1 + arch/sparc64/kernel/ktlb.S | 174 ++++++++++++++++++++++++++++++++++++++++ 4 files changed, 179 insertions(+), 157 deletions(-) create mode 100644 arch/sparc64/kernel/ktlb.S (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/dtlb_base.S b/arch/sparc64/kernel/dtlb_base.S index ded2fed23fc..702d349c1e8 100644 --- a/arch/sparc64/kernel/dtlb_base.S +++ b/arch/sparc64/kernel/dtlb_base.S @@ -71,7 +71,7 @@ from_tl1_trap: rdpr %tl, %g5 ! For TL==3 test CREATE_VPTE_OFFSET1(%g4, %g6) ! Create VPTE offset - be,pn %xcc, 3f ! Yep, special processing + be,pn %xcc, kvmap ! Yep, special processing CREATE_VPTE_OFFSET2(%g4, %g6) ! Create VPTE offset cmp %g5, 4 ! Last trap level? be,pn %xcc, longpath ! Yep, cannot risk VPTE miss @@ -83,9 +83,9 @@ from_tl1_trap: nop ! Delay-slot 9: stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB retry ! Trap return -3: brlz,pt %g4, 9b ! Kernel virtual map? - xor %g2, %g4, %g5 ! Finish bit twiddles - ba,a,pt %xcc, kvmap ! Yep, go check for obp/vmalloc + nop + nop + nop /* DTLB ** ICACHE line 3: winfixups+real_faults */ longpath: diff --git a/arch/sparc64/kernel/entry.S b/arch/sparc64/kernel/entry.S index b4834952785..45cd3bbdb7e 100644 --- a/arch/sparc64/kernel/entry.S +++ b/arch/sparc64/kernel/entry.S @@ -30,159 +30,6 @@ .text .align 32 - .globl sparc64_vpte_patchme1 - .globl sparc64_vpte_patchme2 -/* - * On a second level vpte miss, check whether the original fault is to the OBP - * range (note that this is only possible for instruction miss, data misses to - * obp range do not use vpte). If so, go back directly to the faulting address. - * This is because we want to read the tpc, otherwise we have no way of knowing - * the 8k aligned faulting address if we are using >8k kernel pagesize. This - * also ensures no vpte range addresses are dropped into tlb while obp is - * executing (see inherit_locked_prom_mappings() rant). - */ -sparc64_vpte_nucleus: - /* Note that kvmap below has verified that the address is - * in the range MODULES_VADDR --> VMALLOC_END already. So - * here we need only check if it is an OBP address or not. - */ - sethi %hi(LOW_OBP_ADDRESS), %g5 - cmp %g4, %g5 - blu,pn %xcc, sparc64_vpte_patchme1 - mov 0x1, %g5 - sllx %g5, 32, %g5 - cmp %g4, %g5 - blu,pn %xcc, obp_iaddr_patch - nop - - /* These two instructions are patched by paginig_init(). */ -sparc64_vpte_patchme1: - sethi %hi(0), %g5 -sparc64_vpte_patchme2: - or %g5, %lo(0), %g5 - - /* With kernel PGD in %g5, branch back into dtlb_backend. */ - ba,pt %xcc, sparc64_kpte_continue - andn %g1, 0x3, %g1 /* Finish PMD offset adjustment. */ - -vpte_noent: - /* Restore previous TAG_ACCESS, %g5 is zero, and we will - * skip over the trap instruction so that the top level - * TLB miss handler will thing this %g5 value is just an - * invalid PTE, thus branching to full fault processing. - */ - mov TLB_SFSR, %g1 - stxa %g4, [%g1 + %g1] ASI_DMMU - done - - .globl obp_iaddr_patch -obp_iaddr_patch: - /* These two instructions patched by inherit_prom_mappings(). */ - sethi %hi(0), %g5 - or %g5, %lo(0), %g5 - - /* Behave as if we are at TL0. */ - wrpr %g0, 1, %tl - rdpr %tpc, %g4 /* Find original faulting iaddr */ - srlx %g4, 13, %g4 /* Throw out context bits */ - sllx %g4, 13, %g4 /* g4 has vpn + ctx0 now */ - - /* Restore previous TAG_ACCESS. */ - mov TLB_SFSR, %g1 - stxa %g4, [%g1 + %g1] ASI_IMMU - - /* Get PMD offset. */ - srlx %g4, 23, %g6 - and %g6, 0x7ff, %g6 - sllx %g6, 2, %g6 - - /* Load PMD, is it valid? */ - lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brz,pn %g5, longpath - sllx %g5, 11, %g5 - - /* Get PTE offset. */ - srlx %g4, 13, %g6 - and %g6, 0x3ff, %g6 - sllx %g6, 3, %g6 - - /* Load PTE. */ - ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brgez,pn %g5, longpath - nop - - /* TLB load and return from trap. */ - stxa %g5, [%g0] ASI_ITLB_DATA_IN - retry - - .globl obp_daddr_patch -obp_daddr_patch: - /* These two instructions patched by inherit_prom_mappings(). */ - sethi %hi(0), %g5 - or %g5, %lo(0), %g5 - - /* Get PMD offset. */ - srlx %g4, 23, %g6 - and %g6, 0x7ff, %g6 - sllx %g6, 2, %g6 - - /* Load PMD, is it valid? */ - lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brz,pn %g5, longpath - sllx %g5, 11, %g5 - - /* Get PTE offset. */ - srlx %g4, 13, %g6 - and %g6, 0x3ff, %g6 - sllx %g6, 3, %g6 - - /* Load PTE. */ - ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brgez,pn %g5, longpath - nop - - /* TLB load and return from trap. */ - stxa %g5, [%g0] ASI_DTLB_DATA_IN - retry - -/* - * On a first level data miss, check whether this is to the OBP range (note - * that such accesses can be made by prom, as well as by kernel using - * prom_getproperty on "address"), and if so, do not use vpte access ... - * rather, use information saved during inherit_prom_mappings() using 8k - * pagesize. - */ - .align 32 -kvmap: - sethi %hi(MODULES_VADDR), %g5 - cmp %g4, %g5 - blu,pn %xcc, longpath - mov (VMALLOC_END >> 24), %g5 - sllx %g5, 24, %g5 - cmp %g4, %g5 - bgeu,pn %xcc, longpath - nop - -kvmap_check_obp: - sethi %hi(LOW_OBP_ADDRESS), %g5 - cmp %g4, %g5 - blu,pn %xcc, kvmap_vmalloc_addr - mov 0x1, %g5 - sllx %g5, 32, %g5 - cmp %g4, %g5 - blu,pn %xcc, obp_daddr_patch - nop - -kvmap_vmalloc_addr: - /* If we get here, a vmalloc addr was accessed, load kernel VPTE. */ - ldxa [%g3 + %g6] ASI_N, %g5 - brgez,pn %g5, longpath - nop - - /* PTE is valid, load into TLB and return from trap. */ - stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB - retry - /* This is trivial with the new code... */ .globl do_fpdis do_fpdis: diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index 1fa06c4e3bd..dc3551f46b7 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -762,6 +762,7 @@ bootup_user_stack_end: swapper_pg_dir: .word 0 +#include "ktlb.S" #include "etrap.S" #include "rtrap.S" #include "winfixup.S" diff --git a/arch/sparc64/kernel/ktlb.S b/arch/sparc64/kernel/ktlb.S new file mode 100644 index 00000000000..b7176792c9a --- /dev/null +++ b/arch/sparc64/kernel/ktlb.S @@ -0,0 +1,174 @@ +/* arch/sparc64/kernel/ktlb.S: Kernel mapping TLB miss handling. + * + * Copyright (C) 1995, 1997, 2005 David S. Miller + * Copyright (C) 1996 Eddie C. Dost (ecd@brainaid.de) + * Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx) + * Copyright (C) 1996,98,99 Jakub Jelinek (jj@sunsite.mff.cuni.cz) +*/ + +#include +#include +#include +#include +#include + + .text + .align 32 + + .globl sparc64_vpte_patchme1 + .globl sparc64_vpte_patchme2 +/* + * On a second level vpte miss, check whether the original fault is to the OBP + * range (note that this is only possible for instruction miss, data misses to + * obp range do not use vpte). If so, go back directly to the faulting address. + * This is because we want to read the tpc, otherwise we have no way of knowing + * the 8k aligned faulting address if we are using >8k kernel pagesize. This + * also ensures no vpte range addresses are dropped into tlb while obp is + * executing (see inherit_locked_prom_mappings() rant). + */ +sparc64_vpte_nucleus: + /* Note that kvmap below has verified that the address is + * in the range MODULES_VADDR --> VMALLOC_END already. So + * here we need only check if it is an OBP address or not. + */ + sethi %hi(LOW_OBP_ADDRESS), %g5 + cmp %g4, %g5 + blu,pn %xcc, sparc64_vpte_patchme1 + mov 0x1, %g5 + sllx %g5, 32, %g5 + cmp %g4, %g5 + blu,pn %xcc, obp_iaddr_patch + nop + + /* These two instructions are patched by paginig_init(). */ +sparc64_vpte_patchme1: + sethi %hi(0), %g5 +sparc64_vpte_patchme2: + or %g5, %lo(0), %g5 + + /* With kernel PGD in %g5, branch back into dtlb_backend. */ + ba,pt %xcc, sparc64_kpte_continue + andn %g1, 0x3, %g1 /* Finish PMD offset adjustment. */ + +vpte_noent: + /* Restore previous TAG_ACCESS, %g5 is zero, and we will + * skip over the trap instruction so that the top level + * TLB miss handler will thing this %g5 value is just an + * invalid PTE, thus branching to full fault processing. + */ + mov TLB_SFSR, %g1 + stxa %g4, [%g1 + %g1] ASI_DMMU + done + + .globl obp_iaddr_patch +obp_iaddr_patch: + /* These two instructions patched by inherit_prom_mappings(). */ + sethi %hi(0), %g5 + or %g5, %lo(0), %g5 + + /* Behave as if we are at TL0. */ + wrpr %g0, 1, %tl + rdpr %tpc, %g4 /* Find original faulting iaddr */ + srlx %g4, 13, %g4 /* Throw out context bits */ + sllx %g4, 13, %g4 /* g4 has vpn + ctx0 now */ + + /* Restore previous TAG_ACCESS. */ + mov TLB_SFSR, %g1 + stxa %g4, [%g1 + %g1] ASI_IMMU + + /* Get PMD offset. */ + srlx %g4, 23, %g6 + and %g6, 0x7ff, %g6 + sllx %g6, 2, %g6 + + /* Load PMD, is it valid? */ + lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 + brz,pn %g5, longpath + sllx %g5, 11, %g5 + + /* Get PTE offset. */ + srlx %g4, 13, %g6 + and %g6, 0x3ff, %g6 + sllx %g6, 3, %g6 + + /* Load PTE. */ + ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 + brgez,pn %g5, longpath + nop + + /* TLB load and return from trap. */ + stxa %g5, [%g0] ASI_ITLB_DATA_IN + retry + + .globl obp_daddr_patch +obp_daddr_patch: + /* These two instructions patched by inherit_prom_mappings(). */ + sethi %hi(0), %g5 + or %g5, %lo(0), %g5 + + /* Get PMD offset. */ + srlx %g4, 23, %g6 + and %g6, 0x7ff, %g6 + sllx %g6, 2, %g6 + + /* Load PMD, is it valid? */ + lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 + brz,pn %g5, longpath + sllx %g5, 11, %g5 + + /* Get PTE offset. */ + srlx %g4, 13, %g6 + and %g6, 0x3ff, %g6 + sllx %g6, 3, %g6 + + /* Load PTE. */ + ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 + brgez,pn %g5, longpath + nop + + /* TLB load and return from trap. */ + stxa %g5, [%g0] ASI_DTLB_DATA_IN + retry + +/* + * On a first level data miss, check whether this is to the OBP range (note + * that such accesses can be made by prom, as well as by kernel using + * prom_getproperty on "address"), and if so, do not use vpte access ... + * rather, use information saved during inherit_prom_mappings() using 8k + * pagesize. + */ + .align 32 +kvmap: + brlz,pt %g4, kvmap_load + xor %g2, %g4, %g5 + +kvmap_nonlinear: + sethi %hi(MODULES_VADDR), %g5 + cmp %g4, %g5 + blu,pn %xcc, longpath + mov (VMALLOC_END >> 24), %g5 + sllx %g5, 24, %g5 + cmp %g4, %g5 + bgeu,pn %xcc, longpath + nop + +kvmap_check_obp: + sethi %hi(LOW_OBP_ADDRESS), %g5 + cmp %g4, %g5 + blu,pn %xcc, kvmap_vmalloc_addr + mov 0x1, %g5 + sllx %g5, 32, %g5 + cmp %g4, %g5 + blu,pn %xcc, obp_daddr_patch + nop + +kvmap_vmalloc_addr: + /* If we get here, a vmalloc addr was accessed, load kernel VPTE. */ + ldxa [%g3 + %g6] ASI_N, %g5 + brgez,pn %g5, longpath + nop + +kvmap_load: + /* PTE is valid, load into TLB and return from trap. */ + stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB + retry -- cgit v1.2.3-70-g09d2 From 059deb693ec191e563ec69533d24f3feff0b78cd Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 21 Sep 2005 19:23:48 -0700 Subject: [SPARC64]: Kill SZ_BITS define from dtlb_backend.S This is just a replica of the existing _PAGE_SZBITS, and thus unnecessary. Signed-off-by: David S. Miller --- arch/sparc64/kernel/dtlb_backend.S | 13 +------------ 1 file changed, 1 insertion(+), 12 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/dtlb_backend.S b/arch/sparc64/kernel/dtlb_backend.S index 538522848ad..acc889a7f9c 100644 --- a/arch/sparc64/kernel/dtlb_backend.S +++ b/arch/sparc64/kernel/dtlb_backend.S @@ -9,17 +9,7 @@ #include #include -#if PAGE_SHIFT == 13 -#define SZ_BITS _PAGE_SZ8K -#elif PAGE_SHIFT == 16 -#define SZ_BITS _PAGE_SZ64K -#elif PAGE_SHIFT == 19 -#define SZ_BITS _PAGE_SZ512K -#elif PAGE_SHIFT == 22 -#define SZ_BITS _PAGE_SZ4MB -#endif - -#define VALID_SZ_BITS (_PAGE_VALID | SZ_BITS) +#define VALID_SZ_BITS (_PAGE_VALID | _PAGE_SZBITS) #define VPTE_BITS (_PAGE_CP | _PAGE_CV | _PAGE_P ) #define VPTE_SHIFT (PAGE_SHIFT - 3) @@ -163,7 +153,6 @@ sparc64_vpte_continue: stxa %g4, [%g1 + %g1] ASI_DMMU ! Restore previous TAG_ACCESS retry ! Load PTE once again -#undef SZ_BITS #undef VALID_SZ_BITS #undef VPTE_SHIFT #undef VPTE_BITS -- cgit v1.2.3-70-g09d2 From 1ac4f5ebaa496a23ab4a148c9864d7e30a6c6cd3 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 21 Sep 2005 21:49:32 -0700 Subject: [SPARC64]: Remove ktlb.S instruction patching. This was kind of ugly, and actually buggy. The bug was that we didn't handle a machine with memory starting > 4GB. If the 'prompmd' was allocated in physical memory > 4GB we'd croak because the obp_iaddr_patch and obp_daddr_patch things only supported a 32-bit physical address. So fix this by just loading the appropriate values from two variables in the kernel image, which is locked into the TLB and thus accesses to them can't cause a recursive TLB miss. Signed-off-by: David S. Miller --- arch/sparc64/kernel/ktlb.S | 31 ++++++++++++------------------- arch/sparc64/mm/init.c | 32 ++++++++++++-------------------- 2 files changed, 24 insertions(+), 39 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ktlb.S b/arch/sparc64/kernel/ktlb.S index b7176792c9a..a591bc0ebc7 100644 --- a/arch/sparc64/kernel/ktlb.S +++ b/arch/sparc64/kernel/ktlb.S @@ -15,8 +15,6 @@ .text .align 32 - .globl sparc64_vpte_patchme1 - .globl sparc64_vpte_patchme2 /* * On a second level vpte miss, check whether the original fault is to the OBP * range (note that this is only possible for instruction miss, data misses to @@ -33,18 +31,17 @@ sparc64_vpte_nucleus: */ sethi %hi(LOW_OBP_ADDRESS), %g5 cmp %g4, %g5 - blu,pn %xcc, sparc64_vpte_patchme1 + blu,pn %xcc, kern_vpte mov 0x1, %g5 sllx %g5, 32, %g5 cmp %g4, %g5 - blu,pn %xcc, obp_iaddr_patch + blu,pn %xcc, vpte_insn_obp nop /* These two instructions are patched by paginig_init(). */ -sparc64_vpte_patchme1: - sethi %hi(0), %g5 -sparc64_vpte_patchme2: - or %g5, %lo(0), %g5 +kern_vpte: + sethi %hi(swapper_pgd_zero), %g5 + lduw [%g5 + %lo(swapper_pgd_zero)], %g5 /* With kernel PGD in %g5, branch back into dtlb_backend. */ ba,pt %xcc, sparc64_kpte_continue @@ -60,11 +57,9 @@ vpte_noent: stxa %g4, [%g1 + %g1] ASI_DMMU done - .globl obp_iaddr_patch -obp_iaddr_patch: - /* These two instructions patched by inherit_prom_mappings(). */ - sethi %hi(0), %g5 - or %g5, %lo(0), %g5 +vpte_insn_obp: + sethi %hi(prom_pmd_phys), %g5 + ldx [%g5 + %lo(prom_pmd_phys)], %g5 /* Behave as if we are at TL0. */ wrpr %g0, 1, %tl @@ -100,11 +95,9 @@ obp_iaddr_patch: stxa %g5, [%g0] ASI_ITLB_DATA_IN retry - .globl obp_daddr_patch -obp_daddr_patch: - /* These two instructions patched by inherit_prom_mappings(). */ - sethi %hi(0), %g5 - or %g5, %lo(0), %g5 +kvmap_do_obp: + sethi %hi(prom_pmd_phys), %g5 + ldx [%g5 + %lo(prom_pmd_phys)], %g5 /* Get PMD offset. */ srlx %g4, 23, %g6 @@ -159,7 +152,7 @@ kvmap_check_obp: mov 0x1, %g5 sllx %g5, 32, %g5 cmp %g4, %g5 - blu,pn %xcc, obp_daddr_patch + blu,pn %xcc, kvmap_do_obp nop kvmap_vmalloc_addr: diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index fdb1ebb308c..aaf9a3eee99 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -20,6 +20,7 @@ #include #include #include +#include #include #include @@ -45,10 +46,10 @@ struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS]; unsigned long *sparc64_valid_addr_bitmap; /* Ugly, but necessary... -DaveM */ -unsigned long phys_base; -unsigned long kern_base; -unsigned long kern_size; -unsigned long pfn_base; +unsigned long phys_base __read_mostly; +unsigned long kern_base __read_mostly; +unsigned long kern_size __read_mostly; +unsigned long pfn_base __read_mostly; /* This is even uglier. We have a problem where the kernel may not be * located at phys_base. However, initial __alloc_bootmem() calls need to @@ -73,7 +74,7 @@ extern unsigned long sparc_ramdisk_image64; extern unsigned int sparc_ramdisk_image; extern unsigned int sparc_ramdisk_size; -struct page *mem_map_zero; +struct page *mem_map_zero __read_mostly; int bigkernel = 0; @@ -318,6 +319,10 @@ extern void register_prom_callbacks(void); /* Exported for SMP bootup purposes. */ unsigned long kern_locked_tte_data; +/* Exported for kernel TLB miss handling in ktlb.S */ +unsigned long prom_pmd_phys __read_mostly; +unsigned int swapper_pgd_zero __read_mostly; + void __init early_pgtable_allocfail(char *type) { prom_printf("inherit_prom_mappings: Cannot alloc kernel %s.\n", type); @@ -364,7 +369,6 @@ static void inherit_prom_mappings(void) pmd_t *pmdp; pte_t *ptep; int node, n, i, tsz; - extern unsigned int obp_iaddr_patch[2], obp_daddr_patch[2]; node = prom_finddevice("/virtual-memory"); n = prom_getproplen(node, "translations"); @@ -434,13 +438,7 @@ static void inherit_prom_mappings(void) } } } - phys_page = __pa(prompmd); - obp_iaddr_patch[0] |= (phys_page >> 10); - obp_iaddr_patch[1] |= (phys_page & 0x3ff); - flushi((long)&obp_iaddr_patch[0]); - obp_daddr_patch[0] |= (phys_page >> 10); - obp_daddr_patch[1] |= (phys_page & 0x3ff); - flushi((long)&obp_daddr_patch[0]); + prom_pmd_phys = __pa(prompmd); /* Now fixup OBP's idea about where we really are mapped. */ prom_printf("Remapping the kernel... "); @@ -1407,8 +1405,6 @@ static unsigned long last_valid_pfn; void __init paging_init(void) { extern pmd_t swapper_pmd_dir[1024]; - extern unsigned int sparc64_vpte_patchme1[1]; - extern unsigned int sparc64_vpte_patchme2[1]; unsigned long alias_base = kern_base + PAGE_OFFSET; unsigned long second_alias_page = 0; unsigned long pt, flags, end_pfn, pages_avail; @@ -1502,11 +1498,7 @@ void __init paging_init(void) pud_set(pud_offset(&swapper_pg_dir[0], 0), swapper_pmd_dir + (shift / sizeof(pgd_t))); - sparc64_vpte_patchme1[0] |= - (((unsigned long)pgd_val(init_mm.pgd[0])) >> 10); - sparc64_vpte_patchme2[0] |= - (((unsigned long)pgd_val(init_mm.pgd[0])) & 0x3ff); - flushi((long)&sparc64_vpte_patchme1[0]); + swapper_pgd_zero = pgd_val(init_mm.pgd[0]); /* Setup bootmem... */ pages_avail = 0; -- cgit v1.2.3-70-g09d2 From bff06d552240ba7f5b49482a4865871d7bc03dc2 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 22 Sep 2005 20:11:33 -0700 Subject: [SPARC64]: Rewrite bootup sequence. Instead of all of this cpu-specific code to remap the kernel to the correct location, use portable firmware calls to do this instead. What we do now is the following in position independant assembler: chosen_node = prom_finddevice("/chosen"); prom_mmu_ihandle_cache = prom_getint(chosen_node, "mmu"); vaddr = 4MB_ALIGN(current_text_addr()); prom_translate(vaddr, &paddr_high, &paddr_low, &mode); prom_boot_mapping_mode = mode; prom_boot_mapping_phys_high = paddr_high; prom_boot_mapping_phys_low = paddr_low; prom_map(-1, 8 * 1024 * 1024, KERNBASE, paddr_low); and that replaces the massive amount of by-hand TLB probing and programming we used to do here. The new code should also handle properly the case where the kernel is mapped at the correct address already (think: future kexec support). Consequently, the bulk of remap_kernel() dies as does the entirety of arch/sparc64/prom/map.S We try to share some strings in the PROM library with the ones used at bootup, and while we're here mark input strings to oplib.h routines with "const" when appropriate. There are many more simplifications now possible. For one thing, we can consolidate the two copies we now have of a lot of cpu setup code sitting in head.S and trampoline.S. This is a significant step towards CONFIG_DEBUG_PAGEALLOC support. Signed-off-by: David S. Miller --- arch/sparc64/kernel/head.S | 549 ++++++++++++--------------------------- arch/sparc64/kernel/setup.c | 15 +- arch/sparc64/kernel/trampoline.S | 16 +- arch/sparc64/mm/init.c | 98 +------ arch/sparc64/prom/Makefile | 2 +- arch/sparc64/prom/console.c | 2 +- arch/sparc64/prom/devops.c | 2 +- arch/sparc64/prom/init.c | 2 +- arch/sparc64/prom/map.S | 72 ----- arch/sparc64/prom/misc.c | 34 ++- arch/sparc64/prom/p1275.c | 2 +- arch/sparc64/prom/printf.c | 2 +- arch/sparc64/prom/tree.c | 50 ++-- include/asm-sparc64/oplib.h | 52 ++-- 14 files changed, 254 insertions(+), 644 deletions(-) delete mode 100644 arch/sparc64/prom/map.S (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index dc3551f46b7..8b83520e427 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -80,15 +80,165 @@ sparc_ramdisk_image64: .xword 0 .word _end - /* We must be careful, 32-bit OpenBOOT will get confused if it - * tries to save away a register window to a 64-bit kernel - * stack address. Flush all windows, disable interrupts, - * remap if necessary, jump onto kernel trap table, then kernel - * stack, or else we die. + /* PROM cif handler code address is in %o4. */ +sparc64_boot: +1: rd %pc, %g7 + set 1b, %g1 + cmp %g1, %g7 + be,pn %xcc, sparc64_boot_after_remap + mov %o4, %l7 + + /* We need to remap the kernel. Use position independant + * code to remap us to KERNBASE. * - * PROM entry point is on %o4 + * SILO can invoke us with 32-bit address masking enabled, + * so make sure that's clear. */ -sparc64_boot: + rdpr %pstate, %g1 + andn %g1, PSTATE_AM, %g1 + wrpr %g1, 0x0, %pstate + ba,a,pt %xcc, 1f + + .globl prom_finddev_name, prom_chosen_path + .globl prom_getprop_name, prom_mmu_name + .globl prom_callmethod_name, prom_translate_name + .globl prom_map_name, prom_unmap_name, prom_mmu_ihandle_cache + .globl prom_boot_mapped_pc, prom_boot_mapping_mode + .globl prom_boot_mapping_phys_high, prom_boot_mapping_phys_low +prom_finddev_name: + .asciz "finddevice" +prom_chosen_path: + .asciz "/chosen" +prom_getprop_name: + .asciz "getprop" +prom_mmu_name: + .asciz "mmu" +prom_callmethod_name: + .asciz "call-method" +prom_translate_name: + .asciz "translate" +prom_map_name: + .asciz "map" +prom_unmap_name: + .asciz "unmap" + .align 4 +prom_mmu_ihandle_cache: + .word 0 +prom_boot_mapped_pc: + .word 0 +prom_boot_mapping_mode: + .word 0 + .align 8 +prom_boot_mapping_phys_high: + .xword 0 +prom_boot_mapping_phys_low: + .xword 0 +1: + rd %pc, %l0 + mov (1b - prom_finddev_name), %l1 + mov (1b - prom_chosen_path), %l2 + mov (1b - prom_boot_mapped_pc), %l3 + sub %l0, %l1, %l1 + sub %l0, %l2, %l2 + sub %l0, %l3, %l3 + stw %l0, [%l3] + sub %sp, (192 + 128), %sp + + /* chosen_node = prom_finddevice("/chosen") */ + stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "finddevice" + mov 1, %l3 + stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 1 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1 + stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1, "/chosen" + stx %g0, [%sp + 2047 + 128 + 0x20] ! ret1 + call %l7 + add %sp, (2047 + 128), %o0 ! argument array + + ldx [%sp + 2047 + 128 + 0x20], %l4 ! chosen device node + + mov (1b - prom_getprop_name), %l1 + mov (1b - prom_mmu_name), %l2 + mov (1b - prom_mmu_ihandle_cache), %l5 + sub %l0, %l1, %l1 + sub %l0, %l2, %l2 + sub %l0, %l5, %l5 + + /* prom_mmu_ihandle_cache = prom_getint(chosen_node, "mmu") */ + stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "getprop" + mov 4, %l3 + stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 4 + mov 1, %l3 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1 + stx %l4, [%sp + 2047 + 128 + 0x18] ! arg1, chosen_node + stx %l2, [%sp + 2047 + 128 + 0x20] ! arg2, "mmu" + stx %l5, [%sp + 2047 + 128 + 0x28] ! arg3, &prom_mmu_ihandle_cache + mov 4, %l3 + stx %l3, [%sp + 2047 + 128 + 0x30] ! arg4, sizeof(arg3) + stx %g0, [%sp + 2047 + 128 + 0x38] ! ret1 + call %l7 + add %sp, (2047 + 128), %o0 ! argument array + + mov (1b - prom_callmethod_name), %l1 + mov (1b - prom_translate_name), %l2 + sub %l0, %l1, %l1 + sub %l0, %l2, %l2 + lduw [%l5], %l5 ! prom_mmu_ihandle_cache + + stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "call-method" + mov 3, %l3 + stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 3 + mov 5, %l3 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 5 + stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1: "translate" + stx %l5, [%sp + 2047 + 128 + 0x20] ! arg2: prom_mmu_ihandle_cache + srlx %l0, 22, %l3 + sllx %l3, 22, %l3 + stx %l3, [%sp + 2047 + 128 + 0x28] ! arg3: vaddr, our PC + stx %g0, [%sp + 2047 + 128 + 0x30] ! res1 + stx %g0, [%sp + 2047 + 128 + 0x38] ! res2 + stx %g0, [%sp + 2047 + 128 + 0x40] ! res3 + stx %g0, [%sp + 2047 + 128 + 0x48] ! res4 + stx %g0, [%sp + 2047 + 128 + 0x50] ! res5 + call %l7 + add %sp, (2047 + 128), %o0 ! argument array + + ldx [%sp + 2047 + 128 + 0x40], %l1 ! translation mode + mov (1b - prom_boot_mapping_mode), %l4 + sub %l0, %l4, %l4 + stw %l1, [%l4] + mov (1b - prom_boot_mapping_phys_high), %l4 + sub %l0, %l4, %l4 + ldx [%sp + 2047 + 128 + 0x48], %l2 ! physaddr high + stx %l2, [%l4 + 0x0] + ldx [%sp + 2047 + 128 + 0x50], %l3 ! physaddr low + stx %l3, [%l4 + 0x8] + + /* Leave service as-is, "call-method" */ + mov 7, %l3 + stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 7 + mov 1, %l3 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 7 + mov (1b - prom_map_name), %l3 + sub %l0, %l3, %l3 + stx %l3, [%sp + 2047 + 128 + 0x18] ! arg1: "map" + /* Leave arg2 as-is, prom_mmu_ihandle_cache */ + mov -1, %l3 + stx %l3, [%sp + 2047 + 128 + 0x28] ! arg3: mode (-1 default) + sethi %hi(8 * 1024 * 1024), %l3 + stx %l3, [%sp + 2047 + 128 + 0x30] ! arg4: size (8MB) + sethi %hi(KERNBASE), %l3 + stx %l3, [%sp + 2047 + 128 + 0x38] ! arg5: vaddr (KERNBASE) + stx %g0, [%sp + 2047 + 128 + 0x40] ! arg6: empty + mov (1b - prom_boot_mapping_phys_low), %l3 + sub %l0, %l3, %l3 + ldx [%l3], %l3 + stx %l3, [%sp + 2047 + 128 + 0x48] ! arg7: phys addr + call %l7 + add %sp, (2047 + 128), %o0 ! argument array + + add %sp, (192 + 128), %sp + +sparc64_boot_after_remap: BRANCH_IF_CHEETAH_BASE(g1,g7,cheetah_boot) BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g7,cheetah_plus_boot) ba,pt %xcc, spitfire_boot @@ -125,185 +275,7 @@ cheetah_generic_boot: stxa %g0, [%g3] ASI_IMMU membar #Sync - wrpr %g0, (PSTATE_PRIV|PSTATE_PEF|PSTATE_IE), %pstate - wr %g0, 0, %fprs - - /* Just like for Spitfire, we probe itlb-2 for a mapping which - * matches our current %pc. We take the physical address in - * that mapping and use it to make our own. - */ - - /* %g5 holds the tlb data */ - sethi %uhi(_PAGE_VALID | _PAGE_SZ4MB), %g5 - sllx %g5, 32, %g5 - or %g5, (_PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W | _PAGE_G), %g5 - - /* Put PADDR tlb data mask into %g3. */ - sethi %uhi(_PAGE_PADDR), %g3 - or %g3, %ulo(_PAGE_PADDR), %g3 - sllx %g3, 32, %g3 - sethi %hi(_PAGE_PADDR), %g7 - or %g7, %lo(_PAGE_PADDR), %g7 - or %g3, %g7, %g3 - - set 2 << 16, %l0 /* TLB entry walker. */ - set 0x1fff, %l2 /* Page mask. */ - rd %pc, %l3 - andn %l3, %l2, %g2 /* vaddr comparator */ - -1: ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g2 - be,pn %xcc, cheetah_got_tlbentry - nop - and %l0, (127 << 3), %g1 - cmp %g1, (127 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* Search the small TLB. OBP never maps us like that but - * newer SILO can. - */ - clr %l0 - -1: ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g2 - be,pn %xcc, cheetah_got_tlbentry - nop - cmp %l0, (15 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* BUG() if we get here... */ - ta 0x5 - -cheetah_got_tlbentry: - ldxa [%l0] ASI_ITLB_DATA_ACCESS, %g0 - ldxa [%l0] ASI_ITLB_DATA_ACCESS, %g1 - membar #Sync - and %g1, %g3, %g1 - set 0x5fff, %l0 - andn %g1, %l0, %g1 - or %g5, %g1, %g5 - - /* Clear out any KERNBASE area entries. */ - set 2 << 16, %l0 - sethi %hi(KERNBASE), %g3 - sethi %hi(KERNBASE<<1), %g7 - mov TLB_TAG_ACCESS, %l7 - - /* First, check ITLB */ -1: ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_IMMU - membar #Sync - stxa %g0, [%l0] ASI_ITLB_DATA_ACCESS - membar #Sync - -2: and %l0, (127 << 3), %g1 - cmp %g1, (127 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* Next, check DTLB */ - set 2 << 16, %l0 -1: ldxa [%l0] ASI_DTLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_DMMU - membar #Sync - stxa %g0, [%l0] ASI_DTLB_DATA_ACCESS - membar #Sync - -2: and %l0, (511 << 3), %g1 - cmp %g1, (511 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* On Cheetah+, have to check second DTLB. */ - BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,l0,2f) - ba,pt %xcc, 9f - nop - -2: set 3 << 16, %l0 -1: ldxa [%l0] ASI_DTLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_DMMU - membar #Sync - stxa %g0, [%l0] ASI_DTLB_DATA_ACCESS - membar #Sync - -2: and %l0, (511 << 3), %g1 - cmp %g1, (511 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - -9: - - /* Now lock the TTE we created into ITLB-0 and DTLB-0, - * entry 15 (and maybe 14 too). - */ - sethi %hi(KERNBASE), %g3 - set (0 << 16) | (15 << 3), %g7 - stxa %g3, [%l7] ASI_DMMU - membar #Sync - stxa %g5, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - stxa %g3, [%l7] ASI_IMMU - membar #Sync - stxa %g5, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - flush %g3 - membar #Sync - sethi %hi(_end), %g3 /* Check for bigkernel case */ - or %g3, %lo(_end), %g3 - srl %g3, 23, %g3 /* Check if _end > 8M */ - brz,pt %g3, 1f - sethi %hi(KERNBASE), %g3 /* Restore for fixup code below */ - sethi %hi(0x400000), %g3 - or %g3, %lo(0x400000), %g3 - add %g5, %g3, %g5 /* New tte data */ - andn %g5, (_PAGE_G), %g5 - sethi %hi(KERNBASE+0x400000), %g3 - or %g3, %lo(KERNBASE+0x400000), %g3 - set (0 << 16) | (14 << 3), %g7 - stxa %g3, [%l7] ASI_DMMU - membar #Sync - stxa %g5, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - stxa %g3, [%l7] ASI_IMMU - membar #Sync - stxa %g5, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - flush %g3 - membar #Sync - sethi %hi(KERNBASE), %g3 /* Restore for fixup code below */ - ba,pt %xcc, 1f - nop - -1: set sun4u_init, %g2 - jmpl %g2 + %g0, %g0 - nop + ba,a,pt %xcc, jump_to_sun4u_init spitfire_boot: /* Typically PROM has already enabled both MMU's and both on-chip @@ -313,6 +285,7 @@ spitfire_boot: stxa %g1, [%g0] ASI_LSU_CONTROL membar #Sync +jump_to_sun4u_init: /* * Make sure we are in privileged mode, have address masking, * using the ordinary globals and have enabled floating @@ -324,151 +297,6 @@ spitfire_boot: wrpr %g0, (PSTATE_PRIV|PSTATE_PEF|PSTATE_IE), %pstate wr %g0, 0, %fprs -spitfire_create_mappings: - /* %g5 holds the tlb data */ - sethi %uhi(_PAGE_VALID | _PAGE_SZ4MB), %g5 - sllx %g5, 32, %g5 - or %g5, (_PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W | _PAGE_G), %g5 - - /* Base of physical memory cannot reliably be assumed to be - * at 0x0! Figure out where it happens to be. -DaveM - */ - - /* Put PADDR tlb data mask into %g3. */ - sethi %uhi(_PAGE_PADDR_SF), %g3 - or %g3, %ulo(_PAGE_PADDR_SF), %g3 - sllx %g3, 32, %g3 - sethi %hi(_PAGE_PADDR_SF), %g7 - or %g7, %lo(_PAGE_PADDR_SF), %g7 - or %g3, %g7, %g3 - - /* Walk through entire ITLB, looking for entry which maps - * our %pc currently, stick PADDR from there into %g5 tlb data. - */ - clr %l0 /* TLB entry walker. */ - set 0x1fff, %l2 /* Page mask. */ - rd %pc, %l3 - andn %l3, %l2, %g2 /* vaddr comparator */ -1: - /* Yes, the nops seem to be necessary for now, don't ask me why. -DaveM */ - ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - nop - nop - nop - andn %g1, %l2, %g1 /* Get vaddr */ - cmp %g1, %g2 - be,a,pn %xcc, spitfire_got_tlbentry - ldxa [%l0] ASI_ITLB_DATA_ACCESS, %g1 - cmp %l0, (63 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* BUG() if we get here... */ - ta 0x5 - -spitfire_got_tlbentry: - /* Nops here again, perhaps Cheetah/Blackbird are better behaved... */ - nop - nop - nop - and %g1, %g3, %g1 /* Mask to just get paddr bits. */ - set 0x5fff, %l3 /* Mask offset to get phys base. */ - andn %g1, %l3, %g1 - - /* NOTE: We hold on to %g1 paddr base as we need it below to lock - * NOTE: the PROM cif code into the TLB. - */ - - or %g5, %g1, %g5 /* Or it into TAG being built. */ - - clr %l0 /* TLB entry walker. */ - sethi %hi(KERNBASE), %g3 /* 4M lower limit */ - sethi %hi(KERNBASE<<1), %g7 /* 8M upper limit */ - mov TLB_TAG_ACCESS, %l7 -1: - /* Yes, the nops seem to be necessary for now, don't ask me why. -DaveM */ - ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - nop - nop - nop - andn %g1, %l2, %g1 /* Get vaddr */ - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_IMMU - stxa %g0, [%l0] ASI_ITLB_DATA_ACCESS - membar #Sync -2: - cmp %l0, (63 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - nop; nop; nop - - clr %l0 /* TLB entry walker. */ -1: - /* Yes, the nops seem to be necessary for now, don't ask me why. -DaveM */ - ldxa [%l0] ASI_DTLB_TAG_READ, %g1 - nop - nop - nop - andn %g1, %l2, %g1 /* Get vaddr */ - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_DMMU - stxa %g0, [%l0] ASI_DTLB_DATA_ACCESS - membar #Sync -2: - cmp %l0, (63 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - nop; nop; nop - - - /* PROM never puts any TLB entries into the MMU with the lock bit - * set. So we gladly use tlb entry 63 for KERNBASE. And maybe 62 too. - */ - - sethi %hi(KERNBASE), %g3 - mov (63 << 3), %g7 - stxa %g3, [%l7] ASI_DMMU /* KERNBASE into TLB TAG */ - stxa %g5, [%g7] ASI_DTLB_DATA_ACCESS /* TTE into TLB DATA */ - membar #Sync - stxa %g3, [%l7] ASI_IMMU /* KERNBASE into TLB TAG */ - stxa %g5, [%g7] ASI_ITLB_DATA_ACCESS /* TTE into TLB DATA */ - membar #Sync - flush %g3 - membar #Sync - sethi %hi(_end), %g3 /* Check for bigkernel case */ - or %g3, %lo(_end), %g3 - srl %g3, 23, %g3 /* Check if _end > 8M */ - brz,pt %g3, 2f - sethi %hi(KERNBASE), %g3 /* Restore for fixup code below */ - sethi %hi(0x400000), %g3 - or %g3, %lo(0x400000), %g3 - add %g5, %g3, %g5 /* New tte data */ - andn %g5, (_PAGE_G), %g5 - sethi %hi(KERNBASE+0x400000), %g3 - or %g3, %lo(KERNBASE+0x400000), %g3 - mov (62 << 3), %g7 - stxa %g3, [%l7] ASI_DMMU - stxa %g5, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - stxa %g3, [%l7] ASI_IMMU - stxa %g5, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - flush %g3 - membar #Sync - sethi %hi(KERNBASE), %g3 /* Restore for fixup code below */ -2: ba,pt %xcc, 1f - nop -1: set sun4u_init, %g2 jmpl %g2 + %g0, %g0 nop @@ -483,38 +311,12 @@ sun4u_init: stxa %g0, [%g7] ASI_DMMU membar #Sync - /* We are now safely (we hope) in Nucleus context (0), rewrite - * the KERNBASE TTE's so they no longer have the global bit set. - * Don't forget to setup TAG_ACCESS first 8-) - */ - mov TLB_TAG_ACCESS, %g2 - stxa %g3, [%g2] ASI_IMMU - stxa %g3, [%g2] ASI_DMMU - membar #Sync - BRANCH_IF_ANY_CHEETAH(g1,g7,cheetah_tlb_fixup) ba,pt %xcc, spitfire_tlb_fixup nop cheetah_tlb_fixup: - set (0 << 16) | (15 << 3), %g7 - ldxa [%g7] ASI_ITLB_DATA_ACCESS, %g0 - ldxa [%g7] ASI_ITLB_DATA_ACCESS, %g1 - andn %g1, (_PAGE_G), %g1 - stxa %g1, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - - ldxa [%g7] ASI_DTLB_DATA_ACCESS, %g0 - ldxa [%g7] ASI_DTLB_DATA_ACCESS, %g1 - andn %g1, (_PAGE_G), %g1 - stxa %g1, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - - /* Kill instruction prefetch queues. */ - flush %g3 - membar #Sync - mov 2, %g2 /* Set TLB type to cheetah+. */ BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g7,1f) @@ -551,21 +353,6 @@ cheetah_tlb_fixup: nop spitfire_tlb_fixup: - mov (63 << 3), %g7 - ldxa [%g7] ASI_ITLB_DATA_ACCESS, %g1 - andn %g1, (_PAGE_G), %g1 - stxa %g1, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - - ldxa [%g7] ASI_DTLB_DATA_ACCESS, %g1 - andn %g1, (_PAGE_G), %g1 - stxa %g1, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - - /* Kill instruction prefetch queues. */ - flush %g3 - membar #Sync - /* Set TLB type to spitfire. */ mov 0, %g2 sethi %hi(tlb_type), %g1 @@ -578,24 +365,6 @@ tlb_fixup_done: mov %sp, %l6 mov %o4, %l7 -#if 0 /* We don't do it like this anymore, but for historical hack value - * I leave this snippet here to show how crazy we can be sometimes. 8-) - */ - - /* Setup "Linux Current Register", thanks Sun 8-) */ - wr %g0, 0x1, %pcr - - /* Blackbird errata workaround. See commentary in - * smp.c:smp_percpu_timer_interrupt() for more - * information. - */ - ba,pt %xcc, 99f - nop - .align 64 -99: wr %g6, %g0, %pic - rd %pic, %g0 -#endif - wr %g0, ASI_P, %asi mov 1, %g1 sllx %g1, THREAD_SHIFT, %g1 diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index ddbed3341a2..0296cb00fc9 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -536,20 +536,7 @@ void __init setup_arch(char **cmdline_p) } pfn_base = phys_base >> PAGE_SHIFT; - switch (tlb_type) { - default: - case spitfire: - kern_base = spitfire_get_itlb_data(sparc64_highest_locked_tlbent()); - kern_base &= _PAGE_PADDR_SF; - break; - - case cheetah: - case cheetah_plus: - kern_base = cheetah_get_litlb_data(sparc64_highest_locked_tlbent()); - kern_base &= _PAGE_PADDR; - break; - }; - + kern_base = (prom_boot_mapping_phys_low >> 22UL) << 22UL; kern_size = (unsigned long)&_end - (unsigned long)KERNBASE; if (!root_flags) diff --git a/arch/sparc64/kernel/trampoline.S b/arch/sparc64/kernel/trampoline.S index 3a145fc39cf..89f2fcfcd66 100644 --- a/arch/sparc64/kernel/trampoline.S +++ b/arch/sparc64/kernel/trampoline.S @@ -119,8 +119,8 @@ startup_continue: sethi %hi(itlb_load), %g2 or %g2, %lo(itlb_load), %g2 stx %g2, [%sp + 2047 + 128 + 0x18] - sethi %hi(mmu_ihandle_cache), %g2 - lduw [%g2 + %lo(mmu_ihandle_cache)], %g2 + sethi %hi(prom_mmu_ihandle_cache), %g2 + lduw [%g2 + %lo(prom_mmu_ihandle_cache)], %g2 stx %g2, [%sp + 2047 + 128 + 0x20] sethi %hi(KERNBASE), %g2 stx %g2, [%sp + 2047 + 128 + 0x28] @@ -156,8 +156,8 @@ startup_continue: sethi %hi(itlb_load), %g2 or %g2, %lo(itlb_load), %g2 stx %g2, [%sp + 2047 + 128 + 0x18] - sethi %hi(mmu_ihandle_cache), %g2 - lduw [%g2 + %lo(mmu_ihandle_cache)], %g2 + sethi %hi(prom_mmu_ihandle_cache), %g2 + lduw [%g2 + %lo(prom_mmu_ihandle_cache)], %g2 stx %g2, [%sp + 2047 + 128 + 0x20] sethi %hi(KERNBASE + 0x400000), %g2 stx %g2, [%sp + 2047 + 128 + 0x28] @@ -190,8 +190,8 @@ do_dtlb: sethi %hi(dtlb_load), %g2 or %g2, %lo(dtlb_load), %g2 stx %g2, [%sp + 2047 + 128 + 0x18] - sethi %hi(mmu_ihandle_cache), %g2 - lduw [%g2 + %lo(mmu_ihandle_cache)], %g2 + sethi %hi(prom_mmu_ihandle_cache), %g2 + lduw [%g2 + %lo(prom_mmu_ihandle_cache)], %g2 stx %g2, [%sp + 2047 + 128 + 0x20] sethi %hi(KERNBASE), %g2 stx %g2, [%sp + 2047 + 128 + 0x28] @@ -228,8 +228,8 @@ do_dtlb: sethi %hi(dtlb_load), %g2 or %g2, %lo(dtlb_load), %g2 stx %g2, [%sp + 2047 + 128 + 0x18] - sethi %hi(mmu_ihandle_cache), %g2 - lduw [%g2 + %lo(mmu_ihandle_cache)], %g2 + sethi %hi(prom_mmu_ihandle_cache), %g2 + lduw [%g2 + %lo(prom_mmu_ihandle_cache)], %g2 stx %g2, [%sp + 2047 + 128 + 0x20] sethi %hi(KERNBASE + 0x400000), %g2 stx %g2, [%sp + 2047 + 128 + 0x28] diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index ec47de494c1..e0b9eebf21c 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -505,108 +505,20 @@ static int read_obp_translations(void) return n; } -static inline void early_spitfire_errata32(void) -{ - /* Spitfire Errata #32 workaround */ - /* NOTE: Using plain zero for the context value is - * correct here, we are not using the Linux trap - * tables yet so we should not use the special - * UltraSPARC-III+ page size encodings yet. - */ - __asm__ __volatile__("stxa %0, [%1] %2\n\t" - "flush %%g6" - : /* No outputs */ - : "r" (0), "r" (PRIMARY_CONTEXT), - "i" (ASI_DMMU)); -} - -static void lock_remap_func_page(unsigned long phys_page) -{ - unsigned long tte_data = (phys_page | pgprot_val(PAGE_KERNEL)); - - if (tlb_type == spitfire) { - /* Lock this into i/d tlb entry 59 */ - __asm__ __volatile__( - "stxa %%g0, [%2] %3\n\t" - "stxa %0, [%1] %4\n\t" - "membar #Sync\n\t" - "flush %%g6\n\t" - "stxa %%g0, [%2] %5\n\t" - "stxa %0, [%1] %6\n\t" - "membar #Sync\n\t" - "flush %%g6" - : /* no outputs */ - : "r" (tte_data), "r" (59 << 3), "r" (TLB_TAG_ACCESS), - "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), - "i" (ASI_IMMU), "i" (ASI_ITLB_DATA_ACCESS) - : "memory"); - } else { - /* Lock this into i/d tlb-0 entry 11 */ - __asm__ __volatile__( - "stxa %%g0, [%2] %3\n\t" - "stxa %0, [%1] %4\n\t" - "membar #Sync\n\t" - "flush %%g6\n\t" - "stxa %%g0, [%2] %5\n\t" - "stxa %0, [%1] %6\n\t" - "membar #Sync\n\t" - "flush %%g6" - : /* no outputs */ - : "r" (tte_data), "r" ((0 << 16) | (11 << 3)), - "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU), - "i" (ASI_DTLB_DATA_ACCESS), "i" (ASI_IMMU), - "i" (ASI_ITLB_DATA_ACCESS) - : "memory"); - } -} - static void remap_kernel(void) { unsigned long phys_page, tte_vaddr, tte_data; - void (*remap_func)(unsigned long, unsigned long, int); int tlb_ent = sparc64_highest_locked_tlbent(); - early_spitfire_errata32(); - - if (tlb_type == spitfire) - phys_page = spitfire_get_dtlb_data(tlb_ent); - else - phys_page = cheetah_get_ldtlb_data(tlb_ent); - - phys_page &= _PAGE_PADDR; - phys_page += ((unsigned long)&prom_boot_page - - (unsigned long)KERNBASE); - - lock_remap_func_page(phys_page); - tte_vaddr = (unsigned long) KERNBASE; - - early_spitfire_errata32(); - - if (tlb_type == spitfire) - tte_data = spitfire_get_dtlb_data(tlb_ent); - else - tte_data = cheetah_get_ldtlb_data(tlb_ent); + phys_page = (prom_boot_mapping_phys_low >> 22UL) << 22UL; + tte_data = (phys_page | (_PAGE_VALID | _PAGE_SZ4MB | + _PAGE_CP | _PAGE_CV | _PAGE_P | + _PAGE_L | _PAGE_W)); kern_locked_tte_data = tte_data; - remap_func = (void *) ((unsigned long) &prom_remap - - (unsigned long) &prom_boot_page); - - early_spitfire_errata32(); - - phys_page = tte_data & _PAGE_PADDR; - remap_func(phys_page, KERNBASE, prom_get_mmu_ihandle()); - if (bigkernel) - remap_func(phys_page + 0x400000, - KERNBASE + 0x400000, - prom_get_mmu_ihandle()); - - /* Flush out that temporary mapping. */ - spitfire_flush_dtlb_nucleus_page(0x0); - spitfire_flush_itlb_nucleus_page(0x0); - - /* Now lock us back into the TLBs via OBP. */ + /* Now lock us into the TLBs via OBP. */ prom_dtlb_load(tlb_ent, tte_data, tte_vaddr); prom_itlb_load(tlb_ent, tte_data, tte_vaddr); if (bigkernel) { diff --git a/arch/sparc64/prom/Makefile b/arch/sparc64/prom/Makefile index 8f2420d9e9e..c7898a5ee45 100644 --- a/arch/sparc64/prom/Makefile +++ b/arch/sparc64/prom/Makefile @@ -7,4 +7,4 @@ EXTRA_AFLAGS := -ansi EXTRA_CFLAGS := -Werror lib-y := bootstr.o devops.o init.o memory.o misc.o \ - tree.o console.o printf.o p1275.o map.o cif.o + tree.o console.o printf.o p1275.o cif.o diff --git a/arch/sparc64/prom/console.c b/arch/sparc64/prom/console.c index 028a53fcb1e..eae5db8dda5 100644 --- a/arch/sparc64/prom/console.c +++ b/arch/sparc64/prom/console.c @@ -67,7 +67,7 @@ prom_putchar(char c) } void -prom_puts(char *s, int len) +prom_puts(const char *s, int len) { p1275_cmd("write", P1275_ARG(1,P1275_ARG_IN_BUF)| P1275_INOUT(3,1), diff --git a/arch/sparc64/prom/devops.c b/arch/sparc64/prom/devops.c index 2c99b21b698..4641839eb39 100644 --- a/arch/sparc64/prom/devops.c +++ b/arch/sparc64/prom/devops.c @@ -16,7 +16,7 @@ * Returns 0 on failure. */ int -prom_devopen(char *dstr) +prom_devopen(const char *dstr) { return p1275_cmd ("open", P1275_ARG(0,P1275_ARG_IN_STRING)| P1275_INOUT(1,1), diff --git a/arch/sparc64/prom/init.c b/arch/sparc64/prom/init.c index 817faae058c..8b4b622d090 100644 --- a/arch/sparc64/prom/init.c +++ b/arch/sparc64/prom/init.c @@ -46,7 +46,7 @@ void __init prom_init(void *cif_handler, void *cif_stack) if((prom_root_node == 0) || (prom_root_node == -1)) prom_halt(); - prom_chosen_node = prom_finddevice("/chosen"); + prom_chosen_node = prom_finddevice(prom_chosen_path); if (!prom_chosen_node || prom_chosen_node == -1) prom_halt(); diff --git a/arch/sparc64/prom/map.S b/arch/sparc64/prom/map.S deleted file mode 100644 index 21b3f9c99ea..00000000000 --- a/arch/sparc64/prom/map.S +++ /dev/null @@ -1,72 +0,0 @@ -/* $Id: map.S,v 1.2 1999/11/19 05:53:02 davem Exp $ - * map.S: Tricky coding required to fixup the kernel OBP maps - * properly. - * - * Copyright (C) 1999 David S. Miller (davem@redhat.com) - */ - - .text - .align 8192 - .globl prom_boot_page -prom_boot_page: -call_method: - .asciz "call-method" - .align 8 -map: - .asciz "map" - .align 8 - - /* When we are invoked, our caller has remapped us to - * page zero, therefore we must use PC relative addressing - * for everything after we begin performing the unmap/map - * calls. - */ - .globl prom_remap -prom_remap: /* %o0 = physpage, %o1 = virtpage, %o2 = mmu_ihandle */ - rd %pc, %g1 - srl %o2, 0, %o2 ! kill sign extension - sethi %hi(p1275buf), %g2 - or %g2, %lo(p1275buf), %g2 - ldx [%g2 + 0x10], %g3 ! prom_cif_stack - save %g3, -(192 + 128), %sp - ldx [%g2 + 0x08], %l0 ! prom_cif_handler - mov %g6, %i3 - mov %g4, %i4 - mov %g5, %i5 - flushw - - sethi %hi(prom_remap - call_method), %g7 - or %g7, %lo(prom_remap - call_method), %g7 - sub %g1, %g7, %l2 ! call-method string - sethi %hi(prom_remap - map), %g7 - or %g7, %lo(prom_remap - map), %g7 - sub %g1, %g7, %l4 ! map string - - /* OK, map the 4MB region we really live at. */ - stx %l2, [%sp + 2047 + 128 + 0x00] ! call-method - mov 7, %l5 - stx %l5, [%sp + 2047 + 128 + 0x08] ! num_args - mov 1, %l5 - stx %l5, [%sp + 2047 + 128 + 0x10] ! num_rets - stx %l4, [%sp + 2047 + 128 + 0x18] ! map - stx %i2, [%sp + 2047 + 128 + 0x20] ! mmu_ihandle - mov -1, %l5 - stx %l5, [%sp + 2047 + 128 + 0x28] ! mode == default - sethi %hi(4 * 1024 * 1024), %l5 - stx %l5, [%sp + 2047 + 128 + 0x30] ! size - stx %i1, [%sp + 2047 + 128 + 0x38] ! vaddr - stx %g0, [%sp + 2047 + 128 + 0x40] ! filler - stx %i0, [%sp + 2047 + 128 + 0x48] ! paddr - call %l0 - add %sp, (2047 + 128), %o0 ! argument array - - /* Restore hard-coded globals. */ - mov %i3, %g6 - mov %i4, %g4 - mov %i5, %g5 - - /* Wheee.... we are done. */ - ret - restore - - .align 8192 diff --git a/arch/sparc64/prom/misc.c b/arch/sparc64/prom/misc.c index 19c44e97e9e..9b895faf077 100644 --- a/arch/sparc64/prom/misc.c +++ b/arch/sparc64/prom/misc.c @@ -17,14 +17,14 @@ #include /* Reset and reboot the machine with the command 'bcommand'. */ -void prom_reboot(char *bcommand) +void prom_reboot(const char *bcommand) { p1275_cmd("boot", P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_INOUT(1, 0), bcommand); } /* Forth evaluate the expression contained in 'fstring'. */ -void prom_feval(char *fstring) +void prom_feval(const char *fstring) { if (!fstring || fstring[0] == 0) return; @@ -148,21 +148,19 @@ void prom_set_trap_table(unsigned long tba) p1275_cmd("SUNW,set-trap-table", P1275_INOUT(1, 0), tba); } -int mmu_ihandle_cache = 0; - int prom_get_mmu_ihandle(void) { int node, ret; - if (mmu_ihandle_cache != 0) - return mmu_ihandle_cache; + if (prom_mmu_ihandle_cache != 0) + return prom_mmu_ihandle_cache; - node = prom_finddevice("/chosen"); - ret = prom_getint(node, "mmu"); + node = prom_finddevice(prom_chosen_path); + ret = prom_getint(node, prom_mmu_name); if (ret == -1 || ret == 0) - mmu_ihandle_cache = -1; + prom_mmu_ihandle_cache = -1; else - mmu_ihandle_cache = ret; + prom_mmu_ihandle_cache = ret; return ret; } @@ -190,7 +188,7 @@ long prom_itlb_load(unsigned long index, unsigned long tte_data, unsigned long vaddr) { - return p1275_cmd("call-method", + return p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(2, P1275_ARG_IN_64B) | P1275_ARG(3, P1275_ARG_IN_64B) | @@ -207,7 +205,7 @@ long prom_dtlb_load(unsigned long index, unsigned long tte_data, unsigned long vaddr) { - return p1275_cmd("call-method", + return p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(2, P1275_ARG_IN_64B) | P1275_ARG(3, P1275_ARG_IN_64B) | @@ -223,13 +221,13 @@ long prom_dtlb_load(unsigned long index, int prom_map(int mode, unsigned long size, unsigned long vaddr, unsigned long paddr) { - int ret = p1275_cmd("call-method", + int ret = p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(3, P1275_ARG_IN_64B) | P1275_ARG(4, P1275_ARG_IN_64B) | P1275_ARG(6, P1275_ARG_IN_64B) | P1275_INOUT(7, 1)), - "map", + prom_map_name, prom_get_mmu_ihandle(), mode, size, @@ -244,12 +242,12 @@ int prom_map(int mode, unsigned long size, void prom_unmap(unsigned long size, unsigned long vaddr) { - p1275_cmd("call-method", + p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(2, P1275_ARG_IN_64B) | P1275_ARG(3, P1275_ARG_IN_64B) | P1275_INOUT(4, 0)), - "unmap", + prom_unmap_name, prom_get_mmu_ihandle(), size, vaddr); @@ -258,7 +256,7 @@ void prom_unmap(unsigned long size, unsigned long vaddr) /* Set aside physical memory which is not touched or modified * across soft resets. */ -unsigned long prom_retain(char *name, +unsigned long prom_retain(const char *name, unsigned long pa_low, unsigned long pa_high, long size, long align) { @@ -290,7 +288,7 @@ int prom_getunumber(int syndrome_code, unsigned long phys_addr, char *buf, int buflen) { - return p1275_cmd("call-method", + return p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(3, P1275_ARG_OUT_BUF) | P1275_ARG(6, P1275_ARG_IN_64B) | diff --git a/arch/sparc64/prom/p1275.c b/arch/sparc64/prom/p1275.c index 59fe38bba39..a5a7c571202 100644 --- a/arch/sparc64/prom/p1275.c +++ b/arch/sparc64/prom/p1275.c @@ -46,7 +46,7 @@ static inline unsigned long spitfire_get_primary_context(void) */ DEFINE_SPINLOCK(prom_entry_lock); -long p1275_cmd (char *service, long fmt, ...) +long p1275_cmd(const char *service, long fmt, ...) { char *p, *q; unsigned long flags; diff --git a/arch/sparc64/prom/printf.c b/arch/sparc64/prom/printf.c index a6df82cafa0..660943ee4c2 100644 --- a/arch/sparc64/prom/printf.c +++ b/arch/sparc64/prom/printf.c @@ -34,7 +34,7 @@ prom_write(const char *buf, unsigned int n) } void -prom_printf(char *fmt, ...) +prom_printf(const char *fmt, ...) { va_list args; int i; diff --git a/arch/sparc64/prom/tree.c b/arch/sparc64/prom/tree.c index ccf73258ebf..b1ff9e87dcc 100644 --- a/arch/sparc64/prom/tree.c +++ b/arch/sparc64/prom/tree.c @@ -69,7 +69,7 @@ prom_getsibling(int node) * Return -1 on error. */ __inline__ int -prom_getproplen(int node, char *prop) +prom_getproplen(int node, const char *prop) { if((!node) || (!prop)) return -1; return p1275_cmd ("getproplen", @@ -83,20 +83,20 @@ prom_getproplen(int node, char *prop) * was successful the length will be returned, else -1 is returned. */ __inline__ int -prom_getproperty(int node, char *prop, char *buffer, int bufsize) +prom_getproperty(int node, const char *prop, char *buffer, int bufsize) { int plen; plen = prom_getproplen(node, prop); - if((plen > bufsize) || (plen == 0) || (plen == -1)) + if ((plen > bufsize) || (plen == 0) || (plen == -1)) { return -1; - else { + } else { /* Ok, things seem all right. */ - return p1275_cmd ("getprop", - P1275_ARG(1,P1275_ARG_IN_STRING)| - P1275_ARG(2,P1275_ARG_OUT_BUF)| - P1275_INOUT(4, 1), - node, prop, buffer, P1275_SIZE(plen)); + return p1275_cmd(prom_getprop_name, + P1275_ARG(1,P1275_ARG_IN_STRING)| + P1275_ARG(2,P1275_ARG_OUT_BUF)| + P1275_INOUT(4, 1), + node, prop, buffer, P1275_SIZE(plen)); } } @@ -104,7 +104,7 @@ prom_getproperty(int node, char *prop, char *buffer, int bufsize) * on failure. */ __inline__ int -prom_getint(int node, char *prop) +prom_getint(int node, const char *prop) { int intprop; @@ -119,7 +119,7 @@ prom_getint(int node, char *prop) */ int -prom_getintdefault(int node, char *property, int deflt) +prom_getintdefault(int node, const char *property, int deflt) { int retval; @@ -131,7 +131,7 @@ prom_getintdefault(int node, char *property, int deflt) /* Acquire a boolean property, 1=TRUE 0=FALSE. */ int -prom_getbool(int node, char *prop) +prom_getbool(int node, const char *prop) { int retval; @@ -145,7 +145,7 @@ prom_getbool(int node, char *prop) * buffer. */ void -prom_getstring(int node, char *prop, char *user_buf, int ubuf_size) +prom_getstring(int node, const char *prop, char *user_buf, int ubuf_size) { int len; @@ -160,7 +160,7 @@ prom_getstring(int node, char *prop, char *user_buf, int ubuf_size) * YES = 1 NO = 0 */ int -prom_nodematch(int node, char *name) +prom_nodematch(int node, const char *name) { char namebuf[128]; prom_getproperty(node, "name", namebuf, sizeof(namebuf)); @@ -172,7 +172,7 @@ prom_nodematch(int node, char *name) * 'nodename'. Return node if successful, zero if not. */ int -prom_searchsiblings(int node_start, char *nodename) +prom_searchsiblings(int node_start, const char *nodename) { int thisnode, error; @@ -294,7 +294,7 @@ prom_firstprop(int node, char *buffer) * property types for this node. */ __inline__ char * -prom_nextprop(int node, char *oprop, char *buffer) +prom_nextprop(int node, const char *oprop, char *buffer) { char buf[32]; @@ -314,15 +314,17 @@ prom_nextprop(int node, char *oprop, char *buffer) } int -prom_finddevice(char *name) +prom_finddevice(const char *name) { - if(!name) return 0; - return p1275_cmd ("finddevice", P1275_ARG(0,P1275_ARG_IN_STRING)| - P1275_INOUT(1, 1), - name); + if (!name) + return 0; + return p1275_cmd(prom_finddev_name, + P1275_ARG(0,P1275_ARG_IN_STRING)| + P1275_INOUT(1, 1), + name); } -int prom_node_has_property(int node, char *prop) +int prom_node_has_property(int node, const char *prop) { char buf [32]; @@ -339,7 +341,7 @@ int prom_node_has_property(int node, char *prop) * of 'size' bytes. Return the number of bytes the prom accepted. */ int -prom_setprop(int node, char *pname, char *value, int size) +prom_setprop(int node, const char *pname, char *value, int size) { if(size == 0) return 0; if((pname == 0) || (value == 0)) return 0; @@ -364,7 +366,7 @@ prom_inst2pkg(int inst) * FIXME: Should work for v0 as well */ int -prom_pathtoinode(char *path) +prom_pathtoinode(const char *path) { int node, inst; diff --git a/include/asm-sparc64/oplib.h b/include/asm-sparc64/oplib.h index a432d9e7daa..c628189b6c8 100644 --- a/include/asm-sparc64/oplib.h +++ b/include/asm-sparc64/oplib.h @@ -38,6 +38,20 @@ extern int prom_stdin, prom_stdout; */ extern int prom_chosen_node; +/* Helper values and strings in arch/sparc64/kernel/head.S */ +extern const char prom_finddev_name[]; +extern const char prom_chosen_path[]; +extern const char prom_getprop_name[]; +extern const char prom_mmu_name[]; +extern const char prom_callmethod_name[]; +extern const char prom_translate_name[]; +extern const char prom_map_name[]; +extern const char prom_unmap_name[]; +extern int prom_mmu_ihandle_cache; +extern unsigned int prom_boot_mapped_pc; +extern unsigned int prom_boot_mapping_mode; +extern unsigned long prom_boot_mapping_phys_high, prom_boot_mapping_phys_low; + struct linux_mlist_p1275 { struct linux_mlist_p1275 *theres_more; unsigned long start_adr; @@ -68,7 +82,7 @@ extern char *prom_getbootargs(void); * of the string is different on V0 vs. V2->higher proms. The caller must * know what he/she is doing! Returns the device descriptor, an int. */ -extern int prom_devopen(char *device_string); +extern int prom_devopen(const char *device_string); /* Close a previously opened device described by the passed integer * descriptor. @@ -98,10 +112,10 @@ extern struct linux_mem_p1275 *prom_meminfo(void); /* Miscellaneous routines, don't really fit in any category per se. */ /* Reboot the machine with the command line passed. */ -extern void prom_reboot(char *boot_command); +extern void prom_reboot(const char *boot_command); /* Evaluate the forth string passed. */ -extern void prom_feval(char *forth_string); +extern void prom_feval(const char *forth_string); /* Enter the prom, with possibility of continuation with the 'go' * command in newer proms. @@ -154,7 +168,7 @@ extern char prom_getchar(void); extern void prom_putchar(char character); /* Prom's internal routines, don't use in kernel/boot code. */ -extern void prom_printf(char *fmt, ...); +extern void prom_printf(const char *fmt, ...); extern void prom_write(const char *buf, unsigned int len); /* Query for input device type */ @@ -215,7 +229,7 @@ extern int prom_getunumber(int syndrome_code, char *buf, int buflen); /* Retain physical memory to the caller across soft resets. */ -extern unsigned long prom_retain(char *name, +extern unsigned long prom_retain(const char *name, unsigned long pa_low, unsigned long pa_high, long size, long align); @@ -269,28 +283,28 @@ extern int prom_getsibling(int node); /* Get the length, at the passed node, of the given property type. * Returns -1 on error (ie. no such property at this node). */ -extern int prom_getproplen(int thisnode, char *property); +extern int prom_getproplen(int thisnode, const char *property); /* Fetch the requested property using the given buffer. Returns * the number of bytes the prom put into your buffer or -1 on error. */ -extern int prom_getproperty(int thisnode, char *property, +extern int prom_getproperty(int thisnode, const char *property, char *prop_buffer, int propbuf_size); /* Acquire an integer property. */ -extern int prom_getint(int node, char *property); +extern int prom_getint(int node, const char *property); /* Acquire an integer property, with a default value. */ -extern int prom_getintdefault(int node, char *property, int defval); +extern int prom_getintdefault(int node, const char *property, int defval); /* Acquire a boolean property, 0=FALSE 1=TRUE. */ -extern int prom_getbool(int node, char *prop); +extern int prom_getbool(int node, const char *prop); /* Acquire a string property, null string on error. */ -extern void prom_getstring(int node, char *prop, char *buf, int bufsize); +extern void prom_getstring(int node, const char *prop, char *buf, int bufsize); /* Does the passed node have the given "name"? YES=1 NO=0 */ -extern int prom_nodematch(int thisnode, char *name); +extern int prom_nodematch(int thisnode, const char *name); /* Puts in buffer a prom name in the form name@x,y or name (x for which_io * and y for first regs phys address @@ -300,7 +314,7 @@ extern int prom_getname(int node, char *buf, int buflen); /* Search all siblings starting at the passed node for "name" matching * the given string. Returns the node on success, zero on failure. */ -extern int prom_searchsiblings(int node_start, char *name); +extern int prom_searchsiblings(int node_start, const char *name); /* Return the first property type, as a string, for the given node. * Returns a null string on error. Buffer should be at least 32B long. @@ -310,21 +324,21 @@ extern char *prom_firstprop(int node, char *buffer); /* Returns the next property after the passed property for the given * node. Returns null string on failure. Buffer should be at least 32B long. */ -extern char *prom_nextprop(int node, char *prev_property, char *buffer); +extern char *prom_nextprop(int node, const char *prev_property, char *buffer); /* Returns 1 if the specified node has given property. */ -extern int prom_node_has_property(int node, char *property); +extern int prom_node_has_property(int node, const char *property); /* Returns phandle of the path specified */ -extern int prom_finddevice(char *name); +extern int prom_finddevice(const char *name); /* Set the indicated property at the given node with the passed value. * Returns the number of bytes of your value that the prom took. */ -extern int prom_setprop(int node, char *prop_name, char *prop_value, +extern int prom_setprop(int node, const char *prop_name, char *prop_value, int value_size); -extern int prom_pathtoinode(char *path); +extern int prom_pathtoinode(const char *path); extern int prom_inst2pkg(int); /* CPU probing helpers. */ @@ -334,7 +348,7 @@ int cpu_find_by_mid(int mid, int *prom_node); /* Client interface level routines. */ extern void prom_set_trap_table(unsigned long tba); -extern long p1275_cmd (char *, long, ...); +extern long p1275_cmd(const char *, long, ...); #if 0 -- cgit v1.2.3-70-g09d2 From a8201c61065f3b5a4fe2fde1db1c817e288916be Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 22 Sep 2005 20:31:29 -0700 Subject: [SPARC64]: Fix comment typo in head.S Signed-off-by: David S. Miller --- arch/sparc64/kernel/head.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index 8b83520e427..56af714f5f1 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -217,7 +217,7 @@ prom_boot_mapping_phys_low: mov 7, %l3 stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 7 mov 1, %l3 - stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 7 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1 mov (1b - prom_map_name), %l3 sub %l0, %l3, %l3 stx %l3, [%sp + 2047 + 128 + 0x18] ! arg1: "map" -- cgit v1.2.3-70-g09d2 From 1c9ea5db001142a96fec8fed0f92f26892a1b6ac Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Fri, 23 Sep 2005 11:54:43 -0700 Subject: [SPARC64]: Kill unused variable in setup_arch() 'highest_paddr' is set, but never actually used. Signed-off-by: David S. Miller --- arch/sparc64/kernel/setup.c | 7 ------- 1 file changed, 7 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index 0296cb00fc9..f4345d83728 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -496,7 +496,6 @@ extern void paging_init(void); void __init setup_arch(char **cmdline_p) { - unsigned long highest_paddr; int i; /* Initialize PROM console and command line. */ @@ -519,11 +518,7 @@ void __init setup_arch(char **cmdline_p) idprom_init(); (void) prom_probe_memory(); - /* In paging_init() we tip off this value to see if we need - * to change init_mm.pgd to point to the real alias mapping. - */ phys_base = 0xffffffffffffffffUL; - highest_paddr = 0UL; for (i = 0; sp_banks[i].num_bytes != 0; i++) { unsigned long top; @@ -531,8 +526,6 @@ void __init setup_arch(char **cmdline_p) phys_base = sp_banks[i].base_addr; top = sp_banks[i].base_addr + sp_banks[i].num_bytes; - if (highest_paddr < top) - highest_paddr = top; } pfn_base = phys_base >> PAGE_SHIFT; -- cgit v1.2.3-70-g09d2 From 52f26deb7c67d5f34910660200b925c1a2b8df8c Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Sat, 24 Sep 2005 23:06:14 -0700 Subject: [SPARC64]: Fix mask formation in tomatillo_wsync_handler() "1" needs to be "1UL", this is a 64-bit mask we're creating. Signed-off-by: David S. Miller --- arch/sparc64/kernel/pci_schizo.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/pci_schizo.c b/arch/sparc64/kernel/pci_schizo.c index 331382e1a75..cae5b61fe2f 100644 --- a/arch/sparc64/kernel/pci_schizo.c +++ b/arch/sparc64/kernel/pci_schizo.c @@ -330,7 +330,7 @@ static int schizo_ino_to_pil(struct pci_dev *pdev, unsigned int ino) static void tomatillo_wsync_handler(struct ino_bucket *bucket, void *_arg1, void *_arg2) { unsigned long sync_reg = (unsigned long) _arg2; - u64 mask = 1 << (__irq_ino(__irq(bucket)) & IMAP_INO); + u64 mask = 1UL << (__irq_ino(__irq(bucket)) & IMAP_INO); u64 val; int limit; -- cgit v1.2.3-70-g09d2 From 56425306517ef28a9b480161cdb96d182172bc1d Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Sun, 25 Sep 2005 16:46:57 -0700 Subject: [SPARC64]: Add CONFIG_DEBUG_PAGEALLOC support. The trick is that we do the kernel linear mapping TLB miss starting with an instruction sequence like this: ba,pt %xcc, kvmap_load xor %g2, %g4, %g5 succeeded by an instruction sequence which performs a full page table walk starting at swapper_pg_dir. We first take over the trap table from the firmware. Then, using this constant PTE generation for the linear mapping area above, we build the kernel page tables for the linear mapping. After this is setup, we patch that branch above into a "nop", which will cause TLB misses to fall through to the full page table walk. With this, the page unmapping for CONFIG_DEBUG_PAGEALLOC is trivial. Signed-off-by: David S. Miller --- arch/sparc64/Kconfig.debug | 8 +++ arch/sparc64/kernel/head.S | 6 --- arch/sparc64/kernel/ktlb.S | 33 +++++++++++- arch/sparc64/kernel/vmlinux.lds.S | 3 +- arch/sparc64/mm/init.c | 109 ++++++++++++++++++++++++++++++++++++-- include/asm-sparc64/cacheflush.h | 5 ++ include/asm-sparc64/pgtable.h | 7 +-- 7 files changed, 156 insertions(+), 15 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/Kconfig.debug b/arch/sparc64/Kconfig.debug index af0e9411b83..fa06ea04837 100644 --- a/arch/sparc64/Kconfig.debug +++ b/arch/sparc64/Kconfig.debug @@ -33,6 +33,14 @@ config DEBUG_BOOTMEM depends on DEBUG_KERNEL bool "Debug BOOTMEM initialization" +config DEBUG_PAGEALLOC + bool "Page alloc debugging" + depends on DEBUG_KERNEL && !SOFTWARE_SUSPEND + help + Unmap pages from the kernel linear mapping after free_pages(). + This results in a large slowdown, but helps to find certain types + of memory corruptions. + config MCOUNT bool depends on STACK_DEBUG diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index 56af714f5f1..ecc748fb9ad 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -525,12 +525,6 @@ bootup_user_stack_end: #include "ttable.S" #include "systbls.S" - - .align 1024 - .globl swapper_pg_dir -swapper_pg_dir: - .word 0 - #include "ktlb.S" #include "etrap.S" #include "rtrap.S" diff --git a/arch/sparc64/kernel/ktlb.S b/arch/sparc64/kernel/ktlb.S index a591bc0ebc7..7796b37f478 100644 --- a/arch/sparc64/kernel/ktlb.S +++ b/arch/sparc64/kernel/ktlb.S @@ -132,9 +132,40 @@ kvmap_do_obp: */ .align 32 kvmap: - brlz,pt %g4, kvmap_load + brgez,pn %g4, kvmap_nonlinear + nop + +#ifdef CONFIG_DEBUG_PAGEALLOC + .globl kvmap_linear_patch +kvmap_linear_patch: +#endif + ba,pt %xcc, kvmap_load xor %g2, %g4, %g5 +#ifdef CONFIG_DEBUG_PAGEALLOC + sethi %hi(swapper_pg_dir), %g5 + or %g5, %lo(swapper_pg_dir), %g5 + sllx %g4, 64 - (PGDIR_SHIFT + PGDIR_BITS), %g6 + srlx %g6, 64 - PAGE_SHIFT, %g6 + andn %g6, 0x3, %g6 + lduw [%g5 + %g6], %g5 + brz,pn %g5, longpath + sllx %g4, 64 - (PMD_SHIFT + PMD_BITS), %g6 + srlx %g6, 64 - PAGE_SHIFT, %g6 + sllx %g5, 11, %g5 + andn %g6, 0x3, %g6 + lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 + brz,pn %g5, longpath + sllx %g4, 64 - PMD_SHIFT, %g6 + srlx %g6, 64 - PAGE_SHIFT, %g6 + sllx %g5, 11, %g5 + andn %g6, 0x7, %g6 + ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 + brz,pn %g5, longpath + nop + ba,a,pt %xcc, kvmap_load +#endif + kvmap_nonlinear: sethi %hi(MODULES_VADDR), %g5 cmp %g4, %g5 diff --git a/arch/sparc64/kernel/vmlinux.lds.S b/arch/sparc64/kernel/vmlinux.lds.S index f47d0be3937..2af0cf0a864 100644 --- a/arch/sparc64/kernel/vmlinux.lds.S +++ b/arch/sparc64/kernel/vmlinux.lds.S @@ -9,8 +9,7 @@ ENTRY(_start) jiffies = jiffies_64; SECTIONS { - swapper_pmd_dir = 0x0000000000402000; - empty_pg_dir = 0x0000000000403000; + swapper_low_pmd_dir = 0x0000000000402000; . = 0x4000; .text 0x0000000000404000 : { diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index cf747372f0c..8d72f8a1268 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -1332,15 +1332,114 @@ unsigned long __init bootmem_init(unsigned long *pages_avail) return end_pfn; } +#ifdef CONFIG_DEBUG_PAGEALLOC +static unsigned long kernel_map_range(unsigned long pstart, unsigned long pend, pgprot_t prot) +{ + unsigned long vstart = PAGE_OFFSET + pstart; + unsigned long vend = PAGE_OFFSET + pend; + unsigned long alloc_bytes = 0UL; + + if ((vstart & ~PAGE_MASK) || (vend & ~PAGE_MASK)) { + prom_printf("kernel_map: Unaligned sp_banks[%lx:%lx]\n", + vstart, vend); + prom_halt(); + } + + while (vstart < vend) { + unsigned long this_end, paddr = __pa(vstart); + pgd_t *pgd = pgd_offset_k(vstart); + pud_t *pud; + pmd_t *pmd; + pte_t *pte; + + pud = pud_offset(pgd, vstart); + if (pud_none(*pud)) { + pmd_t *new; + + new = __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, PAGE_SIZE); + alloc_bytes += PAGE_SIZE; + pud_populate(&init_mm, pud, new); + } + + pmd = pmd_offset(pud, vstart); + if (!pmd_present(*pmd)) { + pte_t *new; + + new = __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, PAGE_SIZE); + alloc_bytes += PAGE_SIZE; + pmd_populate_kernel(&init_mm, pmd, new); + } + + pte = pte_offset_kernel(pmd, vstart); + this_end = (vstart + PMD_SIZE) & PMD_MASK; + if (this_end > vend) + this_end = vend; + + while (vstart < this_end) { + pte_val(*pte) = (paddr | pgprot_val(prot)); + + vstart += PAGE_SIZE; + paddr += PAGE_SIZE; + pte++; + } + } + + return alloc_bytes; +} + +extern struct linux_mlist_p1275 *prom_ptot_ptr; +extern unsigned int kvmap_linear_patch[1]; + +static void __init kernel_physical_mapping_init(void) +{ + struct linux_mlist_p1275 *p = prom_ptot_ptr; + unsigned long mem_alloced = 0UL; + + while (p) { + unsigned long phys_start, phys_end; + + phys_start = p->start_adr; + phys_end = phys_start + p->num_bytes; + mem_alloced += kernel_map_range(phys_start, phys_end, + PAGE_KERNEL); + + p = p->theres_more; + } + + printk("Allocated %ld bytes for kernel page tables.\n", + mem_alloced); + + kvmap_linear_patch[0] = 0x01000000; /* nop */ + flushi(&kvmap_linear_patch[0]); + + __flush_tlb_all(); +} + +void kernel_map_pages(struct page *page, int numpages, int enable) +{ + unsigned long phys_start = page_to_pfn(page) << PAGE_SHIFT; + unsigned long phys_end = phys_start + (numpages * PAGE_SIZE); + + kernel_map_range(phys_start, phys_end, + (enable ? PAGE_KERNEL : __pgprot(0))); + + /* we should perform an IPI and flush all tlbs, + * but that can deadlock->flush only current cpu. + */ + __flush_tlb_kernel_range(PAGE_OFFSET + phys_start, + PAGE_OFFSET + phys_end); +} +#endif + /* paging_init() sets up the page tables */ extern void cheetah_ecache_flush_init(void); static unsigned long last_valid_pfn; +pgd_t swapper_pg_dir[2048]; void __init paging_init(void) { - extern pmd_t swapper_pmd_dir[1024]; unsigned long end_pfn, pages_avail, shift; unsigned long real_end; @@ -1361,11 +1460,11 @@ void __init paging_init(void) */ init_mm.pgd += ((shift) / (sizeof(pgd_t))); - memset(swapper_pmd_dir, 0, sizeof(swapper_pmd_dir)); + memset(swapper_low_pmd_dir, 0, sizeof(swapper_low_pmd_dir)); /* Now can init the kernel/bad page tables. */ pud_set(pud_offset(&swapper_pg_dir[0], 0), - swapper_pmd_dir + (shift / sizeof(pgd_t))); + swapper_low_pmd_dir + (shift / sizeof(pgd_t))); swapper_pgd_zero = pgd_val(swapper_pg_dir[0]); @@ -1390,6 +1489,10 @@ void __init paging_init(void) pages_avail = 0; last_valid_pfn = end_pfn = bootmem_init(&pages_avail); +#ifdef CONFIG_DEBUG_PAGEALLOC + kernel_physical_mapping_init(); +#endif + { unsigned long zones_size[MAX_NR_ZONES]; unsigned long zholes_size[MAX_NR_ZONES]; diff --git a/include/asm-sparc64/cacheflush.h b/include/asm-sparc64/cacheflush.h index ededd2659ea..b3f61659ba8 100644 --- a/include/asm-sparc64/cacheflush.h +++ b/include/asm-sparc64/cacheflush.h @@ -66,6 +66,11 @@ extern void flush_ptrace_access(struct vm_area_struct *, struct page *, #define flush_cache_vmap(start, end) do { } while (0) #define flush_cache_vunmap(start, end) do { } while (0) +#ifdef CONFIG_DEBUG_PAGEALLOC +/* internal debugging function */ +void kernel_map_pages(struct page *page, int numpages, int enable); +#endif + #endif /* !__ASSEMBLY__ */ #endif /* _SPARC64_CACHEFLUSH_H */ diff --git a/include/asm-sparc64/pgtable.h b/include/asm-sparc64/pgtable.h index a297f6144f0..43cbb089cde 100644 --- a/include/asm-sparc64/pgtable.h +++ b/include/asm-sparc64/pgtable.h @@ -60,13 +60,13 @@ * table can map */ #define PMD_SHIFT (PAGE_SHIFT + (PAGE_SHIFT-3)) -#define PMD_SIZE (1UL << PMD_SHIFT) +#define PMD_SIZE (_AC(1,UL) << PMD_SHIFT) #define PMD_MASK (~(PMD_SIZE-1)) #define PMD_BITS (PAGE_SHIFT - 2) /* PGDIR_SHIFT determines what a third-level page table entry can map */ #define PGDIR_SHIFT (PAGE_SHIFT + (PAGE_SHIFT-3) + PMD_BITS) -#define PGDIR_SIZE (1UL << PGDIR_SHIFT) +#define PGDIR_SIZE (_AC(1,UL) << PGDIR_SHIFT) #define PGDIR_MASK (~(PGDIR_SIZE-1)) #define PGDIR_BITS (PAGE_SHIFT - 2) @@ -336,7 +336,8 @@ static inline void set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *p #define pte_clear(mm,addr,ptep) \ set_pte_at((mm), (addr), (ptep), __pte(0UL)) -extern pgd_t swapper_pg_dir[1]; +extern pgd_t swapper_pg_dir[2048]; +extern pmd_t swapper_low_pmd_dir[2048]; /* These do nothing with the way I have things setup. */ #define mmu_lockarea(vaddr, len) (vaddr) -- cgit v1.2.3-70-g09d2 From 80dc0d6b44ce0f01df58d8899e46612690ed7d81 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Mon, 26 Sep 2005 00:32:17 -0700 Subject: [SPARC64]: Probe D/I/E-cache config and use. At boot time, determine the D-cache, I-cache and E-cache size and line-size. Use them in cache flushes when appropriate. This change was motivated by discovering that the D-cache on UltraSparc-IIIi and later are 64K not 32K, and the flushes done by the Cheetah error handlers were assuming a 32K size. There are still some pieces of code that are hard coding things and will need to be fixed up at some point. While we're here, fix the D-cache and I-cache parity error handlers to run with interrupts disabled, and when the trap occurs at trap level > 1 log the event via a counter displayed in /proc/cpuinfo. Signed-off-by: David S. Miller --- arch/sparc64/kernel/devices.c | 22 ++++++++++++++++++++++ arch/sparc64/kernel/entry.S | 27 ++++++++++++++++++--------- arch/sparc64/kernel/setup.c | 9 ++++++++- arch/sparc64/kernel/smp.c | 21 +++++++++++++++++++++ arch/sparc64/kernel/traps.c | 40 +++++++++++++++++++++++++++------------- include/asm-sparc64/cpudata.h | 10 ++++++++++ 6 files changed, 106 insertions(+), 23 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/devices.c b/arch/sparc64/kernel/devices.c index d710274e516..df9a1ca8fd7 100644 --- a/arch/sparc64/kernel/devices.c +++ b/arch/sparc64/kernel/devices.c @@ -135,6 +135,28 @@ void __init device_scan(void) cpu_data(0).clock_tick = prom_getintdefault(cpu_node, "clock-frequency", 0); + cpu_data(0).dcache_size = prom_getintdefault(cpu_node, + "dcache-size", + 16 * 1024); + cpu_data(0).dcache_line_size = + prom_getintdefault(cpu_node, "dcache-line-size", 32); + cpu_data(0).icache_size = prom_getintdefault(cpu_node, + "icache-size", + 16 * 1024); + cpu_data(0).icache_line_size = + prom_getintdefault(cpu_node, "icache-line-size", 32); + cpu_data(0).ecache_size = prom_getintdefault(cpu_node, + "ecache-size", + 4 * 1024 * 1024); + cpu_data(0).ecache_line_size = + prom_getintdefault(cpu_node, "ecache-line-size", 64); + printk("CPU[0]: Caches " + "D[sz(%d):line_sz(%d)] " + "I[sz(%d):line_sz(%d)] " + "E[sz(%d):line_sz(%d)]\n", + cpu_data(0).dcache_size, cpu_data(0).dcache_line_size, + cpu_data(0).icache_size, cpu_data(0).icache_line_size, + cpu_data(0).ecache_size, cpu_data(0).ecache_line_size); } #endif diff --git a/arch/sparc64/kernel/entry.S b/arch/sparc64/kernel/entry.S index 45cd3bbdb7e..2879b107292 100644 --- a/arch/sparc64/kernel/entry.S +++ b/arch/sparc64/kernel/entry.S @@ -372,14 +372,13 @@ cheetah_plus_patch_fpdis: * * DATA 0: [low 32-bits] Address of function to call, jmp to this * [high 32-bits] MMU Context Argument 0, place in %g5 - * DATA 1: Address Argument 1, place in %g6 + * DATA 1: Address Argument 1, place in %g1 * DATA 2: Address Argument 2, place in %g7 * * With this method we can do most of the cross-call tlb/cache * flushing very quickly. * - * Current CPU's IRQ worklist table is locked into %g1, - * don't touch. + * Current CPU's IRQ worklist table is locked into %g6, don't touch. */ .text .align 32 @@ -853,13 +852,14 @@ cheetah_plus_dcpe_trap_vector: nop do_cheetah_plus_data_parity: - ba,pt %xcc, etrap + rdpr %pil, %g2 + wrpr %g0, 15, %pil + ba,pt %xcc, etrap_irq rd %pc, %g7 mov 0x0, %o0 call cheetah_plus_parity_error add %sp, PTREGS_OFF, %o1 - ba,pt %xcc, rtrap - clr %l6 + ba,a,pt %xcc, rtrap_irq cheetah_plus_dcpe_trap_vector_tl1: membar #Sync @@ -883,13 +883,14 @@ cheetah_plus_icpe_trap_vector: nop do_cheetah_plus_insn_parity: - ba,pt %xcc, etrap + rdpr %pil, %g2 + wrpr %g0, 15, %pil + ba,pt %xcc, etrap_irq rd %pc, %g7 mov 0x1, %o0 call cheetah_plus_parity_error add %sp, PTREGS_OFF, %o1 - ba,pt %xcc, rtrap - clr %l6 + ba,a,pt %xcc, rtrap_irq cheetah_plus_icpe_trap_vector_tl1: membar #Sync @@ -922,6 +923,10 @@ do_dcpe_tl1: nop wrpr %g1, %tl ! Restore original trap level do_dcpe_tl1_nonfatal: /* Ok we may use interrupt globals safely. */ + sethi %hi(dcache_parity_tl1_occurred), %g2 + lduw [%g2 + %lo(dcache_parity_tl1_occurred)], %g1 + add %g1, 1, %g1 + stw %g1, [%g2 + %lo(dcache_parity_tl1_occurred)] /* Reset D-cache parity */ sethi %hi(1 << 16), %g1 ! D-cache size mov (1 << 5), %g2 ! D-cache line size @@ -968,6 +973,10 @@ do_icpe_tl1: nop wrpr %g1, %tl ! Restore original trap level do_icpe_tl1_nonfatal: /* Ok we may use interrupt globals safely. */ + sethi %hi(icache_parity_tl1_occurred), %g2 + lduw [%g2 + %lo(icache_parity_tl1_occurred)], %g1 + add %g1, 1, %g1 + stw %g1, [%g2 + %lo(icache_parity_tl1_occurred)] /* Flush I-cache */ sethi %hi(1 << 15), %g1 ! I-cache size mov (1 << 5), %g2 ! I-cache line size diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index f4345d83728..8e8baf2354d 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -605,6 +605,9 @@ extern void smp_info(struct seq_file *); extern void smp_bogo(struct seq_file *); extern void mmu_info(struct seq_file *); +unsigned int dcache_parity_tl1_occurred; +unsigned int icache_parity_tl1_occurred; + static int show_cpuinfo(struct seq_file *m, void *__unused) { seq_printf(m, @@ -615,6 +618,8 @@ static int show_cpuinfo(struct seq_file *m, void *__unused) "type\t\t: sun4u\n" "ncpus probed\t: %ld\n" "ncpus active\t: %ld\n" + "D$ parity tl1\t: %u\n" + "I$ parity tl1\t: %u\n" #ifndef CONFIG_SMP "Cpu0Bogo\t: %lu.%02lu\n" "Cpu0ClkTck\t: %016lx\n" @@ -627,7 +632,9 @@ static int show_cpuinfo(struct seq_file *m, void *__unused) (prom_prev >> 8) & 0xff, prom_prev & 0xff, (long)num_possible_cpus(), - (long)num_online_cpus() + (long)num_online_cpus(), + dcache_parity_tl1_occurred, + icache_parity_tl1_occurred #ifndef CONFIG_SMP , cpu_data(0).udelay_val/(500000/HZ), (cpu_data(0).udelay_val/(5000/HZ)) % 100, diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index b4fc6a5462b..590df5a16f5 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -93,6 +93,27 @@ void __init smp_store_cpu_info(int id) cpu_data(id).pte_cache[1] = NULL; cpu_data(id).pgd_cache = NULL; cpu_data(id).idle_volume = 1; + + cpu_data(id).dcache_size = prom_getintdefault(cpu_node, "dcache-size", + 16 * 1024); + cpu_data(id).dcache_line_size = + prom_getintdefault(cpu_node, "dcache-line-size", 32); + cpu_data(id).icache_size = prom_getintdefault(cpu_node, "icache-size", + 16 * 1024); + cpu_data(id).icache_line_size = + prom_getintdefault(cpu_node, "icache-line-size", 32); + cpu_data(id).ecache_size = prom_getintdefault(cpu_node, "ecache-size", + 4 * 1024 * 1024); + cpu_data(id).ecache_line_size = + prom_getintdefault(cpu_node, "ecache-line-size", 64); + printk("CPU[%d]: Caches " + "D[sz(%d):line_sz(%d)] " + "I[sz(%d):line_sz(%d)] " + "E[sz(%d):line_sz(%d)]\n", + id, + cpu_data(id).dcache_size, cpu_data(id).dcache_line_size, + cpu_data(id).icache_size, cpu_data(id).icache_line_size, + cpu_data(id).ecache_size, cpu_data(id).ecache_line_size); } static void smp_setup_percpu_timer(void); diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c index b280b2ef674..f8e7005fede 100644 --- a/arch/sparc64/kernel/traps.c +++ b/arch/sparc64/kernel/traps.c @@ -869,14 +869,19 @@ static void cheetah_flush_ecache_line(unsigned long physaddr) */ static void __cheetah_flush_icache(void) { - unsigned long i; + unsigned int icache_size, icache_line_size; + unsigned long addr; + + icache_size = local_cpu_data().icache_size; + icache_line_size = local_cpu_data().icache_line_size; /* Clear the valid bits in all the tags. */ - for (i = 0; i < (1 << 15); i += (1 << 5)) { + for (addr = 0; addr < icache_size; addr += icache_line_size) { __asm__ __volatile__("stxa %%g0, [%0] %1\n\t" "membar #Sync" : /* no outputs */ - : "r" (i | (2 << 3)), "i" (ASI_IC_TAG)); + : "r" (addr | (2 << 3)), + "i" (ASI_IC_TAG)); } } @@ -904,13 +909,17 @@ static void cheetah_flush_icache(void) static void cheetah_flush_dcache(void) { - unsigned long i; + unsigned int dcache_size, dcache_line_size; + unsigned long addr; + + dcache_size = local_cpu_data().dcache_size; + dcache_line_size = local_cpu_data().dcache_line_size; - for (i = 0; i < (1 << 16); i += (1 << 5)) { + for (addr = 0; addr < dcache_size; addr += dcache_line_size) { __asm__ __volatile__("stxa %%g0, [%0] %1\n\t" "membar #Sync" : /* no outputs */ - : "r" (i), "i" (ASI_DCACHE_TAG)); + : "r" (addr), "i" (ASI_DCACHE_TAG)); } } @@ -921,24 +930,29 @@ static void cheetah_flush_dcache(void) */ static void cheetah_plus_zap_dcache_parity(void) { - unsigned long i; + unsigned int dcache_size, dcache_line_size; + unsigned long addr; + + dcache_size = local_cpu_data().dcache_size; + dcache_line_size = local_cpu_data().dcache_line_size; - for (i = 0; i < (1 << 16); i += (1 << 5)) { - unsigned long tag = (i >> 14); - unsigned long j; + for (addr = 0; addr < dcache_size; addr += dcache_line_size) { + unsigned long tag = (addr >> 14); + unsigned long line; __asm__ __volatile__("membar #Sync\n\t" "stxa %0, [%1] %2\n\t" "membar #Sync" : /* no outputs */ - : "r" (tag), "r" (i), + : "r" (tag), "r" (addr), "i" (ASI_DCACHE_UTAG)); - for (j = i; j < i + (1 << 5); j += (1 << 3)) + for (line = addr; line < addr + dcache_line_size; line += 8) __asm__ __volatile__("membar #Sync\n\t" "stxa %%g0, [%0] %1\n\t" "membar #Sync" : /* no outputs */ - : "r" (j), "i" (ASI_DCACHE_DATA)); + : "r" (line), + "i" (ASI_DCACHE_DATA)); } } diff --git a/include/asm-sparc64/cpudata.h b/include/asm-sparc64/cpudata.h index 9a3a81f1cc5..74de79dca91 100644 --- a/include/asm-sparc64/cpudata.h +++ b/include/asm-sparc64/cpudata.h @@ -22,6 +22,16 @@ typedef struct { unsigned int __pad1; unsigned long *pte_cache[2]; unsigned long *pgd_cache; + + /* Dcache line 3, rarely used */ + unsigned int dcache_size; + unsigned int dcache_line_size; + unsigned int icache_size; + unsigned int icache_line_size; + unsigned int ecache_size; + unsigned int ecache_line_size; + unsigned int __pad2; + unsigned int __pad3; } cpuinfo_sparc; DECLARE_PER_CPU(cpuinfo_sparc, __cpu_data); -- cgit v1.2.3-70-g09d2 From d2212bc7db13268bef0799d9ff4b2e511c284885 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Tue, 27 Sep 2005 22:50:06 -0700 Subject: [SPARC64]: Add missing IDs for newer cpus. Also, the us3_cpufreq driver can work on Ultra-IV and IV+. They use the SAFARI bus register to control the clock divider just like Ultra-III and III+ do. Signed-off-by: David S. Miller --- arch/sparc64/kernel/cpu.c | 4 ++++ arch/sparc64/kernel/us3_cpufreq.c | 5 ++++- include/asm-sparc64/head.h | 9 ++++++--- 3 files changed, 14 insertions(+), 4 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/cpu.c b/arch/sparc64/kernel/cpu.c index 48756958116..77ef5df4e5a 100644 --- a/arch/sparc64/kernel/cpu.c +++ b/arch/sparc64/kernel/cpu.c @@ -39,6 +39,8 @@ struct cpu_fp_info linux_sparc_fpu[] = { { 0x3e, 0x15, 0, "UltraSparc III+ integrated FPU"}, { 0x3e, 0x16, 0, "UltraSparc IIIi integrated FPU"}, { 0x3e, 0x18, 0, "UltraSparc IV integrated FPU"}, + { 0x3e, 0x19, 0, "UltraSparc IV+ integrated FPU"}, + { 0x3e, 0x22, 0, "UltraSparc IIIi+ integrated FPU"}, }; #define NSPARCFPU (sizeof(linux_sparc_fpu)/sizeof(struct cpu_fp_info)) @@ -53,6 +55,8 @@ struct cpu_iu_info linux_sparc_chips[] = { { 0x3e, 0x15, "TI UltraSparc III+ (Cheetah+)"}, { 0x3e, 0x16, "TI UltraSparc IIIi (Jalapeno)"}, { 0x3e, 0x18, "TI UltraSparc IV (Jaguar)"}, + { 0x3e, 0x19, "TI UltraSparc IV+ (Panther)"}, + { 0x3e, 0x22, "TI UltraSparc IIIi+ (Serrano)"}, }; #define NSPARCCHIPS (sizeof(linux_sparc_chips)/sizeof(struct cpu_iu_info)) diff --git a/arch/sparc64/kernel/us3_cpufreq.c b/arch/sparc64/kernel/us3_cpufreq.c index 9080e7cd4bb..0340041f614 100644 --- a/arch/sparc64/kernel/us3_cpufreq.c +++ b/arch/sparc64/kernel/us3_cpufreq.c @@ -208,7 +208,10 @@ static int __init us3_freq_init(void) impl = ((ver >> 32) & 0xffff); if (manuf == CHEETAH_MANUF && - (impl == CHEETAH_IMPL || impl == CHEETAH_PLUS_IMPL)) { + (impl == CHEETAH_IMPL || + impl == CHEETAH_PLUS_IMPL || + impl == JAGUAR_IMPL || + impl == PANTHER_IMPL)) { struct cpufreq_driver *driver; ret = -ENOMEM; diff --git a/include/asm-sparc64/head.h b/include/asm-sparc64/head.h index b63a33cf497..0abd3a674e8 100644 --- a/include/asm-sparc64/head.h +++ b/include/asm-sparc64/head.h @@ -12,9 +12,12 @@ #define __JALAPENO_ID 0x003e0016 #define CHEETAH_MANUF 0x003e -#define CHEETAH_IMPL 0x0014 -#define CHEETAH_PLUS_IMPL 0x0015 -#define JALAPENO_IMPL 0x0016 +#define CHEETAH_IMPL 0x0014 /* Ultra-III */ +#define CHEETAH_PLUS_IMPL 0x0015 /* Ultra-III+ */ +#define JALAPENO_IMPL 0x0016 /* Ultra-IIIi */ +#define JAGUAR_IMPL 0x0018 /* Ultra-IV */ +#define PANTHER_IMPL 0x0019 /* Ultra-IV+ */ +#define SERRANO_IMPL 0x0022 /* Ultra-IIIi+ */ #define BRANCH_IF_CHEETAH_BASE(tmp1,tmp2,label) \ rdpr %ver, %tmp1; \ -- cgit v1.2.3-70-g09d2 From 705747ab87c96f1b4b8e73ba617c323d9087f6ac Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 28 Sep 2005 16:48:40 -0700 Subject: [SPARC64]: Fix bug in unaligned load endianness swapping The in-memory value was being swapped, not the value we loaded into the register. Signed-off-by: David S. Miller --- arch/sparc64/kernel/unaligned.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/unaligned.c b/arch/sparc64/kernel/unaligned.c index 42718f6a7d3..02af08ffec8 100644 --- a/arch/sparc64/kernel/unaligned.c +++ b/arch/sparc64/kernel/unaligned.c @@ -294,7 +294,7 @@ asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn, u kernel_mna_trap_fault(); } else { - unsigned long addr; + unsigned long addr, *reg_addr; int orig_asi, asi; addr = compute_effective_address(regs, insn, @@ -319,11 +319,11 @@ asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn, u }; switch (dir) { case load: - do_int_load(fetch_reg_addr(((insn>>25)&0x1f), regs), - size, (unsigned long *) addr, + reg_addr = fetch_reg_addr(((insn>>25)&0x1f), regs); + do_int_load(reg_addr, size, (unsigned long *) addr, decode_signedness(insn), asi); if (unlikely(asi != orig_asi)) { - unsigned long val_in = *(unsigned long *) addr; + unsigned long val_in = *reg_addr; switch (size) { case 2: val_in = swab16(val_in); @@ -339,7 +339,7 @@ asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn, u BUG(); break; }; - *(unsigned long *) addr = val_in; + *reg_addr = val_in; } break; -- cgit v1.2.3-70-g09d2 From 8cf14af0a740fb7e9f94a203b5a989beb875d58f Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 28 Sep 2005 20:21:11 -0700 Subject: [SPARC64]: Convert to use generic exception table support. The funny "range" exception table entries we had were only used by the compat layer socketcall assembly, and it wasn't even needed there. For free we now get proper exception table sorting and fast binary searching. Signed-off-by: David S. Miller --- arch/sparc64/kernel/sys32.S | 145 ++++++++++++++++++++++------------------ arch/sparc64/kernel/traps.c | 24 +++---- arch/sparc64/kernel/unaligned.c | 9 ++- arch/sparc64/mm/Makefile | 2 +- arch/sparc64/mm/extable.c | 80 ---------------------- arch/sparc64/mm/fault.c | 10 +-- include/asm-sparc64/uaccess.h | 17 +---- 7 files changed, 101 insertions(+), 186 deletions(-) delete mode 100644 arch/sparc64/mm/extable.c (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/sys32.S b/arch/sparc64/kernel/sys32.S index 5f9e4fae612..4fb99e0bc7c 100644 --- a/arch/sparc64/kernel/sys32.S +++ b/arch/sparc64/kernel/sys32.S @@ -158,163 +158,163 @@ sys32_socketcall: /* %o0=call, %o1=args */ jmpl %g2 + %o0, %g0 nop - /* Each entry is exactly 32 bytes. */ .align 32 __socketcall_table_begin: + + /* Each entry is exactly 32 bytes. */ do_sys_socket: /* sys_socket(int, int, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +1: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_socket), %g1 - ldswa [%o1 + 0x8] %asi, %o2 +2: ldswa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_socket), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +3: ldswa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_bind: /* sys_bind(int fd, struct sockaddr *, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +4: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_bind), %g1 - ldswa [%o1 + 0x8] %asi, %o2 +5: ldswa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_bind), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +6: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_connect: /* sys_connect(int, struct sockaddr *, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +7: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_connect), %g1 - ldswa [%o1 + 0x8] %asi, %o2 +8: ldswa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_connect), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +9: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_listen: /* sys_listen(int, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +10: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_listen), %g1 jmpl %g1 + %lo(sys_listen), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +11: ldswa [%o1 + 0x4] %asi, %o1 nop nop nop nop do_sys_accept: /* sys_accept(int, struct sockaddr *, int *) */ - ldswa [%o1 + 0x0] %asi, %o0 +12: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_accept), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +13: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_accept), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +14: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_getsockname: /* sys_getsockname(int, struct sockaddr *, int *) */ - ldswa [%o1 + 0x0] %asi, %o0 +15: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_getsockname), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +16: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_getsockname), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +17: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_getpeername: /* sys_getpeername(int, struct sockaddr *, int *) */ - ldswa [%o1 + 0x0] %asi, %o0 +18: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_getpeername), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +19: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_getpeername), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +20: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_socketpair: /* sys_socketpair(int, int, int, int *) */ - ldswa [%o1 + 0x0] %asi, %o0 +21: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_socketpair), %g1 - ldswa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 +22: ldswa [%o1 + 0x8] %asi, %o2 +23: lduwa [%o1 + 0xc] %asi, %o3 jmpl %g1 + %lo(sys_socketpair), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +24: ldswa [%o1 + 0x4] %asi, %o1 nop nop do_sys_send: /* sys_send(int, void *, size_t, unsigned int) */ - ldswa [%o1 + 0x0] %asi, %o0 +25: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_send), %g1 - lduwa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 +26: lduwa [%o1 + 0x8] %asi, %o2 +27: lduwa [%o1 + 0xc] %asi, %o3 jmpl %g1 + %lo(sys_send), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +28: lduwa [%o1 + 0x4] %asi, %o1 nop nop do_sys_recv: /* sys_recv(int, void *, size_t, unsigned int) */ - ldswa [%o1 + 0x0] %asi, %o0 +29: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_recv), %g1 - lduwa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 +30: lduwa [%o1 + 0x8] %asi, %o2 +31: lduwa [%o1 + 0xc] %asi, %o3 jmpl %g1 + %lo(sys_recv), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +32: lduwa [%o1 + 0x4] %asi, %o1 nop nop do_sys_sendto: /* sys_sendto(int, u32, compat_size_t, unsigned int, u32, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +33: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_sendto), %g1 - lduwa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 - lduwa [%o1 + 0x10] %asi, %o4 - ldswa [%o1 + 0x14] %asi, %o5 +34: lduwa [%o1 + 0x8] %asi, %o2 +35: lduwa [%o1 + 0xc] %asi, %o3 +36: lduwa [%o1 + 0x10] %asi, %o4 +37: ldswa [%o1 + 0x14] %asi, %o5 jmpl %g1 + %lo(sys_sendto), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +38: lduwa [%o1 + 0x4] %asi, %o1 do_sys_recvfrom: /* sys_recvfrom(int, u32, compat_size_t, unsigned int, u32, u32) */ - ldswa [%o1 + 0x0] %asi, %o0 +39: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_recvfrom), %g1 - lduwa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 - lduwa [%o1 + 0x10] %asi, %o4 - lduwa [%o1 + 0x14] %asi, %o5 +40: lduwa [%o1 + 0x8] %asi, %o2 +41: lduwa [%o1 + 0xc] %asi, %o3 +42: lduwa [%o1 + 0x10] %asi, %o4 +43: lduwa [%o1 + 0x14] %asi, %o5 jmpl %g1 + %lo(sys_recvfrom), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +44: lduwa [%o1 + 0x4] %asi, %o1 do_sys_shutdown: /* sys_shutdown(int, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +45: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_shutdown), %g1 jmpl %g1 + %lo(sys_shutdown), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +46: ldswa [%o1 + 0x4] %asi, %o1 nop nop nop nop do_sys_setsockopt: /* compat_sys_setsockopt(int, int, int, char *, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +47: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(compat_sys_setsockopt), %g1 - ldswa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 - ldswa [%o1 + 0x10] %asi, %o4 +48: ldswa [%o1 + 0x8] %asi, %o2 +49: lduwa [%o1 + 0xc] %asi, %o3 +50: ldswa [%o1 + 0x10] %asi, %o4 jmpl %g1 + %lo(compat_sys_setsockopt), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +51: ldswa [%o1 + 0x4] %asi, %o1 nop do_sys_getsockopt: /* compat_sys_getsockopt(int, int, int, u32, u32) */ - ldswa [%o1 + 0x0] %asi, %o0 +52: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(compat_sys_getsockopt), %g1 - ldswa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 - lduwa [%o1 + 0x10] %asi, %o4 +53: ldswa [%o1 + 0x8] %asi, %o2 +54: lduwa [%o1 + 0xc] %asi, %o3 +55: lduwa [%o1 + 0x10] %asi, %o4 jmpl %g1 + %lo(compat_sys_getsockopt), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +56: ldswa [%o1 + 0x4] %asi, %o1 nop do_sys_sendmsg: /* compat_sys_sendmsg(int, struct compat_msghdr *, unsigned int) */ - ldswa [%o1 + 0x0] %asi, %o0 +57: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(compat_sys_sendmsg), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +58: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(compat_sys_sendmsg), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +59: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_recvmsg: /* compat_sys_recvmsg(int, struct compat_msghdr *, unsigned int) */ - ldswa [%o1 + 0x0] %asi, %o0 +60: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(compat_sys_recvmsg), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +61: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(compat_sys_recvmsg), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +62: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop -__socketcall_table_end: do_einval: retl @@ -325,5 +325,20 @@ do_efault: .section __ex_table .align 4 - .word __socketcall_table_begin, 0, __socketcall_table_end, do_efault + .word 1b, do_efault, 2b, do_efault, 3b, do_efault, 4b, do_efault + .word 5b, do_efault, 6b, do_efault, 7b, do_efault, 8b, do_efault + .word 9b, do_efault, 10b, do_efault, 11b, do_efault, 12b, do_efault + .word 13b, do_efault, 14b, do_efault, 15b, do_efault, 16b, do_efault + .word 17b, do_efault, 18b, do_efault, 19b, do_efault, 20b, do_efault + .word 21b, do_efault, 22b, do_efault, 23b, do_efault, 24b, do_efault + .word 25b, do_efault, 26b, do_efault, 27b, do_efault, 28b, do_efault + .word 29b, do_efault, 30b, do_efault, 31b, do_efault, 32b, do_efault + .word 33b, do_efault, 34b, do_efault, 35b, do_efault, 36b, do_efault + .word 37b, do_efault, 38b, do_efault, 39b, do_efault, 40b, do_efault + .word 41b, do_efault, 42b, do_efault, 43b, do_efault, 44b, do_efault + .word 45b, do_efault, 46b, do_efault, 47b, do_efault, 48b, do_efault + .word 49b, do_efault, 50b, do_efault, 51b, do_efault, 52b, do_efault + .word 53b, do_efault, 54b, do_efault, 55b, do_efault, 56b, do_efault + .word 57b, do_efault, 58b, do_efault, 59b, do_efault, 60b, do_efault + .word 61b, do_efault, 62b, do_efault .previous diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c index f8e7005fede..1aa15990f5a 100644 --- a/arch/sparc64/kernel/traps.c +++ b/arch/sparc64/kernel/traps.c @@ -189,19 +189,18 @@ void spitfire_data_access_exception(struct pt_regs *regs, unsigned long sfsr, un if (regs->tstate & TSTATE_PRIV) { /* Test if this comes from uaccess places. */ - unsigned long fixup; - unsigned long g2 = regs->u_regs[UREG_G2]; + const struct exception_table_entry *entry; - if ((fixup = search_extables_range(regs->tpc, &g2))) { - /* Ouch, somebody is trying ugly VM hole tricks on us... */ + entry = search_exception_tables(regs->tpc); + if (entry) { + /* Ouch, somebody is trying VM hole tricks on us... */ #ifdef DEBUG_EXCEPTIONS printk("Exception: PC<%016lx> faddr\n", regs->tpc); - printk("EX_TABLE: insn<%016lx> fixup<%016lx> " - "g2<%016lx>\n", regs->tpc, fixup, g2); + printk("EX_TABLE: insn<%016lx> fixup<%016lx>\n", + regs->tpc, entry->fixup); #endif - regs->tpc = fixup; + regs->tpc = entry->fixup; regs->tnpc = regs->tpc + 4; - regs->u_regs[UREG_G2] = g2; return; } /* Shit... */ @@ -1610,10 +1609,10 @@ void cheetah_deferred_handler(struct pt_regs *regs, unsigned long afsr, unsigned /* OK, usermode access. */ recoverable = 1; } else { - unsigned long g2 = regs->u_regs[UREG_G2]; - unsigned long fixup = search_extables_range(regs->tpc, &g2); + const struct exception_table_entry *entry; - if (fixup != 0UL) { + entry = search_exception_tables(regs->tpc); + if (entry) { /* OK, kernel access to userspace. */ recoverable = 1; @@ -1632,9 +1631,8 @@ void cheetah_deferred_handler(struct pt_regs *regs, unsigned long afsr, unsigned * recoverable condition. */ if (recoverable) { - regs->tpc = fixup; + regs->tpc = entry->fixup; regs->tnpc = regs->tpc + 4; - regs->u_regs[UREG_G2] = g2; } } } diff --git a/arch/sparc64/kernel/unaligned.c b/arch/sparc64/kernel/unaligned.c index 02af08ffec8..9d6be20f412 100644 --- a/arch/sparc64/kernel/unaligned.c +++ b/arch/sparc64/kernel/unaligned.c @@ -246,10 +246,10 @@ void kernel_mna_trap_fault(void) { struct pt_regs *regs = current_thread_info()->kern_una_regs; unsigned int insn = current_thread_info()->kern_una_insn; - unsigned long g2 = regs->u_regs[UREG_G2]; - unsigned long fixup = search_extables_range(regs->tpc, &g2); + const struct exception_table_entry *entry; - if (!fixup) { + entry = search_exception_tables(regs->tpc); + if (!entry) { unsigned long address; address = compute_effective_address(regs, insn, @@ -270,9 +270,8 @@ void kernel_mna_trap_fault(void) die_if_kernel("Oops", regs); /* Not reached */ } - regs->tpc = fixup; + regs->tpc = entry->fixup; regs->tnpc = regs->tpc + 4; - regs->u_regs [UREG_G2] = g2; regs->tstate &= ~TSTATE_ASI; regs->tstate |= (ASI_AIUS << 24UL); diff --git a/arch/sparc64/mm/Makefile b/arch/sparc64/mm/Makefile index cda87333a77..9d0960e69f4 100644 --- a/arch/sparc64/mm/Makefile +++ b/arch/sparc64/mm/Makefile @@ -5,6 +5,6 @@ EXTRA_AFLAGS := -ansi EXTRA_CFLAGS := -Werror -obj-y := ultra.o tlb.o fault.o init.o generic.o extable.o +obj-y := ultra.o tlb.o fault.o init.o generic.o obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o diff --git a/arch/sparc64/mm/extable.c b/arch/sparc64/mm/extable.c deleted file mode 100644 index ec334297ff4..00000000000 --- a/arch/sparc64/mm/extable.c +++ /dev/null @@ -1,80 +0,0 @@ -/* - * linux/arch/sparc64/mm/extable.c - */ - -#include -#include -#include - -extern const struct exception_table_entry __start___ex_table[]; -extern const struct exception_table_entry __stop___ex_table[]; - -void sort_extable(struct exception_table_entry *start, - struct exception_table_entry *finish) -{ -} - -/* Caller knows they are in a range if ret->fixup == 0 */ -const struct exception_table_entry * -search_extable(const struct exception_table_entry *start, - const struct exception_table_entry *last, - unsigned long value) -{ - const struct exception_table_entry *walk; - - /* Single insn entries are encoded as: - * word 1: insn address - * word 2: fixup code address - * - * Range entries are encoded as: - * word 1: first insn address - * word 2: 0 - * word 3: last insn address + 4 bytes - * word 4: fixup code address - * - * See asm/uaccess.h for more details. - */ - - /* 1. Try to find an exact match. */ - for (walk = start; walk <= last; walk++) { - if (walk->fixup == 0) { - /* A range entry, skip both parts. */ - walk++; - continue; - } - - if (walk->insn == value) - return walk; - } - - /* 2. Try to find a range match. */ - for (walk = start; walk <= (last - 1); walk++) { - if (walk->fixup) - continue; - - if (walk[0].insn <= value && walk[1].insn > value) - return walk; - - walk++; - } - - return NULL; -} - -/* Special extable search, which handles ranges. Returns fixup */ -unsigned long search_extables_range(unsigned long addr, unsigned long *g2) -{ - const struct exception_table_entry *entry; - - entry = search_exception_tables(addr); - if (!entry) - return 0; - - /* Inside range? Fix g2 and return correct fixup */ - if (!entry->fixup) { - *g2 = (addr - entry->insn) / 4; - return (entry + 1)->fixup; - } - - return entry->fixup; -} diff --git a/arch/sparc64/mm/fault.c b/arch/sparc64/mm/fault.c index db1e3310e90..59dc9a2ece5 100644 --- a/arch/sparc64/mm/fault.c +++ b/arch/sparc64/mm/fault.c @@ -242,7 +242,6 @@ static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn) static void do_kernel_fault(struct pt_regs *regs, int si_code, int fault_code, unsigned int insn, unsigned long address) { - unsigned long g2; unsigned char asi = ASI_P; if ((!insn) && (regs->tstate & TSTATE_PRIV)) @@ -273,11 +272,9 @@ static void do_kernel_fault(struct pt_regs *regs, int si_code, int fault_code, } } - g2 = regs->u_regs[UREG_G2]; - /* Is this in ex_table? */ if (regs->tstate & TSTATE_PRIV) { - unsigned long fixup; + const struct exception_table_entry *entry; if (asi == ASI_P && (insn & 0xc0800000) == 0xc0800000) { if (insn & 0x2000) @@ -288,10 +285,9 @@ static void do_kernel_fault(struct pt_regs *regs, int si_code, int fault_code, /* Look in asi.h: All _S asis have LS bit set */ if ((asi & 0x1) && - (fixup = search_extables_range(regs->tpc, &g2))) { - regs->tpc = fixup; + (entry = search_exception_tables(regs->tpc))) { + regs->tpc = entry->fixup; regs->tnpc = regs->tpc + 4; - regs->u_regs[UREG_G2] = g2; return; } } else { diff --git a/include/asm-sparc64/uaccess.h b/include/asm-sparc64/uaccess.h index 80a65d7e3db..c099aa33978 100644 --- a/include/asm-sparc64/uaccess.h +++ b/include/asm-sparc64/uaccess.h @@ -70,25 +70,12 @@ static inline int access_ok(int type, const void __user * addr, unsigned long si * with the main instruction path. This means when everything is well, * we don't even have to jump over them. Further, they do not intrude * on our cache or tlb entries. - * - * There is a special way how to put a range of potentially faulting - * insns (like twenty ldd/std's with now intervening other instructions) - * You specify address of first in insn and 0 in fixup and in the next - * exception_table_entry you specify last potentially faulting insn + 1 - * and in fixup the routine which should handle the fault. - * That fixup code will get - * (faulting_insn_address - first_insn_in_the_range_address)/4 - * in %g2 (ie. index of the faulting instruction in the range). */ -struct exception_table_entry -{ - unsigned insn, fixup; +struct exception_table_entry { + unsigned int insn, fixup; }; -/* Special exable search, which handles ranges. Returns fixup */ -unsigned long search_extables_range(unsigned long addr, unsigned long *g2); - extern void __ret_efault(void); /* Uh, these should become the main single-value transfer routines.. -- cgit v1.2.3-70-g09d2 From 5fd29752f09cabff582f65c0ce35518db4c64937 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 28 Sep 2005 20:41:45 -0700 Subject: [SPARC64]: Fix fault handling in unaligned trap handler. We were not calling kernel_mna_trap_fault() correctly. Instead of being fancy, just return 0 vs. -EFAULT from the assembler stubs, and handle that return value as appropriate. Create an "__retl_efault" stub for assembler exception table entries and use it where possible. Signed-off-by: David S. Miller --- arch/sparc64/kernel/head.S | 7 ++-- arch/sparc64/kernel/sys32.S | 57 ++++++++++++++++++------------- arch/sparc64/kernel/una_asm.S | 65 ++++++++++++++++-------------------- arch/sparc64/kernel/unaligned.c | 36 +++++++++++--------- arch/sparc64/lib/strncpy_from_user.S | 16 ++++----- include/asm-sparc64/uaccess.h | 1 + 6 files changed, 95 insertions(+), 87 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index ecc748fb9ad..89406f9649a 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -540,8 +540,11 @@ bootup_user_stack_end: prom_tba: .xword 0 tlb_type: .word 0 /* Must NOT end up in BSS */ .section ".fixup",#alloc,#execinstr - .globl __ret_efault + + .globl __ret_efault, __retl_efault __ret_efault: ret restore %g0, -EFAULT, %o0 - +__retl_efault: + retl + mov -EFAULT, %o0 diff --git a/arch/sparc64/kernel/sys32.S b/arch/sparc64/kernel/sys32.S index 4fb99e0bc7c..9cd272ac3ac 100644 --- a/arch/sparc64/kernel/sys32.S +++ b/arch/sparc64/kernel/sys32.S @@ -157,6 +157,9 @@ sys32_socketcall: /* %o0=call, %o1=args */ or %g2, %lo(__socketcall_table_begin), %g2 jmpl %g2 + %o0, %g0 nop +do_einval: + retl + mov -EINVAL, %o0 .align 32 __socketcall_table_begin: @@ -316,29 +319,37 @@ do_sys_recvmsg: /* compat_sys_recvmsg(int, struct compat_msghdr *, unsigned int) nop nop -do_einval: - retl - mov -EINVAL, %o0 -do_efault: - retl - mov -EFAULT, %o0 - .section __ex_table .align 4 - .word 1b, do_efault, 2b, do_efault, 3b, do_efault, 4b, do_efault - .word 5b, do_efault, 6b, do_efault, 7b, do_efault, 8b, do_efault - .word 9b, do_efault, 10b, do_efault, 11b, do_efault, 12b, do_efault - .word 13b, do_efault, 14b, do_efault, 15b, do_efault, 16b, do_efault - .word 17b, do_efault, 18b, do_efault, 19b, do_efault, 20b, do_efault - .word 21b, do_efault, 22b, do_efault, 23b, do_efault, 24b, do_efault - .word 25b, do_efault, 26b, do_efault, 27b, do_efault, 28b, do_efault - .word 29b, do_efault, 30b, do_efault, 31b, do_efault, 32b, do_efault - .word 33b, do_efault, 34b, do_efault, 35b, do_efault, 36b, do_efault - .word 37b, do_efault, 38b, do_efault, 39b, do_efault, 40b, do_efault - .word 41b, do_efault, 42b, do_efault, 43b, do_efault, 44b, do_efault - .word 45b, do_efault, 46b, do_efault, 47b, do_efault, 48b, do_efault - .word 49b, do_efault, 50b, do_efault, 51b, do_efault, 52b, do_efault - .word 53b, do_efault, 54b, do_efault, 55b, do_efault, 56b, do_efault - .word 57b, do_efault, 58b, do_efault, 59b, do_efault, 60b, do_efault - .word 61b, do_efault, 62b, do_efault + .word 1b, __retl_efault, 2b, __retl_efault + .word 3b, __retl_efault, 4b, __retl_efault + .word 5b, __retl_efault, 6b, __retl_efault + .word 7b, __retl_efault, 8b, __retl_efault + .word 9b, __retl_efault, 10b, __retl_efault + .word 11b, __retl_efault, 12b, __retl_efault + .word 13b, __retl_efault, 14b, __retl_efault + .word 15b, __retl_efault, 16b, __retl_efault + .word 17b, __retl_efault, 18b, __retl_efault + .word 19b, __retl_efault, 20b, __retl_efault + .word 21b, __retl_efault, 22b, __retl_efault + .word 23b, __retl_efault, 24b, __retl_efault + .word 25b, __retl_efault, 26b, __retl_efault + .word 27b, __retl_efault, 28b, __retl_efault + .word 29b, __retl_efault, 30b, __retl_efault + .word 31b, __retl_efault, 32b, __retl_efault + .word 33b, __retl_efault, 34b, __retl_efault + .word 35b, __retl_efault, 36b, __retl_efault + .word 37b, __retl_efault, 38b, __retl_efault + .word 39b, __retl_efault, 40b, __retl_efault + .word 41b, __retl_efault, 42b, __retl_efault + .word 43b, __retl_efault, 44b, __retl_efault + .word 45b, __retl_efault, 46b, __retl_efault + .word 47b, __retl_efault, 48b, __retl_efault + .word 49b, __retl_efault, 50b, __retl_efault + .word 51b, __retl_efault, 52b, __retl_efault + .word 53b, __retl_efault, 54b, __retl_efault + .word 55b, __retl_efault, 56b, __retl_efault + .word 57b, __retl_efault, 58b, __retl_efault + .word 59b, __retl_efault, 60b, __retl_efault + .word 61b, __retl_efault, 62b, __retl_efault .previous diff --git a/arch/sparc64/kernel/una_asm.S b/arch/sparc64/kernel/una_asm.S index da48400bcc9..1f5b5b708ce 100644 --- a/arch/sparc64/kernel/una_asm.S +++ b/arch/sparc64/kernel/una_asm.S @@ -6,13 +6,6 @@ .text -kernel_unaligned_trap_fault: - call kernel_mna_trap_fault - nop - retl - nop - .size kern_unaligned_trap_fault, .-kern_unaligned_trap_fault - .globl __do_int_store __do_int_store: rd %asi, %o4 @@ -51,24 +44,24 @@ __do_int_store: 0: wr %o4, 0x0, %asi retl - nop + mov 0, %o0 .size __do_int_store, .-__do_int_store .section __ex_table - .word 4b, kernel_unaligned_trap_fault - .word 5b, kernel_unaligned_trap_fault - .word 6b, kernel_unaligned_trap_fault - .word 7b, kernel_unaligned_trap_fault - .word 8b, kernel_unaligned_trap_fault - .word 9b, kernel_unaligned_trap_fault - .word 10b, kernel_unaligned_trap_fault - .word 11b, kernel_unaligned_trap_fault - .word 12b, kernel_unaligned_trap_fault - .word 13b, kernel_unaligned_trap_fault - .word 14b, kernel_unaligned_trap_fault - .word 15b, kernel_unaligned_trap_fault - .word 16b, kernel_unaligned_trap_fault - .word 17b, kernel_unaligned_trap_fault + .word 4b, __retl_efault + .word 5b, __retl_efault + .word 6b, __retl_efault + .word 7b, __retl_efault + .word 8b, __retl_efault + .word 9b, __retl_efault + .word 10b, __retl_efault + .word 11b, __retl_efault + .word 12b, __retl_efault + .word 13b, __retl_efault + .word 14b, __retl_efault + .word 15b, __retl_efault + .word 16b, __retl_efault + .word 17b, __retl_efault .previous .globl do_int_load @@ -133,21 +126,21 @@ do_int_load: 0: wr %o5, 0x0, %asi retl - nop + mov 0, %o0 .size __do_int_load, .-__do_int_load .section __ex_table - .word 4b, kernel_unaligned_trap_fault - .word 5b, kernel_unaligned_trap_fault - .word 6b, kernel_unaligned_trap_fault - .word 7b, kernel_unaligned_trap_fault - .word 8b, kernel_unaligned_trap_fault - .word 9b, kernel_unaligned_trap_fault - .word 10b, kernel_unaligned_trap_fault - .word 11b, kernel_unaligned_trap_fault - .word 12b, kernel_unaligned_trap_fault - .word 13b, kernel_unaligned_trap_fault - .word 14b, kernel_unaligned_trap_fault - .word 15b, kernel_unaligned_trap_fault - .word 16b, kernel_unaligned_trap_fault + .word 4b, __retl_efault + .word 5b, __retl_efault + .word 6b, __retl_efault + .word 7b, __retl_efault + .word 8b, __retl_efault + .word 9b, __retl_efault + .word 10b, __retl_efault + .word 11b, __retl_efault + .word 12b, __retl_efault + .word 13b, __retl_efault + .word 14b, __retl_efault + .word 15b, __retl_efault + .word 16b, __retl_efault .previous diff --git a/arch/sparc64/kernel/unaligned.c b/arch/sparc64/kernel/unaligned.c index 9d6be20f412..70faf630603 100644 --- a/arch/sparc64/kernel/unaligned.c +++ b/arch/sparc64/kernel/unaligned.c @@ -180,14 +180,14 @@ static void __attribute_used__ unaligned_panic(char *str, struct pt_regs *regs) die_if_kernel(str, regs); } -extern void do_int_load(unsigned long *dest_reg, int size, - unsigned long *saddr, int is_signed, int asi); +extern int do_int_load(unsigned long *dest_reg, int size, + unsigned long *saddr, int is_signed, int asi); -extern void __do_int_store(unsigned long *dst_addr, int size, - unsigned long src_val, int asi); +extern int __do_int_store(unsigned long *dst_addr, int size, + unsigned long src_val, int asi); -static inline void do_int_store(int reg_num, int size, unsigned long *dst_addr, - struct pt_regs *regs, int asi, int orig_asi) +static inline int do_int_store(int reg_num, int size, unsigned long *dst_addr, + struct pt_regs *regs, int asi, int orig_asi) { unsigned long zero = 0; unsigned long *src_val_p = &zero; @@ -219,7 +219,7 @@ static inline void do_int_store(int reg_num, int size, unsigned long *dst_addr, break; }; } - __do_int_store(dst_addr, size, src_val, asi); + return __do_int_store(dst_addr, size, src_val, asi); } static inline void advance(struct pt_regs *regs) @@ -242,7 +242,7 @@ static inline int ok_for_kernel(unsigned int insn) return !floating_point_load_or_store_p(insn); } -void kernel_mna_trap_fault(void) +static void kernel_mna_trap_fault(void) { struct pt_regs *regs = current_thread_info()->kern_una_regs; unsigned int insn = current_thread_info()->kern_una_insn; @@ -294,7 +294,7 @@ asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn, u kernel_mna_trap_fault(); } else { unsigned long addr, *reg_addr; - int orig_asi, asi; + int orig_asi, asi, err; addr = compute_effective_address(regs, insn, ((insn >> 25) & 0x1f)); @@ -319,9 +319,10 @@ asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn, u switch (dir) { case load: reg_addr = fetch_reg_addr(((insn>>25)&0x1f), regs); - do_int_load(reg_addr, size, (unsigned long *) addr, - decode_signedness(insn), asi); - if (unlikely(asi != orig_asi)) { + err = do_int_load(reg_addr, size, + (unsigned long *) addr, + decode_signedness(insn), asi); + if (likely(!err) && unlikely(asi != orig_asi)) { unsigned long val_in = *reg_addr; switch (size) { case 2: @@ -343,16 +344,19 @@ asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn, u break; case store: - do_int_store(((insn>>25)&0x1f), size, - (unsigned long *) addr, regs, - asi, orig_asi); + err = do_int_store(((insn>>25)&0x1f), size, + (unsigned long *) addr, regs, + asi, orig_asi); break; default: panic("Impossible kernel unaligned trap."); /* Not reached... */ } - advance(regs); + if (unlikely(err)) + kernel_mna_trap_fault(); + else + advance(regs); } } diff --git a/arch/sparc64/lib/strncpy_from_user.S b/arch/sparc64/lib/strncpy_from_user.S index 09cbbaa0ebf..e1264650ca7 100644 --- a/arch/sparc64/lib/strncpy_from_user.S +++ b/arch/sparc64/lib/strncpy_from_user.S @@ -125,15 +125,11 @@ __strncpy_from_user: add %o2, %o3, %o0 .size __strncpy_from_user, .-__strncpy_from_user - .section .fixup,#alloc,#execinstr - .align 4 -4: retl - mov -EFAULT, %o0 - .section __ex_table,#alloc .align 4 - .word 60b, 4b - .word 61b, 4b - .word 62b, 4b - .word 63b, 4b - .word 64b, 4b + .word 60b, __retl_efault + .word 61b, __retl_efault + .word 62b, __retl_efault + .word 63b, __retl_efault + .word 64b, __retl_efault + .previous diff --git a/include/asm-sparc64/uaccess.h b/include/asm-sparc64/uaccess.h index c099aa33978..bc8ddbb1cbe 100644 --- a/include/asm-sparc64/uaccess.h +++ b/include/asm-sparc64/uaccess.h @@ -77,6 +77,7 @@ struct exception_table_entry { }; extern void __ret_efault(void); +extern void __retl_efault(void); /* Uh, these should become the main single-value transfer routines.. * They automatically use the right size if we just have the right -- cgit v1.2.3-70-g09d2 From 801ab3c731e77324c055769491711e620100dbfb Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 28 Sep 2005 21:31:25 -0700 Subject: [SPARC]: Declare paging_init() in asm/pgtable.h Signed-off-by: David S. Miller --- arch/sparc/kernel/setup.c | 2 -- arch/sparc64/kernel/setup.c | 2 -- include/asm-sparc/pgtable.h | 2 ++ include/asm-sparc64/pgtable.h | 2 ++ 4 files changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc/kernel/setup.c b/arch/sparc/kernel/setup.c index 53c192a4982..3509e430553 100644 --- a/arch/sparc/kernel/setup.c +++ b/arch/sparc/kernel/setup.c @@ -249,8 +249,6 @@ struct tt_entry *sparc_ttable; struct pt_regs fake_swapper_regs; -extern void paging_init(void); - void __init setup_arch(char **cmdline_p) { int i; diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index 8e8baf2354d..516f4854955 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -492,8 +492,6 @@ void register_prom_callbacks(void) "' linux-.soft2 to .soft2"); } -extern void paging_init(void); - void __init setup_arch(char **cmdline_p) { int i; diff --git a/include/asm-sparc/pgtable.h b/include/asm-sparc/pgtable.h index 8f4f6a95965..8395ad2f1c0 100644 --- a/include/asm-sparc/pgtable.h +++ b/include/asm-sparc/pgtable.h @@ -82,6 +82,8 @@ extern unsigned long page_kernel; /* Top-level page directory */ extern pgd_t swapper_pg_dir[1024]; +extern void paging_init(void); + /* Page table for 0-4MB for everybody, on the Sparc this * holds the same as on the i386. */ diff --git a/include/asm-sparc64/pgtable.h b/include/asm-sparc64/pgtable.h index 53d612aba8d..82273c801b0 100644 --- a/include/asm-sparc64/pgtable.h +++ b/include/asm-sparc64/pgtable.h @@ -341,6 +341,8 @@ static inline void set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *p extern pgd_t swapper_pg_dir[2048]; extern pmd_t swapper_low_pmd_dir[2048]; +extern void paging_init(void); + /* These do nothing with the way I have things setup. */ #define mmu_lockarea(vaddr, len) (vaddr) #define mmu_unlockarea(vaddr, len) do { } while(0) -- cgit v1.2.3-70-g09d2 From 0836a0eb4073c3e0a09c5965833b9dec19f5abc7 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 28 Sep 2005 21:38:08 -0700 Subject: [SPARC64]: Move phys_base, kern_{base,size}, and sp_banks[] init to paging_init Also, move prom_probe_memory() into arch/sparc64/mm/init.c Signed-off-by: David S. Miller --- arch/sparc64/kernel/setup.c | 19 -------------- arch/sparc64/mm/fault.c | 47 ---------------------------------- arch/sparc64/mm/init.c | 62 ++++++++++++++++++++++++++++++++++++++++++++- 3 files changed, 61 insertions(+), 67 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index 516f4854955..4c9c8f24174 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -464,8 +464,6 @@ static void __init boot_flags_init(char *commands) } } -extern int prom_probe_memory(void); -extern unsigned long start, end; extern void panic_setup(char *, int *); extern unsigned short root_flags; @@ -494,8 +492,6 @@ void register_prom_callbacks(void) void __init setup_arch(char **cmdline_p) { - int i; - /* Initialize PROM console and command line. */ *cmdline_p = prom_getbootargs(); strcpy(saved_command_line, *cmdline_p); @@ -514,21 +510,6 @@ void __init setup_arch(char **cmdline_p) boot_flags_init(*cmdline_p); idprom_init(); - (void) prom_probe_memory(); - - phys_base = 0xffffffffffffffffUL; - for (i = 0; sp_banks[i].num_bytes != 0; i++) { - unsigned long top; - - if (sp_banks[i].base_addr < phys_base) - phys_base = sp_banks[i].base_addr; - top = sp_banks[i].base_addr + - sp_banks[i].num_bytes; - } - pfn_base = phys_base >> PAGE_SHIFT; - - kern_base = (prom_boot_mapping_phys_low >> 22UL) << 22UL; - kern_size = (unsigned long)&_end - (unsigned long)KERNBASE; if (!root_flags) root_mountflags &= ~MS_RDONLY; diff --git a/arch/sparc64/mm/fault.c b/arch/sparc64/mm/fault.c index 4a52e79d515..80793d61f31 100644 --- a/arch/sparc64/mm/fault.c +++ b/arch/sparc64/mm/fault.c @@ -71,53 +71,6 @@ void set_brkpt(unsigned long addr, unsigned char mask, int flags, int mode) : "memory"); } -/* Nice, simple, prom library does all the sweating for us. ;) */ -unsigned long __init prom_probe_memory (void) -{ - register struct linux_mlist_p1275 *mlist; - register unsigned long bytes, base_paddr, tally; - register int i; - - i = 0; - mlist = *prom_meminfo()->p1275_available; - bytes = tally = mlist->num_bytes; - base_paddr = mlist->start_adr; - - sp_banks[0].base_addr = base_paddr; - sp_banks[0].num_bytes = bytes; - - while (mlist->theres_more != (void *) 0) { - i++; - mlist = mlist->theres_more; - bytes = mlist->num_bytes; - tally += bytes; - if (i >= SPARC_PHYS_BANKS-1) { - printk ("The machine has more banks than " - "this kernel can support\n" - "Increase the SPARC_PHYS_BANKS " - "setting (currently %d)\n", - SPARC_PHYS_BANKS); - i = SPARC_PHYS_BANKS-1; - break; - } - - sp_banks[i].base_addr = mlist->start_adr; - sp_banks[i].num_bytes = mlist->num_bytes; - } - - i++; - sp_banks[i].base_addr = 0xdeadbeefbeefdeadUL; - sp_banks[i].num_bytes = 0; - - /* Now mask all bank sizes on a page boundary, it is all we can - * use anyways. - */ - for (i = 0; sp_banks[i].num_bytes != 0; i++) - sp_banks[i].num_bytes &= PAGE_MASK; - - return tally; -} - static void __kprobes unhandled_fault(unsigned long address, struct task_struct *tsk, struct pt_regs *regs) diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index 9f6ca624892..63a7b9bafaf 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -1425,6 +1425,50 @@ void kernel_map_pages(struct page *page, int numpages, int enable) } #endif +static void __init prom_probe_memory(void) +{ + struct linux_mlist_p1275 *mlist; + unsigned long bytes, base_paddr, tally; + int i; + + i = 0; + mlist = *prom_meminfo()->p1275_available; + bytes = tally = mlist->num_bytes; + base_paddr = mlist->start_adr; + + sp_banks[0].base_addr = base_paddr; + sp_banks[0].num_bytes = bytes; + + while (mlist->theres_more != (void *) 0) { + i++; + mlist = mlist->theres_more; + bytes = mlist->num_bytes; + tally += bytes; + if (i >= SPARC_PHYS_BANKS-1) { + printk ("The machine has more banks than " + "this kernel can support\n" + "Increase the SPARC_PHYS_BANKS " + "setting (currently %d)\n", + SPARC_PHYS_BANKS); + i = SPARC_PHYS_BANKS-1; + break; + } + + sp_banks[i].base_addr = mlist->start_adr; + sp_banks[i].num_bytes = mlist->num_bytes; + } + + i++; + sp_banks[i].base_addr = 0xdeadbeefbeefdeadUL; + sp_banks[i].num_bytes = 0; + + /* Now mask all bank sizes on a page boundary, it is all we can + * use anyways. + */ + for (i = 0; sp_banks[i].num_bytes != 0; i++) + sp_banks[i].num_bytes &= PAGE_MASK; +} + /* paging_init() sets up the page tables */ extern void cheetah_ecache_flush_init(void); @@ -1435,7 +1479,23 @@ pgd_t swapper_pg_dir[2048]; void __init paging_init(void) { unsigned long end_pfn, pages_avail, shift; - unsigned long real_end; + unsigned long real_end, i; + + prom_probe_memory(); + + phys_base = 0xffffffffffffffffUL; + for (i = 0; sp_banks[i].num_bytes != 0; i++) { + unsigned long top; + + if (sp_banks[i].base_addr < phys_base) + phys_base = sp_banks[i].base_addr; + top = sp_banks[i].base_addr + + sp_banks[i].num_bytes; + } + pfn_base = phys_base >> PAGE_SHIFT; + + kern_base = (prom_boot_mapping_phys_low >> 22UL) << 22UL; + kern_size = (unsigned long)&_end - (unsigned long)KERNBASE; set_bit(0, mmu_context_bmap); -- cgit v1.2.3-70-g09d2 From 10147570f9eaff3920f0c67bad7244c2eb958d4f Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 28 Sep 2005 21:46:43 -0700 Subject: [SPARC64]: Kill all external references to sp_banks[] Thus, we can mark sp_banks[] static in arch/sparc64/mm/init.c Signed-off-by: David S. Miller --- arch/sparc64/kernel/traps.c | 33 ++++++--------------------------- arch/sparc64/mm/fault.c | 2 -- arch/sparc64/mm/init.c | 23 ++++++++++++++++++++++- include/asm-sparc64/page.h | 17 ----------------- include/asm-sparc64/pgtable.h | 1 + 5 files changed, 29 insertions(+), 47 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c index 1aa15990f5a..eeb1e835c42 100644 --- a/arch/sparc64/kernel/traps.c +++ b/arch/sparc64/kernel/traps.c @@ -757,26 +757,12 @@ void __init cheetah_ecache_flush_init(void) ecache_flush_size = (2 * largest_size); ecache_flush_linesize = smallest_linesize; - /* Discover a physically contiguous chunk of physical - * memory in 'sp_banks' of size ecache_flush_size calculated - * above. Store the physical base of this area at - * ecache_flush_physbase. - */ - for (node = 0; ; node++) { - if (sp_banks[node].num_bytes == 0) - break; - if (sp_banks[node].num_bytes >= ecache_flush_size) { - ecache_flush_physbase = sp_banks[node].base_addr; - break; - } - } + ecache_flush_physbase = find_ecache_flush_span(ecache_flush_size); - /* Note: Zero would be a valid value of ecache_flush_physbase so - * don't use that as the success test. :-) - */ - if (sp_banks[node].num_bytes == 0) { + if (ecache_flush_physbase == ~0UL) { prom_printf("cheetah_ecache_flush_init: Cannot find %d byte " - "contiguous physical memory.\n", ecache_flush_size); + "contiguous physical memory.\n", + ecache_flush_size); prom_halt(); } @@ -1345,16 +1331,9 @@ static int cheetah_fix_ce(unsigned long physaddr) /* Return non-zero if PADDR is a valid physical memory address. */ static int cheetah_check_main_memory(unsigned long paddr) { - int i; + unsigned long vaddr = PAGE_OFFSET + paddr; - for (i = 0; ; i++) { - if (sp_banks[i].num_bytes == 0) - break; - if (paddr >= sp_banks[i].base_addr && - paddr < (sp_banks[i].base_addr + sp_banks[i].num_bytes)) - return 1; - } - return 0; + return kern_addr_valid(vaddr); } void cheetah_cee_handler(struct pt_regs *regs, unsigned long afsr, unsigned long afar) diff --git a/arch/sparc64/mm/fault.c b/arch/sparc64/mm/fault.c index 80793d61f31..31fbc67719a 100644 --- a/arch/sparc64/mm/fault.c +++ b/arch/sparc64/mm/fault.c @@ -32,8 +32,6 @@ #define ELEMENTS(arr) (sizeof (arr)/sizeof (arr[0])) -extern struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS]; - /* * To debug kernel to catch accesses to certain virtual/physical addresses. * Mode = 0 selects physical watchpoints, mode = 1 selects virtual watchpoints. diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index 63a7b9bafaf..48851a2e4fe 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -41,7 +41,14 @@ extern void device_scan(void); -struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS]; +struct sparc_phys_banks { + unsigned long base_addr; + unsigned long num_bytes; +}; + +#define SPARC_PHYS_BANKS 32 + +static struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS]; unsigned long *sparc64_valid_addr_bitmap __read_mostly; @@ -1425,6 +1432,20 @@ void kernel_map_pages(struct page *page, int numpages, int enable) } #endif +unsigned long __init find_ecache_flush_span(unsigned long size) +{ + unsigned long i; + + for (i = 0; ; i++) { + if (sp_banks[i].num_bytes == 0) + break; + if (sp_banks[i].num_bytes >= size) + return sp_banks[i].base_addr; + } + + return ~0UL; +} + static void __init prom_probe_memory(void) { struct linux_mlist_p1275 *mlist; diff --git a/include/asm-sparc64/page.h b/include/asm-sparc64/page.h index 7f8d764abc4..5426bb28a99 100644 --- a/include/asm-sparc64/page.h +++ b/include/asm-sparc64/page.h @@ -140,23 +140,6 @@ extern unsigned long page_to_pfn(struct page *); #define virt_to_phys __pa #define phys_to_virt __va -/* The following structure is used to hold the physical - * memory configuration of the machine. This is filled in - * probe_memory() and is later used by mem_init() to set up - * mem_map[]. We statically allocate SPARC_PHYS_BANKS of - * these structs, this is arbitrary. The entry after the - * last valid one has num_bytes==0. - */ - -struct sparc_phys_banks { - unsigned long base_addr; - unsigned long num_bytes; -}; - -#define SPARC_PHYS_BANKS 32 - -extern struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS]; - #endif /* !(__ASSEMBLY__) */ #define VM_DATA_DEFAULT_FLAGS (VM_READ | VM_WRITE | VM_EXEC | \ diff --git a/include/asm-sparc64/pgtable.h b/include/asm-sparc64/pgtable.h index 82273c801b0..8c6dfc6c7af 100644 --- a/include/asm-sparc64/pgtable.h +++ b/include/asm-sparc64/pgtable.h @@ -342,6 +342,7 @@ extern pgd_t swapper_pg_dir[2048]; extern pmd_t swapper_low_pmd_dir[2048]; extern void paging_init(void); +extern unsigned long find_ecache_flush_span(unsigned long size); /* These do nothing with the way I have things setup. */ #define mmu_lockarea(vaddr, len) (vaddr) -- cgit v1.2.3-70-g09d2 From ed3ffaf7b5e0262cb860f106a6632933671cc88f Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 28 Sep 2005 21:48:25 -0700 Subject: [SPARC64]: Solidify check in cheetah_check_main_memory(). Need to make sure the address is below high_memory before passing it to kern_addr_valid(). Signed-off-by: David S. Miller --- arch/sparc64/kernel/traps.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c index eeb1e835c42..7f190fc5754 100644 --- a/arch/sparc64/kernel/traps.c +++ b/arch/sparc64/kernel/traps.c @@ -1333,6 +1333,9 @@ static int cheetah_check_main_memory(unsigned long paddr) { unsigned long vaddr = PAGE_OFFSET + paddr; + if (vaddr > high_memory) + return 0; + return kern_addr_valid(vaddr); } -- cgit v1.2.3-70-g09d2 From 13edad7a5cef1c952459742482482a6b05e1a8a1 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 29 Sep 2005 17:58:26 -0700 Subject: [SPARC64]: Rewrite convoluted physical memory probing. Delete all of the code working with sp_banks[] and replace with clean acquisition and sorting of physical memory parameters from the firmware. Signed-off-by: David S. Miller --- arch/sparc64/kernel/traps.c | 2 +- arch/sparc64/mm/init.c | 302 +++++++++++++++-------------------------- include/asm-sparc64/openprom.h | 4 +- 3 files changed, 114 insertions(+), 194 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c index 7f190fc5754..5570e7bb22b 100644 --- a/arch/sparc64/kernel/traps.c +++ b/arch/sparc64/kernel/traps.c @@ -1333,7 +1333,7 @@ static int cheetah_check_main_memory(unsigned long paddr) { unsigned long vaddr = PAGE_OFFSET + paddr; - if (vaddr > high_memory) + if (vaddr > (unsigned long) high_memory) return 0; return kern_addr_valid(vaddr); diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index 48851a2e4fe..5db50524f20 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -21,6 +21,7 @@ #include #include #include +#include #include #include @@ -41,14 +42,72 @@ extern void device_scan(void); -struct sparc_phys_banks { - unsigned long base_addr; - unsigned long num_bytes; -}; +#define MAX_BANKS 32 + +static struct linux_prom64_registers pavail[MAX_BANKS] __initdata; +static struct linux_prom64_registers pavail_rescan[MAX_BANKS] __initdata; +static int pavail_ents __initdata; +static int pavail_rescan_ents __initdata; + +static int cmp_p64(const void *a, const void *b) +{ + const struct linux_prom64_registers *x = a, *y = b; + + if (x->phys_addr > y->phys_addr) + return 1; + if (x->phys_addr < y->phys_addr) + return -1; + return 0; +} + +static void __init read_obp_memory(const char *property, + struct linux_prom64_registers *regs, + int *num_ents) +{ + int node = prom_finddevice("/memory"); + int prop_size = prom_getproplen(node, property); + int ents, ret, i; + + ents = prop_size / sizeof(struct linux_prom64_registers); + if (ents > MAX_BANKS) { + prom_printf("The machine has more %s property entries than " + "this kernel can support (%d).\n", + property, MAX_BANKS); + prom_halt(); + } + + ret = prom_getproperty(node, property, (char *) regs, prop_size); + if (ret == -1) { + prom_printf("Couldn't get %s property from /memory.\n"); + prom_halt(); + } + + *num_ents = ents; -#define SPARC_PHYS_BANKS 32 + /* Sanitize what we got from the firmware, by page aligning + * everything. + */ + for (i = 0; i < ents; i++) { + unsigned long base, size; + + base = regs[i].phys_addr; + size = regs[i].reg_size; -static struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS]; + size &= PAGE_MASK; + if (base & ~PAGE_MASK) { + unsigned long new_base = PAGE_ALIGN(base); + + size -= new_base - base; + if ((long) size < 0L) + size = 0UL; + base = new_base; + } + regs[i].phys_addr = base; + regs[i].reg_size = size; + } + sort(regs, ents, sizeof(struct linux_prom64_registers), + cmp_p64, NULL); +} unsigned long *sparc64_valid_addr_bitmap __read_mostly; @@ -1213,14 +1272,14 @@ unsigned long __init bootmem_init(unsigned long *pages_avail) int i; #ifdef CONFIG_DEBUG_BOOTMEM - prom_printf("bootmem_init: Scan sp_banks, "); + prom_printf("bootmem_init: Scan pavail, "); #endif bytes_avail = 0UL; - for (i = 0; sp_banks[i].num_bytes != 0; i++) { - end_of_phys_memory = sp_banks[i].base_addr + - sp_banks[i].num_bytes; - bytes_avail += sp_banks[i].num_bytes; + for (i = 0; i < pavail_ents; i++) { + end_of_phys_memory = pavail[i].phys_addr + + pavail[i].reg_size; + bytes_avail += pavail[i].reg_size; if (cmdline_memory_size) { if (bytes_avail > cmdline_memory_size) { unsigned long slack = bytes_avail - cmdline_memory_size; @@ -1228,12 +1287,15 @@ unsigned long __init bootmem_init(unsigned long *pages_avail) bytes_avail -= slack; end_of_phys_memory -= slack; - sp_banks[i].num_bytes -= slack; - if (sp_banks[i].num_bytes == 0) { - sp_banks[i].base_addr = 0xdeadbeef; + pavail[i].reg_size -= slack; + if ((long)pavail[i].reg_size <= 0L) { + pavail[i].phys_addr = 0xdeadbeefUL; + pavail[i].reg_size = 0UL; + pavail_ents = i; } else { - sp_banks[i+1].num_bytes = 0; - sp_banks[i+1].base_addr = 0xdeadbeef; + pavail[i+1].reg_size = 0Ul; + pavail[i+1].phys_addr = 0xdeadbeefUL; + pavail_ents = i + 1; } break; } @@ -1287,12 +1349,12 @@ unsigned long __init bootmem_init(unsigned long *pages_avail) /* Now register the available physical memory with the * allocator. */ - for (i = 0; sp_banks[i].num_bytes != 0; i++) { + for (i = 0; i < pavail_ents; i++) { #ifdef CONFIG_DEBUG_BOOTMEM - prom_printf("free_bootmem(sp_banks:%d): base[%lx] size[%lx]\n", - i, sp_banks[i].base_addr, sp_banks[i].num_bytes); + prom_printf("free_bootmem(pavail:%d): base[%lx] size[%lx]\n", + i, pavail[i].phys_addr, pavail[i].reg_size); #endif - free_bootmem(sp_banks[i].base_addr, sp_banks[i].num_bytes); + free_bootmem(pavail[i].phys_addr, pavail[i].reg_size); } #ifdef CONFIG_BLK_DEV_INITRD @@ -1341,7 +1403,7 @@ static unsigned long kernel_map_range(unsigned long pstart, unsigned long pend, unsigned long alloc_bytes = 0UL; if ((vstart & ~PAGE_MASK) || (vend & ~PAGE_MASK)) { - prom_printf("kernel_map: Unaligned sp_banks[%lx:%lx]\n", + prom_printf("kernel_map: Unaligned physmem[%lx:%lx]\n", vstart, vend); prom_halt(); } @@ -1388,23 +1450,24 @@ static unsigned long kernel_map_range(unsigned long pstart, unsigned long pend, return alloc_bytes; } -extern struct linux_mlist_p1275 *prom_ptot_ptr; +static struct linux_prom64_registers pall[MAX_BANKS] __initdata; +static int pall_ents __initdata; + extern unsigned int kvmap_linear_patch[1]; static void __init kernel_physical_mapping_init(void) { - struct linux_mlist_p1275 *p = prom_ptot_ptr; - unsigned long mem_alloced = 0UL; + unsigned long i, mem_alloced = 0UL; - while (p) { + read_obp_memory("reg", &pall[0], &pall_ents); + + for (i = 0; i < pall_ents; i++) { unsigned long phys_start, phys_end; - phys_start = p->start_adr; - phys_end = phys_start + p->num_bytes; + phys_start = pall[i].phys_addr; + phys_end = phys_start + pall[i].reg_size; mem_alloced += kernel_map_range(phys_start, phys_end, PAGE_KERNEL); - - p = p->theres_more; } printk("Allocated %ld bytes for kernel page tables.\n", @@ -1434,60 +1497,14 @@ void kernel_map_pages(struct page *page, int numpages, int enable) unsigned long __init find_ecache_flush_span(unsigned long size) { - unsigned long i; - - for (i = 0; ; i++) { - if (sp_banks[i].num_bytes == 0) - break; - if (sp_banks[i].num_bytes >= size) - return sp_banks[i].base_addr; - } - - return ~0UL; -} - -static void __init prom_probe_memory(void) -{ - struct linux_mlist_p1275 *mlist; - unsigned long bytes, base_paddr, tally; int i; - i = 0; - mlist = *prom_meminfo()->p1275_available; - bytes = tally = mlist->num_bytes; - base_paddr = mlist->start_adr; - - sp_banks[0].base_addr = base_paddr; - sp_banks[0].num_bytes = bytes; - - while (mlist->theres_more != (void *) 0) { - i++; - mlist = mlist->theres_more; - bytes = mlist->num_bytes; - tally += bytes; - if (i >= SPARC_PHYS_BANKS-1) { - printk ("The machine has more banks than " - "this kernel can support\n" - "Increase the SPARC_PHYS_BANKS " - "setting (currently %d)\n", - SPARC_PHYS_BANKS); - i = SPARC_PHYS_BANKS-1; - break; - } - - sp_banks[i].base_addr = mlist->start_adr; - sp_banks[i].num_bytes = mlist->num_bytes; + for (i = 0; i < pavail_ents; i++) { + if (pavail[i].reg_size >= size) + return pavail[i].phys_addr; } - i++; - sp_banks[i].base_addr = 0xdeadbeefbeefdeadUL; - sp_banks[i].num_bytes = 0; - - /* Now mask all bank sizes on a page boundary, it is all we can - * use anyways. - */ - for (i = 0; sp_banks[i].num_bytes != 0; i++) - sp_banks[i].num_bytes &= PAGE_MASK; + return ~0UL; } /* paging_init() sets up the page tables */ @@ -1502,17 +1519,13 @@ void __init paging_init(void) unsigned long end_pfn, pages_avail, shift; unsigned long real_end, i; - prom_probe_memory(); + /* Find available physical memory... */ + read_obp_memory("available", &pavail[0], &pavail_ents); phys_base = 0xffffffffffffffffUL; - for (i = 0; sp_banks[i].num_bytes != 0; i++) { - unsigned long top; + for (i = 0; i < pavail_ents; i++) + phys_base = min(phys_base, pavail[i].phys_addr); - if (sp_banks[i].base_addr < phys_base) - phys_base = sp_banks[i].base_addr; - top = sp_banks[i].base_addr + - sp_banks[i].num_bytes; - } pfn_base = phys_base >> PAGE_SHIFT; kern_base = (prom_boot_mapping_phys_low >> 22UL) << 22UL; @@ -1588,128 +1601,35 @@ void __init paging_init(void) device_scan(); } -/* Ok, it seems that the prom can allocate some more memory chunks - * as a side effect of some prom calls we perform during the - * boot sequence. My most likely theory is that it is from the - * prom_set_traptable() call, and OBP is allocating a scratchpad - * for saving client program register state etc. - */ -static void __init sort_memlist(struct linux_mlist_p1275 *thislist) -{ - int swapi = 0; - int i, mitr; - unsigned long tmpaddr, tmpsize; - unsigned long lowest; - - for (i = 0; thislist[i].theres_more != 0; i++) { - lowest = thislist[i].start_adr; - for (mitr = i+1; thislist[mitr-1].theres_more != 0; mitr++) - if (thislist[mitr].start_adr < lowest) { - lowest = thislist[mitr].start_adr; - swapi = mitr; - } - if (lowest == thislist[i].start_adr) - continue; - tmpaddr = thislist[swapi].start_adr; - tmpsize = thislist[swapi].num_bytes; - for (mitr = swapi; mitr > i; mitr--) { - thislist[mitr].start_adr = thislist[mitr-1].start_adr; - thislist[mitr].num_bytes = thislist[mitr-1].num_bytes; - } - thislist[i].start_adr = tmpaddr; - thislist[i].num_bytes = tmpsize; - } -} - -void __init rescan_sp_banks(void) -{ - struct linux_prom64_registers memlist[64]; - struct linux_mlist_p1275 avail[64], *mlist; - unsigned long bytes, base_paddr; - int num_regs, node = prom_finddevice("/memory"); - int i; - - num_regs = prom_getproperty(node, "available", - (char *) memlist, sizeof(memlist)); - num_regs = (num_regs / sizeof(struct linux_prom64_registers)); - for (i = 0; i < num_regs; i++) { - avail[i].start_adr = memlist[i].phys_addr; - avail[i].num_bytes = memlist[i].reg_size; - avail[i].theres_more = &avail[i + 1]; - } - avail[i - 1].theres_more = NULL; - sort_memlist(avail); - - mlist = &avail[0]; - i = 0; - bytes = mlist->num_bytes; - base_paddr = mlist->start_adr; - - sp_banks[0].base_addr = base_paddr; - sp_banks[0].num_bytes = bytes; - - while (mlist->theres_more != NULL){ - i++; - mlist = mlist->theres_more; - bytes = mlist->num_bytes; - if (i >= SPARC_PHYS_BANKS-1) { - printk ("The machine has more banks than " - "this kernel can support\n" - "Increase the SPARC_PHYS_BANKS " - "setting (currently %d)\n", - SPARC_PHYS_BANKS); - i = SPARC_PHYS_BANKS-1; - break; - } - - sp_banks[i].base_addr = mlist->start_adr; - sp_banks[i].num_bytes = mlist->num_bytes; - } - - i++; - sp_banks[i].base_addr = 0xdeadbeefbeefdeadUL; - sp_banks[i].num_bytes = 0; - - for (i = 0; sp_banks[i].num_bytes != 0; i++) - sp_banks[i].num_bytes &= PAGE_MASK; -} - static void __init taint_real_pages(void) { - struct sparc_phys_banks saved_sp_banks[SPARC_PHYS_BANKS]; int i; - for (i = 0; i < SPARC_PHYS_BANKS; i++) { - saved_sp_banks[i].base_addr = - sp_banks[i].base_addr; - saved_sp_banks[i].num_bytes = - sp_banks[i].num_bytes; - } - - rescan_sp_banks(); + read_obp_memory("available", &pavail_rescan[0], &pavail_rescan_ents); - /* Find changes discovered in the sp_bank rescan and + /* Find changes discovered in the physmem available rescan and * reserve the lost portions in the bootmem maps. */ - for (i = 0; saved_sp_banks[i].num_bytes; i++) { + for (i = 0; i < pavail_ents; i++) { unsigned long old_start, old_end; - old_start = saved_sp_banks[i].base_addr; + old_start = pavail[i].phys_addr; old_end = old_start + - saved_sp_banks[i].num_bytes; + pavail[i].reg_size; while (old_start < old_end) { int n; - for (n = 0; sp_banks[n].num_bytes; n++) { + for (n = 0; pavail_rescan_ents; n++) { unsigned long new_start, new_end; - new_start = sp_banks[n].base_addr; - new_end = new_start + sp_banks[n].num_bytes; + new_start = pavail_rescan[n].phys_addr; + new_end = new_start + + pavail_rescan[n].reg_size; if (new_start <= old_start && new_end >= (old_start + PAGE_SIZE)) { - set_bit (old_start >> 22, - sparc64_valid_addr_bitmap); + set_bit(old_start >> 22, + sparc64_valid_addr_bitmap); goto do_next_page; } } diff --git a/include/asm-sparc64/openprom.h b/include/asm-sparc64/openprom.h index 0a336901d58..b4959d2b0d9 100644 --- a/include/asm-sparc64/openprom.h +++ b/include/asm-sparc64/openprom.h @@ -186,8 +186,8 @@ struct linux_prom_registers { }; struct linux_prom64_registers { - long phys_addr; - long reg_size; + unsigned long phys_addr; + unsigned long reg_size; }; struct linux_prom_irqs { -- cgit v1.2.3-70-g09d2 From 717463d806a53380a691eeb0136a4b750a9f9ae0 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 29 Sep 2005 18:50:34 -0700 Subject: [SPARC64]: Fix several bugs in flush_ptrace_access(). 1) Use cpudata cache line sizes, not magic constants. 2) Align start address in cheetah case so we do not get unaligned address traps. (pgrep was good at triggering this, via /proc/${pid}/cmdline accesses) Signed-off-by: David S. Miller --- arch/sparc64/kernel/ptrace.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ptrace.c b/arch/sparc64/kernel/ptrace.c index 5efbff90d66..774ecbb8a03 100644 --- a/arch/sparc64/kernel/ptrace.c +++ b/arch/sparc64/kernel/ptrace.c @@ -31,6 +31,7 @@ #include #include #include +#include /* Returning from ptrace is a bit tricky because the syscall return * low level code assumes any value returned which is negative and @@ -132,12 +133,16 @@ void flush_ptrace_access(struct vm_area_struct *vma, struct page *page, if ((uaddr ^ (unsigned long) kaddr) & (1UL << 13)) { unsigned long start = __pa(kaddr); unsigned long end = start + len; + unsigned long dcache_line_size; + + dcache_line_size = local_cpu_data().dcache_line_size; if (tlb_type == spitfire) { - for (; start < end; start += 32) + for (; start < end; start += dcache_line_size) spitfire_put_dcache_tag(start & 0x3fe0, 0x0); } else { - for (; start < end; start += 32) + start &= ~(dcache_line_size - 1); + for (; start < end; start += dcache_line_size) __asm__ __volatile__( "stxa %%g0, [%0] %1\n\t" "membar #Sync" @@ -150,8 +155,11 @@ void flush_ptrace_access(struct vm_area_struct *vma, struct page *page, if (write && tlb_type == spitfire) { unsigned long start = (unsigned long) kaddr; unsigned long end = start + len; + unsigned long icache_line_size; + + icache_line_size = local_cpu_data().icache_line_size; - for (; start < end; start += 32) + for (; start < end; start += icache_line_size) flushi(start); } } -- cgit v1.2.3-70-g09d2 From 0835ae0f27c0bfde67613d189ef6c537e004a6de Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Tue, 4 Oct 2005 15:23:20 -0700 Subject: [SPARC64]: Replace cheetah+ code patching with variables. Instead of code patching to handle the page size fields in the context registers, just use variables from which we get the proper values. Signed-off-by: David S. Miller --- arch/sparc64/kernel/entry.S | 43 +++++++-------------------------- arch/sparc64/kernel/etrap.S | 51 ++++------------------------------------ arch/sparc64/kernel/head.S | 33 ++++---------------------- arch/sparc64/kernel/rtrap.S | 23 ++---------------- arch/sparc64/kernel/setup.c | 8 ++----- arch/sparc64/kernel/trampoline.S | 15 ++++-------- arch/sparc64/kernel/winfixup.S | 33 ++------------------------ arch/sparc64/mm/init.c | 26 +++++++++++++++----- 8 files changed, 47 insertions(+), 185 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/entry.S b/arch/sparc64/kernel/entry.S index 2879b107292..f685035dbdb 100644 --- a/arch/sparc64/kernel/entry.S +++ b/arch/sparc64/kernel/entry.S @@ -97,8 +97,8 @@ do_fpdis: faddd %f0, %f2, %f4 fmuld %f0, %f2, %f6 ldxa [%g3] ASI_DMMU, %g5 -cplus_fptrap_insn_1: - sethi %hi(0), %g2 + sethi %hi(sparc64_kern_sec_context), %g2 + ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2 stxa %g2, [%g3] ASI_DMMU membar #Sync add %g6, TI_FPREGS + 0xc0, %g2 @@ -126,8 +126,8 @@ cplus_fptrap_insn_1: fzero %f34 ldxa [%g3] ASI_DMMU, %g5 add %g6, TI_FPREGS, %g1 -cplus_fptrap_insn_2: - sethi %hi(0), %g2 + sethi %hi(sparc64_kern_sec_context), %g2 + ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2 stxa %g2, [%g3] ASI_DMMU membar #Sync add %g6, TI_FPREGS + 0x40, %g2 @@ -153,8 +153,8 @@ cplus_fptrap_insn_2: 3: mov SECONDARY_CONTEXT, %g3 add %g6, TI_FPREGS, %g1 ldxa [%g3] ASI_DMMU, %g5 -cplus_fptrap_insn_3: - sethi %hi(0), %g2 + sethi %hi(sparc64_kern_sec_context), %g2 + ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2 stxa %g2, [%g3] ASI_DMMU membar #Sync mov 0x40, %g2 @@ -319,8 +319,8 @@ do_fptrap_after_fsr: stx %g3, [%g6 + TI_GSR] mov SECONDARY_CONTEXT, %g3 ldxa [%g3] ASI_DMMU, %g5 -cplus_fptrap_insn_4: - sethi %hi(0), %g2 + sethi %hi(sparc64_kern_sec_context), %g2 + ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2 stxa %g2, [%g3] ASI_DMMU membar #Sync add %g6, TI_FPREGS, %g2 @@ -341,33 +341,6 @@ cplus_fptrap_insn_4: ba,pt %xcc, etrap wr %g0, 0, %fprs -cplus_fptrap_1: - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - - .globl cheetah_plus_patch_fpdis -cheetah_plus_patch_fpdis: - /* We configure the dTLB512_0 for 4MB pages and the - * dTLB512_1 for 8K pages when in context zero. - */ - sethi %hi(cplus_fptrap_1), %o0 - lduw [%o0 + %lo(cplus_fptrap_1)], %o1 - - set cplus_fptrap_insn_1, %o2 - stw %o1, [%o2] - flush %o2 - set cplus_fptrap_insn_2, %o2 - stw %o1, [%o2] - flush %o2 - set cplus_fptrap_insn_3, %o2 - stw %o1, [%o2] - flush %o2 - set cplus_fptrap_insn_4, %o2 - stw %o1, [%o2] - flush %o2 - - retl - nop - /* The registers for cross calls will be: * * DATA 0: [low 32-bits] Address of function to call, jmp to this diff --git a/arch/sparc64/kernel/etrap.S b/arch/sparc64/kernel/etrap.S index 50d2af1d98a..0d8eba21111 100644 --- a/arch/sparc64/kernel/etrap.S +++ b/arch/sparc64/kernel/etrap.S @@ -68,12 +68,8 @@ etrap_irq: wrpr %g3, 0, %otherwin wrpr %g2, 0, %wstate -cplus_etrap_insn_1: - sethi %hi(0), %g3 - sllx %g3, 32, %g3 -cplus_etrap_insn_2: - sethi %hi(0), %g2 - or %g3, %g2, %g3 + sethi %hi(sparc64_kern_pri_context), %g2 + ldx [%g2 + %lo(sparc64_kern_pri_context)], %g3 stxa %g3, [%l4] ASI_DMMU flush %l6 wr %g0, ASI_AIUS, %asi @@ -215,12 +211,8 @@ scetrap: rdpr %pil, %g2 mov PRIMARY_CONTEXT, %l4 wrpr %g3, 0, %otherwin wrpr %g2, 0, %wstate -cplus_etrap_insn_3: - sethi %hi(0), %g3 - sllx %g3, 32, %g3 -cplus_etrap_insn_4: - sethi %hi(0), %g2 - or %g3, %g2, %g3 + sethi %hi(sparc64_kern_pri_context), %g2 + ldx [%g2 + %lo(sparc64_kern_pri_context)], %g3 stxa %g3, [%l4] ASI_DMMU flush %l6 @@ -264,38 +256,3 @@ cplus_etrap_insn_4: #undef TASK_REGOFF #undef ETRAP_PSTATE1 - -cplus_einsn_1: - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %g3 -cplus_einsn_2: - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - - .globl cheetah_plus_patch_etrap -cheetah_plus_patch_etrap: - /* We configure the dTLB512_0 for 4MB pages and the - * dTLB512_1 for 8K pages when in context zero. - */ - sethi %hi(cplus_einsn_1), %o0 - sethi %hi(cplus_etrap_insn_1), %o2 - lduw [%o0 + %lo(cplus_einsn_1)], %o1 - or %o2, %lo(cplus_etrap_insn_1), %o2 - stw %o1, [%o2] - flush %o2 - sethi %hi(cplus_etrap_insn_3), %o2 - or %o2, %lo(cplus_etrap_insn_3), %o2 - stw %o1, [%o2] - flush %o2 - - sethi %hi(cplus_einsn_2), %o0 - sethi %hi(cplus_etrap_insn_2), %o2 - lduw [%o0 + %lo(cplus_einsn_2)], %o1 - or %o2, %lo(cplus_etrap_insn_2), %o2 - stw %o1, [%o2] - flush %o2 - sethi %hi(cplus_etrap_insn_4), %o2 - or %o2, %lo(cplus_etrap_insn_4), %o2 - stw %o1, [%o2] - flush %o2 - - retl - nop diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index 89406f9649a..24340496cdd 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -325,23 +325,7 @@ cheetah_tlb_fixup: 1: sethi %hi(tlb_type), %g1 stw %g2, [%g1 + %lo(tlb_type)] - BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g7,1f) - ba,pt %xcc, 2f - nop - -1: /* Patch context register writes to support nucleus page - * size correctly. - */ - call cheetah_plus_patch_etrap - nop - call cheetah_plus_patch_rtrap - nop - call cheetah_plus_patch_fpdis - nop - call cheetah_plus_patch_winfixup - nop - -2: /* Patch copy/page operations to cheetah optimized versions. */ + /* Patch copy/page operations to cheetah optimized versions. */ call cheetah_patch_copyops nop call cheetah_patch_copy_page @@ -484,20 +468,13 @@ spitfire_vpte_base: call prom_set_trap_table sethi %hi(sparc64_ttable_tl0), %o0 - BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g2,g3,1f) - ba,pt %xcc, 2f - nop - -1: /* Start using proper page size encodings in ctx register. */ - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %g3 + /* Start using proper page size encodings in ctx register. */ + sethi %hi(sparc64_kern_pri_context), %g3 + ldx [%g3 + %lo(sparc64_kern_pri_context)], %g2 mov PRIMARY_CONTEXT, %g1 - sllx %g3, 32, %g3 - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - or %g3, %g2, %g3 - stxa %g3, [%g1] ASI_DMMU + stxa %g2, [%g1] ASI_DMMU membar #Sync -2: rdpr %pstate, %o1 or %o1, PSTATE_IE, %o1 wrpr %o1, 0, %pstate diff --git a/arch/sparc64/kernel/rtrap.S b/arch/sparc64/kernel/rtrap.S index fafd227735f..ecfb42a69a4 100644 --- a/arch/sparc64/kernel/rtrap.S +++ b/arch/sparc64/kernel/rtrap.S @@ -256,9 +256,8 @@ rt_continue: ldx [%sp + PTREGS_OFF + PT_V9_G1], %g1 brnz,pn %l3, kern_rtt mov PRIMARY_CONTEXT, %l7 ldxa [%l7 + %l7] ASI_DMMU, %l0 -cplus_rtrap_insn_1: - sethi %hi(0), %l1 - sllx %l1, 32, %l1 + sethi %hi(sparc64_kern_pri_nuc_bits), %l1 + ldx [%l1 + %lo(sparc64_kern_pri_nuc_bits)], %l1 or %l0, %l1, %l0 stxa %l0, [%l7] ASI_DMMU flush %g6 @@ -345,21 +344,3 @@ kern_fpucheck: ldub [%g6 + TI_FPDEPTH], %l5 wr %g0, FPRS_DU, %fprs ba,pt %xcc, rt_continue stb %l5, [%g6 + TI_FPDEPTH] - -cplus_rinsn_1: - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %l1 - - .globl cheetah_plus_patch_rtrap -cheetah_plus_patch_rtrap: - /* We configure the dTLB512_0 for 4MB pages and the - * dTLB512_1 for 8K pages when in context zero. - */ - sethi %hi(cplus_rinsn_1), %o0 - sethi %hi(cplus_rtrap_insn_1), %o2 - lduw [%o0 + %lo(cplus_rinsn_1)], %o1 - or %o2, %lo(cplus_rtrap_insn_1), %o2 - stw %o1, [%o2] - flush %o2 - - retl - nop diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index 4c9c8f24174..c1f34237cdf 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -187,17 +187,13 @@ int prom_callback(long *args) } if ((va >= KERNBASE) && (va < (KERNBASE + (4 * 1024 * 1024)))) { - unsigned long kernel_pctx = 0; - - if (tlb_type == cheetah_plus) - kernel_pctx |= (CTX_CHEETAH_PLUS_NUC | - CTX_CHEETAH_PLUS_CTX0); + extern unsigned long sparc64_kern_pri_context; /* Spitfire Errata #32 workaround */ __asm__ __volatile__("stxa %0, [%1] %2\n\t" "flush %%g6" : /* No outputs */ - : "r" (kernel_pctx), + : "r" (sparc64_kern_pri_context), "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU)); diff --git a/arch/sparc64/kernel/trampoline.S b/arch/sparc64/kernel/trampoline.S index 89f2fcfcd66..9478551cb02 100644 --- a/arch/sparc64/kernel/trampoline.S +++ b/arch/sparc64/kernel/trampoline.S @@ -336,20 +336,13 @@ do_unlock: call init_irqwork_curcpu nop - BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g2,g3,1f) - ba,pt %xcc, 2f - nop - -1: /* Start using proper page size encodings in ctx register. */ - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %g3 + /* Start using proper page size encodings in ctx register. */ + sethi %hi(sparc64_kern_pri_context), %g3 + ldx [%g3 + %lo(sparc64_kern_pri_context)], %g2 mov PRIMARY_CONTEXT, %g1 - sllx %g3, 32, %g3 - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - or %g3, %g2, %g3 - stxa %g3, [%g1] ASI_DMMU + stxa %g2, [%g1] ASI_DMMU membar #Sync -2: rdpr %pstate, %o1 or %o1, PSTATE_IE, %o1 wrpr %o1, 0, %pstate diff --git a/arch/sparc64/kernel/winfixup.S b/arch/sparc64/kernel/winfixup.S index 99c809a1e5a..39160926267 100644 --- a/arch/sparc64/kernel/winfixup.S +++ b/arch/sparc64/kernel/winfixup.S @@ -16,23 +16,14 @@ .text set_pcontext: -cplus_winfixup_insn_1: - sethi %hi(0), %l1 + sethi %hi(sparc64_kern_pri_context), %l1 + ldx [%l1 + %lo(sparc64_kern_pri_context)], %l1 mov PRIMARY_CONTEXT, %g1 - sllx %l1, 32, %l1 -cplus_winfixup_insn_2: - sethi %hi(0), %g2 - or %l1, %g2, %l1 stxa %l1, [%g1] ASI_DMMU flush %g6 retl nop -cplus_wfinsn_1: - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %l1 -cplus_wfinsn_2: - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - .align 32 /* Here are the rules, pay attention. @@ -395,23 +386,3 @@ window_dax_from_user_common: add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap clr %l6 - - - .globl cheetah_plus_patch_winfixup -cheetah_plus_patch_winfixup: - sethi %hi(cplus_wfinsn_1), %o0 - sethi %hi(cplus_winfixup_insn_1), %o2 - lduw [%o0 + %lo(cplus_wfinsn_1)], %o1 - or %o2, %lo(cplus_winfixup_insn_1), %o2 - stw %o1, [%o2] - flush %o2 - - sethi %hi(cplus_wfinsn_2), %o0 - sethi %hi(cplus_winfixup_insn_2), %o2 - lduw [%o0 + %lo(cplus_wfinsn_2)], %o1 - or %o2, %lo(cplus_winfixup_insn_2), %o2 - stw %o1, [%o2] - flush %o2 - - retl - nop diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index 5db50524f20..4e2f71e0abc 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -133,6 +133,12 @@ extern unsigned int sparc_ramdisk_size; struct page *mem_map_zero __read_mostly; +unsigned int sparc64_highest_unlocked_tlb_ent __read_mostly; + +unsigned long sparc64_kern_pri_context __read_mostly; +unsigned long sparc64_kern_pri_nuc_bits __read_mostly; +unsigned long sparc64_kern_sec_context __read_mostly; + int bigkernel = 0; /* XXX Tune this... */ @@ -582,13 +588,21 @@ static void __init remap_kernel(void) prom_dtlb_load(tlb_ent, tte_data, tte_vaddr); prom_itlb_load(tlb_ent, tte_data, tte_vaddr); if (bigkernel) { - prom_dtlb_load(tlb_ent - 1, + tlb_ent -= 1; + prom_dtlb_load(tlb_ent, tte_data + 0x400000, tte_vaddr + 0x400000); - prom_itlb_load(tlb_ent - 1, + prom_itlb_load(tlb_ent, tte_data + 0x400000, tte_vaddr + 0x400000); } + sparc64_highest_unlocked_tlb_ent = tlb_ent - 1; + if (tlb_type == cheetah_plus) { + sparc64_kern_pri_context = (CTX_CHEETAH_PLUS_CTX0 | + CTX_CHEETAH_PLUS_NUC); + sparc64_kern_pri_nuc_bits = CTX_CHEETAH_PLUS_NUC; + sparc64_kern_sec_context = CTX_CHEETAH_PLUS_CTX0; + } } static void __init inherit_prom_mappings(void) @@ -788,8 +802,8 @@ void inherit_locked_prom_mappings(int save_p) } } if (tlb_type == spitfire) { - int high = SPITFIRE_HIGHEST_LOCKED_TLBENT - bigkernel; - for (i = 0; i < high; i++) { + int high = sparc64_highest_unlocked_tlb_ent; + for (i = 0; i <= high; i++) { unsigned long data; /* Spitfire Errata #32 workaround */ @@ -877,9 +891,9 @@ void inherit_locked_prom_mappings(int save_p) } } } else if (tlb_type == cheetah || tlb_type == cheetah_plus) { - int high = CHEETAH_HIGHEST_LOCKED_TLBENT - bigkernel; + int high = sparc64_highest_unlocked_tlb_ent; - for (i = 0; i < high; i++) { + for (i = 0; i <= high; i++) { unsigned long data; data = cheetah_get_ldtlb_data(i); -- cgit v1.2.3-70-g09d2 From 2256c13b992b09f1f9563c26457aa048da2865df Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 6 Oct 2005 20:43:54 -0700 Subject: [SPARC64]: Probe for power device on ISA bus too. Signed-off-by: David S. Miller --- arch/sparc64/kernel/power.c | 64 ++++++++++++++++++++++++++++++++++++--------- 1 file changed, 51 insertions(+), 13 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/power.c b/arch/sparc64/kernel/power.c index 946cee0257e..9e8362ea310 100644 --- a/arch/sparc64/kernel/power.c +++ b/arch/sparc64/kernel/power.c @@ -17,6 +17,7 @@ #include #include +#include #include #include @@ -100,46 +101,83 @@ again: return 0; } -static int __init has_button_interrupt(struct linux_ebus_device *edev) +static int __init has_button_interrupt(unsigned int irq, int prom_node) { - if (edev->irqs[0] == PCI_IRQ_NONE) + if (irq == PCI_IRQ_NONE) return 0; - if (!prom_node_has_property(edev->prom_node, "button")) + if (!prom_node_has_property(prom_node, "button")) return 0; return 1; } -void __init power_init(void) +static int __init power_probe_ebus(struct resource **resp, unsigned int *irq_p, int *prom_node_p) { struct linux_ebus *ebus; struct linux_ebus_device *edev; + + for_each_ebus(ebus) { + for_each_ebusdev(edev, ebus) { + if (!strcmp(edev->prom_name, "power")) { + *resp = &edev->resource[0]; + *irq_p = edev->irqs[0]; + *prom_node_p = edev->prom_node; + return 0; + } + } + } + return -ENODEV; +} + +static int __init power_probe_isa(struct resource **resp, unsigned int *irq_p, int *prom_node_p) +{ + struct sparc_isa_bridge *isa_bus; + struct sparc_isa_device *isa_dev; + + for_each_isa(isa_bus) { + for_each_isadev(isa_dev, isa_bus) { + if (!strcmp(isa_dev->prom_name, "power")) { + *resp = &isa_dev->resource; + *irq_p = isa_dev->irq; + *prom_node_p = isa_dev->prom_node; + return 0; + } + } + } + return -ENODEV; +} + +void __init power_init(void) +{ + struct resource *res = NULL; + unsigned int irq; + int prom_node; static int invoked; if (invoked) return; invoked = 1; - for_each_ebus(ebus) { - for_each_ebusdev(edev, ebus) { - if (!strcmp(edev->prom_name, "power")) - goto found; - } - } + if (!power_probe_ebus(&res, &irq, &prom_node)) + goto found; + + if (!power_probe_isa(&res, &irq, &prom_node)) + goto found; + return; found: - power_reg = ioremap(edev->resource[0].start, 0x4); + power_reg = ioremap(res->start, 0x4); printk("power: Control reg at %p ... ", power_reg); poweroff_method = machine_halt; /* able to use the standard halt */ - if (has_button_interrupt(edev)) { + if (has_button_interrupt(irq, prom_node)) { if (kernel_thread(powerd, NULL, CLONE_FS) < 0) { printk("Failed to start power daemon.\n"); return; } printk("powerd running.\n"); - if (request_irq(edev->irqs[0], + if (request_irq(irq, power_handler, SA_SHIRQ, "power", NULL) < 0) printk("power: Error, cannot register IRQ handler.\n"); } else { -- cgit v1.2.3-70-g09d2 From ba6399334dd8a75bd295de26496196c720abae0a Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Fri, 7 Oct 2005 13:30:49 -0700 Subject: [SPARC64]: Fix userland FPU state corruption. We need to use stricter memory barriers around the block load and store instructions we use to save and restore the FPU register file. Signed-off-by: David S. Miller --- arch/sparc64/kernel/entry.S | 39 +++++++++++++++++++++------------------ arch/sparc64/kernel/rtrap.S | 7 ++++--- arch/sparc64/lib/VISsave.S | 8 +++++--- 3 files changed, 30 insertions(+), 24 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/entry.S b/arch/sparc64/kernel/entry.S index f685035dbdb..11a848402fb 100644 --- a/arch/sparc64/kernel/entry.S +++ b/arch/sparc64/kernel/entry.S @@ -33,7 +33,7 @@ /* This is trivial with the new code... */ .globl do_fpdis do_fpdis: - sethi %hi(TSTATE_PEF), %g4 ! IEU0 + sethi %hi(TSTATE_PEF), %g4 rdpr %tstate, %g5 andcc %g5, %g4, %g0 be,pt %xcc, 1f @@ -50,18 +50,18 @@ do_fpdis: add %g0, %g0, %g0 ba,a,pt %xcc, rtrap_clr_l6 -1: ldub [%g6 + TI_FPSAVED], %g5 ! Load Group - wr %g0, FPRS_FEF, %fprs ! LSU Group+4bubbles - andcc %g5, FPRS_FEF, %g0 ! IEU1 Group - be,a,pt %icc, 1f ! CTI - clr %g7 ! IEU0 - ldx [%g6 + TI_GSR], %g7 ! Load Group -1: andcc %g5, FPRS_DL, %g0 ! IEU1 - bne,pn %icc, 2f ! CTI - fzero %f0 ! FPA - andcc %g5, FPRS_DU, %g0 ! IEU1 Group - bne,pn %icc, 1f ! CTI - fzero %f2 ! FPA +1: ldub [%g6 + TI_FPSAVED], %g5 + wr %g0, FPRS_FEF, %fprs + andcc %g5, FPRS_FEF, %g0 + be,a,pt %icc, 1f + clr %g7 + ldx [%g6 + TI_GSR], %g7 +1: andcc %g5, FPRS_DL, %g0 + bne,pn %icc, 2f + fzero %f0 + andcc %g5, FPRS_DU, %g0 + bne,pn %icc, 1f + fzero %f2 faddd %f0, %f2, %f4 fmuld %f0, %f2, %f6 faddd %f0, %f2, %f8 @@ -104,8 +104,10 @@ do_fpdis: add %g6, TI_FPREGS + 0xc0, %g2 faddd %f0, %f2, %f8 fmuld %f0, %f2, %f10 - ldda [%g1] ASI_BLK_S, %f32 ! grrr, where is ASI_BLK_NUCLEUS 8-( + membar #Sync + ldda [%g1] ASI_BLK_S, %f32 ldda [%g2] ASI_BLK_S, %f48 + membar #Sync faddd %f0, %f2, %f12 fmuld %f0, %f2, %f14 faddd %f0, %f2, %f16 @@ -116,7 +118,6 @@ do_fpdis: fmuld %f0, %f2, %f26 faddd %f0, %f2, %f28 fmuld %f0, %f2, %f30 - membar #Sync b,pt %xcc, fpdis_exit nop 2: andcc %g5, FPRS_DU, %g0 @@ -133,8 +134,10 @@ do_fpdis: add %g6, TI_FPREGS + 0x40, %g2 faddd %f32, %f34, %f36 fmuld %f32, %f34, %f38 - ldda [%g1] ASI_BLK_S, %f0 ! grrr, where is ASI_BLK_NUCLEUS 8-( + membar #Sync + ldda [%g1] ASI_BLK_S, %f0 ldda [%g2] ASI_BLK_S, %f16 + membar #Sync faddd %f32, %f34, %f40 fmuld %f32, %f34, %f42 faddd %f32, %f34, %f44 @@ -147,7 +150,6 @@ do_fpdis: fmuld %f32, %f34, %f58 faddd %f32, %f34, %f60 fmuld %f32, %f34, %f62 - membar #Sync ba,pt %xcc, fpdis_exit nop 3: mov SECONDARY_CONTEXT, %g3 @@ -158,7 +160,8 @@ do_fpdis: stxa %g2, [%g3] ASI_DMMU membar #Sync mov 0x40, %g2 - ldda [%g1] ASI_BLK_S, %f0 ! grrr, where is ASI_BLK_NUCLEUS 8-( + membar #Sync + ldda [%g1] ASI_BLK_S, %f0 ldda [%g1 + %g2] ASI_BLK_S, %f16 add %g1, 0x80, %g1 ldda [%g1] ASI_BLK_S, %f32 diff --git a/arch/sparc64/kernel/rtrap.S b/arch/sparc64/kernel/rtrap.S index ecfb42a69a4..090dcca00d2 100644 --- a/arch/sparc64/kernel/rtrap.S +++ b/arch/sparc64/kernel/rtrap.S @@ -312,32 +312,33 @@ kern_fpucheck: ldub [%g6 + TI_FPDEPTH], %l5 wr %g1, FPRS_FEF, %fprs ldx [%o1 + %o5], %g1 add %g6, TI_XFSR, %o1 - membar #StoreLoad | #LoadLoad sll %o0, 8, %o2 add %g6, TI_FPREGS, %o3 brz,pn %l6, 1f add %g6, TI_FPREGS+0x40, %o4 + membar #Sync ldda [%o3 + %o2] ASI_BLK_P, %f0 ldda [%o4 + %o2] ASI_BLK_P, %f16 + membar #Sync 1: andcc %l2, FPRS_DU, %g0 be,pn %icc, 1f wr %g1, 0, %gsr add %o2, 0x80, %o2 + membar #Sync ldda [%o3 + %o2] ASI_BLK_P, %f32 ldda [%o4 + %o2] ASI_BLK_P, %f48 - 1: membar #Sync ldx [%o1 + %o5], %fsr 2: stb %l5, [%g6 + TI_FPDEPTH] ba,pt %xcc, rt_continue nop 5: wr %g0, FPRS_FEF, %fprs - membar #StoreLoad | #LoadLoad sll %o0, 8, %o2 add %g6, TI_FPREGS+0x80, %o3 add %g6, TI_FPREGS+0xc0, %o4 + membar #Sync ldda [%o3 + %o2] ASI_BLK_P, %f32 ldda [%o4 + %o2] ASI_BLK_P, %f48 membar #Sync diff --git a/arch/sparc64/lib/VISsave.S b/arch/sparc64/lib/VISsave.S index 4e18989bd60..a0ded5c5aa5 100644 --- a/arch/sparc64/lib/VISsave.S +++ b/arch/sparc64/lib/VISsave.S @@ -59,15 +59,17 @@ vis1: ldub [%g6 + TI_FPSAVED], %g3 be,pn %icc, 9b add %g6, TI_FPREGS, %g2 andcc %o5, FPRS_DL, %g0 - membar #StoreStore | #LoadStore be,pn %icc, 4f add %g6, TI_FPREGS+0x40, %g3 + membar #Sync stda %f0, [%g2 + %g1] ASI_BLK_P stda %f16, [%g3 + %g1] ASI_BLK_P + membar #Sync andcc %o5, FPRS_DU, %g0 be,pn %icc, 5f 4: add %g1, 128, %g1 + membar #Sync stda %f32, [%g2 + %g1] ASI_BLK_P stda %f48, [%g3 + %g1] ASI_BLK_P @@ -87,7 +89,7 @@ vis1: ldub [%g6 + TI_FPSAVED], %g3 sll %g1, 5, %g1 add %g6, TI_FPREGS+0xc0, %g3 wr %g0, FPRS_FEF, %fprs - membar #StoreStore | #LoadStore + membar #Sync stda %f32, [%g2 + %g1] ASI_BLK_P stda %f48, [%g3 + %g1] ASI_BLK_P membar #Sync @@ -128,8 +130,8 @@ VISenterhalf: be,pn %icc, 4f add %g6, TI_FPREGS, %g2 - membar #StoreStore | #LoadStore add %g6, TI_FPREGS+0x40, %g3 + membar #Sync stda %f0, [%g2 + %g1] ASI_BLK_P stda %f16, [%g3 + %g1] ASI_BLK_P membar #Sync -- cgit v1.2.3-70-g09d2 From 2e457ef667158840c1be511f5d10dd42c6dbbe46 Mon Sep 17 00:00:00 2001 From: Sven Hartge Date: Sat, 8 Oct 2005 21:12:04 -0700 Subject: [SPARC64]: Fix compile error in irq.c irq.c is missing the inclusion of asm/io.h, which causes readb() and writeb() the be undefined. Signed-off-by: Sven Hartge Signed-off-by: David S. Miller --- arch/sparc64/kernel/irq.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/irq.c b/arch/sparc64/kernel/irq.c index c9b69167632..233526ba3ab 100644 --- a/arch/sparc64/kernel/irq.c +++ b/arch/sparc64/kernel/irq.c @@ -27,6 +27,7 @@ #include #include #include +#include #include #include #include -- cgit v1.2.3-70-g09d2 From 5d8e1b181c4ad63e6ca90d51287b31afd400d2eb Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Mon, 10 Oct 2005 16:12:13 -0700 Subject: [SPARC64]: Fix Ultra5, Ultra60, et al. boot failures. On the boot processor, we need to do the move onto the Linux trap table a little bit differently else we'll take unhandlable faults in the firmware address space. Previously we would do the following: 1) Disable PSTATE_IE in %pstate. 2) Set %tba by hand to sparc64_ttable_tl0 3) Initialize alternate, mmu, and interrupt global trap registers. 4) Call prom_set_traptable() That doesn't work very well actually with the way we boot the kernel VM these days. It worked by luck on many systems because the firmware accesses for the prom_set_traptable() call happened to be loaded into the TLB already, something we cannot assume. So the new scheme is this: 1) Clear PSTATE_IE in %pstate and set %pil to 15 2) Call prom_set_traptable() 3) Initialize alternate, mmu, and interrupt global trap registers. and this works quite well. This sequence has been moved into a callable function in assembler named setup-trap_table(). The idea is that eventually trampoline.S can use this code as well. That isn't possible currently due to some complications, but eventually we should be able to do it. Thanks to Meelis Roos for the Ultra5 boot failure report. Signed-off-by: David S. Miller --- arch/sparc64/kernel/head.S | 127 ++++++++++++++++++++++++++++++++------------- 1 file changed, 91 insertions(+), 36 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index 24340496cdd..f1dcdf8f743 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -382,32 +382,79 @@ tlb_fixup_done: nop /* Not reached... */ -/* IMPORTANT NOTE: Whenever making changes here, check - * trampoline.S as well. -jj */ - .globl setup_tba -setup_tba: /* i0 = is_starfire */ - save %sp, -160, %sp + /* This is meant to allow the sharing of this code between + * boot processor invocation (via setup_tba() below) and + * secondary processor startup (via trampoline.S). The + * former does use this code, the latter does not yet due + * to some complexities. That should be fixed up at some + * point. + */ + .globl setup_trap_table +setup_trap_table: + save %sp, -192, %sp + + /* Force interrupts to be disabled. Transferring over to + * the Linux trap table is a very delicate operation. + * Until we are actually on the Linux trap table, we cannot + * get the PAGE_OFFSET linear mappings translated. We need + * that mapping to be setup in order to initialize the firmware + * page tables. + * + * So there is this window of time, from the return from + * prom_set_trap_table() until inherit_prom_mappings_post() + * (in arch/sparc64/mm/init.c) completes, during which no + * firmware address space accesses can be made. + */ + rdpr %pstate, %o1 + andn %o1, PSTATE_IE, %o1 + wrpr %o1, 0x0, %pstate + wrpr %g0, 15, %pil - rdpr %tba, %g7 - sethi %hi(prom_tba), %o1 - or %o1, %lo(prom_tba), %o1 - stx %g7, [%o1] + /* Ok, now make the final valid firmware call to jump over + * to the Linux trap table. + */ + call prom_set_trap_table + sethi %hi(sparc64_ttable_tl0), %o0 + + /* Start using proper page size encodings in ctx register. */ + sethi %hi(sparc64_kern_pri_context), %g3 + ldx [%g3 + %lo(sparc64_kern_pri_context)], %g2 + mov PRIMARY_CONTEXT, %g1 + stxa %g2, [%g1] ASI_DMMU + membar #Sync + + /* The Linux trap handlers expect various trap global registers + * to be setup with some fixed values. So here we set these + * up very carefully. These globals are: + * + * Alternate Globals (PSTATE_AG): + * + * %g6 --> current_thread_info() + * + * MMU Globals (PSTATE_MG): + * + * %g1 --> TLB_SFSR + * %g2 --> ((_PAGE_VALID | _PAGE_SZ4MB | + * _PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_W) + * ^ 0xfffff80000000000) + * (this %g2 value is used for computing the PAGE_OFFSET kernel + * TLB entries quickly, the virtual address of the fault XOR'd + * with this %g2 value is the PTE to load into the TLB) + * %g3 --> VPTE_BASE_CHEETAH or VPTE_BASE_SPITFIRE + * + * Interrupt Globals (PSTATE_IG, setup by init_irqwork_curcpu()): + * + * %g6 --> __irq_work[smp_processor_id()] + */ - /* Setup "Linux" globals 8-) */ rdpr %pstate, %o1 mov %g6, %o2 - wrpr %o1, (PSTATE_AG|PSTATE_IE), %pstate - sethi %hi(sparc64_ttable_tl0), %g1 - wrpr %g1, %tba + wrpr %o1, PSTATE_AG, %pstate mov %o2, %g6 - /* Set up MMU globals */ - wrpr %o1, (PSTATE_MG|PSTATE_IE), %pstate - - /* Set fixed globals used by dTLB miss handler. */ #define KERN_HIGHBITS ((_PAGE_VALID|_PAGE_SZ4MB)^0xfffff80000000000) #define KERN_LOWBITS (_PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_W) - + wrpr %o1, PSTATE_MG, %pstate mov TSB_REG, %g1 stxa %g0, [%g1] ASI_DMMU membar #Sync @@ -419,17 +466,17 @@ setup_tba: /* i0 = is_starfire */ sllx %g2, 32, %g2 or %g2, KERN_LOWBITS, %g2 - BRANCH_IF_ANY_CHEETAH(g3,g7,cheetah_vpte_base) - ba,pt %xcc, spitfire_vpte_base + BRANCH_IF_ANY_CHEETAH(g3,g7,8f) + ba,pt %xcc, 9f nop -cheetah_vpte_base: +8: sethi %uhi(VPTE_BASE_CHEETAH), %g3 or %g3, %ulo(VPTE_BASE_CHEETAH), %g3 ba,pt %xcc, 2f sllx %g3, 32, %g3 -spitfire_vpte_base: +9: sethi %uhi(VPTE_BASE_SPITFIRE), %g3 or %g3, %ulo(VPTE_BASE_SPITFIRE), %g3 sllx %g3, 32, %g3 @@ -455,29 +502,37 @@ spitfire_vpte_base: sllx %o2, 32, %o2 wr %o2, %asr25 - /* Ok, we're done setting up all the state our trap mechanims needs, - * now get back into normal globals and let the PROM know what is up. - */ 2: wrpr %g0, %g0, %wstate - wrpr %o1, PSTATE_IE, %pstate + wrpr %o1, 0x0, %pstate call init_irqwork_curcpu nop - call prom_set_trap_table - sethi %hi(sparc64_ttable_tl0), %o0 - - /* Start using proper page size encodings in ctx register. */ - sethi %hi(sparc64_kern_pri_context), %g3 - ldx [%g3 + %lo(sparc64_kern_pri_context)], %g2 - mov PRIMARY_CONTEXT, %g1 - stxa %g2, [%g1] ASI_DMMU - membar #Sync - + /* Now we can turn interrupts back on. */ rdpr %pstate, %o1 or %o1, PSTATE_IE, %o1 wrpr %o1, 0, %pstate + wrpr %g0, 0x0, %pil + + ret + restore + + .globl setup_tba +setup_tba: /* i0 = is_starfire */ + save %sp, -192, %sp + + /* The boot processor is the only cpu which invokes this + * routine, the other cpus set things up via trampoline.S. + * So save the OBP trap table address here. + */ + rdpr %tba, %g7 + sethi %hi(prom_tba), %o1 + or %o1, %lo(prom_tba), %o1 + stx %g7, [%o1] + + call setup_trap_table + nop ret restore -- cgit v1.2.3-70-g09d2 From b1b510aa284af1908d5d369d52f7dae16aaabd71 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Tue, 11 Oct 2005 15:45:16 -0700 Subject: [SPARC64]: Fix net booting on Ultra5 We were not doing alignment properly when remapping the kernel image. What we want is a 4MB aligned physical address to map at KERNBASE. Mistakedly we were 4MB aligning the virtual address where the kernel initially sits, that's wrong. Instead, we should PAGE align the virtual address, then 4MB align the physical address result the prom gives to us. Signed-off-by: David S. Miller --- arch/sparc64/kernel/head.S | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index f1dcdf8f743..4c942f71184 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -191,8 +191,9 @@ prom_boot_mapping_phys_low: stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 5 stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1: "translate" stx %l5, [%sp + 2047 + 128 + 0x20] ! arg2: prom_mmu_ihandle_cache - srlx %l0, 22, %l3 - sllx %l3, 22, %l3 + /* PAGE align */ + srlx %l0, 13, %l3 + sllx %l3, 13, %l3 stx %l3, [%sp + 2047 + 128 + 0x28] ! arg3: vaddr, our PC stx %g0, [%sp + 2047 + 128 + 0x30] ! res1 stx %g0, [%sp + 2047 + 128 + 0x38] ! res2 @@ -211,6 +212,9 @@ prom_boot_mapping_phys_low: ldx [%sp + 2047 + 128 + 0x48], %l2 ! physaddr high stx %l2, [%l4 + 0x0] ldx [%sp + 2047 + 128 + 0x50], %l3 ! physaddr low + /* 4MB align */ + srlx %l3, 22, %l3 + sllx %l3, 22, %l3 stx %l3, [%l4 + 0x8] /* Leave service as-is, "call-method" */ -- cgit v1.2.3-70-g09d2 From c9c10830740df1b5e7848d6fbb68c93a73e8f7cd Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Wed, 12 Oct 2005 12:22:46 -0700 Subject: [SPARC64]: Fix boot failures on SunBlade-150 The sequence to move over to the Linux trap tables from the firmware ones needs to be more air tight. It turns out that to be %100 safe we do need to be able to translate OBP mappings in our TLB miss handlers early. In order not to eat up a lot of kernel image memory with static page tables, just use the translations array in the OBP TLB miss handlers. That solves the bulk of the problem. Furthermore, to make sure the OBP TLB miss path will work even before the fixed MMU globals are loaded, explicitly load %g1 to TLB_SFSR at the beginning of the i-TLB and d-TLB miss handlers. To ease the OBP TLB miss walking of the prom_trans[] array, we sort it then delete all of the non-OBP entries in there (for example, there are entries for the kernel image itself which we're not interested in at all). We also save about 32K of kernel image size with this change. Not a bad side effect :-) There are still some reasons why trampoline.S can't use the setup_trap_table() yet. The most noteworthy are: 1) OBP boots secondary processors with non-bias'd stack for some reason. This is easily fixed by using a small bootup stack in the kernel image explicitly for this purpose. 2) Doing a firmware call via the normal C call prom_set_trap_table() goes through the whole OBP enter/exit sequence that saves and restores OBP and Linux kernel state in the MMUs. This path unfortunately does a "flush %g6" while loading up the OBP locked TLB entries for the firmware call. If we setup the %g6 in the trampoline.S code properly, that is in the PAGE_OFFSET linear mapping, but we're not on the kernel trap table yet so those addresses won't translate properly. One idea is to do a by-hand firmware call like we do in the early bootup code and elsewhere here in trampoline.S But this fails as well, as aparently the secondary processors are not booted with OBP's special locked TLB entries loaded. These are necessary for the firwmare to processes TLB misses correctly up until the point where we take over the trap table. This does need to be resolved at some point. Signed-off-by: David S. Miller --- arch/sparc64/kernel/dtlb_base.S | 14 ++- arch/sparc64/kernel/dtlb_prot.S | 12 +-- arch/sparc64/kernel/head.S | 61 ++++++------- arch/sparc64/kernel/itlb_base.S | 26 +++--- arch/sparc64/kernel/ktlb.S | 92 ++++++++++---------- arch/sparc64/mm/init.c | 187 ++++++++++++++++------------------------ 6 files changed, 166 insertions(+), 226 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/dtlb_base.S b/arch/sparc64/kernel/dtlb_base.S index 702d349c1e8..6528786840c 100644 --- a/arch/sparc64/kernel/dtlb_base.S +++ b/arch/sparc64/kernel/dtlb_base.S @@ -53,19 +53,18 @@ * be guaranteed to be 0 ... mmu_context.h does guarantee this * by only using 10 bits in the hwcontext value. */ -#define CREATE_VPTE_OFFSET1(r1, r2) +#define CREATE_VPTE_OFFSET1(r1, r2) nop #define CREATE_VPTE_OFFSET2(r1, r2) \ srax r1, 10, r2 -#define CREATE_VPTE_NOP nop #else #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, PAGE_SHIFT, r2 #define CREATE_VPTE_OFFSET2(r1, r2) \ sllx r2, 3, r2 -#define CREATE_VPTE_NOP #endif /* DTLB ** ICACHE line 1: Quick user TLB misses */ + mov TLB_SFSR, %g1 ldxa [%g1 + %g1] ASI_DMMU, %g4 ! Get TAG_ACCESS andcc %g4, TAG_CONTEXT_BITS, %g0 ! From Nucleus? from_tl1_trap: @@ -74,18 +73,16 @@ from_tl1_trap: be,pn %xcc, kvmap ! Yep, special processing CREATE_VPTE_OFFSET2(%g4, %g6) ! Create VPTE offset cmp %g5, 4 ! Last trap level? - be,pn %xcc, longpath ! Yep, cannot risk VPTE miss - nop ! delay slot /* DTLB ** ICACHE line 2: User finish + quick kernel TLB misses */ + be,pn %xcc, longpath ! Yep, cannot risk VPTE miss + nop ! delay slot ldxa [%g3 + %g6] ASI_S, %g5 ! Load VPTE 1: brgez,pn %g5, longpath ! Invalid, branch out nop ! Delay-slot 9: stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB retry ! Trap return nop - nop - nop /* DTLB ** ICACHE line 3: winfixups+real_faults */ longpath: @@ -106,8 +103,7 @@ longpath: nop nop nop - CREATE_VPTE_NOP + nop #undef CREATE_VPTE_OFFSET1 #undef CREATE_VPTE_OFFSET2 -#undef CREATE_VPTE_NOP diff --git a/arch/sparc64/kernel/dtlb_prot.S b/arch/sparc64/kernel/dtlb_prot.S index d848bb7374b..e0a92016260 100644 --- a/arch/sparc64/kernel/dtlb_prot.S +++ b/arch/sparc64/kernel/dtlb_prot.S @@ -14,14 +14,14 @@ */ /* PROT ** ICACHE line 1: User DTLB protection trap */ - stxa %g0, [%g1] ASI_DMMU ! Clear SFSR FaultValid bit - membar #Sync ! Synchronize ASI stores - rdpr %pstate, %g5 ! Move into alternate globals + mov TLB_SFSR, %g1 + stxa %g0, [%g1] ASI_DMMU ! Clear FaultValid bit + membar #Sync ! Synchronize stores + rdpr %pstate, %g5 ! Move into alt-globals wrpr %g5, PSTATE_AG|PSTATE_MG, %pstate - rdpr %tl, %g1 ! Need to do a winfixup? + rdpr %tl, %g1 ! Need a winfixup? cmp %g1, 1 ! Trap level >1? - mov TLB_TAG_ACCESS, %g4 ! Prepare reload of vaddr - nop + mov TLB_TAG_ACCESS, %g4 ! For reload of vaddr /* PROT ** ICACHE line 2: More real fault processing */ bgu,pn %xcc, winfix_trampoline ! Yes, perform winfixup diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index 4c942f71184..b49dcd4504b 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -28,19 +28,14 @@ #include /* This section from from _start to sparc64_boot_end should fit into - * 0x0000.0000.0040.4000 to 0x0000.0000.0040.8000 and will be sharing space - * with bootup_user_stack, which is from 0x0000.0000.0040.4000 to - * 0x0000.0000.0040.6000 and empty_bad_page, which is from - * 0x0000.0000.0040.6000 to 0x0000.0000.0040.8000. + * 0x0000000000404000 to 0x0000000000408000. */ - .text .globl start, _start, stext, _stext _start: start: _stext: stext: -bootup_user_stack: ! 0x0000000000404000 b sparc64_boot flushw /* Flush register file. */ @@ -392,31 +387,30 @@ tlb_fixup_done: * former does use this code, the latter does not yet due * to some complexities. That should be fixed up at some * point. + * + * There used to be enormous complexity wrt. transferring + * over from the firwmare's trap table to the Linux kernel's. + * For example, there was a chicken & egg problem wrt. building + * the OBP page tables, yet needing to be on the Linux kernel + * trap table (to translate PAGE_OFFSET addresses) in order to + * do that. + * + * We now handle OBP tlb misses differently, via linear lookups + * into the prom_trans[] array. So that specific problem no + * longer exists. Yet, unfortunately there are still some issues + * preventing trampoline.S from using this code... ho hum. */ .globl setup_trap_table setup_trap_table: save %sp, -192, %sp - /* Force interrupts to be disabled. Transferring over to - * the Linux trap table is a very delicate operation. - * Until we are actually on the Linux trap table, we cannot - * get the PAGE_OFFSET linear mappings translated. We need - * that mapping to be setup in order to initialize the firmware - * page tables. - * - * So there is this window of time, from the return from - * prom_set_trap_table() until inherit_prom_mappings_post() - * (in arch/sparc64/mm/init.c) completes, during which no - * firmware address space accesses can be made. - */ + /* Force interrupts to be disabled. */ rdpr %pstate, %o1 andn %o1, PSTATE_IE, %o1 wrpr %o1, 0x0, %pstate wrpr %g0, 15, %pil - /* Ok, now make the final valid firmware call to jump over - * to the Linux trap table. - */ + /* Make the firmware call to jump over to the Linux trap table. */ call prom_set_trap_table sethi %hi(sparc64_ttable_tl0), %o0 @@ -540,15 +534,21 @@ setup_tba: /* i0 = is_starfire */ ret restore +sparc64_boot_end: + +#include "systbls.S" +#include "ktlb.S" +#include "etrap.S" +#include "rtrap.S" +#include "winfixup.S" +#include "entry.S" /* - * The following skips make sure the trap table in ttable.S is aligned + * The following skip makes sure the trap table in ttable.S is aligned * on a 32K boundary as required by the v9 specs for TBA register. */ -sparc64_boot_end: - .skip 0x2000 + _start - sparc64_boot_end -bootup_user_stack_end: - .skip 0x2000 +1: + .skip 0x4000 + _start - 1b #ifdef CONFIG_SBUS /* This is just a hack to fool make depend config.h discovering @@ -560,15 +560,6 @@ bootup_user_stack_end: ! 0x0000000000408000 #include "ttable.S" -#include "systbls.S" -#include "ktlb.S" -#include "etrap.S" -#include "rtrap.S" -#include "winfixup.S" -#include "entry.S" - - /* This is just anal retentiveness on my part... */ - .align 16384 .data .align 8 diff --git a/arch/sparc64/kernel/itlb_base.S b/arch/sparc64/kernel/itlb_base.S index b5e32dfa4fb..4951ff8f687 100644 --- a/arch/sparc64/kernel/itlb_base.S +++ b/arch/sparc64/kernel/itlb_base.S @@ -15,14 +15,12 @@ */ #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, 10, r2 -#define CREATE_VPTE_OFFSET2(r1, r2) -#define CREATE_VPTE_NOP nop +#define CREATE_VPTE_OFFSET2(r1, r2) nop #else /* PAGE_SHIFT */ #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, PAGE_SHIFT, r2 #define CREATE_VPTE_OFFSET2(r1, r2) \ sllx r2, 3, r2 -#define CREATE_VPTE_NOP #endif /* PAGE_SHIFT */ @@ -36,6 +34,7 @@ */ /* ITLB ** ICACHE line 1: Quick user TLB misses */ + mov TLB_SFSR, %g1 ldxa [%g1 + %g1] ASI_IMMU, %g4 ! Get TAG_ACCESS CREATE_VPTE_OFFSET1(%g4, %g6) ! Create VPTE offset CREATE_VPTE_OFFSET2(%g4, %g6) ! Create VPTE offset @@ -43,41 +42,38 @@ 1: brgez,pn %g5, 3f ! Not valid, branch out sethi %hi(_PAGE_EXEC), %g4 ! Delay-slot andcc %g5, %g4, %g0 ! Executable? + +/* ITLB ** ICACHE line 2: Real faults */ be,pn %xcc, 3f ! Nope, branch. nop ! Delay-slot 2: stxa %g5, [%g0] ASI_ITLB_DATA_IN ! Load PTE into TLB retry ! Trap return -3: rdpr %pstate, %g4 ! Move into alternate globals - -/* ITLB ** ICACHE line 2: Real faults */ +3: rdpr %pstate, %g4 ! Move into alt-globals wrpr %g4, PSTATE_AG|PSTATE_MG, %pstate rdpr %tpc, %g5 ! And load faulting VA mov FAULT_CODE_ITLB, %g4 ! It was read from ITLB -sparc64_realfault_common: ! Called by TL0 dtlb_miss too + +/* ITLB ** ICACHE line 3: Finish faults */ +sparc64_realfault_common: ! Called by dtlb_miss stb %g4, [%g6 + TI_FAULT_CODE] stx %g5, [%g6 + TI_FAULT_ADDR] ba,pt %xcc, etrap ! Save state 1: rd %pc, %g7 ! ... - nop - -/* ITLB ** ICACHE line 3: Finish faults + window fixups */ call do_sparc64_fault ! Call fault handler add %sp, PTREGS_OFF, %o0! Compute pt_regs arg ba,pt %xcc, rtrap_clr_l6 ! Restore cpu state nop + +/* ITLB ** ICACHE line 4: Window fixups */ winfix_trampoline: rdpr %tpc, %g3 ! Prepare winfixup TNPC - or %g3, 0x7c, %g3 ! Compute offset to branch + or %g3, 0x7c, %g3 ! Compute branch offset wrpr %g3, %tnpc ! Write it into TNPC done ! Do it to it - -/* ITLB ** ICACHE line 4: Unused... */ nop nop nop nop - CREATE_VPTE_NOP #undef CREATE_VPTE_OFFSET1 #undef CREATE_VPTE_OFFSET2 -#undef CREATE_VPTE_NOP diff --git a/arch/sparc64/kernel/ktlb.S b/arch/sparc64/kernel/ktlb.S index 7796b37f478..d9244d3c9f7 100644 --- a/arch/sparc64/kernel/ktlb.S +++ b/arch/sparc64/kernel/ktlb.S @@ -58,9 +58,6 @@ vpte_noent: done vpte_insn_obp: - sethi %hi(prom_pmd_phys), %g5 - ldx [%g5 + %lo(prom_pmd_phys)], %g5 - /* Behave as if we are at TL0. */ wrpr %g0, 1, %tl rdpr %tpc, %g4 /* Find original faulting iaddr */ @@ -71,58 +68,57 @@ vpte_insn_obp: mov TLB_SFSR, %g1 stxa %g4, [%g1 + %g1] ASI_IMMU - /* Get PMD offset. */ - srlx %g4, 23, %g6 - and %g6, 0x7ff, %g6 - sllx %g6, 2, %g6 - - /* Load PMD, is it valid? */ - lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brz,pn %g5, longpath - sllx %g5, 11, %g5 - - /* Get PTE offset. */ - srlx %g4, 13, %g6 - and %g6, 0x3ff, %g6 - sllx %g6, 3, %g6 - - /* Load PTE. */ - ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brgez,pn %g5, longpath - nop - - /* TLB load and return from trap. */ + sethi %hi(prom_trans), %g5 + or %g5, %lo(prom_trans), %g5 + +1: ldx [%g5 + 0x00], %g6 ! base + brz,a,pn %g6, longpath ! no more entries, fail + mov TLB_SFSR, %g1 ! and restore %g1 + ldx [%g5 + 0x08], %g1 ! len + add %g6, %g1, %g1 ! end + cmp %g6, %g4 + bgu,pt %xcc, 2f + cmp %g4, %g1 + bgeu,pt %xcc, 2f + ldx [%g5 + 0x10], %g1 ! PTE + + /* TLB load, restore %g1, and return from trap. */ + sub %g4, %g6, %g6 + add %g1, %g6, %g5 + mov TLB_SFSR, %g1 stxa %g5, [%g0] ASI_ITLB_DATA_IN retry -kvmap_do_obp: - sethi %hi(prom_pmd_phys), %g5 - ldx [%g5 + %lo(prom_pmd_phys)], %g5 - - /* Get PMD offset. */ - srlx %g4, 23, %g6 - and %g6, 0x7ff, %g6 - sllx %g6, 2, %g6 - - /* Load PMD, is it valid? */ - lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brz,pn %g5, longpath - sllx %g5, 11, %g5 - - /* Get PTE offset. */ - srlx %g4, 13, %g6 - and %g6, 0x3ff, %g6 - sllx %g6, 3, %g6 - - /* Load PTE. */ - ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brgez,pn %g5, longpath - nop +2: ba,pt %xcc, 1b + add %g5, (3 * 8), %g5 ! next entry - /* TLB load and return from trap. */ +kvmap_do_obp: + sethi %hi(prom_trans), %g5 + or %g5, %lo(prom_trans), %g5 + srlx %g4, 13, %g4 + sllx %g4, 13, %g4 + +1: ldx [%g5 + 0x00], %g6 ! base + brz,a,pn %g6, longpath ! no more entries, fail + mov TLB_SFSR, %g1 ! and restore %g1 + ldx [%g5 + 0x08], %g1 ! len + add %g6, %g1, %g1 ! end + cmp %g6, %g4 + bgu,pt %xcc, 2f + cmp %g4, %g1 + bgeu,pt %xcc, 2f + ldx [%g5 + 0x10], %g1 ! PTE + + /* TLB load, restore %g1, and return from trap. */ + sub %g4, %g6, %g6 + add %g1, %g6, %g5 + mov TLB_SFSR, %g1 stxa %g5, [%g0] ASI_DTLB_DATA_IN retry +2: ba,pt %xcc, 1b + add %g5, (3 * 8), %g5 ! next entry + /* * On a first level data miss, check whether this is to the OBP range (note * that such accesses can be made by prom, as well as by kernel using diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index 0d2e967c720..1e44ee26cee 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -105,7 +105,7 @@ static void __init read_obp_memory(const char *property, regs[i].phys_addr = base; regs[i].reg_size = size; } - sort(regs, ents, sizeof(struct linux_prom64_registers), + sort(regs, ents, sizeof(struct linux_prom64_registers), cmp_p64, NULL); } @@ -367,8 +367,11 @@ struct linux_prom_translation { unsigned long size; unsigned long data; }; -static struct linux_prom_translation prom_trans[512] __initdata; -static unsigned int prom_trans_ents __initdata; + +/* Exported for kernel TLB miss handling in ktlb.S */ +struct linux_prom_translation prom_trans[512] __read_mostly; +unsigned int prom_trans_ents __read_mostly; +unsigned int swapper_pgd_zero __read_mostly; extern unsigned long prom_boot_page; extern void prom_remap(unsigned long physpage, unsigned long virtpage, int mmu_ihandle); @@ -378,122 +381,57 @@ extern void register_prom_callbacks(void); /* Exported for SMP bootup purposes. */ unsigned long kern_locked_tte_data; -/* Exported for kernel TLB miss handling in ktlb.S */ -unsigned long prom_pmd_phys __read_mostly; -unsigned int swapper_pgd_zero __read_mostly; - -static pmd_t *prompmd __read_mostly; - -#define BASE_PAGE_SIZE 8192 - /* * Translate PROM's mapping we capture at boot time into physical address. * The second parameter is only set from prom_callback() invocations. */ unsigned long prom_virt_to_phys(unsigned long promva, int *error) { - pmd_t *pmdp = prompmd + ((promva >> 23) & 0x7ff); - pte_t *ptep; - unsigned long base; - - if (pmd_none(*pmdp)) { - if (error) - *error = 1; - return 0; - } - ptep = (pte_t *)__pmd_page(*pmdp) + ((promva >> 13) & 0x3ff); - if (!pte_present(*ptep)) { - if (error) - *error = 1; - return 0; - } - if (error) { - *error = 0; - return pte_val(*ptep); - } - base = pte_val(*ptep) & _PAGE_PADDR; - - return base + (promva & (BASE_PAGE_SIZE - 1)); -} + int i; -/* The obp translations are saved based on 8k pagesize, since obp can - * use a mixture of pagesizes. Misses to the LOW_OBP_ADDRESS -> - * HI_OBP_ADDRESS range are handled in entry.S and do not use the vpte - * scheme (also, see rant in inherit_locked_prom_mappings()). - */ -static void __init build_obp_range(unsigned long start, unsigned long end, unsigned long data) -{ - unsigned long vaddr; + for (i = 0; i < prom_trans_ents; i++) { + struct linux_prom_translation *p = &prom_trans[i]; - for (vaddr = start; vaddr < end; vaddr += BASE_PAGE_SIZE) { - unsigned long val; - pmd_t *pmd; - pte_t *pte; + if (promva >= p->virt && + promva < (p->virt + p->size)) { + unsigned long base = p->data & _PAGE_PADDR; - pmd = prompmd + ((vaddr >> 23) & 0x7ff); - if (pmd_none(*pmd)) { - pte = __alloc_bootmem(BASE_PAGE_SIZE, BASE_PAGE_SIZE, - PAGE_SIZE); - if (!pte) - prom_halt(); - memset(pte, 0, BASE_PAGE_SIZE); - pmd_set(pmd, pte); + if (error) + *error = 0; + return base + (promva & (8192 - 1)); } - pte = (pte_t *) __pmd_page(*pmd) + ((vaddr >> 13) & 0x3ff); - - val = data; - - /* Clear diag TTE bits. */ - if (tlb_type == spitfire) - val &= ~0x0003fe0000000000UL; - - set_pte_at(&init_mm, vaddr, pte, - __pte(val | _PAGE_MODIFIED)); - - data += BASE_PAGE_SIZE; } + if (error) + *error = 1; + return 0UL; } +/* The obp translations are saved based on 8k pagesize, since obp can + * use a mixture of pagesizes. Misses to the LOW_OBP_ADDRESS -> + * HI_OBP_ADDRESS range are handled in ktlb.S and do not use the vpte + * scheme (also, see rant in inherit_locked_prom_mappings()). + */ static inline int in_obp_range(unsigned long vaddr) { return (vaddr >= LOW_OBP_ADDRESS && vaddr < HI_OBP_ADDRESS); } -#define OBP_PMD_SIZE 2048 -static void __init build_obp_pgtable(void) +static int cmp_ptrans(const void *a, const void *b) { - unsigned long i; - - prompmd = __alloc_bootmem(OBP_PMD_SIZE, OBP_PMD_SIZE, PAGE_SIZE); - if (!prompmd) - prom_halt(); - - memset(prompmd, 0, OBP_PMD_SIZE); - - prom_pmd_phys = __pa(prompmd); - - for (i = 0; i < prom_trans_ents; i++) { - unsigned long start, end; - - if (!in_obp_range(prom_trans[i].virt)) - continue; + const struct linux_prom_translation *x = a, *y = b; - start = prom_trans[i].virt; - end = start + prom_trans[i].size; - if (end > HI_OBP_ADDRESS) - end = HI_OBP_ADDRESS; - - build_obp_range(start, end, prom_trans[i].data); - } + if (x->virt > y->virt) + return 1; + if (x->virt < y->virt) + return -1; + return 0; } -/* Read OBP translations property into 'prom_trans[]'. - * Return the number of entries. - */ +/* Read OBP translations property into 'prom_trans[]'. */ static void __init read_obp_translations(void) { - int n, node; + int n, node, ents, first, last, i; node = prom_finddevice("/virtual-memory"); n = prom_getproplen(node, "translations"); @@ -515,7 +453,41 @@ static void __init read_obp_translations(void) n = n / sizeof(struct linux_prom_translation); - prom_trans_ents = n; + ents = n; + + sort(prom_trans, ents, sizeof(struct linux_prom_translation), + cmp_ptrans, NULL); + + /* Now kick out all the non-OBP entries. */ + for (i = 0; i < ents; i++) { + if (in_obp_range(prom_trans[i].virt)) + break; + } + first = i; + for (; i < ents; i++) { + if (!in_obp_range(prom_trans[i].virt)) + break; + } + last = i; + + for (i = 0; i < (last - first); i++) { + struct linux_prom_translation *src = &prom_trans[i + first]; + struct linux_prom_translation *dest = &prom_trans[i]; + + *dest = *src; + } + for (; i < ents; i++) { + struct linux_prom_translation *dest = &prom_trans[i]; + dest->virt = dest->size = dest->data = 0x0UL; + } + + prom_trans_ents = last - first; + + if (tlb_type == spitfire) { + /* Clear diag TTE bits. */ + for (i = 0; i < prom_trans_ents; i++) + prom_trans[i].data &= ~0x0003fe0000000000UL; + } } static void __init remap_kernel(void) @@ -553,21 +525,18 @@ static void __init remap_kernel(void) } -static void __init inherit_prom_mappings_pre(void) +static void __init inherit_prom_mappings(void) { read_obp_translations(); /* Now fixup OBP's idea about where we really are mapped. */ prom_printf("Remapping the kernel... "); remap_kernel(); - prom_printf("done.\n"); -} -static void __init inherit_prom_mappings_post(void) -{ - build_obp_pgtable(); + prom_printf("Registering callbacks... "); register_prom_callbacks(); + prom_printf("done.\n"); } /* The OBP specifications for sun4u mark 0xfffffffc00000000 and @@ -1519,7 +1488,7 @@ void __init paging_init(void) swapper_pgd_zero = pgd_val(swapper_pg_dir[0]); - inherit_prom_mappings_pre(); + inherit_prom_mappings(); /* Ok, we can use our TLB miss and window trap handlers safely. * We need to do a quick peek here to see if we are on StarFire @@ -1530,23 +1499,15 @@ void __init paging_init(void) extern void setup_tba(int); setup_tba(this_is_starfire); } - __flush_tlb_all(); - /* Everything from this point forward, until we are done with - * inherit_prom_mappings_post(), must complete successfully - * without calling into the firmware. The firwmare page tables - * have not been built, but we are running on the Linux kernel's - * trap table. - */ + inherit_locked_prom_mappings(1); + + __flush_tlb_all(); /* Setup bootmem... */ pages_avail = 0; last_valid_pfn = end_pfn = bootmem_init(&pages_avail); - inherit_prom_mappings_post(); - - inherit_locked_prom_mappings(1); - #ifdef CONFIG_DEBUG_PAGEALLOC kernel_physical_mapping_init(); #endif -- cgit v1.2.3-70-g09d2 From 51e8513615ed8202b22ba9a43b0c7376ea4f6868 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 13 Oct 2005 21:10:08 -0700 Subject: [SPARC64]: Consolidate common PCI IOMMU init code. All the PCI controller drivers were doing the same thing setting up the IOMMU software state, put it all in one spot. Signed-off-by: David S. Miller --- arch/sparc64/kernel/pci_iommu.c | 54 ++++++++++++++++++++++++++++++++++--- arch/sparc64/kernel/pci_psycho.c | 44 ++++--------------------------- arch/sparc64/kernel/pci_sabre.c | 39 +++------------------------ arch/sparc64/kernel/pci_schizo.c | 57 +++------------------------------------- include/asm-sparc64/pbm.h | 2 +- 5 files changed, 63 insertions(+), 133 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/pci_iommu.c b/arch/sparc64/kernel/pci_iommu.c index 425c60cfea1..cdee7d9fed7 100644 --- a/arch/sparc64/kernel/pci_iommu.c +++ b/arch/sparc64/kernel/pci_iommu.c @@ -80,13 +80,59 @@ static void inline iopte_make_dummy(struct pci_iommu *iommu, iopte_t *iopte) iopte_val(*iopte) = val; } -void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize) +void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize, u32 dma_offset, u32 dma_addr_mask) { - int i; + unsigned long i, tsbbase, order; + + /* Setup initial software IOMMU state. */ + spin_lock_init(&iommu->lock); + iommu->ctx_lowest_free = 1; + iommu->page_table_map_base = dma_offset; + iommu->dma_addr_mask = dma_addr_mask; + + switch (tsbsize / (8 * 1024)) { + case 64: + iommu->page_table_sz_bits = 16; + break; + case 128: + iommu->page_table_sz_bits = 17; + break; + default: + prom_printf("PCI_IOMMU: Illegal TSB size %d\n", + tsbsize / (8 * 1024)); + prom_halt(); + break; + }; + + iommu->lowest_consistent_map = + 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); + + for (i = 0; i < PBM_NCLUSTERS; i++) { + iommu->alloc_info[i].flush = 0; + iommu->alloc_info[i].next = 0; + } - tsbsize /= sizeof(iopte_t); + /* Allocate and initialize the dummy page which we + * set inactive IO PTEs to point to. + */ + iommu->dummy_page = __get_free_pages(GFP_KERNEL, 0); + if (!iommu->dummy_page) { + prom_printf("PCI_IOMMU: Error, gfp(dummy_page) failed.\n"); + prom_halt(); + } + memset((void *)iommu->dummy_page, 0, PAGE_SIZE); + iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); + + /* Now allocate and setup the IOMMU page table itself. */ + order = get_order(tsbsize); + tsbbase = __get_free_pages(GFP_KERNEL, order); + if (!tsbbase) { + prom_printf("PCI_IOMMU: Error, gfp(tsb) failed.\n"); + prom_halt(); + } + iommu->page_table = (iopte_t *)tsbbase; - for (i = 0; i < tsbsize; i++) + for (i = 0; i < tsbsize / sizeof(iopte_t); i++) iopte_make_dummy(iommu, &iommu->page_table[i]); } diff --git a/arch/sparc64/kernel/pci_psycho.c b/arch/sparc64/kernel/pci_psycho.c index 6ed1ef25e0a..c03ed5f49d3 100644 --- a/arch/sparc64/kernel/pci_psycho.c +++ b/arch/sparc64/kernel/pci_psycho.c @@ -1207,13 +1207,9 @@ static void psycho_scan_bus(struct pci_controller_info *p) static void psycho_iommu_init(struct pci_controller_info *p) { struct pci_iommu *iommu = p->pbm_A.iommu; - unsigned long tsbbase, i; + unsigned long i; u64 control; - /* Setup initial software IOMMU state. */ - spin_lock_init(&iommu->lock); - iommu->ctx_lowest_free = 1; - /* Register addresses. */ iommu->iommu_control = p->pbm_A.controller_regs + PSYCHO_IOMMU_CONTROL; iommu->iommu_tsbbase = p->pbm_A.controller_regs + PSYCHO_IOMMU_TSBBASE; @@ -1240,40 +1236,10 @@ static void psycho_iommu_init(struct pci_controller_info *p) /* Leave diag mode enabled for full-flushing done * in pci_iommu.c */ + pci_iommu_table_init(iommu, IO_TSB_SIZE, 0xc0000000, 0xffffffff); - iommu->dummy_page = __get_free_pages(GFP_KERNEL, 0); - if (!iommu->dummy_page) { - prom_printf("PSYCHO_IOMMU: Error, gfp(dummy_page) failed.\n"); - prom_halt(); - } - memset((void *)iommu->dummy_page, 0, PAGE_SIZE); - iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); - - /* Using assumed page size 8K with 128K entries we need 1MB iommu page - * table (128K ioptes * 8 bytes per iopte). This is - * page order 7 on UltraSparc. - */ - tsbbase = __get_free_pages(GFP_KERNEL, get_order(IO_TSB_SIZE)); - if (!tsbbase) { - prom_printf("PSYCHO_IOMMU: Error, gfp(tsb) failed.\n"); - prom_halt(); - } - iommu->page_table = (iopte_t *)tsbbase; - iommu->page_table_sz_bits = 17; - iommu->page_table_map_base = 0xc0000000; - iommu->dma_addr_mask = 0xffffffff; - pci_iommu_table_init(iommu, IO_TSB_SIZE); - - /* We start with no consistent mappings. */ - iommu->lowest_consistent_map = - 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); - - for (i = 0; i < PBM_NCLUSTERS; i++) { - iommu->alloc_info[i].flush = 0; - iommu->alloc_info[i].next = 0; - } - - psycho_write(p->pbm_A.controller_regs + PSYCHO_IOMMU_TSBBASE, __pa(tsbbase)); + psycho_write(p->pbm_A.controller_regs + PSYCHO_IOMMU_TSBBASE, + __pa(iommu->page_table)); control = psycho_read(p->pbm_A.controller_regs + PSYCHO_IOMMU_CONTROL); control &= ~(PSYCHO_IOMMU_CTRL_TSBSZ | PSYCHO_IOMMU_CTRL_TBWSZ); @@ -1281,7 +1247,7 @@ static void psycho_iommu_init(struct pci_controller_info *p) psycho_write(p->pbm_A.controller_regs + PSYCHO_IOMMU_CONTROL, control); /* If necessary, hook us up for starfire IRQ translations. */ - if(this_is_starfire) + if (this_is_starfire) p->starfire_cookie = starfire_hookup(p->pbm_A.portid); else p->starfire_cookie = NULL; diff --git a/arch/sparc64/kernel/pci_sabre.c b/arch/sparc64/kernel/pci_sabre.c index 0ee6bd5b9ac..da8e1364194 100644 --- a/arch/sparc64/kernel/pci_sabre.c +++ b/arch/sparc64/kernel/pci_sabre.c @@ -1267,13 +1267,9 @@ static void sabre_iommu_init(struct pci_controller_info *p, u32 dma_mask) { struct pci_iommu *iommu = p->pbm_A.iommu; - unsigned long tsbbase, i, order; + unsigned long i; u64 control; - /* Setup initial software IOMMU state. */ - spin_lock_init(&iommu->lock); - iommu->ctx_lowest_free = 1; - /* Register addresses. */ iommu->iommu_control = p->pbm_A.controller_regs + SABRE_IOMMU_CONTROL; iommu->iommu_tsbbase = p->pbm_A.controller_regs + SABRE_IOMMU_TSBBASE; @@ -1295,26 +1291,10 @@ static void sabre_iommu_init(struct pci_controller_info *p, /* Leave diag mode enabled for full-flushing done * in pci_iommu.c */ + pci_iommu_table_init(iommu, tsbsize * 1024 * 8, dvma_offset, dma_mask); - iommu->dummy_page = __get_free_pages(GFP_KERNEL, 0); - if (!iommu->dummy_page) { - prom_printf("PSYCHO_IOMMU: Error, gfp(dummy_page) failed.\n"); - prom_halt(); - } - memset((void *)iommu->dummy_page, 0, PAGE_SIZE); - iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); - - tsbbase = __get_free_pages(GFP_KERNEL, order = get_order(tsbsize * 1024 * 8)); - if (!tsbbase) { - prom_printf("SABRE_IOMMU: Error, gfp(tsb) failed.\n"); - prom_halt(); - } - iommu->page_table = (iopte_t *)tsbbase; - iommu->page_table_map_base = dvma_offset; - iommu->dma_addr_mask = dma_mask; - pci_iommu_table_init(iommu, PAGE_SIZE << order); - - sabre_write(p->pbm_A.controller_regs + SABRE_IOMMU_TSBBASE, __pa(tsbbase)); + sabre_write(p->pbm_A.controller_regs + SABRE_IOMMU_TSBBASE, + __pa(iommu->page_table)); control = sabre_read(p->pbm_A.controller_regs + SABRE_IOMMU_CONTROL); control &= ~(SABRE_IOMMUCTRL_TSBSZ | SABRE_IOMMUCTRL_TBWSZ); @@ -1322,11 +1302,9 @@ static void sabre_iommu_init(struct pci_controller_info *p, switch(tsbsize) { case 64: control |= SABRE_IOMMU_TSBSZ_64K; - iommu->page_table_sz_bits = 16; break; case 128: control |= SABRE_IOMMU_TSBSZ_128K; - iommu->page_table_sz_bits = 17; break; default: prom_printf("iommu_init: Illegal TSB size %d\n", tsbsize); @@ -1334,15 +1312,6 @@ static void sabre_iommu_init(struct pci_controller_info *p, break; } sabre_write(p->pbm_A.controller_regs + SABRE_IOMMU_CONTROL, control); - - /* We start with no consistent mappings. */ - iommu->lowest_consistent_map = - 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); - - for (i = 0; i < PBM_NCLUSTERS; i++) { - iommu->alloc_info[i].flush = 0; - iommu->alloc_info[i].next = 0; - } } static void pbm_register_toplevel_resources(struct pci_controller_info *p, diff --git a/arch/sparc64/kernel/pci_schizo.c b/arch/sparc64/kernel/pci_schizo.c index cae5b61fe2f..d8c4e0919b4 100644 --- a/arch/sparc64/kernel/pci_schizo.c +++ b/arch/sparc64/kernel/pci_schizo.c @@ -1765,7 +1765,7 @@ static void schizo_pbm_strbuf_init(struct pci_pbm_info *pbm) static void schizo_pbm_iommu_init(struct pci_pbm_info *pbm) { struct pci_iommu *iommu = pbm->iommu; - unsigned long tsbbase, i, tagbase, database, order; + unsigned long i, tagbase, database; u32 vdma[2], dma_mask; u64 control; int err, tsbsize; @@ -1800,10 +1800,6 @@ static void schizo_pbm_iommu_init(struct pci_pbm_info *pbm) prom_halt(); }; - /* Setup initial software IOMMU state. */ - spin_lock_init(&iommu->lock); - iommu->ctx_lowest_free = 1; - /* Register addresses, SCHIZO has iommu ctx flushing. */ iommu->iommu_control = pbm->pbm_regs + SCHIZO_IOMMU_CONTROL; iommu->iommu_tsbbase = pbm->pbm_regs + SCHIZO_IOMMU_TSBBASE; @@ -1832,56 +1828,9 @@ static void schizo_pbm_iommu_init(struct pci_pbm_info *pbm) /* Leave diag mode enabled for full-flushing done * in pci_iommu.c */ + pci_iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask); - iommu->dummy_page = __get_free_pages(GFP_KERNEL, 0); - if (!iommu->dummy_page) { - prom_printf("PSYCHO_IOMMU: Error, gfp(dummy_page) failed.\n"); - prom_halt(); - } - memset((void *)iommu->dummy_page, 0, PAGE_SIZE); - iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); - - /* Using assumed page size 8K with 128K entries we need 1MB iommu page - * table (128K ioptes * 8 bytes per iopte). This is - * page order 7 on UltraSparc. - */ - order = get_order(tsbsize * 8 * 1024); - tsbbase = __get_free_pages(GFP_KERNEL, order); - if (!tsbbase) { - prom_printf("%s: Error, gfp(tsb) failed.\n", pbm->name); - prom_halt(); - } - - iommu->page_table = (iopte_t *)tsbbase; - iommu->page_table_map_base = vdma[0]; - iommu->dma_addr_mask = dma_mask; - pci_iommu_table_init(iommu, PAGE_SIZE << order); - - switch (tsbsize) { - case 64: - iommu->page_table_sz_bits = 16; - break; - - case 128: - iommu->page_table_sz_bits = 17; - break; - - default: - prom_printf("iommu_init: Illegal TSB size %d\n", tsbsize); - prom_halt(); - break; - }; - - /* We start with no consistent mappings. */ - iommu->lowest_consistent_map = - 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); - - for (i = 0; i < PBM_NCLUSTERS; i++) { - iommu->alloc_info[i].flush = 0; - iommu->alloc_info[i].next = 0; - } - - schizo_write(iommu->iommu_tsbbase, __pa(tsbbase)); + schizo_write(iommu->iommu_tsbbase, __pa(iommu->page_table)); control = schizo_read(iommu->iommu_control); control &= ~(SCHIZO_IOMMU_CTRL_TSBSZ | SCHIZO_IOMMU_CTRL_TBWSZ); diff --git a/include/asm-sparc64/pbm.h b/include/asm-sparc64/pbm.h index 38bbbccb406..c067407de0b 100644 --- a/include/asm-sparc64/pbm.h +++ b/include/asm-sparc64/pbm.h @@ -102,7 +102,7 @@ struct pci_iommu { u32 dma_addr_mask; }; -extern void pci_iommu_table_init(struct pci_iommu *, int); +extern void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize, u32 dma_offset, u32 dma_addr_mask); /* This describes a PCI bus module's streaming buffer. */ struct pci_strbuf { -- cgit v1.2.3-70-g09d2 From 688cb30bdc3e398d97682a6a58f825821ee838c2 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 13 Oct 2005 22:15:24 -0700 Subject: [SPARC64]: Eliminate PCI IOMMU dma mapping size limit. The hairy fast allocator in the sparc64 PCI IOMMU code has a hard limit of 256 pages. Certain devices can exceed this when performing very large I/Os. So replace with a more simple allocator, based largely upon the arch/ppc64/kernel/iommu.c code. Signed-off-by: David S. Miller --- arch/sparc64/kernel/pci_iommu.c | 365 +++++++++++++++------------------------- include/asm-sparc64/pbm.h | 28 +-- 2 files changed, 145 insertions(+), 248 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/pci_iommu.c b/arch/sparc64/kernel/pci_iommu.c index cdee7d9fed7..a11910be101 100644 --- a/arch/sparc64/kernel/pci_iommu.c +++ b/arch/sparc64/kernel/pci_iommu.c @@ -49,12 +49,6 @@ static void __iommu_flushall(struct pci_iommu *iommu) /* Ensure completion of previous PIO writes. */ (void) pci_iommu_read(iommu->write_complete_reg); - - /* Now update everyone's flush point. */ - for (entry = 0; entry < PBM_NCLUSTERS; entry++) { - iommu->alloc_info[entry].flush = - iommu->alloc_info[entry].next; - } } #define IOPTE_CONSISTENT(CTX) \ @@ -80,9 +74,61 @@ static void inline iopte_make_dummy(struct pci_iommu *iommu, iopte_t *iopte) iopte_val(*iopte) = val; } +/* Based largely upon the ppc64 iommu allocator. */ +static long pci_arena_alloc(struct pci_iommu *iommu, unsigned long npages) +{ + struct pci_iommu_arena *arena = &iommu->arena; + unsigned long n, i, start, end, limit; + int pass; + + limit = arena->limit; + start = arena->hint; + pass = 0; + +again: + n = find_next_zero_bit(arena->map, limit, start); + end = n + npages; + if (unlikely(end >= limit)) { + if (likely(pass < 1)) { + limit = start; + start = 0; + __iommu_flushall(iommu); + pass++; + goto again; + } else { + /* Scanned the whole thing, give up. */ + return -1; + } + } + + for (i = n; i < end; i++) { + if (test_bit(i, arena->map)) { + start = i + 1; + goto again; + } + } + + for (i = n; i < end; i++) + __set_bit(i, arena->map); + + arena->hint = end; + + return n; +} + +static void pci_arena_free(struct pci_iommu_arena *arena, unsigned long base, unsigned long npages) +{ + unsigned long i; + + for (i = base; i < (base + npages); i++) + __clear_bit(i, arena->map); +} + void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize, u32 dma_offset, u32 dma_addr_mask) { - unsigned long i, tsbbase, order; + unsigned long i, tsbbase, order, sz, num_tsb_entries; + + num_tsb_entries = tsbsize / sizeof(iopte_t); /* Setup initial software IOMMU state. */ spin_lock_init(&iommu->lock); @@ -90,27 +136,16 @@ void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize, u32 dma_offset, iommu->page_table_map_base = dma_offset; iommu->dma_addr_mask = dma_addr_mask; - switch (tsbsize / (8 * 1024)) { - case 64: - iommu->page_table_sz_bits = 16; - break; - case 128: - iommu->page_table_sz_bits = 17; - break; - default: - prom_printf("PCI_IOMMU: Illegal TSB size %d\n", - tsbsize / (8 * 1024)); + /* Allocate and initialize the free area map. */ + sz = num_tsb_entries / 8; + sz = (sz + 7UL) & ~7UL; + iommu->arena.map = kmalloc(sz, GFP_KERNEL); + if (!iommu->arena.map) { + prom_printf("PCI_IOMMU: Error, kmalloc(arena.map) failed.\n"); prom_halt(); - break; - }; - - iommu->lowest_consistent_map = - 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); - - for (i = 0; i < PBM_NCLUSTERS; i++) { - iommu->alloc_info[i].flush = 0; - iommu->alloc_info[i].next = 0; } + memset(iommu->arena.map, 0, sz); + iommu->arena.limit = num_tsb_entries; /* Allocate and initialize the dummy page which we * set inactive IO PTEs to point to. @@ -132,114 +167,24 @@ void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize, u32 dma_offset, } iommu->page_table = (iopte_t *)tsbbase; - for (i = 0; i < tsbsize / sizeof(iopte_t); i++) + for (i = 0; i < num_tsb_entries; i++) iopte_make_dummy(iommu, &iommu->page_table[i]); } -static iopte_t *alloc_streaming_cluster(struct pci_iommu *iommu, unsigned long npages) +static inline iopte_t *alloc_npages(struct pci_iommu *iommu, unsigned long npages) { - iopte_t *iopte, *limit, *first; - unsigned long cnum, ent, flush_point; - - cnum = 0; - while ((1UL << cnum) < npages) - cnum++; - iopte = (iommu->page_table + - (cnum << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS))); - - if (cnum == 0) - limit = (iommu->page_table + - iommu->lowest_consistent_map); - else - limit = (iopte + - (1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS))); + long entry; - iopte += ((ent = iommu->alloc_info[cnum].next) << cnum); - flush_point = iommu->alloc_info[cnum].flush; - - first = iopte; - for (;;) { - if (IOPTE_IS_DUMMY(iommu, iopte)) { - if ((iopte + (1 << cnum)) >= limit) - ent = 0; - else - ent = ent + 1; - iommu->alloc_info[cnum].next = ent; - if (ent == flush_point) - __iommu_flushall(iommu); - break; - } - iopte += (1 << cnum); - ent++; - if (iopte >= limit) { - iopte = (iommu->page_table + - (cnum << - (iommu->page_table_sz_bits - PBM_LOGCLUSTERS))); - ent = 0; - } - if (ent == flush_point) - __iommu_flushall(iommu); - if (iopte == first) - goto bad; - } - - /* I've got your streaming cluster right here buddy boy... */ - return iopte; + entry = pci_arena_alloc(iommu, npages); + if (unlikely(entry < 0)) + return NULL; -bad: - printk(KERN_EMERG "pci_iommu: alloc_streaming_cluster of npages(%ld) failed!\n", - npages); - return NULL; + return iommu->page_table + entry; } -static void free_streaming_cluster(struct pci_iommu *iommu, dma_addr_t base, - unsigned long npages, unsigned long ctx) +static inline void free_npages(struct pci_iommu *iommu, dma_addr_t base, unsigned long npages) { - unsigned long cnum, ent; - - cnum = 0; - while ((1UL << cnum) < npages) - cnum++; - - ent = (base << (32 - IO_PAGE_SHIFT + PBM_LOGCLUSTERS - iommu->page_table_sz_bits)) - >> (32 + PBM_LOGCLUSTERS + cnum - iommu->page_table_sz_bits); - - /* If the global flush might not have caught this entry, - * adjust the flush point such that we will flush before - * ever trying to reuse it. - */ -#define between(X,Y,Z) (((Z) - (Y)) >= ((X) - (Y))) - if (between(ent, iommu->alloc_info[cnum].next, iommu->alloc_info[cnum].flush)) - iommu->alloc_info[cnum].flush = ent; -#undef between -} - -/* We allocate consistent mappings from the end of cluster zero. */ -static iopte_t *alloc_consistent_cluster(struct pci_iommu *iommu, unsigned long npages) -{ - iopte_t *iopte; - - iopte = iommu->page_table + (1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS)); - while (iopte > iommu->page_table) { - iopte--; - if (IOPTE_IS_DUMMY(iommu, iopte)) { - unsigned long tmp = npages; - - while (--tmp) { - iopte--; - if (!IOPTE_IS_DUMMY(iommu, iopte)) - break; - } - if (tmp == 0) { - u32 entry = (iopte - iommu->page_table); - - if (entry < iommu->lowest_consistent_map) - iommu->lowest_consistent_map = entry; - return iopte; - } - } - } - return NULL; + pci_arena_free(&iommu->arena, base >> IO_PAGE_SHIFT, npages); } static int iommu_alloc_ctx(struct pci_iommu *iommu) @@ -279,7 +224,7 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_ad struct pcidev_cookie *pcp; struct pci_iommu *iommu; iopte_t *iopte; - unsigned long flags, order, first_page, ctx; + unsigned long flags, order, first_page; void *ret; int npages; @@ -297,9 +242,10 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_ad iommu = pcp->pbm->iommu; spin_lock_irqsave(&iommu->lock, flags); - iopte = alloc_consistent_cluster(iommu, size >> IO_PAGE_SHIFT); - if (iopte == NULL) { - spin_unlock_irqrestore(&iommu->lock, flags); + iopte = alloc_npages(iommu, size >> IO_PAGE_SHIFT); + spin_unlock_irqrestore(&iommu->lock, flags); + + if (unlikely(iopte == NULL)) { free_pages(first_page, order); return NULL; } @@ -308,31 +254,15 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_ad ((iopte - iommu->page_table) << IO_PAGE_SHIFT)); ret = (void *) first_page; npages = size >> IO_PAGE_SHIFT; - ctx = 0; - if (iommu->iommu_ctxflush) - ctx = iommu_alloc_ctx(iommu); first_page = __pa(first_page); while (npages--) { - iopte_val(*iopte) = (IOPTE_CONSISTENT(ctx) | + iopte_val(*iopte) = (IOPTE_CONSISTENT(0UL) | IOPTE_WRITE | (first_page & IOPTE_PAGE)); iopte++; first_page += IO_PAGE_SIZE; } - { - int i; - u32 daddr = *dma_addrp; - - npages = size >> IO_PAGE_SHIFT; - for (i = 0; i < npages; i++) { - pci_iommu_write(iommu->iommu_flush, daddr); - daddr += IO_PAGE_SIZE; - } - } - - spin_unlock_irqrestore(&iommu->lock, flags); - return ret; } @@ -342,7 +272,7 @@ void pci_free_consistent(struct pci_dev *pdev, size_t size, void *cpu, dma_addr_ struct pcidev_cookie *pcp; struct pci_iommu *iommu; iopte_t *iopte; - unsigned long flags, order, npages, i, ctx; + unsigned long flags, order, npages; npages = IO_PAGE_ALIGN(size) >> IO_PAGE_SHIFT; pcp = pdev->sysdata; @@ -352,46 +282,7 @@ void pci_free_consistent(struct pci_dev *pdev, size_t size, void *cpu, dma_addr_ spin_lock_irqsave(&iommu->lock, flags); - if ((iopte - iommu->page_table) == - iommu->lowest_consistent_map) { - iopte_t *walk = iopte + npages; - iopte_t *limit; - - limit = (iommu->page_table + - (1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS))); - while (walk < limit) { - if (!IOPTE_IS_DUMMY(iommu, walk)) - break; - walk++; - } - iommu->lowest_consistent_map = - (walk - iommu->page_table); - } - - /* Data for consistent mappings cannot enter the streaming - * buffers, so we only need to update the TSB. We flush - * the IOMMU here as well to prevent conflicts with the - * streaming mapping deferred tlb flush scheme. - */ - - ctx = 0; - if (iommu->iommu_ctxflush) - ctx = (iopte_val(*iopte) & IOPTE_CONTEXT) >> 47UL; - - for (i = 0; i < npages; i++, iopte++) - iopte_make_dummy(iommu, iopte); - - if (iommu->iommu_ctxflush) { - pci_iommu_write(iommu->iommu_ctxflush, ctx); - } else { - for (i = 0; i < npages; i++) { - u32 daddr = dvma + (i << IO_PAGE_SHIFT); - - pci_iommu_write(iommu->iommu_flush, daddr); - } - } - - iommu_free_ctx(iommu, ctx); + free_npages(iommu, dvma, npages); spin_unlock_irqrestore(&iommu->lock, flags); @@ -418,25 +309,27 @@ dma_addr_t pci_map_single(struct pci_dev *pdev, void *ptr, size_t sz, int direct iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; - if (direction == PCI_DMA_NONE) - BUG(); + if (unlikely(direction == PCI_DMA_NONE)) + goto bad_no_ctx; oaddr = (unsigned long)ptr; npages = IO_PAGE_ALIGN(oaddr + sz) - (oaddr & IO_PAGE_MASK); npages >>= IO_PAGE_SHIFT; spin_lock_irqsave(&iommu->lock, flags); + base = alloc_npages(iommu, npages); + ctx = 0; + if (iommu->iommu_ctxflush) + ctx = iommu_alloc_ctx(iommu); + spin_unlock_irqrestore(&iommu->lock, flags); - base = alloc_streaming_cluster(iommu, npages); - if (base == NULL) + if (unlikely(!base)) goto bad; + bus_addr = (iommu->page_table_map_base + ((base - iommu->page_table) << IO_PAGE_SHIFT)); ret = bus_addr | (oaddr & ~IO_PAGE_MASK); base_paddr = __pa(oaddr & IO_PAGE_MASK); - ctx = 0; - if (iommu->iommu_ctxflush) - ctx = iommu_alloc_ctx(iommu); if (strbuf->strbuf_enabled) iopte_protection = IOPTE_STREAMING(ctx); else @@ -447,12 +340,13 @@ dma_addr_t pci_map_single(struct pci_dev *pdev, void *ptr, size_t sz, int direct for (i = 0; i < npages; i++, base++, base_paddr += IO_PAGE_SIZE) iopte_val(*base) = iopte_protection | base_paddr; - spin_unlock_irqrestore(&iommu->lock, flags); - return ret; bad: - spin_unlock_irqrestore(&iommu->lock, flags); + iommu_free_ctx(iommu, ctx); +bad_no_ctx: + if (printk_ratelimit()) + WARN_ON(1); return PCI_DMA_ERROR_CODE; } @@ -527,10 +421,13 @@ void pci_unmap_single(struct pci_dev *pdev, dma_addr_t bus_addr, size_t sz, int struct pci_iommu *iommu; struct pci_strbuf *strbuf; iopte_t *base; - unsigned long flags, npages, ctx; + unsigned long flags, npages, ctx, i; - if (direction == PCI_DMA_NONE) - BUG(); + if (unlikely(direction == PCI_DMA_NONE)) { + if (printk_ratelimit()) + WARN_ON(1); + return; + } pcp = pdev->sysdata; iommu = pcp->pbm->iommu; @@ -556,13 +453,14 @@ void pci_unmap_single(struct pci_dev *pdev, dma_addr_t bus_addr, size_t sz, int /* Step 1: Kick data out of streaming buffers if necessary. */ if (strbuf->strbuf_enabled) - pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction); + pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, + npages, direction); - /* Step 2: Clear out first TSB entry. */ - iopte_make_dummy(iommu, base); + /* Step 2: Clear out TSB entries. */ + for (i = 0; i < npages; i++) + iopte_make_dummy(iommu, base + i); - free_streaming_cluster(iommu, bus_addr - iommu->page_table_map_base, - npages, ctx); + free_npages(iommu, bus_addr - iommu->page_table_map_base, npages); iommu_free_ctx(iommu, ctx); @@ -667,6 +565,8 @@ int pci_map_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, int pci_map_single(pdev, (page_address(sglist->page) + sglist->offset), sglist->length, direction); + if (unlikely(sglist->dma_address == PCI_DMA_ERROR_CODE)) + return 0; sglist->dma_length = sglist->length; return 1; } @@ -675,21 +575,29 @@ int pci_map_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, int iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; - if (direction == PCI_DMA_NONE) - BUG(); + if (unlikely(direction == PCI_DMA_NONE)) + goto bad_no_ctx; /* Step 1: Prepare scatter list. */ npages = prepare_sg(sglist, nelems); - /* Step 2: Allocate a cluster. */ + /* Step 2: Allocate a cluster and context, if necessary. */ spin_lock_irqsave(&iommu->lock, flags); - base = alloc_streaming_cluster(iommu, npages); + base = alloc_npages(iommu, npages); + ctx = 0; + if (iommu->iommu_ctxflush) + ctx = iommu_alloc_ctx(iommu); + + spin_unlock_irqrestore(&iommu->lock, flags); + if (base == NULL) goto bad; - dma_base = iommu->page_table_map_base + ((base - iommu->page_table) << IO_PAGE_SHIFT); + + dma_base = iommu->page_table_map_base + + ((base - iommu->page_table) << IO_PAGE_SHIFT); /* Step 3: Normalize DMA addresses. */ used = nelems; @@ -702,30 +610,28 @@ int pci_map_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, int } used = nelems - used; - /* Step 4: Choose a context if necessary. */ - ctx = 0; - if (iommu->iommu_ctxflush) - ctx = iommu_alloc_ctx(iommu); - - /* Step 5: Create the mappings. */ + /* Step 4: Create the mappings. */ if (strbuf->strbuf_enabled) iopte_protection = IOPTE_STREAMING(ctx); else iopte_protection = IOPTE_CONSISTENT(ctx); if (direction != PCI_DMA_TODEVICE) iopte_protection |= IOPTE_WRITE; - fill_sg (base, sglist, used, nelems, iopte_protection); + + fill_sg(base, sglist, used, nelems, iopte_protection); + #ifdef VERIFY_SG verify_sglist(sglist, nelems, base, npages); #endif - spin_unlock_irqrestore(&iommu->lock, flags); - return used; bad: - spin_unlock_irqrestore(&iommu->lock, flags); - return PCI_DMA_ERROR_CODE; + iommu_free_ctx(iommu, ctx); +bad_no_ctx: + if (printk_ratelimit()) + WARN_ON(1); + return 0; } /* Unmap a set of streaming mode DMA translations. */ @@ -738,8 +644,10 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, unsigned long flags, ctx, i, npages; u32 bus_addr; - if (direction == PCI_DMA_NONE) - BUG(); + if (unlikely(direction == PCI_DMA_NONE)) { + if (printk_ratelimit()) + WARN_ON(1); + } pcp = pdev->sysdata; iommu = pcp->pbm->iommu; @@ -751,7 +659,8 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, if (sglist[i].dma_length == 0) break; i--; - npages = (IO_PAGE_ALIGN(sglist[i].dma_address + sglist[i].dma_length) - bus_addr) >> IO_PAGE_SHIFT; + npages = (IO_PAGE_ALIGN(sglist[i].dma_address + sglist[i].dma_length) - + bus_addr) >> IO_PAGE_SHIFT; base = iommu->page_table + ((bus_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT); @@ -772,11 +681,11 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, if (strbuf->strbuf_enabled) pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction); - /* Step 2: Clear out first TSB entry. */ - iopte_make_dummy(iommu, base); + /* Step 2: Clear out the TSB entries. */ + for (i = 0; i < npages; i++) + iopte_make_dummy(iommu, base + i); - free_streaming_cluster(iommu, bus_addr - iommu->page_table_map_base, - npages, ctx); + free_npages(iommu, bus_addr - iommu->page_table_map_base, npages); iommu_free_ctx(iommu, ctx); diff --git a/include/asm-sparc64/pbm.h b/include/asm-sparc64/pbm.h index c067407de0b..dd35a2c7798 100644 --- a/include/asm-sparc64/pbm.h +++ b/include/asm-sparc64/pbm.h @@ -27,23 +27,27 @@ * PCI bus. */ -#define PBM_LOGCLUSTERS 3 -#define PBM_NCLUSTERS (1 << PBM_LOGCLUSTERS) - struct pci_controller_info; /* This contains the software state necessary to drive a PCI * controller's IOMMU. */ +struct pci_iommu_arena { + unsigned long *map; + unsigned int hint; + unsigned int limit; +}; + struct pci_iommu { /* This protects the controller's IOMMU and all * streaming buffers underneath. */ spinlock_t lock; + struct pci_iommu_arena arena; + /* IOMMU page table, a linear array of ioptes. */ iopte_t *page_table; /* The page table itself. */ - int page_table_sz_bits; /* log2 of ow many pages does it map? */ /* Base PCI memory space address where IOMMU mappings * begin. @@ -62,12 +66,6 @@ struct pci_iommu { */ unsigned long write_complete_reg; - /* The lowest used consistent mapping entry. Since - * we allocate consistent maps out of cluster 0 this - * is relative to the beginning of closter 0. - */ - u32 lowest_consistent_map; - /* In order to deal with some buggy third-party PCI bridges that * do wrong prefetching, we never mark valid mappings as invalid. * Instead we point them at this dummy page. @@ -75,16 +73,6 @@ struct pci_iommu { unsigned long dummy_page; unsigned long dummy_page_pa; - /* If PBM_NCLUSTERS is ever decreased to 4 or lower, - * or if largest supported page_table_sz * 8K goes above - * 2GB, you must increase the size of the type of - * these counters. You have been duly warned. -DaveM - */ - struct { - u16 next; - u16 flush; - } alloc_info[PBM_NCLUSTERS]; - /* CTX allocation. */ unsigned long ctx_lowest_free; unsigned long ctx_bitmap[IOMMU_NUM_CTXS / (sizeof(unsigned long) * 8)]; -- cgit v1.2.3-70-g09d2 From b4d1b825785847cddee6d104113da913f2ca8efb Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Fri, 14 Oct 2005 15:26:08 -0700 Subject: [SPARC64]: Fix powering off on SMP. Doing a "SUNW,stop-self" firmware call on the other cpus is not the correct thing to do when dropping into the firmware for a halt, reboot, or power-off. For now, just do nothing to quiet the other cpus, as the system should be quiescent enough. Later we may decide to implement smp_send_stop() like the other SMP platforms do. Based upon a report from Christopher Zimmermann. Signed-off-by: David S. Miller --- arch/sparc64/kernel/smp.c | 7 ------- arch/sparc64/mm/ultra.S | 16 ---------------- arch/sparc64/prom/misc.c | 12 ------------ 3 files changed, 35 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index 590df5a16f5..b137fd63f5e 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -1001,13 +1001,6 @@ void smp_penguin_jailcell(int irq, struct pt_regs *regs) preempt_enable(); } -extern unsigned long xcall_promstop; - -void smp_promstop_others(void) -{ - smp_cross_call(&xcall_promstop, 0, 0, 0); -} - #define prof_multiplier(__cpu) cpu_data(__cpu).multiplier #define prof_counter(__cpu) cpu_data(__cpu).counter diff --git a/arch/sparc64/mm/ultra.S b/arch/sparc64/mm/ultra.S index 058b8126c1a..e4c9151fa11 100644 --- a/arch/sparc64/mm/ultra.S +++ b/arch/sparc64/mm/ultra.S @@ -453,22 +453,6 @@ xcall_flush_dcache_page_spitfire: /* %g1 == physical page address nop nop - .globl xcall_promstop -xcall_promstop: - rdpr %pstate, %g2 - wrpr %g2, PSTATE_IG | PSTATE_AG, %pstate - rdpr %pil, %g2 - wrpr %g0, 15, %pil - sethi %hi(109f), %g7 - b,pt %xcc, etrap_irq -109: or %g7, %lo(109b), %g7 - flushw - call prom_stopself - nop - /* We should not return, just spin if we do... */ -1: b,a,pt %xcc, 1b - nop - .data errata32_hwbug: diff --git a/arch/sparc64/prom/misc.c b/arch/sparc64/prom/misc.c index 9b895faf077..87f5cfce23b 100644 --- a/arch/sparc64/prom/misc.c +++ b/arch/sparc64/prom/misc.c @@ -68,19 +68,11 @@ void prom_cmdline(void) local_irq_restore(flags); } -#ifdef CONFIG_SMP -extern void smp_promstop_others(void); -#endif - /* Drop into the prom, but completely terminate the program. * No chance of continuing. */ void prom_halt(void) { -#ifdef CONFIG_SMP - smp_promstop_others(); - udelay(8000); -#endif again: p1275_cmd("exit", P1275_INOUT(0, 0)); goto again; /* PROM is out to get me -DaveM */ @@ -88,10 +80,6 @@ again: void prom_halt_power_off(void) { -#ifdef CONFIG_SMP - smp_promstop_others(); - udelay(8000); -#endif p1275_cmd("SUNW,power-off", P1275_INOUT(0, 0)); /* if nothing else helps, we just halt */ -- cgit v1.2.3-70-g09d2 From 404351e67a9facb475abf1492245374a28d13e90 Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Sat, 29 Oct 2005 18:16:04 -0700 Subject: [PATCH] mm: mm_init set_mm_counters How is anon_rss initialized? In dup_mmap, and by mm_alloc's memset; but that's not so good if an mm_counter_t is a special type. And how is rss initialized? By set_mm_counter, all over the place. Come on, we just need to initialize them both at once by set_mm_counter in mm_init (which follows the memcpy when forking). Signed-off-by: Hugh Dickins Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/mips/kernel/irixelf.c | 1 - arch/sparc64/kernel/binfmt_aout32.c | 1 - arch/x86_64/ia32/ia32_aout.c | 1 - fs/binfmt_aout.c | 1 - fs/binfmt_elf.c | 1 - fs/binfmt_elf_fdpic.c | 7 ------- fs/binfmt_flat.c | 1 - fs/binfmt_som.c | 1 - kernel/fork.c | 4 ++-- 9 files changed, 2 insertions(+), 16 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/mips/kernel/irixelf.c b/arch/mips/kernel/irixelf.c index 99262fe6456..7ce34d4aa22 100644 --- a/arch/mips/kernel/irixelf.c +++ b/arch/mips/kernel/irixelf.c @@ -697,7 +697,6 @@ static int load_irix_binary(struct linux_binprm * bprm, struct pt_regs * regs) /* Do this so that we can load the interpreter, if need be. We will * change some of these later. */ - set_mm_counter(current->mm, rss, 0); setup_arg_pages(bprm, STACK_TOP, EXSTACK_DEFAULT); current->mm->start_stack = bprm->p; diff --git a/arch/sparc64/kernel/binfmt_aout32.c b/arch/sparc64/kernel/binfmt_aout32.c index b2854ef221d..edf52d06b28 100644 --- a/arch/sparc64/kernel/binfmt_aout32.c +++ b/arch/sparc64/kernel/binfmt_aout32.c @@ -241,7 +241,6 @@ static int load_aout32_binary(struct linux_binprm * bprm, struct pt_regs * regs) current->mm->brk = ex.a_bss + (current->mm->start_brk = N_BSSADDR(ex)); - set_mm_counter(current->mm, rss, 0); current->mm->mmap = NULL; compute_creds(bprm); current->flags &= ~PF_FORKNOEXEC; diff --git a/arch/x86_64/ia32/ia32_aout.c b/arch/x86_64/ia32/ia32_aout.c index 3e6780fa018..93c60f4aa47 100644 --- a/arch/x86_64/ia32/ia32_aout.c +++ b/arch/x86_64/ia32/ia32_aout.c @@ -314,7 +314,6 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs) current->mm->free_area_cache = TASK_UNMAPPED_BASE; current->mm->cached_hole_size = 0; - set_mm_counter(current->mm, rss, 0); current->mm->mmap = NULL; compute_creds(bprm); current->flags &= ~PF_FORKNOEXEC; diff --git a/fs/binfmt_aout.c b/fs/binfmt_aout.c index dd9baabaf01..72011826f0c 100644 --- a/fs/binfmt_aout.c +++ b/fs/binfmt_aout.c @@ -318,7 +318,6 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs) current->mm->free_area_cache = current->mm->mmap_base; current->mm->cached_hole_size = 0; - set_mm_counter(current->mm, rss, 0); current->mm->mmap = NULL; compute_creds(bprm); current->flags &= ~PF_FORKNOEXEC; diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c index d4b15576e58..918ccc267e4 100644 --- a/fs/binfmt_elf.c +++ b/fs/binfmt_elf.c @@ -773,7 +773,6 @@ static int load_elf_binary(struct linux_binprm * bprm, struct pt_regs * regs) /* Do this so that we can load the interpreter, if need be. We will change some of these later */ - set_mm_counter(current->mm, rss, 0); current->mm->free_area_cache = current->mm->mmap_base; current->mm->cached_hole_size = 0; retval = setup_arg_pages(bprm, randomize_stack_top(STACK_TOP), diff --git a/fs/binfmt_elf_fdpic.c b/fs/binfmt_elf_fdpic.c index 134c9c0d1f5..dda87c4c82a 100644 --- a/fs/binfmt_elf_fdpic.c +++ b/fs/binfmt_elf_fdpic.c @@ -294,14 +294,7 @@ static int load_elf_fdpic_binary(struct linux_binprm *bprm, struct pt_regs *regs &interp_params, ¤t->mm->start_stack, ¤t->mm->start_brk); -#endif - - /* do this so that we can load the interpreter, if need be - * - we will change some of these later - */ - set_mm_counter(current->mm, rss, 0); -#ifdef CONFIG_MMU retval = setup_arg_pages(bprm, current->mm->start_stack, executable_stack); if (retval < 0) { send_sig(SIGKILL, current, 0); diff --git a/fs/binfmt_flat.c b/fs/binfmt_flat.c index 7974efa107b..9d6625829b9 100644 --- a/fs/binfmt_flat.c +++ b/fs/binfmt_flat.c @@ -650,7 +650,6 @@ static int load_flat_file(struct linux_binprm * bprm, current->mm->start_brk = datapos + data_len + bss_len; current->mm->brk = (current->mm->start_brk + 3) & ~3; current->mm->context.end_brk = memp + ksize((void *) memp) - stack_len; - set_mm_counter(current->mm, rss, 0); } if (flags & FLAT_FLAG_KTRACE) diff --git a/fs/binfmt_som.c b/fs/binfmt_som.c index 227a2682d2b..00a91dc25d1 100644 --- a/fs/binfmt_som.c +++ b/fs/binfmt_som.c @@ -259,7 +259,6 @@ load_som_binary(struct linux_binprm * bprm, struct pt_regs * regs) create_som_tables(bprm); current->mm->start_stack = bprm->p; - set_mm_counter(current->mm, rss, 0); #if 0 printk("(start_brk) %08lx\n" , (unsigned long) current->mm->start_brk); diff --git a/kernel/fork.c b/kernel/fork.c index e2ff11f8c1b..25caa02e2ea 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -198,8 +198,6 @@ static inline int dup_mmap(struct mm_struct * mm, struct mm_struct * oldmm) mm->free_area_cache = oldmm->mmap_base; mm->cached_hole_size = ~0UL; mm->map_count = 0; - set_mm_counter(mm, rss, 0); - set_mm_counter(mm, anon_rss, 0); cpus_clear(mm->cpu_vm_mask); mm->mm_rb = RB_ROOT; rb_link = &mm->mm_rb.rb_node; @@ -323,6 +321,8 @@ static struct mm_struct * mm_init(struct mm_struct * mm) INIT_LIST_HEAD(&mm->mmlist); mm->core_waiters = 0; mm->nr_ptes = 0; + set_mm_counter(mm, rss, 0); + set_mm_counter(mm, anon_rss, 0); spin_lock_init(&mm->page_table_lock); rwlock_init(&mm->ioctx_list_lock); mm->ioctx_list = NULL; -- cgit v1.2.3-70-g09d2 From 9c0cbd54ce0397017a823484f9a8054ab369b8a2 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Sun, 30 Oct 2005 15:01:41 -0800 Subject: [PATCH] TIOC* compat ioctl handling TIOCSTART and TIOCSTOP are defined in asm/ioctls.h and asm/termios.h by various architectures but not actually implemented anywhere but in the IRIX compatibility layer, so remove their COMPATIBLE_IOCTL from parisc, ppc64 and sparc64. Move the TIOCSLTC COMPATIBLE_IOCTL to common code, guided by an ifdef to only show up on architectures that support it (same as the code handling it in tty_ioctl.c), aswell as it's brother TIOCGLTC that wasn't handled so far. Signed-off-by: Christoph Hellwig Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/parisc/kernel/ioctl32.c | 5 ----- arch/ppc64/kernel/ioctl32.c | 4 +--- arch/sparc64/kernel/ioctl32.c | 3 --- fs/compat_ioctl.c | 4 ++++ 4 files changed, 5 insertions(+), 11 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/parisc/kernel/ioctl32.c b/arch/parisc/kernel/ioctl32.c index 8cad8f004f0..0a331104ad5 100644 --- a/arch/parisc/kernel/ioctl32.c +++ b/arch/parisc/kernel/ioctl32.c @@ -561,11 +561,6 @@ IOCTL_TABLE_START #define DECLARES #include "compat_ioctl.c" -/* Might be moved to compat_ioctl.h with some ifdefs... */ -COMPATIBLE_IOCTL(TIOCSTART) -COMPATIBLE_IOCTL(TIOCSTOP) -COMPATIBLE_IOCTL(TIOCSLTC) - /* PA-specific ioctls */ COMPATIBLE_IOCTL(PA_PERF_ON) COMPATIBLE_IOCTL(PA_PERF_OFF) diff --git a/arch/ppc64/kernel/ioctl32.c b/arch/ppc64/kernel/ioctl32.c index a8005db23ec..ba4a899045c 100644 --- a/arch/ppc64/kernel/ioctl32.c +++ b/arch/ppc64/kernel/ioctl32.c @@ -39,9 +39,7 @@ IOCTL_TABLE_START #include #define DECLARES #include "compat_ioctl.c" -COMPATIBLE_IOCTL(TIOCSTART) -COMPATIBLE_IOCTL(TIOCSTOP) -COMPATIBLE_IOCTL(TIOCSLTC) + /* Little p (/dev/rtc, /dev/envctrl, etc.) */ COMPATIBLE_IOCTL(_IOR('p', 20, int[7])) /* RTCGET */ COMPATIBLE_IOCTL(_IOW('p', 21, int[7])) /* RTCSET */ diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 43fc3173d48..e6a00325075 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -475,9 +475,6 @@ IOCTL_TABLE_START #include #define DECLARES #include "compat_ioctl.c" -COMPATIBLE_IOCTL(TIOCSTART) -COMPATIBLE_IOCTL(TIOCSTOP) -COMPATIBLE_IOCTL(TIOCSLTC) COMPATIBLE_IOCTL(FBIOGTYPE) COMPATIBLE_IOCTL(FBIOSATTR) COMPATIBLE_IOCTL(FBIOGATTR) diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c index a327e03753a..43dbcb0b21e 100644 --- a/fs/compat_ioctl.c +++ b/fs/compat_ioctl.c @@ -3046,6 +3046,10 @@ HANDLE_IOCTL(RAW_GETBIND, raw_ioctl) /* Serial */ HANDLE_IOCTL(TIOCGSERIAL, serial_struct_ioctl) HANDLE_IOCTL(TIOCSSERIAL, serial_struct_ioctl) +#ifdef TIOCGLTC +COMPATIBLE_IOCTL(TIOCGLTC) +COMPATIBLE_IOCTL(TIOCSLTC) +#endif /* Usbdevfs */ HANDLE_IOCTL(USBDEVFS_CONTROL32, do_usbdevfs_control) HANDLE_IOCTL(USBDEVFS_BULK32, do_usbdevfs_bulk) -- cgit v1.2.3-70-g09d2 From ecea8d19c9f0ebd62ddaa07fc919ff4e4b820d99 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sun, 30 Oct 2005 15:03:00 -0800 Subject: [PATCH] jiffies_64 cleanup Define jiffies_64 in kernel/timer.c rather than having 24 duplicated defines in each architecture. Signed-off-by: Thomas Gleixner Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/alpha/kernel/time.c | 4 ---- arch/arm/kernel/time.c | 4 ---- arch/arm26/kernel/time.c | 4 ---- arch/cris/kernel/time.c | 4 ---- arch/frv/kernel/time.c | 3 --- arch/h8300/kernel/time.c | 4 ---- arch/i386/kernel/time.c | 4 ---- arch/ia64/kernel/time.c | 4 ---- arch/m32r/kernel/time.c | 4 ---- arch/m68k/kernel/time.c | 4 ---- arch/m68knommu/kernel/time.c | 4 ---- arch/mips/kernel/time.c | 4 ---- arch/parisc/kernel/time.c | 4 ---- arch/ppc/kernel/time.c | 5 ----- arch/ppc64/kernel/time.c | 4 ---- arch/s390/kernel/time.c | 4 ---- arch/sh/kernel/time.c | 4 ---- arch/sh64/kernel/time.c | 2 -- arch/sparc/kernel/time.c | 4 ---- arch/sparc64/kernel/time.c | 4 ---- arch/um/kernel/time_kern.c | 4 ---- arch/v850/kernel/time.c | 4 ---- arch/x86_64/kernel/time.c | 4 ---- arch/xtensa/kernel/time.c | 3 --- kernel/timer.c | 4 ++++ 25 files changed, 4 insertions(+), 93 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/alpha/kernel/time.c b/arch/alpha/kernel/time.c index 67be50b7d80..6b2921be190 100644 --- a/arch/alpha/kernel/time.c +++ b/arch/alpha/kernel/time.c @@ -55,10 +55,6 @@ #include "proto.h" #include "irq_impl.h" -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - extern unsigned long wall_jiffies; /* kernel/timer.c */ static int set_rtc_mmss(unsigned long); diff --git a/arch/arm/kernel/time.c b/arch/arm/kernel/time.c index 69449a818dc..fc4729106a3 100644 --- a/arch/arm/kernel/time.c +++ b/arch/arm/kernel/time.c @@ -36,10 +36,6 @@ #include #include -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - /* * Our system timer. */ diff --git a/arch/arm26/kernel/time.c b/arch/arm26/kernel/time.c index e66aedd02fa..335525339ad 100644 --- a/arch/arm26/kernel/time.c +++ b/arch/arm26/kernel/time.c @@ -34,10 +34,6 @@ #include #include -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - extern unsigned long wall_jiffies; /* this needs a better home */ diff --git a/arch/cris/kernel/time.c b/arch/cris/kernel/time.c index a2d99b4aedc..b863815de78 100644 --- a/arch/cris/kernel/time.c +++ b/arch/cris/kernel/time.c @@ -32,10 +32,6 @@ #include #include -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - int have_rtc; /* used to remember if we have an RTC or not */; #define TICK_SIZE tick diff --git a/arch/frv/kernel/time.c b/arch/frv/kernel/time.c index f43b734482e..2e9741227b7 100644 --- a/arch/frv/kernel/time.c +++ b/arch/frv/kernel/time.c @@ -34,9 +34,6 @@ extern unsigned long wall_jiffies; -u64 jiffies_64 = INITIAL_JIFFIES; -EXPORT_SYMBOL(jiffies_64); - unsigned long __nongprelbss __clkin_clock_speed_HZ; unsigned long __nongprelbss __ext_bus_clock_speed_HZ; unsigned long __nongprelbss __res_bus_clock_speed_HZ; diff --git a/arch/h8300/kernel/time.c b/arch/h8300/kernel/time.c index af8c5d2057d..688a5100604 100644 --- a/arch/h8300/kernel/time.c +++ b/arch/h8300/kernel/time.c @@ -32,10 +32,6 @@ #define TICK_SIZE (tick_nsec / 1000) -u64 jiffies_64; - -EXPORT_SYMBOL(jiffies_64); - /* * timer_interrupt() needs to keep up the real-time clock, * as well as call the "do_timer()" routine every clocktick diff --git a/arch/i386/kernel/time.c b/arch/i386/kernel/time.c index 46c35ec9137..07471bba2dc 100644 --- a/arch/i386/kernel/time.c +++ b/arch/i386/kernel/time.c @@ -74,10 +74,6 @@ int pit_latch_buggy; /* extern */ #include "do_timer.h" -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - unsigned int cpu_khz; /* Detected as we calibrate the TSC */ EXPORT_SYMBOL(cpu_khz); diff --git a/arch/ia64/kernel/time.c b/arch/ia64/kernel/time.c index 8b8a5a45b62..5b7e736f3b4 100644 --- a/arch/ia64/kernel/time.c +++ b/arch/ia64/kernel/time.c @@ -32,10 +32,6 @@ extern unsigned long wall_jiffies; -u64 jiffies_64 __cacheline_aligned_in_smp = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - #define TIME_KEEPER_ID 0 /* smp_processor_id() of time-keeper */ #ifdef CONFIG_IA64_DEBUG_IRQ diff --git a/arch/m32r/kernel/time.c b/arch/m32r/kernel/time.c index 539c562cd54..2ebce2063fe 100644 --- a/arch/m32r/kernel/time.c +++ b/arch/m32r/kernel/time.c @@ -39,10 +39,6 @@ extern void send_IPI_allbutself(int, int); extern void smp_local_timer_interrupt(struct pt_regs *); #endif -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - extern unsigned long wall_jiffies; #define TICK_SIZE (tick_nsec / 1000) diff --git a/arch/m68k/kernel/time.c b/arch/m68k/kernel/time.c index 4ec95e3cb87..98e4b1adfa2 100644 --- a/arch/m68k/kernel/time.c +++ b/arch/m68k/kernel/time.c @@ -27,10 +27,6 @@ #include #include -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - static inline int set_rtc_mmss(unsigned long nowtime) { if (mach_set_clock_mmss) diff --git a/arch/m68knommu/kernel/time.c b/arch/m68knommu/kernel/time.c index b17c1ecba96..b9d8abb4543 100644 --- a/arch/m68knommu/kernel/time.c +++ b/arch/m68knommu/kernel/time.c @@ -27,10 +27,6 @@ #define TICK_SIZE (tick_nsec / 1000) -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - extern unsigned long wall_jiffies; diff --git a/arch/mips/kernel/time.c b/arch/mips/kernel/time.c index a24651dfaab..787ed541d44 100644 --- a/arch/mips/kernel/time.c +++ b/arch/mips/kernel/time.c @@ -45,10 +45,6 @@ #define TICK_SIZE (tick_nsec / 1000) -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - /* * forward reference */ diff --git a/arch/parisc/kernel/time.c b/arch/parisc/kernel/time.c index bc979e1abde..cded2568078 100644 --- a/arch/parisc/kernel/time.c +++ b/arch/parisc/kernel/time.c @@ -33,10 +33,6 @@ #include -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - /* xtime and wall_jiffies keep wall-clock time */ extern unsigned long wall_jiffies; diff --git a/arch/ppc/kernel/time.c b/arch/ppc/kernel/time.c index 22d7fd1e0ae..67797184f4e 100644 --- a/arch/ppc/kernel/time.c +++ b/arch/ppc/kernel/time.c @@ -66,11 +66,6 @@ #include -/* XXX false sharing with below? */ -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - unsigned long disarm_decr[NR_CPUS]; extern struct timezone sys_tz; diff --git a/arch/ppc64/kernel/time.c b/arch/ppc64/kernel/time.c index b56c6a324e1..ab565468578 100644 --- a/arch/ppc64/kernel/time.c +++ b/arch/ppc64/kernel/time.c @@ -68,10 +68,6 @@ #include #include -u64 jiffies_64 __cacheline_aligned_in_smp = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - /* keep track of when we need to update the rtc */ time_t last_rtc_update; extern int piranha_simulator; diff --git a/arch/s390/kernel/time.c b/arch/s390/kernel/time.c index 2fd75da1549..9a1d95894f3 100644 --- a/arch/s390/kernel/time.c +++ b/arch/s390/kernel/time.c @@ -49,10 +49,6 @@ #define TICK_SIZE tick -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - static ext_int_info_t ext_int_info_cc; static u64 init_timer_cc; static u64 jiffies_timer_cc; diff --git a/arch/sh/kernel/time.c b/arch/sh/kernel/time.c index 02ca69918d7..671b876416b 100644 --- a/arch/sh/kernel/time.c +++ b/arch/sh/kernel/time.c @@ -56,10 +56,6 @@ extern unsigned long wall_jiffies; #define TICK_SIZE (tick_nsec / 1000) DEFINE_SPINLOCK(tmu0_lock); -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - /* XXX: Can we initialize this in a routine somewhere? Dreamcast doesn't want * these routines anywhere... */ #ifdef CONFIG_SH_RTC diff --git a/arch/sh64/kernel/time.c b/arch/sh64/kernel/time.c index 43e395a14f4..870fe5327e0 100644 --- a/arch/sh64/kernel/time.c +++ b/arch/sh64/kernel/time.c @@ -116,8 +116,6 @@ extern unsigned long wall_jiffies; -u64 jiffies_64 = INITIAL_JIFFIES; - static unsigned long tmu_base, rtc_base; unsigned long cprc_base; diff --git a/arch/sparc/kernel/time.c b/arch/sparc/kernel/time.c index 279a62627c1..24814d58f9e 100644 --- a/arch/sparc/kernel/time.c +++ b/arch/sparc/kernel/time.c @@ -45,10 +45,6 @@ extern unsigned long wall_jiffies; -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - DEFINE_SPINLOCK(rtc_lock); enum sparc_clock_type sp_clock_typ; DEFINE_SPINLOCK(mostek_lock); diff --git a/arch/sparc64/kernel/time.c b/arch/sparc64/kernel/time.c index 3f08a32f51a..38c5525087a 100644 --- a/arch/sparc64/kernel/time.c +++ b/arch/sparc64/kernel/time.c @@ -55,10 +55,6 @@ unsigned long ds1287_regs = 0UL; extern unsigned long wall_jiffies; -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - static void __iomem *mstk48t08_regs; static void __iomem *mstk48t59_regs; diff --git a/arch/um/kernel/time_kern.c b/arch/um/kernel/time_kern.c index 4e08f7545d6..020ca79b8d3 100644 --- a/arch/um/kernel/time_kern.c +++ b/arch/um/kernel/time_kern.c @@ -22,10 +22,6 @@ #include "mode.h" #include "os.h" -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - int hz(void) { return(HZ); diff --git a/arch/v850/kernel/time.c b/arch/v850/kernel/time.c index ea3fd8844ff..c1e85c2aef6 100644 --- a/arch/v850/kernel/time.c +++ b/arch/v850/kernel/time.c @@ -26,10 +26,6 @@ #include "mach.h" -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - #define TICK_SIZE (tick_nsec / 1000) /* diff --git a/arch/x86_64/kernel/time.c b/arch/x86_64/kernel/time.c index 47d25ad0816..bd5ea09bfdf 100644 --- a/arch/x86_64/kernel/time.c +++ b/arch/x86_64/kernel/time.c @@ -42,10 +42,6 @@ #include #endif -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - #ifdef CONFIG_CPU_FREQ static void cpufreq_delayed_get(void); #endif diff --git a/arch/xtensa/kernel/time.c b/arch/xtensa/kernel/time.c index 8e423d1335c..cb6e38ed2b1 100644 --- a/arch/xtensa/kernel/time.c +++ b/arch/xtensa/kernel/time.c @@ -29,9 +29,6 @@ extern volatile unsigned long wall_jiffies; -u64 jiffies_64 = INITIAL_JIFFIES; -EXPORT_SYMBOL(jiffies_64); - spinlock_t rtc_lock = SPIN_LOCK_UNLOCKED; EXPORT_SYMBOL(rtc_lock); diff --git a/kernel/timer.c b/kernel/timer.c index 562d53eabb4..fd74268d866 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -46,6 +46,10 @@ static void time_interpolator_update(long delta_nsec); #define time_interpolator_update(x) #endif +u64 jiffies_64 __cacheline_aligned_in_smp = INITIAL_JIFFIES; + +EXPORT_SYMBOL(jiffies_64); + /* * per-CPU timer vector definitions: */ -- cgit v1.2.3-70-g09d2 From 66ff2d0691e00e1e7bfdf398a970310c9a0fe671 Mon Sep 17 00:00:00 2001 From: Ananth N Mavinakayanahalli Date: Mon, 7 Nov 2005 01:00:07 -0800 Subject: [PATCH] Kprobes: rearrange preempt_disable/enable() calls The following set of patches are aimed at improving kprobes scalability. We currently serialize kprobe registration, unregistration and handler execution using a single spinlock - kprobe_lock. With these changes, kprobe handlers can run without any locks held. It also allows for simultaneous kprobe handler executions on different processors as we now track kprobe execution on a per processor basis. It is now necessary that the handlers be re-entrant since handlers can run concurrently on multiple processors. All changes have been tested on i386, ia64, ppc64 and x86_64, while sparc64 has been compile tested only. The patches can be viewed as 3 logical chunks: patch 1: Reorder preempt_(dis/en)able calls patches 2-7: Introduce per_cpu data areas to track kprobe execution patches 8-9: Use RCU to synchronize kprobe (un)registration and handler execution. Thanks to Maneesh Soni, James Keniston and Anil Keshavamurthy for their review and suggestions. Thanks again to Anil, Hien Nguyen and Kevin Stafford for testing the patches. This patch: Reorder preempt_disable/enable() calls in arch kprobes files in preparation to introduce locking changes. No functional changes introduced by this patch. Signed-off-by: Ananth N Mavinakayahanalli Signed-off-by: Anil S Keshavamurthy Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/i386/kernel/kprobes.c | 35 ++++++++++++++++++----------------- arch/ia64/kernel/kprobes.c | 22 ++++++++++++++-------- arch/ppc64/kernel/kprobes.c | 11 ++++++----- arch/sparc64/kernel/kprobes.c | 25 +++++++++++++------------ arch/x86_64/kernel/kprobes.c | 28 ++++++++++++++-------------- 5 files changed, 65 insertions(+), 56 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/i386/kernel/kprobes.c b/arch/i386/kernel/kprobes.c index 6345b430b10..fd35039859e 100644 --- a/arch/i386/kernel/kprobes.c +++ b/arch/i386/kernel/kprobes.c @@ -158,8 +158,6 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) kprobe_opcode_t *addr = NULL; unsigned long *lp; - /* We're in an interrupt, but this is clear and BUG()-safe. */ - preempt_disable(); /* Check if the application is using LDT entry for its code segment and * calculate the address by reading the base address from the LDT entry. */ @@ -232,6 +230,11 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) goto no_kprobe; } + /* + * This preempt_disable() matches the preempt_enable_no_resched() + * in post_kprobe_handler() + */ + preempt_disable(); kprobe_status = KPROBE_HIT_ACTIVE; set_current_kprobe(p, regs); @@ -245,7 +248,6 @@ ss_probe: return 1; no_kprobe: - preempt_enable_no_resched(); return ret; } @@ -313,11 +315,11 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) unlock_kprobes(); preempt_enable_no_resched(); - /* - * By returning a non-zero value, we are telling - * kprobe_handler() that we have handled unlocking - * and re-enabling preemption. - */ + /* + * By returning a non-zero value, we are telling + * kprobe_handler() that we have handled unlocking + * and re-enabling preemption + */ return 1; } @@ -453,29 +455,29 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, unsigned long val, void *data) { struct die_args *args = (struct die_args *)data; + int ret = NOTIFY_DONE; + + preempt_disable(); switch (val) { case DIE_INT3: if (kprobe_handler(args->regs)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; case DIE_DEBUG: if (post_kprobe_handler(args->regs)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; case DIE_GPF: - if (kprobe_running() && - kprobe_fault_handler(args->regs, args->trapnr)) - return NOTIFY_STOP; - break; case DIE_PAGE_FAULT: if (kprobe_running() && kprobe_fault_handler(args->regs, args->trapnr)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; default: break; } - return NOTIFY_DONE; + preempt_enable(); + return ret; } int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) @@ -502,7 +504,6 @@ int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) void __kprobes jprobe_return(void) { - preempt_enable_no_resched(); asm volatile (" xchgl %%ebx,%%esp \n" " int3 \n" " .globl jprobe_return_end \n" diff --git a/arch/ia64/kernel/kprobes.c b/arch/ia64/kernel/kprobes.c index 471086b808a..1e80ec80dd2 100644 --- a/arch/ia64/kernel/kprobes.c +++ b/arch/ia64/kernel/kprobes.c @@ -395,7 +395,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) /* * By returning a non-zero value, we are telling * kprobe_handler() that we have handled unlocking - * and re-enabling preemption. + * and re-enabling preemption */ return 1; } @@ -607,8 +607,6 @@ static int __kprobes pre_kprobes_handler(struct die_args *args) struct pt_regs *regs = args->regs; kprobe_opcode_t *addr = (kprobe_opcode_t *)instruction_pointer(regs); - preempt_disable(); - /* Handle recursion cases */ if (kprobe_running()) { p = get_kprobe(addr); @@ -665,6 +663,11 @@ static int __kprobes pre_kprobes_handler(struct die_args *args) goto no_kprobe; } + /* + * This preempt_disable() matches the preempt_enable_no_resched() + * in post_kprobes_handler() + */ + preempt_disable(); kprobe_status = KPROBE_HIT_ACTIVE; set_current_kprobe(p); @@ -682,7 +685,6 @@ ss_probe: return 1; no_kprobe: - preempt_enable_no_resched(); return ret; } @@ -733,22 +735,26 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, unsigned long val, void *data) { struct die_args *args = (struct die_args *)data; + int ret = NOTIFY_DONE; + + preempt_disable(); switch(val) { case DIE_BREAK: if (pre_kprobes_handler(args)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; case DIE_SS: if (post_kprobes_handler(args->regs)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; case DIE_PAGE_FAULT: if (kprobes_fault_handler(args->regs, args->trapnr)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; default: break; } - return NOTIFY_DONE; + preempt_enable(); + return ret; } int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) diff --git a/arch/ppc64/kernel/kprobes.c b/arch/ppc64/kernel/kprobes.c index ed876a5178a..6071ee99f5c 100644 --- a/arch/ppc64/kernel/kprobes.c +++ b/arch/ppc64/kernel/kprobes.c @@ -209,6 +209,11 @@ static inline int kprobe_handler(struct pt_regs *regs) goto no_kprobe; } + /* + * This preempt_disable() matches the preempt_enable_no_resched() + * in post_kprobe_handler(). + */ + preempt_disable(); kprobe_status = KPROBE_HIT_ACTIVE; current_kprobe = p; kprobe_saved_msr = regs->msr; @@ -219,11 +224,6 @@ static inline int kprobe_handler(struct pt_regs *regs) ss_probe: prepare_singlestep(p, regs); kprobe_status = KPROBE_HIT_SS; - /* - * This preempt_disable() matches the preempt_enable_no_resched() - * in post_kprobe_handler(). - */ - preempt_disable(); return 1; no_kprobe: @@ -293,6 +293,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) regs->nip = orig_ret_address; unlock_kprobes(); + preempt_enable_no_resched(); /* * By returning a non-zero value, we are telling diff --git a/arch/sparc64/kernel/kprobes.c b/arch/sparc64/kernel/kprobes.c index 0d66d07c8c6..755a0d7d887 100644 --- a/arch/sparc64/kernel/kprobes.c +++ b/arch/sparc64/kernel/kprobes.c @@ -118,8 +118,6 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) void *addr = (void *) regs->tpc; int ret = 0; - preempt_disable(); - if (kprobe_running()) { /* We *are* holding lock here, so this is safe. * Disarm the probe we just hit, and ignore it. @@ -171,6 +169,11 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) goto no_kprobe; } + /* + * This preempt_disable() matches the preempt_enable_no_resched() + * in post_kprobes_handler() + */ + preempt_disable(); set_current_kprobe(p, regs); kprobe_status = KPROBE_HIT_ACTIVE; if (p->pre_handler && p->pre_handler(p, regs)) @@ -182,7 +185,6 @@ ss_probe: return 1; no_kprobe: - preempt_enable_no_resched(); return ret; } @@ -322,29 +324,29 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, unsigned long val, void *data) { struct die_args *args = (struct die_args *)data; + int ret = NOTIFY_DONE; + + preempt_disable(); switch (val) { case DIE_DEBUG: if (kprobe_handler(args->regs)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; case DIE_DEBUG_2: if (post_kprobe_handler(args->regs)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; case DIE_GPF: - if (kprobe_running() && - kprobe_fault_handler(args->regs, args->trapnr)) - return NOTIFY_STOP; - break; case DIE_PAGE_FAULT: if (kprobe_running() && kprobe_fault_handler(args->regs, args->trapnr)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; default: break; } - return NOTIFY_DONE; + preempt_enable(); + return ret; } asmlinkage void __kprobes kprobe_trap(unsigned long trap_level, @@ -396,7 +398,6 @@ int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) void __kprobes jprobe_return(void) { - preempt_enable_no_resched(); __asm__ __volatile__( ".globl jprobe_return_trap_instruction\n" "jprobe_return_trap_instruction:\n\t" diff --git a/arch/x86_64/kernel/kprobes.c b/arch/x86_64/kernel/kprobes.c index 76a28b007be..ebfa2c9241c 100644 --- a/arch/x86_64/kernel/kprobes.c +++ b/arch/x86_64/kernel/kprobes.c @@ -302,9 +302,6 @@ int __kprobes kprobe_handler(struct pt_regs *regs) int ret = 0; kprobe_opcode_t *addr = (kprobe_opcode_t *)(regs->rip - sizeof(kprobe_opcode_t)); - /* We're in an interrupt, but this is clear and BUG()-safe. */ - preempt_disable(); - /* Check we're not actually recursing */ if (kprobe_running()) { /* We *are* holding lock here, so this is safe. @@ -372,6 +369,11 @@ int __kprobes kprobe_handler(struct pt_regs *regs) goto no_kprobe; } + /* + * This preempt_disable() matches the preempt_enable_no_resched() + * in post_kprobe_handler() + */ + preempt_disable(); kprobe_status = KPROBE_HIT_ACTIVE; set_current_kprobe(p, regs); @@ -385,7 +387,6 @@ ss_probe: return 1; no_kprobe: - preempt_enable_no_resched(); return ret; } @@ -456,7 +457,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) /* * By returning a non-zero value, we are telling * kprobe_handler() that we have handled unlocking - * and re-enabling preemption. + * and re-enabling preemption */ return 1; } @@ -599,29 +600,29 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, unsigned long val, void *data) { struct die_args *args = (struct die_args *)data; + int ret = NOTIFY_DONE; + + preempt_disable(); switch (val) { case DIE_INT3: if (kprobe_handler(args->regs)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; case DIE_DEBUG: if (post_kprobe_handler(args->regs)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; case DIE_GPF: - if (kprobe_running() && - kprobe_fault_handler(args->regs, args->trapnr)) - return NOTIFY_STOP; - break; case DIE_PAGE_FAULT: if (kprobe_running() && kprobe_fault_handler(args->regs, args->trapnr)) - return NOTIFY_STOP; + ret = NOTIFY_STOP; break; default: break; } - return NOTIFY_DONE; + preempt_enable(); + return ret; } int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) @@ -647,7 +648,6 @@ int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) void __kprobes jprobe_return(void) { - preempt_enable_no_resched(); asm volatile (" xchg %%rbx,%%rsp \n" " int3 \n" " .globl jprobe_return_end \n" -- cgit v1.2.3-70-g09d2 From f215d985e936cf493959b365a10593b6d5f80447 Mon Sep 17 00:00:00 2001 From: Ananth N Mavinakayanahalli Date: Mon, 7 Nov 2005 01:00:11 -0800 Subject: [PATCH] Kprobes: Track kprobe on a per_cpu basis - sparc64 changes Sparc64 changes to track kprobe execution on a per-cpu basis. We now track the kprobe state machine independently on each cpu using an arch specific kprobe control block. Signed-off-by: Ananth N Mavinakayanahalli Signed-off-by: Anil S Keshavamurthy Cc: "David S. Miller" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/sparc64/kernel/kprobes.c | 131 +++++++++++++++++++++--------------------- include/asm-sparc64/kprobes.h | 20 +++++++ 2 files changed, 87 insertions(+), 64 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/kprobes.c b/arch/sparc64/kernel/kprobes.c index 755a0d7d887..b95984154db 100644 --- a/arch/sparc64/kernel/kprobes.c +++ b/arch/sparc64/kernel/kprobes.c @@ -38,6 +38,9 @@ * - Mark that we are no longer actively in a kprobe. */ +DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL; +DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk); + int __kprobes arch_prepare_kprobe(struct kprobe *p) { return 0; @@ -66,46 +69,39 @@ void __kprobes arch_remove_kprobe(struct kprobe *p) { } -static struct kprobe *current_kprobe; -static unsigned long current_kprobe_orig_tnpc; -static unsigned long current_kprobe_orig_tstate_pil; -static unsigned int kprobe_status; -static struct kprobe *kprobe_prev; -static unsigned long kprobe_orig_tnpc_prev; -static unsigned long kprobe_orig_tstate_pil_prev; -static unsigned int kprobe_status_prev; - -static inline void save_previous_kprobe(void) +static inline void save_previous_kprobe(struct kprobe_ctlblk *kcb) { - kprobe_status_prev = kprobe_status; - kprobe_orig_tnpc_prev = current_kprobe_orig_tnpc; - kprobe_orig_tstate_pil_prev = current_kprobe_orig_tstate_pil; - kprobe_prev = current_kprobe; + kcb->prev_kprobe.kp = kprobe_running(); + kcb->prev_kprobe.status = kcb->kprobe_status; + kcb->prev_kprobe.orig_tnpc = kcb->kprobe_orig_tnpc; + kcb->prev_kprobe.orig_tstate_pil = kcb->kprobe_orig_tstate_pil; } -static inline void restore_previous_kprobe(void) +static inline void restore_previous_kprobe(struct kprobe_ctlblk *kcb) { - kprobe_status = kprobe_status_prev; - current_kprobe_orig_tnpc = kprobe_orig_tnpc_prev; - current_kprobe_orig_tstate_pil = kprobe_orig_tstate_pil_prev; - current_kprobe = kprobe_prev; + __get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp; + kcb->kprobe_status = kcb->prev_kprobe.status; + kcb->kprobe_orig_tnpc = kcb->prev_kprobe.orig_tnpc; + kcb->kprobe_orig_tstate_pil = kcb->prev_kprobe.orig_tstate_pil; } -static inline void set_current_kprobe(struct kprobe *p, struct pt_regs *regs) +static inline void set_current_kprobe(struct kprobe *p, struct pt_regs *regs, + struct kprobe_ctlblk *kcb) { - current_kprobe_orig_tnpc = regs->tnpc; - current_kprobe_orig_tstate_pil = (regs->tstate & TSTATE_PIL); - current_kprobe = p; + __get_cpu_var(current_kprobe) = p; + kcb->kprobe_orig_tnpc = regs->tnpc; + kcb->kprobe_orig_tstate_pil = (regs->tstate & TSTATE_PIL); } -static inline void prepare_singlestep(struct kprobe *p, struct pt_regs *regs) +static inline void prepare_singlestep(struct kprobe *p, struct pt_regs *regs, + struct kprobe_ctlblk *kcb) { regs->tstate |= TSTATE_PIL; /*single step inline, if it a breakpoint instruction*/ if (p->opcode == BREAKPOINT_INSTRUCTION) { regs->tpc = (unsigned long) p->addr; - regs->tnpc = current_kprobe_orig_tnpc; + regs->tnpc = kcb->kprobe_orig_tnpc; } else { regs->tpc = (unsigned long) &p->ainsn.insn[0]; regs->tnpc = (unsigned long) &p->ainsn.insn[1]; @@ -117,6 +113,7 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) struct kprobe *p; void *addr = (void *) regs->tpc; int ret = 0; + struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); if (kprobe_running()) { /* We *are* holding lock here, so this is safe. @@ -124,9 +121,9 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) */ p = get_kprobe(addr); if (p) { - if (kprobe_status == KPROBE_HIT_SS) { + if (kcb->kprobe_status == KPROBE_HIT_SS) { regs->tstate = ((regs->tstate & ~TSTATE_PIL) | - current_kprobe_orig_tstate_pil); + kcb->kprobe_orig_tstate_pil); unlock_kprobes(); goto no_kprobe; } @@ -136,14 +133,14 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) * just single step on the instruction of the new probe * without calling any user handlers. */ - save_previous_kprobe(); - set_current_kprobe(p, regs); + save_previous_kprobe(kcb); + set_current_kprobe(p, regs, kcb); p->nmissed++; - kprobe_status = KPROBE_REENTER; - prepare_singlestep(p, regs); + kcb->kprobe_status = KPROBE_REENTER; + prepare_singlestep(p, regs, kcb); return 1; } else { - p = current_kprobe; + p = __get_cpu_var(current_kprobe); if (p->break_handler && p->break_handler(p, regs)) goto ss_probe; } @@ -174,14 +171,14 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) * in post_kprobes_handler() */ preempt_disable(); - set_current_kprobe(p, regs); - kprobe_status = KPROBE_HIT_ACTIVE; + set_current_kprobe(p, regs, kcb); + kcb->kprobe_status = KPROBE_HIT_ACTIVE; if (p->pre_handler && p->pre_handler(p, regs)) return 1; ss_probe: - prepare_singlestep(p, regs); - kprobe_status = KPROBE_HIT_SS; + prepare_singlestep(p, regs, kcb); + kcb->kprobe_status = KPROBE_HIT_SS; return 1; no_kprobe: @@ -262,11 +259,12 @@ static void __kprobes retpc_fixup(struct pt_regs *regs, u32 insn, * This function prepares to return from the post-single-step * breakpoint trap. */ -static void __kprobes resume_execution(struct kprobe *p, struct pt_regs *regs) +static void __kprobes resume_execution(struct kprobe *p, + struct pt_regs *regs, struct kprobe_ctlblk *kcb) { u32 insn = p->ainsn.insn[0]; - regs->tpc = current_kprobe_orig_tnpc; + regs->tpc = kcb->kprobe_orig_tnpc; regs->tnpc = relbranch_fixup(insn, (unsigned long) p->addr, (unsigned long) &p->ainsn.insn[0], @@ -274,26 +272,30 @@ static void __kprobes resume_execution(struct kprobe *p, struct pt_regs *regs) retpc_fixup(regs, insn, (unsigned long) p->addr); regs->tstate = ((regs->tstate & ~TSTATE_PIL) | - current_kprobe_orig_tstate_pil); + kcb->kprobe_orig_tstate_pil); } static inline int post_kprobe_handler(struct pt_regs *regs) { - if (!kprobe_running()) + struct kprobe *cur = kprobe_running(); + struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); + + if (!cur) return 0; - if ((kprobe_status != KPROBE_REENTER) && current_kprobe->post_handler) { - kprobe_status = KPROBE_HIT_SSDONE; - current_kprobe->post_handler(current_kprobe, regs, 0); + if ((kcb->kprobe_status != KPROBE_REENTER) && cur->post_handler) { + kcb->kprobe_status = KPROBE_HIT_SSDONE; + cur->post_handler(cur, regs, 0); } - resume_execution(current_kprobe, regs); + resume_execution(cur, regs, kcb); /*Restore back the original saved kprobes variables and continue. */ - if (kprobe_status == KPROBE_REENTER) { - restore_previous_kprobe(); + if (kcb->kprobe_status == KPROBE_REENTER) { + restore_previous_kprobe(kcb); goto out; } + reset_current_kprobe(); unlock_kprobes(); out: preempt_enable_no_resched(); @@ -304,13 +306,16 @@ out: /* Interrupts disabled, kprobe_lock held. */ static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) { - if (current_kprobe->fault_handler - && current_kprobe->fault_handler(current_kprobe, regs, trapnr)) + struct kprobe *cur = kprobe_running(); + struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); + + if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr)) return 1; - if (kprobe_status & KPROBE_HIT_SS) { - resume_execution(current_kprobe, regs); + if (kcb->kprobe_status & KPROBE_HIT_SS) { + resume_execution(cur, regs, kcb); + reset_current_kprobe(); unlock_kprobes(); preempt_enable_no_resched(); } @@ -370,24 +375,21 @@ asmlinkage void __kprobes kprobe_trap(unsigned long trap_level, } /* Jprobes support. */ -static struct pt_regs jprobe_saved_regs; -static struct pt_regs *jprobe_saved_regs_location; -static struct sparc_stackf jprobe_saved_stack; - int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) { struct jprobe *jp = container_of(p, struct jprobe, kp); + struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); - jprobe_saved_regs_location = regs; - memcpy(&jprobe_saved_regs, regs, sizeof(*regs)); + kcb->jprobe_saved_regs_location = regs; + memcpy(&(kcb->jprobe_saved_regs), regs, sizeof(*regs)); /* Save a whole stack frame, this gets arguments * pushed onto the stack after using up all the * arg registers. */ - memcpy(&jprobe_saved_stack, + memcpy(&(kcb->jprobe_saved_stack), (char *) (regs->u_regs[UREG_FP] + STACK_BIAS), - sizeof(jprobe_saved_stack)); + sizeof(kcb->jprobe_saved_stack)); regs->tpc = (unsigned long) jp->entry; regs->tnpc = ((unsigned long) jp->entry) + 0x4UL; @@ -411,14 +413,15 @@ extern void __show_regs(struct pt_regs * regs); int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) { u32 *addr = (u32 *) regs->tpc; + struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); if (addr == (u32 *) jprobe_return_trap_instruction) { - if (jprobe_saved_regs_location != regs) { + if (kcb->jprobe_saved_regs_location != regs) { printk("JPROBE: Current regs (%p) does not match " "saved regs (%p).\n", - regs, jprobe_saved_regs_location); + regs, kcb->jprobe_saved_regs_location); printk("JPROBE: Saved registers\n"); - __show_regs(jprobe_saved_regs_location); + __show_regs(kcb->jprobe_saved_regs_location); printk("JPROBE: Current registers\n"); __show_regs(regs); BUG(); @@ -427,11 +430,11 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) * first so that UREG_FP is the original one for * the stack frame restore. */ - memcpy(regs, &jprobe_saved_regs, sizeof(*regs)); + memcpy(regs, &(kcb->jprobe_saved_regs), sizeof(*regs)); memcpy((char *) (regs->u_regs[UREG_FP] + STACK_BIAS), - &jprobe_saved_stack, - sizeof(jprobe_saved_stack)); + &(kcb->jprobe_saved_stack), + sizeof(kcb->jprobe_saved_stack)); return 1; } diff --git a/include/asm-sparc64/kprobes.h b/include/asm-sparc64/kprobes.h index a8d326a598f..7ba845320f5 100644 --- a/include/asm-sparc64/kprobes.h +++ b/include/asm-sparc64/kprobes.h @@ -3,6 +3,7 @@ #include #include +#include typedef u32 kprobe_opcode_t; @@ -18,6 +19,25 @@ struct arch_specific_insn { kprobe_opcode_t insn[MAX_INSN_SIZE]; }; +struct prev_kprobe { + struct kprobe *kp; + unsigned int status; + unsigned long orig_tnpc; + unsigned long orig_tstate_pil; +}; + +/* per-cpu kprobe control block */ +struct kprobe_ctlblk { + unsigned long kprobe_status; + unsigned long kprobe_orig_tnpc; + unsigned long kprobe_orig_tstate_pil; + long *jprobe_saved_esp; + struct pt_regs jprobe_saved_regs; + struct pt_regs *jprobe_saved_regs_location; + struct sparc_stackf jprobe_saved_stack; + struct prev_kprobe prev_kprobe; +}; + #ifdef CONFIG_KPROBES extern int kprobe_exceptions_notify(struct notifier_block *self, unsigned long val, void *data); -- cgit v1.2.3-70-g09d2 From 991a51d83a3d9bebfafdd1e692cf310899d60791 Mon Sep 17 00:00:00 2001 From: Ananth N Mavinakayanahalli Date: Mon, 7 Nov 2005 01:00:14 -0800 Subject: [PATCH] Kprobes: Use RCU for (un)register synchronization - arch changes Changes to the arch kprobes infrastructure to take advantage of the locking changes introduced by usage of RCU for synchronization. All handlers are now run without any locks held, so they have to be re-entrant or provide their own synchronization. Signed-off-by: Ananth N Mavinakayanahalli Signed-off-by: Anil S Keshavamurthy Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/i386/kernel/kprobes.c | 22 +++++++--------------- arch/ia64/kernel/kprobes.c | 16 ++++++---------- arch/ppc64/kernel/kprobes.c | 24 ++++++------------------ arch/sparc64/kernel/kprobes.c | 14 ++------------ arch/x86_64/kernel/kprobes.c | 25 ++++++------------------- 5 files changed, 27 insertions(+), 74 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/i386/kernel/kprobes.c b/arch/i386/kernel/kprobes.c index 99565a66915..ad469299267 100644 --- a/arch/i386/kernel/kprobes.c +++ b/arch/i386/kernel/kprobes.c @@ -31,7 +31,6 @@ #include #include #include -#include #include #include #include @@ -123,6 +122,7 @@ static inline void prepare_singlestep(struct kprobe *p, struct pt_regs *regs) regs->eip = (unsigned long)&p->ainsn.insn; } +/* Called with kretprobe_lock held */ void __kprobes arch_prepare_kretprobe(struct kretprobe *rp, struct pt_regs *regs) { @@ -168,15 +168,12 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) } /* Check we're not actually recursing */ if (kprobe_running()) { - /* We *are* holding lock here, so this is safe. - Disarm the probe we just hit, and ignore it. */ p = get_kprobe(addr); if (p) { if (kcb->kprobe_status == KPROBE_HIT_SS && *p->ainsn.insn == BREAKPOINT_INSTRUCTION) { regs->eflags &= ~TF_MASK; regs->eflags |= kcb->kprobe_saved_eflags; - unlock_kprobes(); goto no_kprobe; } /* We have reentered the kprobe_handler(), since @@ -197,14 +194,11 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) goto ss_probe; } } - /* If it's not ours, can't be delete race, (we hold lock). */ goto no_kprobe; } - lock_kprobes(); p = get_kprobe(addr); if (!p) { - unlock_kprobes(); if (regs->eflags & VM_MASK) { /* We are in virtual-8086 mode. Return 0 */ goto no_kprobe; @@ -268,9 +262,10 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) struct kretprobe_instance *ri = NULL; struct hlist_head *head; struct hlist_node *node, *tmp; - unsigned long orig_ret_address = 0; + unsigned long flags, orig_ret_address = 0; unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline; + spin_lock_irqsave(&kretprobe_lock, flags); head = kretprobe_inst_table_head(current); /* @@ -310,7 +305,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) regs->eip = orig_ret_address; reset_current_kprobe(); - unlock_kprobes(); + spin_unlock_irqrestore(&kretprobe_lock, flags); preempt_enable_no_resched(); /* @@ -395,7 +390,7 @@ static void __kprobes resume_execution(struct kprobe *p, /* * Interrupts are disabled on entry as trap1 is an interrupt gate and they - * remain disabled thoroughout this function. And we hold kprobe lock. + * remain disabled thoroughout this function. */ static inline int post_kprobe_handler(struct pt_regs *regs) { @@ -419,7 +414,6 @@ static inline int post_kprobe_handler(struct pt_regs *regs) goto out; } reset_current_kprobe(); - unlock_kprobes(); out: preempt_enable_no_resched(); @@ -434,7 +428,6 @@ out: return 1; } -/* Interrupts disabled, kprobe_lock held. */ static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) { struct kprobe *cur = kprobe_running(); @@ -448,7 +441,6 @@ static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) regs->eflags |= kcb->kprobe_old_eflags; reset_current_kprobe(); - unlock_kprobes(); preempt_enable_no_resched(); } return 0; @@ -463,7 +455,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - preempt_disable(); + rcu_read_lock(); switch (val) { case DIE_INT3: if (kprobe_handler(args->regs)) @@ -482,7 +474,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, default: break; } - preempt_enable(); + rcu_read_unlock(); return ret; } diff --git a/arch/ia64/kernel/kprobes.c b/arch/ia64/kernel/kprobes.c index 17e70b1b8d7..fddbac32d44 100644 --- a/arch/ia64/kernel/kprobes.c +++ b/arch/ia64/kernel/kprobes.c @@ -26,7 +26,6 @@ #include #include #include -#include #include #include #include @@ -343,10 +342,11 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) struct kretprobe_instance *ri = NULL; struct hlist_head *head; struct hlist_node *node, *tmp; - unsigned long orig_ret_address = 0; + unsigned long flags, orig_ret_address = 0; unsigned long trampoline_address = ((struct fnptr *)kretprobe_trampoline)->ip; + spin_lock_irqsave(&kretprobe_lock, flags); head = kretprobe_inst_table_head(current); /* @@ -386,7 +386,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) regs->cr_iip = orig_ret_address; reset_current_kprobe(); - unlock_kprobes(); + spin_unlock_irqrestore(&kretprobe_lock, flags); preempt_enable_no_resched(); /* @@ -397,6 +397,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) return 1; } +/* Called with kretprobe_lock held */ void __kprobes arch_prepare_kretprobe(struct kretprobe *rp, struct pt_regs *regs) { @@ -612,7 +613,6 @@ static int __kprobes pre_kprobes_handler(struct die_args *args) if ((kcb->kprobe_status == KPROBE_HIT_SS) && (p->ainsn.inst_flag == INST_FLAG_BREAK_INST)) { ia64_psr(regs)->ss = 0; - unlock_kprobes(); goto no_kprobe; } /* We have reentered the pre_kprobe_handler(), since @@ -641,10 +641,8 @@ static int __kprobes pre_kprobes_handler(struct die_args *args) } } - lock_kprobes(); p = get_kprobe(addr); if (!p) { - unlock_kprobes(); if (!is_ia64_break_inst(regs)) { /* * The breakpoint instruction was removed right @@ -707,7 +705,6 @@ static int __kprobes post_kprobes_handler(struct pt_regs *regs) goto out; } reset_current_kprobe(); - unlock_kprobes(); out: preempt_enable_no_resched(); @@ -728,7 +725,6 @@ static int __kprobes kprobes_fault_handler(struct pt_regs *regs, int trapnr) if (kcb->kprobe_status & KPROBE_HIT_SS) { resume_execution(cur, regs); reset_current_kprobe(); - unlock_kprobes(); preempt_enable_no_resched(); } @@ -741,7 +737,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - preempt_disable(); + rcu_read_lock(); switch(val) { case DIE_BREAK: if (pre_kprobes_handler(args)) @@ -757,7 +753,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, default: break; } - preempt_enable(); + rcu_read_unlock(); return ret; } diff --git a/arch/ppc64/kernel/kprobes.c b/arch/ppc64/kernel/kprobes.c index 3f89f3e5584..e0a25b35437 100644 --- a/arch/ppc64/kernel/kprobes.c +++ b/arch/ppc64/kernel/kprobes.c @@ -30,7 +30,6 @@ #include #include #include -#include #include #include #include @@ -125,6 +124,7 @@ static inline void set_current_kprobe(struct kprobe *p, struct pt_regs *regs, kcb->kprobe_saved_msr = regs->msr; } +/* Called with kretprobe_lock held */ void __kprobes arch_prepare_kretprobe(struct kretprobe *rp, struct pt_regs *regs) { @@ -152,8 +152,6 @@ static inline int kprobe_handler(struct pt_regs *regs) /* Check we're not actually recursing */ if (kprobe_running()) { - /* We *are* holding lock here, so this is safe. - Disarm the probe we just hit, and ignore it. */ p = get_kprobe(addr); if (p) { kprobe_opcode_t insn = *p->ainsn.insn; @@ -161,7 +159,6 @@ static inline int kprobe_handler(struct pt_regs *regs) is_trap(insn)) { regs->msr &= ~MSR_SE; regs->msr |= kcb->kprobe_saved_msr; - unlock_kprobes(); goto no_kprobe; } /* We have reentered the kprobe_handler(), since @@ -183,14 +180,11 @@ static inline int kprobe_handler(struct pt_regs *regs) goto ss_probe; } } - /* If it's not ours, can't be delete race, (we hold lock). */ goto no_kprobe; } - lock_kprobes(); p = get_kprobe(addr); if (!p) { - unlock_kprobes(); if (*addr != BREAKPOINT_INSTRUCTION) { /* * PowerPC has multiple variants of the "trap" @@ -254,9 +248,10 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) struct kretprobe_instance *ri = NULL; struct hlist_head *head; struct hlist_node *node, *tmp; - unsigned long orig_ret_address = 0; + unsigned long flags, orig_ret_address = 0; unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline; + spin_lock_irqsave(&kretprobe_lock, flags); head = kretprobe_inst_table_head(current); /* @@ -296,7 +291,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) regs->nip = orig_ret_address; reset_current_kprobe(); - unlock_kprobes(); + spin_unlock_irqrestore(&kretprobe_lock, flags); preempt_enable_no_resched(); /* @@ -348,7 +343,6 @@ static inline int post_kprobe_handler(struct pt_regs *regs) goto out; } reset_current_kprobe(); - unlock_kprobes(); out: preempt_enable_no_resched(); @@ -363,7 +357,6 @@ out: return 1; } -/* Interrupts disabled, kprobe_lock held. */ static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) { struct kprobe *cur = kprobe_running(); @@ -378,7 +371,6 @@ static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) regs->msr |= kcb->kprobe_saved_msr; reset_current_kprobe(); - unlock_kprobes(); preempt_enable_no_resched(); } return 0; @@ -393,11 +385,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - /* - * Interrupts are not disabled here. We need to disable - * preemption, because kprobe_running() uses smp_processor_id(). - */ - preempt_disable(); + rcu_read_lock(); switch (val) { case DIE_BPT: if (kprobe_handler(args->regs)) @@ -415,7 +403,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, default: break; } - preempt_enable_no_resched(); + rcu_read_unlock(); return ret; } diff --git a/arch/sparc64/kernel/kprobes.c b/arch/sparc64/kernel/kprobes.c index b95984154db..58a815e9037 100644 --- a/arch/sparc64/kernel/kprobes.c +++ b/arch/sparc64/kernel/kprobes.c @@ -116,15 +116,11 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); if (kprobe_running()) { - /* We *are* holding lock here, so this is safe. - * Disarm the probe we just hit, and ignore it. - */ p = get_kprobe(addr); if (p) { if (kcb->kprobe_status == KPROBE_HIT_SS) { regs->tstate = ((regs->tstate & ~TSTATE_PIL) | kcb->kprobe_orig_tstate_pil); - unlock_kprobes(); goto no_kprobe; } /* We have reentered the kprobe_handler(), since @@ -144,14 +140,11 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) if (p->break_handler && p->break_handler(p, regs)) goto ss_probe; } - /* If it's not ours, can't be delete race, (we hold lock). */ goto no_kprobe; } - lock_kprobes(); p = get_kprobe(addr); if (!p) { - unlock_kprobes(); if (*(u32 *)addr != BREAKPOINT_INSTRUCTION) { /* * The breakpoint instruction was removed right @@ -296,14 +289,12 @@ static inline int post_kprobe_handler(struct pt_regs *regs) goto out; } reset_current_kprobe(); - unlock_kprobes(); out: preempt_enable_no_resched(); return 1; } -/* Interrupts disabled, kprobe_lock held. */ static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) { struct kprobe *cur = kprobe_running(); @@ -316,7 +307,6 @@ static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) resume_execution(cur, regs, kcb); reset_current_kprobe(); - unlock_kprobes(); preempt_enable_no_resched(); } return 0; @@ -331,7 +321,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - preempt_disable(); + rcu_read_lock(); switch (val) { case DIE_DEBUG: if (kprobe_handler(args->regs)) @@ -350,7 +340,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, default: break; } - preempt_enable(); + rcu_read_unlock(); return ret; } diff --git a/arch/x86_64/kernel/kprobes.c b/arch/x86_64/kernel/kprobes.c index 6cb40d133b7..9bef2c8dc12 100644 --- a/arch/x86_64/kernel/kprobes.c +++ b/arch/x86_64/kernel/kprobes.c @@ -34,7 +34,6 @@ #include #include #include -#include #include #include #include @@ -266,6 +265,7 @@ static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs) regs->rip = (unsigned long)p->ainsn.insn; } +/* Called with kretprobe_lock held */ void __kprobes arch_prepare_kretprobe(struct kretprobe *rp, struct pt_regs *regs) { @@ -299,15 +299,12 @@ int __kprobes kprobe_handler(struct pt_regs *regs) /* Check we're not actually recursing */ if (kprobe_running()) { - /* We *are* holding lock here, so this is safe. - Disarm the probe we just hit, and ignore it. */ p = get_kprobe(addr); if (p) { if (kcb->kprobe_status == KPROBE_HIT_SS && *p->ainsn.insn == BREAKPOINT_INSTRUCTION) { regs->eflags &= ~TF_MASK; regs->eflags |= kcb->kprobe_saved_rflags; - unlock_kprobes(); goto no_kprobe; } else if (kcb->kprobe_status == KPROBE_HIT_SSDONE) { /* TODO: Provide re-entrancy from @@ -340,14 +337,11 @@ int __kprobes kprobe_handler(struct pt_regs *regs) goto ss_probe; } } - /* If it's not ours, can't be delete race, (we hold lock). */ goto no_kprobe; } - lock_kprobes(); p = get_kprobe(addr); if (!p) { - unlock_kprobes(); if (*addr != BREAKPOINT_INSTRUCTION) { /* * The breakpoint instruction was removed right @@ -406,9 +400,10 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) struct kretprobe_instance *ri = NULL; struct hlist_head *head; struct hlist_node *node, *tmp; - unsigned long orig_ret_address = 0; + unsigned long flags, orig_ret_address = 0; unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline; + spin_lock_irqsave(&kretprobe_lock, flags); head = kretprobe_inst_table_head(current); /* @@ -448,7 +443,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) regs->rip = orig_ret_address; reset_current_kprobe(); - unlock_kprobes(); + spin_unlock_irqrestore(&kretprobe_lock, flags); preempt_enable_no_resched(); /* @@ -536,10 +531,6 @@ static void __kprobes resume_execution(struct kprobe *p, } } -/* - * Interrupts are disabled on entry as trap1 is an interrupt gate and they - * remain disabled thoroughout this function. And we hold kprobe lock. - */ int __kprobes post_kprobe_handler(struct pt_regs *regs) { struct kprobe *cur = kprobe_running(); @@ -560,8 +551,6 @@ int __kprobes post_kprobe_handler(struct pt_regs *regs) if (kcb->kprobe_status == KPROBE_REENTER) { restore_previous_kprobe(kcb); goto out; - } else { - unlock_kprobes(); } reset_current_kprobe(); out: @@ -578,7 +567,6 @@ out: return 1; } -/* Interrupts disabled, kprobe_lock held. */ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr) { struct kprobe *cur = kprobe_running(); @@ -592,7 +580,6 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr) regs->eflags |= kcb->kprobe_old_rflags; reset_current_kprobe(); - unlock_kprobes(); preempt_enable_no_resched(); } return 0; @@ -607,7 +594,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - preempt_disable(); + rcu_read_lock(); switch (val) { case DIE_INT3: if (kprobe_handler(args->regs)) @@ -626,7 +613,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, default: break; } - preempt_enable(); + rcu_read_unlock(); return ret; } -- cgit v1.2.3-70-g09d2 From d217d5450f11d8c907c0458d175b0dc999b4d06d Mon Sep 17 00:00:00 2001 From: Ananth N Mavinakayanahalli Date: Mon, 7 Nov 2005 01:00:14 -0800 Subject: [PATCH] Kprobes: preempt_disable/enable() simplification Reorganize the preempt_disable/enable calls to eliminate the extra preempt depth. Changes based on Paul McKenney's review suggestions for the kprobes RCU changeset. Signed-off-by: Ananth N Mavinakayanahalli Signed-off-by: Anil S Keshavamurthy Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/i386/kernel/kprobes.c | 25 +++++++++++++++---------- arch/ia64/kernel/kprobes.c | 37 ++++++++++++++++++++----------------- arch/ppc64/kernel/kprobes.c | 25 +++++++++++++++---------- arch/sparc64/kernel/kprobes.c | 21 +++++++++++++-------- arch/x86_64/kernel/kprobes.c | 29 +++++++++++++++-------------- include/linux/kprobes.h | 2 +- kernel/kprobes.c | 2 +- 7 files changed, 80 insertions(+), 61 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/i386/kernel/kprobes.c b/arch/i386/kernel/kprobes.c index ad469299267..32b0c24ab9a 100644 --- a/arch/i386/kernel/kprobes.c +++ b/arch/i386/kernel/kprobes.c @@ -153,7 +153,14 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) int ret = 0; kprobe_opcode_t *addr = NULL; unsigned long *lp; - struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); + struct kprobe_ctlblk *kcb; + + /* + * We don't want to be preempted for the entire + * duration of kprobe processing + */ + preempt_disable(); + kcb = get_kprobe_ctlblk(); /* Check if the application is using LDT entry for its code segment and * calculate the address by reading the base address from the LDT entry. @@ -221,11 +228,6 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) goto no_kprobe; } - /* - * This preempt_disable() matches the preempt_enable_no_resched() - * in post_kprobe_handler() - */ - preempt_disable(); set_current_kprobe(p, regs, kcb); kcb->kprobe_status = KPROBE_HIT_ACTIVE; @@ -239,6 +241,7 @@ ss_probe: return 1; no_kprobe: + preempt_enable_no_resched(); return ret; } @@ -310,8 +313,8 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) /* * By returning a non-zero value, we are telling - * kprobe_handler() that we have handled unlocking - * and re-enabling preemption + * kprobe_handler() that we don't want the post_handler + * to run (and have re-enabled preemption) */ return 1; } @@ -455,7 +458,6 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - rcu_read_lock(); switch (val) { case DIE_INT3: if (kprobe_handler(args->regs)) @@ -467,14 +469,16 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, break; case DIE_GPF: case DIE_PAGE_FAULT: + /* kprobe_running() needs smp_processor_id() */ + preempt_disable(); if (kprobe_running() && kprobe_fault_handler(args->regs, args->trapnr)) ret = NOTIFY_STOP; + preempt_enable(); break; default: break; } - rcu_read_unlock(); return ret; } @@ -537,6 +541,7 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) *regs = kcb->jprobe_saved_regs; memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack, MIN_STACK_SIZE(stack_addr)); + preempt_enable_no_resched(); return 1; } return 0; diff --git a/arch/ia64/kernel/kprobes.c b/arch/ia64/kernel/kprobes.c index fddbac32d44..96736a119c9 100644 --- a/arch/ia64/kernel/kprobes.c +++ b/arch/ia64/kernel/kprobes.c @@ -389,11 +389,11 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) spin_unlock_irqrestore(&kretprobe_lock, flags); preempt_enable_no_resched(); - /* - * By returning a non-zero value, we are telling - * kprobe_handler() that we have handled unlocking - * and re-enabling preemption - */ + /* + * By returning a non-zero value, we are telling + * kprobe_handler() that we don't want the post_handler + * to run (and have re-enabled preemption) + */ return 1; } @@ -604,7 +604,14 @@ static int __kprobes pre_kprobes_handler(struct die_args *args) int ret = 0; struct pt_regs *regs = args->regs; kprobe_opcode_t *addr = (kprobe_opcode_t *)instruction_pointer(regs); - struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); + struct kprobe_ctlblk *kcb; + + /* + * We don't want to be preempted for the entire + * duration of kprobe processing + */ + preempt_disable(); + kcb = get_kprobe_ctlblk(); /* Handle recursion cases */ if (kprobe_running()) { @@ -659,11 +666,6 @@ static int __kprobes pre_kprobes_handler(struct die_args *args) goto no_kprobe; } - /* - * This preempt_disable() matches the preempt_enable_no_resched() - * in post_kprobes_handler() - */ - preempt_disable(); set_current_kprobe(p, kcb); kcb->kprobe_status = KPROBE_HIT_ACTIVE; @@ -681,6 +683,7 @@ ss_probe: return 1; no_kprobe: + preempt_enable_no_resched(); return ret; } @@ -716,9 +719,6 @@ static int __kprobes kprobes_fault_handler(struct pt_regs *regs, int trapnr) struct kprobe *cur = kprobe_running(); struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); - if (!cur) - return 0; - if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr)) return 1; @@ -737,7 +737,6 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - rcu_read_lock(); switch(val) { case DIE_BREAK: if (pre_kprobes_handler(args)) @@ -748,12 +747,15 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, ret = NOTIFY_STOP; break; case DIE_PAGE_FAULT: - if (kprobes_fault_handler(args->regs, args->trapnr)) + /* kprobe_running() needs smp_processor_id() */ + preempt_disable(); + if (kprobe_running() && + kprobes_fault_handler(args->regs, args->trapnr)) ret = NOTIFY_STOP; + preempt_enable(); default: break; } - rcu_read_unlock(); return ret; } @@ -785,6 +787,7 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); *regs = kcb->jprobe_saved_regs; + preempt_enable_no_resched(); return 1; } diff --git a/arch/ppc64/kernel/kprobes.c b/arch/ppc64/kernel/kprobes.c index e0a25b35437..511af54e623 100644 --- a/arch/ppc64/kernel/kprobes.c +++ b/arch/ppc64/kernel/kprobes.c @@ -148,7 +148,14 @@ static inline int kprobe_handler(struct pt_regs *regs) struct kprobe *p; int ret = 0; unsigned int *addr = (unsigned int *)regs->nip; - struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); + struct kprobe_ctlblk *kcb; + + /* + * We don't want to be preempted for the entire + * duration of kprobe processing + */ + preempt_disable(); + kcb = get_kprobe_ctlblk(); /* Check we're not actually recursing */ if (kprobe_running()) { @@ -207,11 +214,6 @@ static inline int kprobe_handler(struct pt_regs *regs) goto no_kprobe; } - /* - * This preempt_disable() matches the preempt_enable_no_resched() - * in post_kprobe_handler(). - */ - preempt_disable(); kcb->kprobe_status = KPROBE_HIT_ACTIVE; set_current_kprobe(p, regs, kcb); if (p->pre_handler && p->pre_handler(p, regs)) @@ -224,6 +226,7 @@ ss_probe: return 1; no_kprobe: + preempt_enable_no_resched(); return ret; } @@ -296,8 +299,8 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) /* * By returning a non-zero value, we are telling - * kprobe_handler() that we have handled unlocking - * and re-enabling preemption. + * kprobe_handler() that we don't want the post_handler + * to run (and have re-enabled preemption) */ return 1; } @@ -385,7 +388,6 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - rcu_read_lock(); switch (val) { case DIE_BPT: if (kprobe_handler(args->regs)) @@ -396,14 +398,16 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, ret = NOTIFY_STOP; break; case DIE_PAGE_FAULT: + /* kprobe_running() needs smp_processor_id() */ + preempt_disable(); if (kprobe_running() && kprobe_fault_handler(args->regs, args->trapnr)) ret = NOTIFY_STOP; + preempt_enable(); break; default: break; } - rcu_read_unlock(); return ret; } @@ -440,6 +444,7 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) * saved regs... */ memcpy(regs, &kcb->jprobe_saved_regs, sizeof(struct pt_regs)); + preempt_enable_no_resched(); return 1; } diff --git a/arch/sparc64/kernel/kprobes.c b/arch/sparc64/kernel/kprobes.c index 58a815e9037..96bd09b098f 100644 --- a/arch/sparc64/kernel/kprobes.c +++ b/arch/sparc64/kernel/kprobes.c @@ -113,7 +113,14 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) struct kprobe *p; void *addr = (void *) regs->tpc; int ret = 0; - struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); + struct kprobe_ctlblk *kcb; + + /* + * We don't want to be preempted for the entire + * duration of kprobe processing + */ + preempt_disable(); + kcb = get_kprobe_ctlblk(); if (kprobe_running()) { p = get_kprobe(addr); @@ -159,11 +166,6 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) goto no_kprobe; } - /* - * This preempt_disable() matches the preempt_enable_no_resched() - * in post_kprobes_handler() - */ - preempt_disable(); set_current_kprobe(p, regs, kcb); kcb->kprobe_status = KPROBE_HIT_ACTIVE; if (p->pre_handler && p->pre_handler(p, regs)) @@ -175,6 +177,7 @@ ss_probe: return 1; no_kprobe: + preempt_enable_no_resched(); return ret; } @@ -321,7 +324,6 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - rcu_read_lock(); switch (val) { case DIE_DEBUG: if (kprobe_handler(args->regs)) @@ -333,14 +335,16 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, break; case DIE_GPF: case DIE_PAGE_FAULT: + /* kprobe_running() needs smp_processor_id() */ + preempt_disable(); if (kprobe_running() && kprobe_fault_handler(args->regs, args->trapnr)) ret = NOTIFY_STOP; + preempt_enable(); break; default: break; } - rcu_read_unlock(); return ret; } @@ -426,6 +430,7 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) &(kcb->jprobe_saved_stack), sizeof(kcb->jprobe_saved_stack)); + preempt_enable_no_resched(); return 1; } return 0; diff --git a/arch/x86_64/kernel/kprobes.c b/arch/x86_64/kernel/kprobes.c index 9bef2c8dc12..dddeb678b44 100644 --- a/arch/x86_64/kernel/kprobes.c +++ b/arch/x86_64/kernel/kprobes.c @@ -286,16 +286,19 @@ void __kprobes arch_prepare_kretprobe(struct kretprobe *rp, } } -/* - * Interrupts are disabled on entry as trap3 is an interrupt gate and they - * remain disabled thorough out this function. - */ int __kprobes kprobe_handler(struct pt_regs *regs) { struct kprobe *p; int ret = 0; kprobe_opcode_t *addr = (kprobe_opcode_t *)(regs->rip - sizeof(kprobe_opcode_t)); - struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); + struct kprobe_ctlblk *kcb; + + /* + * We don't want to be preempted for the entire + * duration of kprobe processing + */ + preempt_disable(); + kcb = get_kprobe_ctlblk(); /* Check we're not actually recursing */ if (kprobe_running()) { @@ -359,11 +362,6 @@ int __kprobes kprobe_handler(struct pt_regs *regs) goto no_kprobe; } - /* - * This preempt_disable() matches the preempt_enable_no_resched() - * in post_kprobe_handler() - */ - preempt_disable(); set_current_kprobe(p, regs, kcb); kcb->kprobe_status = KPROBE_HIT_ACTIVE; @@ -377,6 +375,7 @@ ss_probe: return 1; no_kprobe: + preempt_enable_no_resched(); return ret; } @@ -448,8 +447,8 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs) /* * By returning a non-zero value, we are telling - * kprobe_handler() that we have handled unlocking - * and re-enabling preemption + * kprobe_handler() that we don't want the post_handler + * to run (and have re-enabled preemption) */ return 1; } @@ -594,7 +593,6 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, struct die_args *args = (struct die_args *)data; int ret = NOTIFY_DONE; - rcu_read_lock(); switch (val) { case DIE_INT3: if (kprobe_handler(args->regs)) @@ -606,14 +604,16 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self, break; case DIE_GPF: case DIE_PAGE_FAULT: + /* kprobe_running() needs smp_processor_id() */ + preempt_disable(); if (kprobe_running() && kprobe_fault_handler(args->regs, args->trapnr)) ret = NOTIFY_STOP; + preempt_enable(); break; default: break; } - rcu_read_unlock(); return ret; } @@ -675,6 +675,7 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) *regs = kcb->jprobe_saved_regs; memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack, MIN_STACK_SIZE(stack_addr)); + preempt_enable_no_resched(); return 1; } return 0; diff --git a/include/linux/kprobes.h b/include/linux/kprobes.h index cff281cf70c..e373c4a9de5 100644 --- a/include/linux/kprobes.h +++ b/include/linux/kprobes.h @@ -159,7 +159,7 @@ extern void show_registers(struct pt_regs *regs); extern kprobe_opcode_t *get_insn_slot(void); extern void free_insn_slot(kprobe_opcode_t *slot); -/* Get the kprobe at this addr (if any) - called under a rcu_read_lock() */ +/* Get the kprobe at this addr (if any) - called with preemption disabled */ struct kprobe *get_kprobe(void *addr); struct hlist_head * kretprobe_inst_table_head(struct task_struct *tsk); diff --git a/kernel/kprobes.c b/kernel/kprobes.c index cfef426e4cd..5beda378cc7 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -167,7 +167,7 @@ static inline void reset_kprobe_instance(void) * This routine is called either: * - under the kprobe_lock spinlock - during kprobe_[un]register() * OR - * - under an rcu_read_lock() - from arch/xxx/kernel/kprobes.c + * - with preemption disabled - from arch/xxx/kernel/kprobes.c */ struct kprobe __kprobes *get_kprobe(void *addr) { -- cgit v1.2.3-70-g09d2 From b2325fe1b7e5654fac9e9419423aa2c58a3dbd83 Mon Sep 17 00:00:00 2001 From: Jesper Juhl Date: Mon, 7 Nov 2005 01:01:35 -0800 Subject: [PATCH] kfree cleanup: arch This is the arch/ part of the big kfree cleanup patch. Remove pointless checks for NULL prior to calling kfree() in arch/. Signed-off-by: Jesper Juhl Acked-by: Grant Grundler Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/arm/mach-integrator/impd1.c | 3 +-- arch/cris/arch-v32/drivers/cryptocop.c | 14 +++++++------- arch/ia64/kernel/perfmon.c | 2 +- arch/powerpc/platforms/pseries/reconfig.c | 6 ++---- arch/ppc/8xx_io/cs4218_tdm.c | 3 +-- arch/ppc/syslib/prom.c | 6 ++---- arch/ppc64/kernel/lparcfg.c | 4 +--- arch/ppc64/kernel/scanlog.c | 3 +-- arch/s390/mm/extmem.c | 8 ++++---- arch/sparc64/kernel/us2e_cpufreq.c | 7 ++----- arch/sparc64/kernel/us3_cpufreq.c | 7 ++----- arch/um/kernel/sigio_user.c | 2 +- 12 files changed, 25 insertions(+), 40 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/arm/mach-integrator/impd1.c b/arch/arm/mach-integrator/impd1.c index a1b153d1626..a4bafee77a0 100644 --- a/arch/arm/mach-integrator/impd1.c +++ b/arch/arm/mach-integrator/impd1.c @@ -420,8 +420,7 @@ static int impd1_probe(struct lm_device *dev) free_impd1: if (impd1 && impd1->base) iounmap(impd1->base); - if (impd1) - kfree(impd1); + kfree(impd1); release_lm: release_mem_region(dev->resource.start, SZ_4K); return ret; diff --git a/arch/cris/arch-v32/drivers/cryptocop.c b/arch/cris/arch-v32/drivers/cryptocop.c index ca72076c630..501fa52d8d3 100644 --- a/arch/cris/arch-v32/drivers/cryptocop.c +++ b/arch/cris/arch-v32/drivers/cryptocop.c @@ -277,7 +277,7 @@ struct file_operations cryptocop_fops = { static void free_cdesc(struct cryptocop_dma_desc *cdesc) { DEBUG(printk("free_cdesc: cdesc 0x%p, from_pool=%d\n", cdesc, cdesc->from_pool)); - if (cdesc->free_buf) kfree(cdesc->free_buf); + kfree(cdesc->free_buf); if (cdesc->from_pool) { unsigned long int flags; @@ -2950,15 +2950,15 @@ static int cryptocop_ioctl_process(struct inode *inode, struct file *filp, unsig put_page(outpages[i]); } - if (digest_result) kfree(digest_result); - if (inpages) kfree(inpages); - if (outpages) kfree(outpages); + kfree(digest_result); + kfree(inpages); + kfree(outpages); if (cop){ - if (cop->tfrm_op.indata) kfree(cop->tfrm_op.indata); - if (cop->tfrm_op.outdata) kfree(cop->tfrm_op.outdata); + kfree(cop->tfrm_op.indata); + kfree(cop->tfrm_op.outdata); kfree(cop); } - if (jc) kfree(jc); + kfree(jc); DEBUG(print_lock_status()); diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c index f7dfc107cb7..410d4804fa6 100644 --- a/arch/ia64/kernel/perfmon.c +++ b/arch/ia64/kernel/perfmon.c @@ -4940,7 +4940,7 @@ abort_locked: if (call_made && PFM_CMD_RW_ARG(cmd) && copy_to_user(arg, args_k, base_sz*count)) ret = -EFAULT; error_args: - if (args_k) kfree(args_k); + kfree(args_k); DPRINT(("cmd=%s ret=%ld\n", PFM_CMD_NAME(cmd), ret)); diff --git a/arch/powerpc/platforms/pseries/reconfig.c b/arch/powerpc/platforms/pseries/reconfig.c index 58c61219d08..d7d40033945 100644 --- a/arch/powerpc/platforms/pseries/reconfig.c +++ b/arch/powerpc/platforms/pseries/reconfig.c @@ -286,10 +286,8 @@ static struct property *new_property(const char *name, const int length, return new; cleanup: - if (new->name) - kfree(new->name); - if (new->value) - kfree(new->value); + kfree(new->name); + kfree(new->value); kfree(new); return NULL; } diff --git a/arch/ppc/8xx_io/cs4218_tdm.c b/arch/ppc/8xx_io/cs4218_tdm.c index 532caa388dc..49eb2a7e65c 100644 --- a/arch/ppc/8xx_io/cs4218_tdm.c +++ b/arch/ppc/8xx_io/cs4218_tdm.c @@ -1013,8 +1013,7 @@ static void CS_IrqCleanup(void) */ cpm_free_handler(CPMVEC_SMC2); - if (beep_buf) - kfree(beep_buf); + kfree(beep_buf); kd_mksound = orig_mksound; } #endif /* MODULE */ diff --git a/arch/ppc/syslib/prom.c b/arch/ppc/syslib/prom.c index 278da6ee62e..1b9aa0d6a92 100644 --- a/arch/ppc/syslib/prom.c +++ b/arch/ppc/syslib/prom.c @@ -1335,10 +1335,8 @@ release_OF_resource(struct device_node* node, int index) if (!res) return -ENODEV; - if (res->name) { - kfree(res->name); - res->name = NULL; - } + kfree(res->name); + res->name = NULL; release_resource(res); kfree(res); diff --git a/arch/ppc64/kernel/lparcfg.c b/arch/ppc64/kernel/lparcfg.c index e86155770bb..3e7b2f28ec8 100644 --- a/arch/ppc64/kernel/lparcfg.c +++ b/arch/ppc64/kernel/lparcfg.c @@ -599,9 +599,7 @@ int __init lparcfg_init(void) void __exit lparcfg_cleanup(void) { if (proc_ppc64_lparcfg) { - if (proc_ppc64_lparcfg->data) { - kfree(proc_ppc64_lparcfg->data); - } + kfree(proc_ppc64_lparcfg->data); remove_proc_entry("lparcfg", proc_ppc64_lparcfg->parent); } } diff --git a/arch/ppc64/kernel/scanlog.c b/arch/ppc64/kernel/scanlog.c index 215bf890030..2edc947f7c4 100644 --- a/arch/ppc64/kernel/scanlog.c +++ b/arch/ppc64/kernel/scanlog.c @@ -225,8 +225,7 @@ int __init scanlog_init(void) void __exit scanlog_cleanup(void) { if (proc_ppc64_scan_log_dump) { - if (proc_ppc64_scan_log_dump->data) - kfree(proc_ppc64_scan_log_dump->data); + kfree(proc_ppc64_scan_log_dump->data); remove_proc_entry("scan-log-dump", proc_ppc64_scan_log_dump->parent); } } diff --git a/arch/s390/mm/extmem.c b/arch/s390/mm/extmem.c index c5348108ca3..506a33b51e4 100644 --- a/arch/s390/mm/extmem.c +++ b/arch/s390/mm/extmem.c @@ -234,8 +234,8 @@ query_segment_type (struct dcss_segment *seg) rc = 0; out_free: - if (qin) kfree(qin); - if (qout) kfree(qout); + kfree(qin); + kfree(qout); return rc; } @@ -394,7 +394,7 @@ __segment_load (char *name, int do_nonshared, unsigned long *addr, unsigned long segtype_string[seg->vm_segtype]); goto out; out_free: - kfree (seg); + kfree(seg); out: return rc; } @@ -505,7 +505,7 @@ segment_modify_shared (char *name, int do_nonshared) list_del(&seg->list); dcss_diag(DCSS_PURGESEG, seg->dcss_name, &dummy, &dummy); - kfree (seg); + kfree(seg); out_unlock: spin_unlock(&dcss_lock); return rc; diff --git a/arch/sparc64/kernel/us2e_cpufreq.c b/arch/sparc64/kernel/us2e_cpufreq.c index 686e526bec0..b35dc8dc995 100644 --- a/arch/sparc64/kernel/us2e_cpufreq.c +++ b/arch/sparc64/kernel/us2e_cpufreq.c @@ -388,10 +388,8 @@ err_out: kfree(driver); cpufreq_us2e_driver = NULL; } - if (us2e_freq_table) { - kfree(us2e_freq_table); - us2e_freq_table = NULL; - } + kfree(us2e_freq_table); + us2e_freq_table = NULL; return ret; } @@ -402,7 +400,6 @@ static void __exit us2e_freq_exit(void) { if (cpufreq_us2e_driver) { cpufreq_unregister_driver(cpufreq_us2e_driver); - kfree(cpufreq_us2e_driver); cpufreq_us2e_driver = NULL; kfree(us2e_freq_table); diff --git a/arch/sparc64/kernel/us3_cpufreq.c b/arch/sparc64/kernel/us3_cpufreq.c index 0340041f614..6d1f9a3c464 100644 --- a/arch/sparc64/kernel/us3_cpufreq.c +++ b/arch/sparc64/kernel/us3_cpufreq.c @@ -249,10 +249,8 @@ err_out: kfree(driver); cpufreq_us3_driver = NULL; } - if (us3_freq_table) { - kfree(us3_freq_table); - us3_freq_table = NULL; - } + kfree(us3_freq_table); + us3_freq_table = NULL; return ret; } @@ -263,7 +261,6 @@ static void __exit us3_freq_exit(void) { if (cpufreq_us3_driver) { cpufreq_unregister_driver(cpufreq_us3_driver); - kfree(cpufreq_us3_driver); cpufreq_us3_driver = NULL; kfree(us3_freq_table); diff --git a/arch/um/kernel/sigio_user.c b/arch/um/kernel/sigio_user.c index 3fbfd956bfe..48b1f644b9a 100644 --- a/arch/um/kernel/sigio_user.c +++ b/arch/um/kernel/sigio_user.c @@ -224,7 +224,7 @@ static int need_poll(int n) next_poll.used = n; return(0); } - if(next_poll.poll != NULL) kfree(next_poll.poll); + kfree(next_poll.poll); next_poll.poll = um_kmalloc_atomic(n * sizeof(struct pollfd)); if(next_poll.poll == NULL){ printk("need_poll : failed to allocate new pollfds\n"); -- cgit v1.2.3-70-g09d2 From b8ae48656db860d4c83a29aa7b0588fc89361935 Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Mon, 7 Nov 2005 14:08:46 -0800 Subject: [SPARC64] mm: don't re-evaluate *ptep sparc64 prom_callback and new_setup_frame32 each operates on a user page table without holding lock, and no doubt they've good reason. But I'd feel more confident if they were to do a "pte = *ptep" and then operate on pte, rather than re-evaluating *ptep. Signed-off-by: Hugh Dickins Signed-off-by: David S. Miller --- arch/sparc64/kernel/setup.c | 12 ++++++++---- arch/sparc64/kernel/signal32.c | 6 ++++-- 2 files changed, 12 insertions(+), 6 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index c1f34237cdf..bf1849dd9c4 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -154,6 +154,7 @@ int prom_callback(long *args) pud_t *pudp; pmd_t *pmdp; pte_t *ptep; + pte_t pte; for_each_process(p) { mm = p->mm; @@ -178,8 +179,9 @@ int prom_callback(long *args) * being called from inside OBP. */ ptep = pte_offset_map(pmdp, va); - if (pte_present(*ptep)) { - tte = pte_val(*ptep); + pte = *ptep; + if (pte_present(pte)) { + tte = pte_val(pte); res = PROM_TRUE; } pte_unmap(ptep); @@ -218,6 +220,7 @@ int prom_callback(long *args) pud_t *pudp; pmd_t *pmdp; pte_t *ptep; + pte_t pte; int error; if ((va >= LOW_OBP_ADDRESS) && (va < HI_OBP_ADDRESS)) { @@ -240,8 +243,9 @@ int prom_callback(long *args) * being called from inside OBP. */ ptep = pte_offset_kernel(pmdp, va); - if (pte_present(*ptep)) { - tte = pte_val(*ptep); + pte = *ptep; + if (pte_present(pte)) { + tte = pte_val(pte); res = PROM_TRUE; } goto done; diff --git a/arch/sparc64/kernel/signal32.c b/arch/sparc64/kernel/signal32.c index aecccd0df1d..009a86e5ded 100644 --- a/arch/sparc64/kernel/signal32.c +++ b/arch/sparc64/kernel/signal32.c @@ -863,6 +863,7 @@ static void new_setup_frame32(struct k_sigaction *ka, struct pt_regs *regs, pud_t *pudp = pud_offset(pgdp, address); pmd_t *pmdp = pmd_offset(pudp, address); pte_t *ptep; + pte_t pte; regs->u_regs[UREG_I7] = (unsigned long) (&(sf->insns[0]) - 2); @@ -873,9 +874,10 @@ static void new_setup_frame32(struct k_sigaction *ka, struct pt_regs *regs, preempt_disable(); ptep = pte_offset_map(pmdp, address); - if (pte_present(*ptep)) { + pte = *ptep; + if (pte_present(pte)) { unsigned long page = (unsigned long) - page_address(pte_page(*ptep)); + page_address(pte_page(pte)); wmb(); __asm__ __volatile__("flush %0 + %1" -- cgit v1.2.3-70-g09d2 From dedeb0029b9c83420fc1337d4ee53daa7b2a0ad4 Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Mon, 7 Nov 2005 14:09:01 -0800 Subject: [SPARC64] mm: context switch ptlock sparc64 is unique among architectures in taking the page_table_lock in its context switch (well, cris does too, but erroneously, and it's not yet SMP anyway). This seems to be a private affair between switch_mm and activate_mm, using page_table_lock as a per-mm lock, without any relation to its uses elsewhere. That's fine, but comment it as such; and unlock sooner in switch_mm, more like in activate_mm (preemption is disabled here). There is a block of "if (0)"ed code in smp_flush_tlb_pending which would have liked to rely on the page_table_lock, in switch_mm and elsewhere; but its comment explains how dup_mmap's flush_tlb_mm defeated it. And though that could have been changed at any time over the past few years, now the chance vanishes as we push the page_table_lock downwards, and perhaps split it per page table page. Just delete that block of code. Which leaves the mysterious spin_unlock_wait(&oldmm->page_table_lock) in kernel/fork.c copy_mm. Textual analysis (supported by Nick Piggin) suggests that the comment was written by DaveM, and that it relates to the defeated approach in the sparc64 smp_flush_tlb_pending. Just delete this block too. Signed-off-by: Hugh Dickins Signed-off-by: David S. Miller --- arch/sparc64/kernel/smp.c | 31 +++++--------------------- include/asm-sparc64/mmu_context.h | 46 ++++++++++++++++++++------------------- kernel/fork.c | 7 ------ 3 files changed, 29 insertions(+), 55 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index b137fd63f5e..a9089e2140e 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -883,34 +883,13 @@ void smp_flush_tlb_pending(struct mm_struct *mm, unsigned long nr, unsigned long u32 ctx = CTX_HWBITS(mm->context); int cpu = get_cpu(); - if (mm == current->active_mm && atomic_read(&mm->mm_users) == 1) { + if (mm == current->active_mm && atomic_read(&mm->mm_users) == 1) mm->cpu_vm_mask = cpumask_of_cpu(cpu); - goto local_flush_and_out; - } else { - /* This optimization is not valid. Normally - * we will be holding the page_table_lock, but - * there is an exception which is copy_page_range() - * when forking. The lock is held during the individual - * page table updates in the parent, but not at the - * top level, which is where we are invoked. - */ - if (0) { - cpumask_t this_cpu_mask = cpumask_of_cpu(cpu); - - /* By virtue of running under the mm->page_table_lock, - * and mmu_context.h:switch_mm doing the same, the - * following operation is safe. - */ - if (cpus_equal(mm->cpu_vm_mask, this_cpu_mask)) - goto local_flush_and_out; - } - } - - smp_cross_call_masked(&xcall_flush_tlb_pending, - ctx, nr, (unsigned long) vaddrs, - mm->cpu_vm_mask); + else + smp_cross_call_masked(&xcall_flush_tlb_pending, + ctx, nr, (unsigned long) vaddrs, + mm->cpu_vm_mask); -local_flush_and_out: __flush_tlb_pending(ctx, nr, vaddrs); put_cpu(); diff --git a/include/asm-sparc64/mmu_context.h b/include/asm-sparc64/mmu_context.h index 87c43c67866..08ba72d7722 100644 --- a/include/asm-sparc64/mmu_context.h +++ b/include/asm-sparc64/mmu_context.h @@ -87,37 +87,35 @@ extern void __flush_tlb_mm(unsigned long, unsigned long); static inline void switch_mm(struct mm_struct *old_mm, struct mm_struct *mm, struct task_struct *tsk) { unsigned long ctx_valid; + int cpu; + /* Note: page_table_lock is used here to serialize switch_mm + * and activate_mm, and their calls to get_new_mmu_context. + * This use of page_table_lock is unrelated to its other uses. + */ spin_lock(&mm->page_table_lock); - if (CTX_VALID(mm->context)) - ctx_valid = 1; - else - ctx_valid = 0; + ctx_valid = CTX_VALID(mm->context); + if (!ctx_valid) + get_new_mmu_context(mm); + spin_unlock(&mm->page_table_lock); if (!ctx_valid || (old_mm != mm)) { - if (!ctx_valid) - get_new_mmu_context(mm); - load_secondary_context(mm); reload_tlbmiss_state(tsk, mm); } - { - int cpu = smp_processor_id(); - - /* Even if (mm == old_mm) we _must_ check - * the cpu_vm_mask. If we do not we could - * corrupt the TLB state because of how - * smp_flush_tlb_{page,range,mm} on sparc64 - * and lazy tlb switches work. -DaveM - */ - if (!ctx_valid || !cpu_isset(cpu, mm->cpu_vm_mask)) { - cpu_set(cpu, mm->cpu_vm_mask); - __flush_tlb_mm(CTX_HWBITS(mm->context), - SECONDARY_CONTEXT); - } + /* Even if (mm == old_mm) we _must_ check + * the cpu_vm_mask. If we do not we could + * corrupt the TLB state because of how + * smp_flush_tlb_{page,range,mm} on sparc64 + * and lazy tlb switches work. -DaveM + */ + cpu = smp_processor_id(); + if (!ctx_valid || !cpu_isset(cpu, mm->cpu_vm_mask)) { + cpu_set(cpu, mm->cpu_vm_mask); + __flush_tlb_mm(CTX_HWBITS(mm->context), + SECONDARY_CONTEXT); } - spin_unlock(&mm->page_table_lock); } #define deactivate_mm(tsk,mm) do { } while (0) @@ -127,6 +125,10 @@ static inline void activate_mm(struct mm_struct *active_mm, struct mm_struct *mm { int cpu; + /* Note: page_table_lock is used here to serialize switch_mm + * and activate_mm, and their calls to get_new_mmu_context. + * This use of page_table_lock is unrelated to its other uses. + */ spin_lock(&mm->page_table_lock); if (!CTX_VALID(mm->context)) get_new_mmu_context(mm); diff --git a/kernel/fork.c b/kernel/fork.c index efac2c58ec7..158710d2256 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -470,13 +470,6 @@ static int copy_mm(unsigned long clone_flags, struct task_struct * tsk) if (clone_flags & CLONE_VM) { atomic_inc(&oldmm->mm_users); mm = oldmm; - /* - * There are cases where the PTL is held to ensure no - * new threads start up in user mode using an mm, which - * allows optimizing out ipis; the tlb_gather_mmu code - * is an example. - */ - spin_unlock_wait(&oldmm->page_table_lock); goto good_mm; } -- cgit v1.2.3-70-g09d2 From 62dbec78be652c28f63ad5eda3d01c244c916040 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Mon, 7 Nov 2005 14:09:58 -0800 Subject: [SPARC64] mm: Do not flush TLB mm in tlb_finish_mmu() It isn't needed any longer, as noted by Hugh Dickins. We still need the flush routines, due to the one remaining call site in hugetlb_prefault_arch_hook(). That can be eliminated at some later point, however. Signed-off-by: David S. Miller --- arch/sparc64/kernel/smp.c | 48 +++++++++++++++++------------------------------ include/asm-sparc64/tlb.h | 6 ++---- 2 files changed, 19 insertions(+), 35 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index a9089e2140e..5d90ee9aebf 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -839,43 +839,29 @@ void smp_flush_tlb_all(void) * questionable (in theory the big win for threads is the massive sharing of * address space state across processors). */ + +/* This currently is only used by the hugetlb arch pre-fault + * hook on UltraSPARC-III+ and later when changing the pagesize + * bits of the context register for an address space. + */ void smp_flush_tlb_mm(struct mm_struct *mm) { - /* - * This code is called from two places, dup_mmap and exit_mmap. In the - * former case, we really need a flush. In the later case, the callers - * are single threaded exec_mmap (really need a flush), multithreaded - * exec_mmap case (do not need to flush, since the caller gets a new - * context via activate_mm), and all other callers of mmput() whence - * the flush can be optimized since the associated threads are dead and - * the mm is being torn down (__exit_mm and other mmput callers) or the - * owning thread is dissociating itself from the mm. The - * (atomic_read(&mm->mm_users) == 0) check ensures real work is done - * for single thread exec and dup_mmap cases. An alternate check might - * have been (current->mm != mm). - * Kanoj Sarcar - */ - if (atomic_read(&mm->mm_users) == 0) - return; - - { - u32 ctx = CTX_HWBITS(mm->context); - int cpu = get_cpu(); + u32 ctx = CTX_HWBITS(mm->context); + int cpu = get_cpu(); - if (atomic_read(&mm->mm_users) == 1) { - mm->cpu_vm_mask = cpumask_of_cpu(cpu); - goto local_flush_and_out; - } + if (atomic_read(&mm->mm_users) == 1) { + mm->cpu_vm_mask = cpumask_of_cpu(cpu); + goto local_flush_and_out; + } - smp_cross_call_masked(&xcall_flush_tlb_mm, - ctx, 0, 0, - mm->cpu_vm_mask); + smp_cross_call_masked(&xcall_flush_tlb_mm, + ctx, 0, 0, + mm->cpu_vm_mask); - local_flush_and_out: - __flush_tlb_mm(ctx, SECONDARY_CONTEXT); +local_flush_and_out: + __flush_tlb_mm(ctx, SECONDARY_CONTEXT); - put_cpu(); - } + put_cpu(); } void smp_flush_tlb_pending(struct mm_struct *mm, unsigned long nr, unsigned long *vaddrs) diff --git a/include/asm-sparc64/tlb.h b/include/asm-sparc64/tlb.h index 66138d959df..1eda17954f3 100644 --- a/include/asm-sparc64/tlb.h +++ b/include/asm-sparc64/tlb.h @@ -78,11 +78,9 @@ static inline void tlb_finish_mmu(struct mmu_gather *mp, unsigned long start, un { tlb_flush_mmu(mp); - if (mp->fullmm) { - if (CTX_VALID(mp->mm->context)) - do_flush_tlb_mm(mp->mm); + if (mp->fullmm) mp->fullmm = 0; - } else + else flush_tlb_pending(); /* keep the page table cache within bounds */ -- cgit v1.2.3-70-g09d2 From fc3214952fac07fef7e102fdd4a18b3d736f33f1 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Mon, 7 Nov 2005 14:10:10 -0800 Subject: [SPARC64]: Kill off dummy_tick_ops. It only serves to generate false-positive buildcheck warnings. Just set it initially to tick_operations which uses the v9 %tick register which every sparc64 processor has. Signed-off-by: David S. Miller --- arch/sparc64/kernel/time.c | 13 ++----------- 1 file changed, 2 insertions(+), 11 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/time.c b/arch/sparc64/kernel/time.c index 38c5525087a..459c8fbe02b 100644 --- a/arch/sparc64/kernel/time.c +++ b/arch/sparc64/kernel/time.c @@ -60,17 +60,6 @@ static void __iomem *mstk48t59_regs; static int set_rtc_mmss(unsigned long); -static __init unsigned long dummy_get_tick(void) -{ - return 0; -} - -static __initdata struct sparc64_tick_ops dummy_tick_ops = { - .get_tick = dummy_get_tick, -}; - -struct sparc64_tick_ops *tick_ops __read_mostly = &dummy_tick_ops; - #define TICK_PRIV_BIT (1UL << 63) #ifdef CONFIG_SMP @@ -200,6 +189,8 @@ static struct sparc64_tick_ops tick_operations __read_mostly = { .softint_mask = 1UL << 0, }; +struct sparc64_tick_ops *tick_ops __read_mostly = &tick_operations; + static void stick_init_tick(unsigned long offset) { tick_disable_protection(); -- cgit v1.2.3-70-g09d2 From e0436b3164fd071acd30a50339b7b6ba5f053cf6 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:11:02 -0800 Subject: [SPARC64]: remove alloc_user_space() this inline routine in arch/sparc64/kernel/ioctl32.c is completely unused and superceeded by compat_alloc_user_space() Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 11 ----------- 1 file changed, 11 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index e6a00325075..0e587d6de31 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -27,17 +27,6 @@ */ #define A(__x) compat_ptr(__x) -static __inline__ void *alloc_user_space(long len) -{ - struct pt_regs *regs = current_thread_info()->kregs; - unsigned long usp = regs->u_regs[UREG_I6]; - - if (!(test_thread_flag(TIF_32BIT))) - usp += STACK_BIAS; - - return (void *) (usp - len); -} - #define CODE #include "compat_ioctl.c" -- cgit v1.2.3-70-g09d2 From 9d3c7d1bfd41d5082a541666db404aae7699b79e Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:11:14 -0800 Subject: [SPARC]: remove audioio.h The old sound drivers are gone in 2.6, so the only user left are the compat ioctls. Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 7 -- include/asm-sparc/audioio.h | 234 ------------------------------------------ include/asm-sparc64/audioio.h | 234 ------------------------------------------ 3 files changed, 475 deletions(-) delete mode 100644 include/asm-sparc/audioio.h delete mode 100644 include/asm-sparc64/audioio.h (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 0e587d6de31..6fda044a737 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -19,7 +19,6 @@ #include #include #include -#include #include /* Use this to get at 32-bit user passed pointers. @@ -524,12 +523,6 @@ COMPATIBLE_IOCTL(OPROMPATH2NODE) COMPATIBLE_IOCTL(LOOP_SET_STATUS64) COMPATIBLE_IOCTL(LOOP_GET_STATUS64) /* Big A */ -COMPATIBLE_IOCTL(AUDIO_GETINFO) -COMPATIBLE_IOCTL(AUDIO_SETINFO) -COMPATIBLE_IOCTL(AUDIO_DRAIN) -COMPATIBLE_IOCTL(AUDIO_GETDEV) -COMPATIBLE_IOCTL(AUDIO_GETDEV_SUNOS) -COMPATIBLE_IOCTL(AUDIO_FLUSH) COMPATIBLE_IOCTL(AUTOFS_IOC_EXPIRE_MULTI) #if defined(CONFIG_DRM) || defined(CONFIG_DRM_MODULE) COMPATIBLE_IOCTL(DRM_IOCTL_GET_MAGIC) diff --git a/include/asm-sparc/audioio.h b/include/asm-sparc/audioio.h deleted file mode 100644 index cf16173f521..00000000000 --- a/include/asm-sparc/audioio.h +++ /dev/null @@ -1,234 +0,0 @@ -/* - * include/asm-sparc/audioio.h - * - * Sparc Audio Midlayer - * Copyright (C) 1996 Thomas K. Dyas (tdyas@noc.rutgers.edu) - */ - -#ifndef _AUDIOIO_H_ -#define _AUDIOIO_H_ - -/* - * SunOS/Solaris /dev/audio interface - */ - -#if defined(__KERNEL__) || !defined(__GLIBC__) || (__GLIBC__ < 2) -#include -#include -#include -#endif - -/* - * This structure contains state information for audio device IO streams. - */ -typedef struct audio_prinfo { - /* - * The following values describe the audio data encoding. - */ - unsigned int sample_rate; /* samples per second */ - unsigned int channels; /* number of interleaved channels */ - unsigned int precision; /* bit-width of each sample */ - unsigned int encoding; /* data encoding method */ - - /* - * The following values control audio device configuration - */ - unsigned int gain; /* gain level: 0 - 255 */ - unsigned int port; /* selected I/O port (see below) */ - unsigned int avail_ports; /* available I/O ports (see below) */ - unsigned int _xxx[2]; /* Reserved for future use */ - - unsigned int buffer_size; /* I/O buffer size */ - - /* - * The following values describe driver state - */ - unsigned int samples; /* number of samples converted */ - unsigned int eof; /* End Of File counter (play only) */ - - unsigned char pause; /* non-zero for pause, zero to resume */ - unsigned char error; /* non-zero if overflow/underflow */ - unsigned char waiting; /* non-zero if a process wants access */ - unsigned char balance; /* stereo channel balance */ - - unsigned short minordev; - - /* - * The following values are read-only state flags - */ - unsigned char open; /* non-zero if open access permitted */ - unsigned char active; /* non-zero if I/O is active */ -} audio_prinfo_t; - - -/* - * This structure describes the current state of the audio device. - */ -typedef struct audio_info { - /* - * Per-stream information - */ - audio_prinfo_t play; /* output status information */ - audio_prinfo_t record; /* input status information */ - - /* - * Per-unit/channel information - */ - unsigned int monitor_gain; /* input to output mix: 0 - 255 */ - unsigned char output_muted; /* non-zero if output is muted */ - unsigned char _xxx[3]; /* Reserved for future use */ - unsigned int _yyy[3]; /* Reserved for future use */ -} audio_info_t; - - -/* - * Audio encoding types - */ -#define AUDIO_ENCODING_NONE (0) /* no encoding assigned */ -#define AUDIO_ENCODING_ULAW (1) /* u-law encoding */ -#define AUDIO_ENCODING_ALAW (2) /* A-law encoding */ -#define AUDIO_ENCODING_LINEAR (3) /* Linear PCM encoding */ -#define AUDIO_ENCODING_FLOAT (4) /* IEEE float (-1. <-> +1.) */ -#define AUDIO_ENCODING_DVI (104) /* DVI ADPCM */ -#define AUDIO_ENCODING_LINEAR8 (105) /* 8 bit UNSIGNED */ -#define AUDIO_ENCODING_LINEARLE (106) /* Linear PCM LE encoding */ - -/* - * These ranges apply to record, play, and monitor gain values - */ -#define AUDIO_MIN_GAIN (0) /* minimum gain value */ -#define AUDIO_MAX_GAIN (255) /* maximum gain value */ - -/* - * These values apply to the balance field to adjust channel gain values - */ -#define AUDIO_LEFT_BALANCE (0) /* left channel only */ -#define AUDIO_MID_BALANCE (32) /* equal left/right channel */ -#define AUDIO_RIGHT_BALANCE (64) /* right channel only */ -#define AUDIO_BALANCE_SHIFT (3) - -/* - * Generic minimum/maximum limits for number of channels, both modes - */ -#define AUDIO_MIN_PLAY_CHANNELS (1) -#define AUDIO_MAX_PLAY_CHANNELS (4) -#define AUDIO_MIN_REC_CHANNELS (1) -#define AUDIO_MAX_REC_CHANNELS (4) - -/* - * Generic minimum/maximum limits for sample precision - */ -#define AUDIO_MIN_PLAY_PRECISION (8) -#define AUDIO_MAX_PLAY_PRECISION (32) -#define AUDIO_MIN_REC_PRECISION (8) -#define AUDIO_MAX_REC_PRECISION (32) - -/* - * Define some convenient names for typical audio ports - */ -/* - * output ports (several may be enabled simultaneously) - */ -#define AUDIO_SPEAKER 0x01 /* output to built-in speaker */ -#define AUDIO_HEADPHONE 0x02 /* output to headphone jack */ -#define AUDIO_LINE_OUT 0x04 /* output to line out */ - -/* - * input ports (usually only one at a time) - */ -#define AUDIO_MICROPHONE 0x01 /* input from microphone */ -#define AUDIO_LINE_IN 0x02 /* input from line in */ -#define AUDIO_CD 0x04 /* input from on-board CD inputs */ -#define AUDIO_INTERNAL_CD_IN AUDIO_CD /* input from internal CDROM */ -#define AUDIO_ANALOG_LOOPBACK 0x40 /* input from output */ - - -/* - * This macro initializes an audio_info structure to 'harmless' values. - * Note that (~0) might not be a harmless value for a flag that was - * a signed int. - */ -#define AUDIO_INITINFO(i) { \ - unsigned int *__x__; \ - for (__x__ = (unsigned int *)(i); \ - (char *) __x__ < (((char *)(i)) + sizeof (audio_info_t)); \ - *__x__++ = ~0); \ -} - -/* - * These allow testing for what the user wants to set - */ -#define AUD_INITVALUE (~0) -#define Modify(X) ((unsigned int)(X) != AUD_INITVALUE) -#define Modifys(X) ((X) != (unsigned short)AUD_INITVALUE) -#define Modifyc(X) ((X) != (unsigned char)AUD_INITVALUE) - -/* - * Parameter for the AUDIO_GETDEV ioctl to determine current - * audio devices. - */ -#define MAX_AUDIO_DEV_LEN (16) -typedef struct audio_device { - char name[MAX_AUDIO_DEV_LEN]; - char version[MAX_AUDIO_DEV_LEN]; - char config[MAX_AUDIO_DEV_LEN]; -} audio_device_t; - - -/* - * Ioctl calls for the audio device. - */ - -/* - * AUDIO_GETINFO retrieves the current state of the audio device. - * - * AUDIO_SETINFO copies all fields of the audio_info structure whose - * values are not set to the initialized value (-1) to the device state. - * It performs an implicit AUDIO_GETINFO to return the new state of the - * device. Note that the record.samples and play.samples fields are set - * to the last value before the AUDIO_SETINFO took effect. This allows - * an application to reset the counters while atomically retrieving the - * last value. - * - * AUDIO_DRAIN suspends the calling process until the write buffers are - * empty. - * - * AUDIO_GETDEV returns a structure of type audio_device_t which contains - * three strings. The string "name" is a short identifying string (for - * example, the SBus Fcode name string), the string "version" identifies - * the current version of the device, and the "config" string identifies - * the specific configuration of the audio stream. All fields are - * device-dependent -- see the device specific manual pages for details. - * - * AUDIO_GETDEV_SUNOS returns a number which is an audio device defined - * herein (making it not too portable) - * - * AUDIO_FLUSH stops all playback and recording, clears all queued buffers, - * resets error counters, and restarts recording and playback as appropriate - * for the current sampling mode. - */ -#define AUDIO_GETINFO _IOR('A', 1, audio_info_t) -#define AUDIO_SETINFO _IOWR('A', 2, audio_info_t) -#define AUDIO_DRAIN _IO('A', 3) -#define AUDIO_GETDEV _IOR('A', 4, audio_device_t) -#define AUDIO_GETDEV_SUNOS _IOR('A', 4, int) -#define AUDIO_FLUSH _IO('A', 5) - -/* Define possible audio hardware configurations for - * old SunOS-style AUDIO_GETDEV ioctl */ -#define AUDIO_DEV_UNKNOWN (0) /* not defined */ -#define AUDIO_DEV_AMD (1) /* audioamd device */ -#define AUDIO_DEV_SPEAKERBOX (2) /* dbri device with speakerbox */ -#define AUDIO_DEV_CODEC (3) /* dbri device (internal speaker) */ -#define AUDIO_DEV_CS4231 (5) /* cs4231 device */ - -/* - * The following ioctl sets the audio device into an internal loopback mode, - * if the hardware supports this. The argument is TRUE to set loopback, - * FALSE to reset to normal operation. If the hardware does not support - * internal loopback, the ioctl should fail with EINVAL. - * Causes ADC data to be digitally mixed in and sent to the DAC. - */ -#define AUDIO_DIAG_LOOPBACK _IOW('A', 101, int) - -#endif /* _AUDIOIO_H_ */ diff --git a/include/asm-sparc64/audioio.h b/include/asm-sparc64/audioio.h deleted file mode 100644 index cf16173f521..00000000000 --- a/include/asm-sparc64/audioio.h +++ /dev/null @@ -1,234 +0,0 @@ -/* - * include/asm-sparc/audioio.h - * - * Sparc Audio Midlayer - * Copyright (C) 1996 Thomas K. Dyas (tdyas@noc.rutgers.edu) - */ - -#ifndef _AUDIOIO_H_ -#define _AUDIOIO_H_ - -/* - * SunOS/Solaris /dev/audio interface - */ - -#if defined(__KERNEL__) || !defined(__GLIBC__) || (__GLIBC__ < 2) -#include -#include -#include -#endif - -/* - * This structure contains state information for audio device IO streams. - */ -typedef struct audio_prinfo { - /* - * The following values describe the audio data encoding. - */ - unsigned int sample_rate; /* samples per second */ - unsigned int channels; /* number of interleaved channels */ - unsigned int precision; /* bit-width of each sample */ - unsigned int encoding; /* data encoding method */ - - /* - * The following values control audio device configuration - */ - unsigned int gain; /* gain level: 0 - 255 */ - unsigned int port; /* selected I/O port (see below) */ - unsigned int avail_ports; /* available I/O ports (see below) */ - unsigned int _xxx[2]; /* Reserved for future use */ - - unsigned int buffer_size; /* I/O buffer size */ - - /* - * The following values describe driver state - */ - unsigned int samples; /* number of samples converted */ - unsigned int eof; /* End Of File counter (play only) */ - - unsigned char pause; /* non-zero for pause, zero to resume */ - unsigned char error; /* non-zero if overflow/underflow */ - unsigned char waiting; /* non-zero if a process wants access */ - unsigned char balance; /* stereo channel balance */ - - unsigned short minordev; - - /* - * The following values are read-only state flags - */ - unsigned char open; /* non-zero if open access permitted */ - unsigned char active; /* non-zero if I/O is active */ -} audio_prinfo_t; - - -/* - * This structure describes the current state of the audio device. - */ -typedef struct audio_info { - /* - * Per-stream information - */ - audio_prinfo_t play; /* output status information */ - audio_prinfo_t record; /* input status information */ - - /* - * Per-unit/channel information - */ - unsigned int monitor_gain; /* input to output mix: 0 - 255 */ - unsigned char output_muted; /* non-zero if output is muted */ - unsigned char _xxx[3]; /* Reserved for future use */ - unsigned int _yyy[3]; /* Reserved for future use */ -} audio_info_t; - - -/* - * Audio encoding types - */ -#define AUDIO_ENCODING_NONE (0) /* no encoding assigned */ -#define AUDIO_ENCODING_ULAW (1) /* u-law encoding */ -#define AUDIO_ENCODING_ALAW (2) /* A-law encoding */ -#define AUDIO_ENCODING_LINEAR (3) /* Linear PCM encoding */ -#define AUDIO_ENCODING_FLOAT (4) /* IEEE float (-1. <-> +1.) */ -#define AUDIO_ENCODING_DVI (104) /* DVI ADPCM */ -#define AUDIO_ENCODING_LINEAR8 (105) /* 8 bit UNSIGNED */ -#define AUDIO_ENCODING_LINEARLE (106) /* Linear PCM LE encoding */ - -/* - * These ranges apply to record, play, and monitor gain values - */ -#define AUDIO_MIN_GAIN (0) /* minimum gain value */ -#define AUDIO_MAX_GAIN (255) /* maximum gain value */ - -/* - * These values apply to the balance field to adjust channel gain values - */ -#define AUDIO_LEFT_BALANCE (0) /* left channel only */ -#define AUDIO_MID_BALANCE (32) /* equal left/right channel */ -#define AUDIO_RIGHT_BALANCE (64) /* right channel only */ -#define AUDIO_BALANCE_SHIFT (3) - -/* - * Generic minimum/maximum limits for number of channels, both modes - */ -#define AUDIO_MIN_PLAY_CHANNELS (1) -#define AUDIO_MAX_PLAY_CHANNELS (4) -#define AUDIO_MIN_REC_CHANNELS (1) -#define AUDIO_MAX_REC_CHANNELS (4) - -/* - * Generic minimum/maximum limits for sample precision - */ -#define AUDIO_MIN_PLAY_PRECISION (8) -#define AUDIO_MAX_PLAY_PRECISION (32) -#define AUDIO_MIN_REC_PRECISION (8) -#define AUDIO_MAX_REC_PRECISION (32) - -/* - * Define some convenient names for typical audio ports - */ -/* - * output ports (several may be enabled simultaneously) - */ -#define AUDIO_SPEAKER 0x01 /* output to built-in speaker */ -#define AUDIO_HEADPHONE 0x02 /* output to headphone jack */ -#define AUDIO_LINE_OUT 0x04 /* output to line out */ - -/* - * input ports (usually only one at a time) - */ -#define AUDIO_MICROPHONE 0x01 /* input from microphone */ -#define AUDIO_LINE_IN 0x02 /* input from line in */ -#define AUDIO_CD 0x04 /* input from on-board CD inputs */ -#define AUDIO_INTERNAL_CD_IN AUDIO_CD /* input from internal CDROM */ -#define AUDIO_ANALOG_LOOPBACK 0x40 /* input from output */ - - -/* - * This macro initializes an audio_info structure to 'harmless' values. - * Note that (~0) might not be a harmless value for a flag that was - * a signed int. - */ -#define AUDIO_INITINFO(i) { \ - unsigned int *__x__; \ - for (__x__ = (unsigned int *)(i); \ - (char *) __x__ < (((char *)(i)) + sizeof (audio_info_t)); \ - *__x__++ = ~0); \ -} - -/* - * These allow testing for what the user wants to set - */ -#define AUD_INITVALUE (~0) -#define Modify(X) ((unsigned int)(X) != AUD_INITVALUE) -#define Modifys(X) ((X) != (unsigned short)AUD_INITVALUE) -#define Modifyc(X) ((X) != (unsigned char)AUD_INITVALUE) - -/* - * Parameter for the AUDIO_GETDEV ioctl to determine current - * audio devices. - */ -#define MAX_AUDIO_DEV_LEN (16) -typedef struct audio_device { - char name[MAX_AUDIO_DEV_LEN]; - char version[MAX_AUDIO_DEV_LEN]; - char config[MAX_AUDIO_DEV_LEN]; -} audio_device_t; - - -/* - * Ioctl calls for the audio device. - */ - -/* - * AUDIO_GETINFO retrieves the current state of the audio device. - * - * AUDIO_SETINFO copies all fields of the audio_info structure whose - * values are not set to the initialized value (-1) to the device state. - * It performs an implicit AUDIO_GETINFO to return the new state of the - * device. Note that the record.samples and play.samples fields are set - * to the last value before the AUDIO_SETINFO took effect. This allows - * an application to reset the counters while atomically retrieving the - * last value. - * - * AUDIO_DRAIN suspends the calling process until the write buffers are - * empty. - * - * AUDIO_GETDEV returns a structure of type audio_device_t which contains - * three strings. The string "name" is a short identifying string (for - * example, the SBus Fcode name string), the string "version" identifies - * the current version of the device, and the "config" string identifies - * the specific configuration of the audio stream. All fields are - * device-dependent -- see the device specific manual pages for details. - * - * AUDIO_GETDEV_SUNOS returns a number which is an audio device defined - * herein (making it not too portable) - * - * AUDIO_FLUSH stops all playback and recording, clears all queued buffers, - * resets error counters, and restarts recording and playback as appropriate - * for the current sampling mode. - */ -#define AUDIO_GETINFO _IOR('A', 1, audio_info_t) -#define AUDIO_SETINFO _IOWR('A', 2, audio_info_t) -#define AUDIO_DRAIN _IO('A', 3) -#define AUDIO_GETDEV _IOR('A', 4, audio_device_t) -#define AUDIO_GETDEV_SUNOS _IOR('A', 4, int) -#define AUDIO_FLUSH _IO('A', 5) - -/* Define possible audio hardware configurations for - * old SunOS-style AUDIO_GETDEV ioctl */ -#define AUDIO_DEV_UNKNOWN (0) /* not defined */ -#define AUDIO_DEV_AMD (1) /* audioamd device */ -#define AUDIO_DEV_SPEAKERBOX (2) /* dbri device with speakerbox */ -#define AUDIO_DEV_CODEC (3) /* dbri device (internal speaker) */ -#define AUDIO_DEV_CS4231 (5) /* cs4231 device */ - -/* - * The following ioctl sets the audio device into an internal loopback mode, - * if the hardware supports this. The argument is TRUE to set loopback, - * FALSE to reset to normal operation. If the hardware does not support - * internal loopback, the ioctl should fail with EINVAL. - * Causes ADC data to be digitally mixed in and sent to the DAC. - */ -#define AUDIO_DIAG_LOOPBACK _IOW('A', 101, int) - -#endif /* _AUDIOIO_H_ */ -- cgit v1.2.3-70-g09d2 From e1413315b8dfcdebc61416dadc1334619dfb4543 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:11:25 -0800 Subject: [SPARC]: remove kbio.h The old keyboard driver is gone in 2.6, so the only user left are the compat ioctls. Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 11 --------- include/asm-sparc/kbio.h | 56 ------------------------------------------- include/asm-sparc64/kbio.h | 56 ------------------------------------------- 3 files changed, 123 deletions(-) delete mode 100644 include/asm-sparc/kbio.h delete mode 100644 include/asm-sparc64/kbio.h (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 6fda044a737..947bd265aaa 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -14,7 +14,6 @@ #include #include #include -#include #include #include #include @@ -473,16 +472,6 @@ COMPATIBLE_IOCTL(FBIOSCURPOS) COMPATIBLE_IOCTL(FBIOGCURPOS) COMPATIBLE_IOCTL(FBIOGCURMAX) /* Little k */ -COMPATIBLE_IOCTL(KIOCTYPE) -COMPATIBLE_IOCTL(KIOCLAYOUT) -COMPATIBLE_IOCTL(KIOCGTRANS) -COMPATIBLE_IOCTL(KIOCTRANS) -COMPATIBLE_IOCTL(KIOCCMD) -COMPATIBLE_IOCTL(KIOCSDIRECT) -COMPATIBLE_IOCTL(KIOCSLED) -COMPATIBLE_IOCTL(KIOCGLED) -COMPATIBLE_IOCTL(KIOCSRATE) -COMPATIBLE_IOCTL(KIOCGRATE) COMPATIBLE_IOCTL(VUIDSFORMAT) COMPATIBLE_IOCTL(VUIDGFORMAT) /* Little v, the video4linux ioctls */ diff --git a/include/asm-sparc/kbio.h b/include/asm-sparc/kbio.h deleted file mode 100644 index 3cf496bdf39..00000000000 --- a/include/asm-sparc/kbio.h +++ /dev/null @@ -1,56 +0,0 @@ -#ifndef __LINUX_KBIO_H -#define __LINUX_KBIO_H - -/* Return keyboard type */ -#define KIOCTYPE _IOR('k', 9, int) -/* Return Keyboard layout */ -#define KIOCLAYOUT _IOR('k', 20, int) - -enum { - TR_NONE, - TR_ASCII, /* keyboard is in regular state */ - TR_EVENT, /* keystrokes sent as firm events */ - TR_UNTRANS_EVENT /* EVENT+up and down+no translation */ -}; - -/* Return the current keyboard translation */ -#define KIOCGTRANS _IOR('k', 5, int) -/* Set the keyboard translation */ -#define KIOCTRANS _IOW('k', 0, int) - -/* Send a keyboard command */ -#define KIOCCMD _IOW('k', 8, int) - -/* Return if keystrokes are being sent to /dev/kbd */ - -/* Set routing of keystrokes to /dev/kbd */ -#define KIOCSDIRECT _IOW('k', 10, int) - -/* Set keyboard leds */ -#define KIOCSLED _IOW('k', 14, unsigned char) - -/* Get keyboard leds */ -#define KIOCGLED _IOR('k', 15, unsigned char) - -/* Used by KIOC[GS]RATE */ -struct kbd_rate { - unsigned char delay; /* Delay in Hz before first repeat. */ - unsigned char rate; /* In characters per second (0..50). */ -}; - -/* Set keyboard rate */ -#define KIOCSRATE _IOW('k', 40, struct kbd_rate) - -/* Get keyboard rate */ -#define KIOCGRATE _IOW('k', 41, struct kbd_rate) - -/* Top bit records if the key is up or down */ -#define KBD_UP 0x80 - -/* Usable information */ -#define KBD_KEYMASK 0x7f - -/* All keys up */ -#define KBD_IDLE 0x75 - -#endif /* __LINUX_KBIO_H */ diff --git a/include/asm-sparc64/kbio.h b/include/asm-sparc64/kbio.h deleted file mode 100644 index 3cf496bdf39..00000000000 --- a/include/asm-sparc64/kbio.h +++ /dev/null @@ -1,56 +0,0 @@ -#ifndef __LINUX_KBIO_H -#define __LINUX_KBIO_H - -/* Return keyboard type */ -#define KIOCTYPE _IOR('k', 9, int) -/* Return Keyboard layout */ -#define KIOCLAYOUT _IOR('k', 20, int) - -enum { - TR_NONE, - TR_ASCII, /* keyboard is in regular state */ - TR_EVENT, /* keystrokes sent as firm events */ - TR_UNTRANS_EVENT /* EVENT+up and down+no translation */ -}; - -/* Return the current keyboard translation */ -#define KIOCGTRANS _IOR('k', 5, int) -/* Set the keyboard translation */ -#define KIOCTRANS _IOW('k', 0, int) - -/* Send a keyboard command */ -#define KIOCCMD _IOW('k', 8, int) - -/* Return if keystrokes are being sent to /dev/kbd */ - -/* Set routing of keystrokes to /dev/kbd */ -#define KIOCSDIRECT _IOW('k', 10, int) - -/* Set keyboard leds */ -#define KIOCSLED _IOW('k', 14, unsigned char) - -/* Get keyboard leds */ -#define KIOCGLED _IOR('k', 15, unsigned char) - -/* Used by KIOC[GS]RATE */ -struct kbd_rate { - unsigned char delay; /* Delay in Hz before first repeat. */ - unsigned char rate; /* In characters per second (0..50). */ -}; - -/* Set keyboard rate */ -#define KIOCSRATE _IOW('k', 40, struct kbd_rate) - -/* Get keyboard rate */ -#define KIOCGRATE _IOW('k', 41, struct kbd_rate) - -/* Top bit records if the key is up or down */ -#define KBD_UP 0x80 - -/* Usable information */ -#define KBD_KEYMASK 0x7f - -/* All keys up */ -#define KBD_IDLE 0x75 - -#endif /* __LINUX_KBIO_H */ -- cgit v1.2.3-70-g09d2 From 59f85dc95e81281b424b2eb0e7b002cf7f77db03 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:11:38 -0800 Subject: [SPARC]: remove vuid_event.h I don't know if we ever implemented this, but the only user in any 2.6 tree are the compat ioctls. Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 3 --- include/asm-m68k/vuid_event.h | 4 ---- include/asm-sparc/vuid_event.h | 41 ---------------------------------------- include/asm-sparc64/vuid_event.h | 40 --------------------------------------- 4 files changed, 88 deletions(-) delete mode 100644 include/asm-m68k/vuid_event.h delete mode 100644 include/asm-sparc/vuid_event.h delete mode 100644 include/asm-sparc64/vuid_event.h (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 947bd265aaa..94e2b99802c 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -14,7 +14,6 @@ #include #include #include -#include #include #include #include @@ -472,8 +471,6 @@ COMPATIBLE_IOCTL(FBIOSCURPOS) COMPATIBLE_IOCTL(FBIOGCURPOS) COMPATIBLE_IOCTL(FBIOGCURMAX) /* Little k */ -COMPATIBLE_IOCTL(VUIDSFORMAT) -COMPATIBLE_IOCTL(VUIDGFORMAT) /* Little v, the video4linux ioctls */ COMPATIBLE_IOCTL(_IOR('p', 20, int[7])) /* RTCGET */ COMPATIBLE_IOCTL(_IOW('p', 21, int[7])) /* RTCSET */ diff --git a/include/asm-m68k/vuid_event.h b/include/asm-m68k/vuid_event.h deleted file mode 100644 index 52ecb521a39..00000000000 --- a/include/asm-m68k/vuid_event.h +++ /dev/null @@ -1,4 +0,0 @@ -#ifndef _M68K_VUID_EVENT_H -#define _M68K_VUID_EVENT_H -#include -#endif diff --git a/include/asm-sparc/vuid_event.h b/include/asm-sparc/vuid_event.h deleted file mode 100644 index 7781e9f2fdd..00000000000 --- a/include/asm-sparc/vuid_event.h +++ /dev/null @@ -1,41 +0,0 @@ -/* SunOS Virtual User Input Device (VUID) compatibility */ - - -typedef struct firm_event { - unsigned short id; /* tag for this event */ - unsigned char pair_type; /* unused by X11 */ - unsigned char pair; /* unused by X11 */ - int value; /* VKEY_UP, VKEY_DOWN or delta */ - struct timeval time; -} Firm_event; - -enum { - FE_PAIR_NONE, - FE_PAIR_SET, - FE_PAIR_DELTA, - FE_PAIR_ABSOLUTE -}; - -/* VUID stream formats */ -#define VUID_NATIVE 0 /* Native byte stream format */ -#define VUID_FIRM_EVENT 1 /* send firm_event structures */ - -/* ioctls */ - /* Set input device byte stream format (any of VUID_{NATIVE,FIRM_EVENT}) */ -#define VUIDSFORMAT _IOW('v', 1, int) - /* Retrieve input device byte stream format */ -#define VUIDGFORMAT _IOR('v', 2, int) - -/* Possible tag values */ -/* mouse buttons: */ -#define MS_LEFT 0x7f20 -#define MS_MIDDLE 0x7f21 -#define MS_RIGHT 0x7f22 -/* motion: */ -#define LOC_X_DELTA 0x7f80 -#define LOC_Y_DELTA 0x7f81 -#define LOC_X_ABSOLUTE 0x7f82 /* X compat, unsupported */ -#define LOC_Y_ABSOLUTE 0x7f83 /* X compat, unsupported */ - -#define VKEY_UP 0 -#define VKEY_DOWN 1 diff --git a/include/asm-sparc64/vuid_event.h b/include/asm-sparc64/vuid_event.h deleted file mode 100644 index 9ef4d17ad08..00000000000 --- a/include/asm-sparc64/vuid_event.h +++ /dev/null @@ -1,40 +0,0 @@ -/* SunOS Virtual User Input Device (VUID) compatibility */ - -typedef struct firm_event { - unsigned short id; /* tag for this event */ - unsigned char pair_type; /* unused by X11 */ - unsigned char pair; /* unused by X11 */ - int value; /* VKEY_UP, VKEY_DOWN or delta */ - struct timeval time; -} Firm_event; - -enum { - FE_PAIR_NONE, - FE_PAIR_SET, - FE_PAIR_DELTA, - FE_PAIR_ABSOLUTE -}; - -/* VUID stream formats */ -#define VUID_NATIVE 0 /* Native byte stream format */ -#define VUID_FIRM_EVENT 1 /* send firm_event structures */ - -/* ioctls */ - /* Set input device byte stream format (any of VUID_{NATIVE,FIRM_EVENT}) */ -#define VUIDSFORMAT _IOW('v', 1, int) - /* Retrieve input device byte stream format */ -#define VUIDGFORMAT _IOR('v', 2, int) - -/* Possible tag values */ -/* mouse buttons: */ -#define MS_LEFT 0x7f20 -#define MS_MIDDLE 0x7f21 -#define MS_RIGHT 0x7f22 -/* motion: */ -#define LOC_X_DELTA 0x7f80 -#define LOC_Y_DELTA 0x7f81 -#define LOC_X_ABSOLUTE 0x7f82 /* X compat, unsupported */ -#define LOC_Y_ABSOLUTE 0x7f83 /* X compat, unsupported */ - -#define VKEY_UP 0 -#define VKEY_DOWN 1 -- cgit v1.2.3-70-g09d2 From 261b033afc2db37ad371263db2e1316f37c8ed51 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:11:49 -0800 Subject: [SPARC64]: remove duplicated compat ioctl entries all these are handled by fs/compat_ioctls.c already. Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 94e2b99802c..fa486229c61 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -505,11 +505,7 @@ COMPATIBLE_IOCTL(OPROMGETBOOTARGS) COMPATIBLE_IOCTL(OPROMSETCUR) COMPATIBLE_IOCTL(OPROMPCI2NODE) COMPATIBLE_IOCTL(OPROMPATH2NODE) -/* Big L */ -COMPATIBLE_IOCTL(LOOP_SET_STATUS64) -COMPATIBLE_IOCTL(LOOP_GET_STATUS64) /* Big A */ -COMPATIBLE_IOCTL(AUTOFS_IOC_EXPIRE_MULTI) #if defined(CONFIG_DRM) || defined(CONFIG_DRM_MODULE) COMPATIBLE_IOCTL(DRM_IOCTL_GET_MAGIC) COMPATIBLE_IOCTL(DRM_IOCTL_IRQ_BUSID) -- cgit v1.2.3-70-g09d2 From 16cf0d816541fde06ed8f37c0f5cf9940cdfc145 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:12:21 -0800 Subject: [SPARC]: Kill remaining kbio.h references. Would you mind applying the following patch that kills those two + the m68k and Documentation/ references? Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- Documentation/ioctl-number.txt | 2 -- arch/sparc/kernel/sunos_ioctl.c | 1 - arch/sparc64/kernel/sunos_ioctl32.c | 1 - include/asm-m68k/kbio.h | 1 - 4 files changed, 5 deletions(-) delete mode 100644 include/asm-m68k/kbio.h (limited to 'arch/sparc64/kernel') diff --git a/Documentation/ioctl-number.txt b/Documentation/ioctl-number.txt index 769f925c852..87f4d052e39 100644 --- a/Documentation/ioctl-number.txt +++ b/Documentation/ioctl-number.txt @@ -130,8 +130,6 @@ Code Seq# Include File Comments 'i' 00-3F linux/i2o.h 'j' 00-3F linux/joystick.h -'k' all asm-sparc/kbio.h - asm-sparc64/kbio.h 'l' 00-3F linux/tcfs_fs.h transparent cryptographic file system 'l' 40-7F linux/udf_fs_i.h in development: diff --git a/arch/sparc/kernel/sunos_ioctl.c b/arch/sparc/kernel/sunos_ioctl.c index df1c0b31a93..a6ba3d26222 100644 --- a/arch/sparc/kernel/sunos_ioctl.c +++ b/arch/sparc/kernel/sunos_ioctl.c @@ -23,7 +23,6 @@ #include #include #include -#include #if 0 extern char sunkbd_type; diff --git a/arch/sparc64/kernel/sunos_ioctl32.c b/arch/sparc64/kernel/sunos_ioctl32.c index 7654b8a7f03..3f619ead22c 100644 --- a/arch/sparc64/kernel/sunos_ioctl32.c +++ b/arch/sparc64/kernel/sunos_ioctl32.c @@ -24,7 +24,6 @@ #include #include #include -#include #define SUNOS_NR_OPEN 256 diff --git a/include/asm-m68k/kbio.h b/include/asm-m68k/kbio.h deleted file mode 100644 index e1fbf8fba3e..00000000000 --- a/include/asm-m68k/kbio.h +++ /dev/null @@ -1 +0,0 @@ -#include -- cgit v1.2.3-70-g09d2 From 1928f8e541245eae933f8c95b64b2bc3683f9661 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:12:34 -0800 Subject: [SPARC] envctrl: implement ->unlocked_ioctl and ->compat_ioctl all the ioctls in the driver are 32bit compat clean and don't need BKL, so we can switch it to ->unlocked_ioctl and ->compat_ioctl trivially. Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 10 ---------- drivers/sbus/char/envctrl.c | 18 ++++++++++-------- 2 files changed, 10 insertions(+), 18 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index fa486229c61..d20c8098cdf 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -474,16 +474,6 @@ COMPATIBLE_IOCTL(FBIOGCURMAX) /* Little v, the video4linux ioctls */ COMPATIBLE_IOCTL(_IOR('p', 20, int[7])) /* RTCGET */ COMPATIBLE_IOCTL(_IOW('p', 21, int[7])) /* RTCSET */ -COMPATIBLE_IOCTL(ENVCTRL_RD_WARNING_TEMPERATURE) -COMPATIBLE_IOCTL(ENVCTRL_RD_SHUTDOWN_TEMPERATURE) -COMPATIBLE_IOCTL(ENVCTRL_RD_CPU_TEMPERATURE) -COMPATIBLE_IOCTL(ENVCTRL_RD_FAN_STATUS) -COMPATIBLE_IOCTL(ENVCTRL_RD_VOLTAGE_STATUS) -COMPATIBLE_IOCTL(ENVCTRL_RD_SCSI_TEMPERATURE) -COMPATIBLE_IOCTL(ENVCTRL_RD_ETHERNET_TEMPERATURE) -COMPATIBLE_IOCTL(ENVCTRL_RD_MTHRBD_TEMPERATURE) -COMPATIBLE_IOCTL(ENVCTRL_RD_CPU_VOLTAGE) -COMPATIBLE_IOCTL(ENVCTRL_RD_GLOBALADDRESS) /* COMPATIBLE_IOCTL(D7SIOCRD) same value as ENVCTRL_RD_VOLTAGE_STATUS */ COMPATIBLE_IOCTL(D7SIOCWR) COMPATIBLE_IOCTL(D7SIOCTM) diff --git a/drivers/sbus/char/envctrl.c b/drivers/sbus/char/envctrl.c index ba56762b05f..19e8eddf887 100644 --- a/drivers/sbus/char/envctrl.c +++ b/drivers/sbus/char/envctrl.c @@ -654,9 +654,8 @@ envctrl_read(struct file *file, char __user *buf, size_t count, loff_t *ppos) /* Function Description: Command what to read. Mapped to user ioctl(). * Return: Gives 0 for implemented commands, -EINVAL otherwise. */ -static int -envctrl_ioctl(struct inode *inode, struct file *file, - unsigned int cmd, unsigned long arg) +static long +envctrl_ioctl(struct file *file, unsigned int cmd, unsigned long arg) { char __user *infobuf; @@ -715,11 +714,14 @@ envctrl_release(struct inode *inode, struct file *file) } static struct file_operations envctrl_fops = { - .owner = THIS_MODULE, - .read = envctrl_read, - .ioctl = envctrl_ioctl, - .open = envctrl_open, - .release = envctrl_release, + .owner = THIS_MODULE, + .read = envctrl_read, + .unlocked_ioctl = envctrl_ioctl, +#ifdef CONFIG_COMPAT + .compat_ioctl = envctrl_ioctl, +#endif + .open = envctrl_open, + .release = envctrl_release, }; static struct miscdevice envctrl_dev = { -- cgit v1.2.3-70-g09d2 From b31023fc24e5c39d246e9c6fc75dba1a2902c1d6 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:12:47 -0800 Subject: [SPARC] openprom: implement ->compat_ioctl implement a compat_ioctl handle in the driver instead of having table entries in sparc64 ioctl32.c (I plan to get rid of the arch ioctl32.c file eventually) Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 18 ------------------ drivers/sbus/char/openprom.c | 33 +++++++++++++++++++++++++++++++++ 2 files changed, 33 insertions(+), 18 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index d20c8098cdf..ec4e08c523f 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -477,24 +477,6 @@ COMPATIBLE_IOCTL(_IOW('p', 21, int[7])) /* RTCSET */ /* COMPATIBLE_IOCTL(D7SIOCRD) same value as ENVCTRL_RD_VOLTAGE_STATUS */ COMPATIBLE_IOCTL(D7SIOCWR) COMPATIBLE_IOCTL(D7SIOCTM) -/* OPENPROMIO, SunOS/Solaris only, the NetBSD one's have - * embedded pointers in the arg which we'd need to clean up... - */ -COMPATIBLE_IOCTL(OPROMGETOPT) -COMPATIBLE_IOCTL(OPROMSETOPT) -COMPATIBLE_IOCTL(OPROMNXTOPT) -COMPATIBLE_IOCTL(OPROMSETOPT2) -COMPATIBLE_IOCTL(OPROMNEXT) -COMPATIBLE_IOCTL(OPROMCHILD) -COMPATIBLE_IOCTL(OPROMGETPROP) -COMPATIBLE_IOCTL(OPROMNXTPROP) -COMPATIBLE_IOCTL(OPROMU2P) -COMPATIBLE_IOCTL(OPROMGETCONS) -COMPATIBLE_IOCTL(OPROMGETFBNAME) -COMPATIBLE_IOCTL(OPROMGETBOOTARGS) -COMPATIBLE_IOCTL(OPROMSETCUR) -COMPATIBLE_IOCTL(OPROMPCI2NODE) -COMPATIBLE_IOCTL(OPROMPATH2NODE) /* Big A */ #if defined(CONFIG_DRM) || defined(CONFIG_DRM_MODULE) COMPATIBLE_IOCTL(DRM_IOCTL_GET_MAGIC) diff --git a/drivers/sbus/char/openprom.c b/drivers/sbus/char/openprom.c index 58ed3374957..5028ac21432 100644 --- a/drivers/sbus/char/openprom.c +++ b/drivers/sbus/char/openprom.c @@ -39,6 +39,7 @@ #include #include #include +#include #include #include #include @@ -565,6 +566,38 @@ static int openprom_ioctl(struct inode * inode, struct file * file, } } +static long openprom_compat_ioctl(struct file *file, unsigned int cmd, + unsigned long arg) +{ + long rval = -ENOTTY; + + /* + * SunOS/Solaris only, the NetBSD one's have embedded pointers in + * the arg which we'd need to clean up... + */ + switch (cmd) { + case OPROMGETOPT: + case OPROMSETOPT: + case OPROMNXTOPT: + case OPROMSETOPT2: + case OPROMNEXT: + case OPROMCHILD: + case OPROMGETPROP: + case OPROMNXTPROP: + case OPROMU2P: + case OPROMGETCONS: + case OPROMGETFBNAME: + case OPROMGETBOOTARGS: + case OPROMSETCUR: + case OPROMPCI2NODE: + case OPROMPATH2NODE: + lock_kernel(); + rval = openprom_ioctl(file->f_dentry->d_inode, file, cmd, arg); + lock_kernel(); + break; + } +} + static int openprom_open(struct inode * inode, struct file * file) { DATA *data; -- cgit v1.2.3-70-g09d2 From 1d5d00bd9c44ab4730d353ee6ba0c8ebbff295c7 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:13:01 -0800 Subject: [SPARC] display7seg: implement ->unlocked_ioctl and ->compat_ioctl all ioctls are 32bit compat clean, so the driver can use ->compat_ioctl and ->unlocked_ioctl easily. Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 3 --- drivers/sbus/char/display7seg.c | 32 ++++++++++++++++++++------------ 2 files changed, 20 insertions(+), 15 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index ec4e08c523f..f8e9ffb125c 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -474,9 +474,6 @@ COMPATIBLE_IOCTL(FBIOGCURMAX) /* Little v, the video4linux ioctls */ COMPATIBLE_IOCTL(_IOR('p', 20, int[7])) /* RTCGET */ COMPATIBLE_IOCTL(_IOW('p', 21, int[7])) /* RTCSET */ -/* COMPATIBLE_IOCTL(D7SIOCRD) same value as ENVCTRL_RD_VOLTAGE_STATUS */ -COMPATIBLE_IOCTL(D7SIOCWR) -COMPATIBLE_IOCTL(D7SIOCTM) /* Big A */ #if defined(CONFIG_DRM) || defined(CONFIG_DRM_MODULE) COMPATIBLE_IOCTL(DRM_IOCTL_GET_MAGIC) diff --git a/drivers/sbus/char/display7seg.c b/drivers/sbus/char/display7seg.c index 24ed5893b4f..39f54213a6d 100644 --- a/drivers/sbus/char/display7seg.c +++ b/drivers/sbus/char/display7seg.c @@ -15,6 +15,7 @@ #include #include #include /* request_region */ +#include #include #include /* EBus device */ #include /* OpenProm Library */ @@ -114,22 +115,25 @@ static int d7s_release(struct inode *inode, struct file *f) return 0; } -static int d7s_ioctl(struct inode *inode, struct file *f, - unsigned int cmd, unsigned long arg) +static long d7s_ioctl(struct file *file, unsigned int cmd, unsigned long arg) { __u8 regs = readb(d7s_regs); __u8 ireg = 0; + int error = 0 - if (D7S_MINOR != iminor(inode)) + if (D7S_MINOR != iminor(file->f_dentry->d_inode)) return -ENODEV; + lock_kernel(); switch (cmd) { case D7SIOCWR: /* assign device register values * we mask-out D7S_FLIP if in sol_compat mode */ - if (get_user(ireg, (int __user *) arg)) - return -EFAULT; + if (get_user(ireg, (int __user *) arg)) { + error = -EFAULT; + break; + } if (0 != sol_compat) { (regs & D7S_FLIP) ? (ireg |= D7S_FLIP) : (ireg &= ~D7S_FLIP); @@ -144,8 +148,10 @@ static int d7s_ioctl(struct inode *inode, struct file *f, * This driver will not misinform you about the state * of your hardware while in sol_compat mode */ - if (put_user(regs, (int __user *) arg)) - return -EFAULT; + if (put_user(regs, (int __user *) arg)) { + error = -EFAULT; + break; + } break; case D7SIOCTM: @@ -155,15 +161,17 @@ static int d7s_ioctl(struct inode *inode, struct file *f, writeb(regs, d7s_regs); break; }; + lock_kernel(); - return 0; + return error; } static struct file_operations d7s_fops = { - .owner = THIS_MODULE, - .ioctl = d7s_ioctl, - .open = d7s_open, - .release = d7s_release, + .owner = THIS_MODULE, + .unlocked_ioctl = d7s_ioctl, + .compat_ioctl = d7s_ioctl, + .open = d7s_open, + .release = d7s_release, }; static struct miscdevice d7s_miscdev = { D7S_MINOR, D7S_DEVNAME, &d7s_fops }; -- cgit v1.2.3-70-g09d2 From b66621fef30e15810d459212bc8bdc274e08f14f Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:13:14 -0800 Subject: [SPARC] cpwatchdog: implement ->compat_ioctl Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 3 --- drivers/sbus/char/cpwatchdog.c | 24 ++++++++++++++++++++++++ 2 files changed, 24 insertions(+), 3 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index f8e9ffb125c..398ddbffc6a 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -496,9 +496,6 @@ COMPATIBLE_IOCTL(DRM_IOCTL_LOCK) COMPATIBLE_IOCTL(DRM_IOCTL_UNLOCK) COMPATIBLE_IOCTL(DRM_IOCTL_FINISH) #endif /* DRM */ -COMPATIBLE_IOCTL(WIOCSTART) -COMPATIBLE_IOCTL(WIOCSTOP) -COMPATIBLE_IOCTL(WIOCGSTAT) /* And these ioctls need translation */ /* Note SIOCRTMSG is no longer, so this is safe and * the user would have seen just an -EINVAL anyways. */ HANDLE_IOCTL(FBIOPUTCMAP32, fbiogetputcmap) diff --git a/drivers/sbus/char/cpwatchdog.c b/drivers/sbus/char/cpwatchdog.c index c82abeb59d3..071ae24be89 100644 --- a/drivers/sbus/char/cpwatchdog.c +++ b/drivers/sbus/char/cpwatchdog.c @@ -26,6 +26,7 @@ #include #include #include +#include #include #include #include @@ -394,6 +395,28 @@ static int wd_ioctl(struct inode *inode, struct file *file, return(0); } +static long wd_compat_ioctl(struct file *file, unsigned int cmd, + unsigned long arg) +{ + int rval = -ENOIOCTLCMD; + + switch (cmd) { + /* solaris ioctls are specific to this driver */ + case WIOCSTART: + case WIOCSTOP: + case WIOCGSTAT: + lock_kernel(); + rval = wd_ioctl(file->f_dentry->d_inode, file, cmd, arg); + lock_kernel(); + break; + /* everything else is handled by the generic compat layer */ + default: + break; + } + + return rval; +} + static ssize_t wd_write(struct file *file, const char __user *buf, size_t count, @@ -441,6 +464,7 @@ static irqreturn_t wd_interrupt(int irq, void *dev_id, struct pt_regs *regs) static struct file_operations wd_fops = { .owner = THIS_MODULE, .ioctl = wd_ioctl, + .compat_ioctl = wd_compat_ioctl, .open = wd_open, .write = wd_write, .read = wd_read, -- cgit v1.2.3-70-g09d2 From f48497e38331464c25e564d9e76ee915ca55fea8 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 7 Nov 2005 14:13:27 -0800 Subject: [SPARC64]: remove drm compat ioctl handling drivers/drm/ now implements proper ->compat_ioctl methods, so this isn't needed anymore. Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 384 ------------------------------------------ 1 file changed, 384 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 398ddbffc6a..5eab41c1b1c 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -97,357 +97,6 @@ static int fbiogscursor(unsigned int fd, unsigned int cmd, unsigned long arg) return sys_ioctl (fd, FBIOSCURSOR, (unsigned long)p); } -#if defined(CONFIG_DRM) || defined(CONFIG_DRM_MODULE) -/* This really belongs in include/linux/drm.h -DaveM */ -#include "../../../drivers/char/drm/drm.h" - -typedef struct drm32_version { - int version_major; /* Major version */ - int version_minor; /* Minor version */ - int version_patchlevel;/* Patch level */ - int name_len; /* Length of name buffer */ - u32 name; /* Name of driver */ - int date_len; /* Length of date buffer */ - u32 date; /* User-space buffer to hold date */ - int desc_len; /* Length of desc buffer */ - u32 desc; /* User-space buffer to hold desc */ -} drm32_version_t; -#define DRM32_IOCTL_VERSION DRM_IOWR(0x00, drm32_version_t) - -static int drm32_version(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - drm32_version_t __user *uversion = (drm32_version_t __user *)arg; - drm_version_t __user *p = compat_alloc_user_space(sizeof(*p)); - compat_uptr_t addr; - int n; - int ret; - - if (clear_user(p, 3 * sizeof(int)) || - get_user(n, &uversion->name_len) || - put_user(n, &p->name_len) || - get_user(addr, &uversion->name) || - put_user(compat_ptr(addr), &p->name) || - get_user(n, &uversion->date_len) || - put_user(n, &p->date_len) || - get_user(addr, &uversion->date) || - put_user(compat_ptr(addr), &p->date) || - get_user(n, &uversion->desc_len) || - put_user(n, &p->desc_len) || - get_user(addr, &uversion->desc) || - put_user(compat_ptr(addr), &p->desc)) - return -EFAULT; - - ret = sys_ioctl(fd, DRM_IOCTL_VERSION, (unsigned long)p); - if (ret) - return ret; - - if (copy_in_user(uversion, p, 3 * sizeof(int)) || - get_user(n, &p->name_len) || - put_user(n, &uversion->name_len) || - get_user(n, &p->date_len) || - put_user(n, &uversion->date_len) || - get_user(n, &p->desc_len) || - put_user(n, &uversion->desc_len)) - return -EFAULT; - - return 0; -} - -typedef struct drm32_unique { - int unique_len; /* Length of unique */ - u32 unique; /* Unique name for driver instantiation */ -} drm32_unique_t; -#define DRM32_IOCTL_GET_UNIQUE DRM_IOWR(0x01, drm32_unique_t) -#define DRM32_IOCTL_SET_UNIQUE DRM_IOW( 0x10, drm32_unique_t) - -static int drm32_getsetunique(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - drm32_unique_t __user *uarg = (drm32_unique_t __user *)arg; - drm_unique_t __user *p = compat_alloc_user_space(sizeof(*p)); - compat_uptr_t addr; - int n; - int ret; - - if (get_user(n, &uarg->unique_len) || - put_user(n, &p->unique_len) || - get_user(addr, &uarg->unique) || - put_user(compat_ptr(addr), &p->unique)) - return -EFAULT; - - if (cmd == DRM32_IOCTL_GET_UNIQUE) - ret = sys_ioctl (fd, DRM_IOCTL_GET_UNIQUE, (unsigned long)p); - else - ret = sys_ioctl (fd, DRM_IOCTL_SET_UNIQUE, (unsigned long)p); - - if (ret) - return ret; - - if (get_user(n, &p->unique_len) || put_user(n, &uarg->unique_len)) - return -EFAULT; - - return 0; -} - -typedef struct drm32_map { - u32 offset; /* Requested physical address (0 for SAREA)*/ - u32 size; /* Requested physical size (bytes) */ - drm_map_type_t type; /* Type of memory to map */ - drm_map_flags_t flags; /* Flags */ - u32 handle; /* User-space: "Handle" to pass to mmap */ - /* Kernel-space: kernel-virtual address */ - int mtrr; /* MTRR slot used */ - /* Private data */ -} drm32_map_t; -#define DRM32_IOCTL_ADD_MAP DRM_IOWR(0x15, drm32_map_t) - -static int drm32_addmap(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - drm32_map_t __user *uarg = (drm32_map_t __user *) arg; - drm_map_t karg; - mm_segment_t old_fs; - u32 tmp; - int ret; - - ret = get_user(karg.offset, &uarg->offset); - ret |= get_user(karg.size, &uarg->size); - ret |= get_user(karg.type, &uarg->type); - ret |= get_user(karg.flags, &uarg->flags); - ret |= get_user(tmp, &uarg->handle); - ret |= get_user(karg.mtrr, &uarg->mtrr); - if (ret) - return -EFAULT; - - karg.handle = (void *) (unsigned long) tmp; - - old_fs = get_fs(); - set_fs(KERNEL_DS); - ret = sys_ioctl(fd, DRM_IOCTL_ADD_MAP, (unsigned long) &karg); - set_fs(old_fs); - - if (!ret) { - ret = put_user(karg.offset, &uarg->offset); - ret |= put_user(karg.size, &uarg->size); - ret |= put_user(karg.type, &uarg->type); - ret |= put_user(karg.flags, &uarg->flags); - tmp = (u32) (long)karg.handle; - ret |= put_user(tmp, &uarg->handle); - ret |= put_user(karg.mtrr, &uarg->mtrr); - if (ret) - ret = -EFAULT; - } - - return ret; -} - -typedef struct drm32_buf_info { - int count; /* Entries in list */ - u32 list; /* (drm_buf_desc_t *) */ -} drm32_buf_info_t; -#define DRM32_IOCTL_INFO_BUFS DRM_IOWR(0x18, drm32_buf_info_t) - -static int drm32_info_bufs(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - drm32_buf_info_t __user *uarg = (drm32_buf_info_t __user *)arg; - drm_buf_info_t __user *p = compat_alloc_user_space(sizeof(*p)); - compat_uptr_t addr; - int n; - int ret; - - if (get_user(n, &uarg->count) || put_user(n, &p->count) || - get_user(addr, &uarg->list) || put_user(compat_ptr(addr), &p->list)) - return -EFAULT; - - ret = sys_ioctl(fd, DRM_IOCTL_INFO_BUFS, (unsigned long)p); - if (ret) - return ret; - - if (get_user(n, &p->count) || put_user(n, &uarg->count)) - return -EFAULT; - - return 0; -} - -typedef struct drm32_buf_free { - int count; - u32 list; /* (int *) */ -} drm32_buf_free_t; -#define DRM32_IOCTL_FREE_BUFS DRM_IOW( 0x1a, drm32_buf_free_t) - -static int drm32_free_bufs(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - drm32_buf_free_t __user *uarg = (drm32_buf_free_t __user *)arg; - drm_buf_free_t __user *p = compat_alloc_user_space(sizeof(*p)); - compat_uptr_t addr; - int n; - - if (get_user(n, &uarg->count) || put_user(n, &p->count) || - get_user(addr, &uarg->list) || put_user(compat_ptr(addr), &p->list)) - return -EFAULT; - - return sys_ioctl(fd, DRM_IOCTL_FREE_BUFS, (unsigned long)p); -} - -typedef struct drm32_buf_pub { - int idx; /* Index into master buflist */ - int total; /* Buffer size */ - int used; /* Amount of buffer in use (for DMA) */ - u32 address; /* Address of buffer (void *) */ -} drm32_buf_pub_t; - -typedef struct drm32_buf_map { - int count; /* Length of buflist */ - u32 virtual; /* Mmaped area in user-virtual (void *) */ - u32 list; /* Buffer information (drm_buf_pub_t *) */ -} drm32_buf_map_t; -#define DRM32_IOCTL_MAP_BUFS DRM_IOWR(0x19, drm32_buf_map_t) - -static int drm32_map_bufs(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - drm32_buf_map_t __user *uarg = (drm32_buf_map_t __user *)arg; - drm32_buf_pub_t __user *ulist; - drm_buf_map_t __user *arg64; - drm_buf_pub_t __user *list; - int orig_count, ret, i; - int n; - compat_uptr_t addr; - - if (get_user(orig_count, &uarg->count)) - return -EFAULT; - - arg64 = compat_alloc_user_space(sizeof(drm_buf_map_t) + - (size_t)orig_count * sizeof(drm_buf_pub_t)); - list = (void __user *)(arg64 + 1); - - if (put_user(orig_count, &arg64->count) || - put_user(list, &arg64->list) || - get_user(addr, &uarg->virtual) || - put_user(compat_ptr(addr), &arg64->virtual) || - get_user(addr, &uarg->list)) - return -EFAULT; - - ulist = compat_ptr(addr); - - for (i = 0; i < orig_count; i++) { - if (get_user(n, &ulist[i].idx) || - put_user(n, &list[i].idx) || - get_user(n, &ulist[i].total) || - put_user(n, &list[i].total) || - get_user(n, &ulist[i].used) || - put_user(n, &list[i].used) || - get_user(addr, &ulist[i].address) || - put_user(compat_ptr(addr), &list[i].address)) - return -EFAULT; - } - - ret = sys_ioctl(fd, DRM_IOCTL_MAP_BUFS, (unsigned long) arg64); - if (ret) - return ret; - - for (i = 0; i < orig_count; i++) { - void __user *p; - if (get_user(n, &list[i].idx) || - put_user(n, &ulist[i].idx) || - get_user(n, &list[i].total) || - put_user(n, &ulist[i].total) || - get_user(n, &list[i].used) || - put_user(n, &ulist[i].used) || - get_user(p, &list[i].address) || - put_user((unsigned long)p, &ulist[i].address)) - return -EFAULT; - } - - if (get_user(n, &arg64->count) || put_user(n, &uarg->count)) - return -EFAULT; - - return 0; -} - -typedef struct drm32_dma { - /* Indices here refer to the offset into - buflist in drm_buf_get_t. */ - int context; /* Context handle */ - int send_count; /* Number of buffers to send */ - u32 send_indices; /* List of handles to buffers (int *) */ - u32 send_sizes; /* Lengths of data to send (int *) */ - drm_dma_flags_t flags; /* Flags */ - int request_count; /* Number of buffers requested */ - int request_size; /* Desired size for buffers */ - u32 request_indices; /* Buffer information (int *) */ - u32 request_sizes; /* (int *) */ - int granted_count; /* Number of buffers granted */ -} drm32_dma_t; -#define DRM32_IOCTL_DMA DRM_IOWR(0x29, drm32_dma_t) - -/* RED PEN The DRM layer blindly dereferences the send/request - * index/size arrays even though they are userland - * pointers. -DaveM - */ -static int drm32_dma(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - drm32_dma_t __user *uarg = (drm32_dma_t __user *) arg; - drm_dma_t __user *p = compat_alloc_user_space(sizeof(*p)); - compat_uptr_t addr; - int ret; - - if (copy_in_user(p, uarg, 2 * sizeof(int)) || - get_user(addr, &uarg->send_indices) || - put_user(compat_ptr(addr), &p->send_indices) || - get_user(addr, &uarg->send_sizes) || - put_user(compat_ptr(addr), &p->send_sizes) || - copy_in_user(&p->flags, &uarg->flags, sizeof(drm_dma_flags_t)) || - copy_in_user(&p->request_count, &uarg->request_count, sizeof(int))|| - copy_in_user(&p->request_size, &uarg->request_size, sizeof(int)) || - get_user(addr, &uarg->request_indices) || - put_user(compat_ptr(addr), &p->request_indices) || - get_user(addr, &uarg->request_sizes) || - put_user(compat_ptr(addr), &p->request_sizes) || - copy_in_user(&p->granted_count, &uarg->granted_count, sizeof(int))) - return -EFAULT; - - ret = sys_ioctl(fd, DRM_IOCTL_DMA, (unsigned long)p); - if (ret) - return ret; - - if (copy_in_user(uarg, p, 2 * sizeof(int)) || - copy_in_user(&uarg->flags, &p->flags, sizeof(drm_dma_flags_t)) || - copy_in_user(&uarg->request_count, &p->request_count, sizeof(int))|| - copy_in_user(&uarg->request_size, &p->request_size, sizeof(int)) || - copy_in_user(&uarg->granted_count, &p->granted_count, sizeof(int))) - return -EFAULT; - - return 0; -} - -typedef struct drm32_ctx_res { - int count; - u32 contexts; /* (drm_ctx_t *) */ -} drm32_ctx_res_t; -#define DRM32_IOCTL_RES_CTX DRM_IOWR(0x26, drm32_ctx_res_t) - -static int drm32_res_ctx(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - drm32_ctx_res_t __user *uarg = (drm32_ctx_res_t __user *) arg; - drm_ctx_res_t __user *p = compat_alloc_user_space(sizeof(*p)); - compat_uptr_t addr; - int ret; - - if (copy_in_user(p, uarg, sizeof(int)) || - get_user(addr, &uarg->contexts) || - put_user(compat_ptr(addr), &p->contexts)) - return -EFAULT; - - ret = sys_ioctl(fd, DRM_IOCTL_RES_CTX, (unsigned long)p); - if (ret) - return ret; - - if (copy_in_user(uarg, p, sizeof(int))) - return -EFAULT; - - return 0; -} - -#endif - typedef int (* ioctl32_handler_t)(unsigned int, unsigned int, unsigned long, struct file *); #define COMPATIBLE_IOCTL(cmd) HANDLE_IOCTL((cmd),sys_ioctl) @@ -474,44 +123,11 @@ COMPATIBLE_IOCTL(FBIOGCURMAX) /* Little v, the video4linux ioctls */ COMPATIBLE_IOCTL(_IOR('p', 20, int[7])) /* RTCGET */ COMPATIBLE_IOCTL(_IOW('p', 21, int[7])) /* RTCSET */ -/* Big A */ -#if defined(CONFIG_DRM) || defined(CONFIG_DRM_MODULE) -COMPATIBLE_IOCTL(DRM_IOCTL_GET_MAGIC) -COMPATIBLE_IOCTL(DRM_IOCTL_IRQ_BUSID) -COMPATIBLE_IOCTL(DRM_IOCTL_AUTH_MAGIC) -COMPATIBLE_IOCTL(DRM_IOCTL_BLOCK) -COMPATIBLE_IOCTL(DRM_IOCTL_UNBLOCK) -COMPATIBLE_IOCTL(DRM_IOCTL_CONTROL) -COMPATIBLE_IOCTL(DRM_IOCTL_ADD_BUFS) -COMPATIBLE_IOCTL(DRM_IOCTL_MARK_BUFS) -COMPATIBLE_IOCTL(DRM_IOCTL_ADD_CTX) -COMPATIBLE_IOCTL(DRM_IOCTL_RM_CTX) -COMPATIBLE_IOCTL(DRM_IOCTL_MOD_CTX) -COMPATIBLE_IOCTL(DRM_IOCTL_GET_CTX) -COMPATIBLE_IOCTL(DRM_IOCTL_SWITCH_CTX) -COMPATIBLE_IOCTL(DRM_IOCTL_NEW_CTX) -COMPATIBLE_IOCTL(DRM_IOCTL_ADD_DRAW) -COMPATIBLE_IOCTL(DRM_IOCTL_RM_DRAW) -COMPATIBLE_IOCTL(DRM_IOCTL_LOCK) -COMPATIBLE_IOCTL(DRM_IOCTL_UNLOCK) -COMPATIBLE_IOCTL(DRM_IOCTL_FINISH) -#endif /* DRM */ /* And these ioctls need translation */ /* Note SIOCRTMSG is no longer, so this is safe and * the user would have seen just an -EINVAL anyways. */ HANDLE_IOCTL(FBIOPUTCMAP32, fbiogetputcmap) HANDLE_IOCTL(FBIOGETCMAP32, fbiogetputcmap) HANDLE_IOCTL(FBIOSCURSOR32, fbiogscursor) -#if defined(CONFIG_DRM) || defined(CONFIG_DRM_MODULE) -HANDLE_IOCTL(DRM32_IOCTL_VERSION, drm32_version) -HANDLE_IOCTL(DRM32_IOCTL_GET_UNIQUE, drm32_getsetunique) -HANDLE_IOCTL(DRM32_IOCTL_SET_UNIQUE, drm32_getsetunique) -HANDLE_IOCTL(DRM32_IOCTL_ADD_MAP, drm32_addmap) -HANDLE_IOCTL(DRM32_IOCTL_INFO_BUFS, drm32_info_bufs) -HANDLE_IOCTL(DRM32_IOCTL_FREE_BUFS, drm32_free_bufs) -HANDLE_IOCTL(DRM32_IOCTL_MAP_BUFS, drm32_map_bufs) -HANDLE_IOCTL(DRM32_IOCTL_DMA, drm32_dma) -HANDLE_IOCTL(DRM32_IOCTL_RES_CTX, drm32_res_ctx) -#endif /* DRM */ #if 0 HANDLE_IOCTL(RTC32_IRQP_READ, do_rtc_ioctl) HANDLE_IOCTL(RTC32_IRQP_SET, do_rtc_ioctl) -- cgit v1.2.3-70-g09d2 From dd3e2dcf3408843ed35501c28626f389b30be756 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Mon, 7 Nov 2005 14:13:46 -0800 Subject: [SPARC64]: Kill some unnecessary includes from ioctl32.c Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 5 ----- 1 file changed, 5 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 5eab41c1b1c..92e26304de9 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -11,13 +11,8 @@ #define INCLUDES #include "compat_ioctl.c" -#include #include #include -#include -#include -#include -#include /* Use this to get at 32-bit user passed pointers. * See sys_sparc32.c for description about it. -- cgit v1.2.3-70-g09d2 From 7e4c54a2a4f1cec6a652c6014714ee51b7e1311f Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 8 Nov 2005 21:35:07 -0800 Subject: [PATCH] remove ioctl32_handler_t Some architectures define and use this type in their compat_ioctl code, but all of them can easily use the identical ioctl_trans_handler_t type that is defined in common code. Signed-off-by: Christoph Hellwig Cc: Andi Kleen Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/ia64/ia32/ia32_ioctl.c | 4 +--- arch/mips/kernel/ioctl32.c | 4 +--- arch/sparc64/kernel/ioctl32.c | 4 +--- 3 files changed, 3 insertions(+), 9 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/ia64/ia32/ia32_ioctl.c b/arch/ia64/ia32/ia32_ioctl.c index 164b211f417..88739394f6d 100644 --- a/arch/ia64/ia32/ia32_ioctl.c +++ b/arch/ia64/ia32/ia32_ioctl.c @@ -29,10 +29,8 @@ #define CODE #include "compat_ioctl.c" -typedef int (* ioctl32_handler_t)(unsigned int, unsigned int, unsigned long, struct file *); - #define COMPATIBLE_IOCTL(cmd) HANDLE_IOCTL((cmd),sys_ioctl) -#define HANDLE_IOCTL(cmd,handler) { (cmd), (ioctl32_handler_t)(handler), NULL }, +#define HANDLE_IOCTL(cmd,handler) { (cmd), (ioctl_trans_handler_t)(handler), NULL }, #define IOCTL_TABLE_START \ struct ioctl_trans ioctl_start[] = { #define IOCTL_TABLE_END \ diff --git a/arch/mips/kernel/ioctl32.c b/arch/mips/kernel/ioctl32.c index ed9b2da510b..9ea1fc74886 100644 --- a/arch/mips/kernel/ioctl32.c +++ b/arch/mips/kernel/ioctl32.c @@ -26,10 +26,8 @@ long sys_ioctl(unsigned int fd, unsigned int cmd, unsigned long arg); #define CODE #include "compat_ioctl.c" -typedef int (* ioctl32_handler_t)(unsigned int, unsigned int, unsigned long, struct file *); - #define COMPATIBLE_IOCTL(cmd) HANDLE_IOCTL((cmd),sys_ioctl) -#define HANDLE_IOCTL(cmd,handler) { (cmd), (ioctl32_handler_t)(handler), NULL }, +#define HANDLE_IOCTL(cmd,handler) { (cmd), (ioctl_trans_handler_t)(handler), NULL }, #define IOCTL_TABLE_START \ struct ioctl_trans ioctl_start[] = { #define IOCTL_TABLE_END \ diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 92e26304de9..c2e96daa5ab 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -92,10 +92,8 @@ static int fbiogscursor(unsigned int fd, unsigned int cmd, unsigned long arg) return sys_ioctl (fd, FBIOSCURSOR, (unsigned long)p); } -typedef int (* ioctl32_handler_t)(unsigned int, unsigned int, unsigned long, struct file *); - #define COMPATIBLE_IOCTL(cmd) HANDLE_IOCTL((cmd),sys_ioctl) -#define HANDLE_IOCTL(cmd,handler) { (cmd), (ioctl32_handler_t)(handler), NULL }, +#define HANDLE_IOCTL(cmd,handler) { (cmd), (ioctl_trans_handler_t)(handler), NULL }, #define IOCTL_TABLE_START \ struct ioctl_trans ioctl_start[] = { #define IOCTL_TABLE_END \ -- cgit v1.2.3-70-g09d2 From 5bfb5d690f36d316a5f3b4f7775fda996faa6b12 Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Tue, 8 Nov 2005 21:39:01 -0800 Subject: [PATCH] sched: disable preempt in idle tasks Run idle threads with preempt disabled. Also corrected a bugs in arm26's cpu_idle (make it actually call schedule()). How did it ever work before? Might fix the CPU hotplugging hang which Nigel Cunningham noted. We think the bug hits if the idle thread is preempted after checking need_resched() and before going to sleep, then the CPU offlined. After calling stop_machine_run, the CPU eventually returns from preemption and into the idle thread and goes to sleep. The CPU will continue executing previous idle and have no chance to call play_dead. By disabling preemption until we are ready to explicitly schedule, this bug is fixed and the idle threads generally become more robust. From: alexs PPC build fix From: Yoichi Yuasa MIPS build fix Signed-off-by: Nick Piggin Signed-off-by: Yoichi Yuasa Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/arm/kernel/process.c | 4 ++-- arch/arm/kernel/smp.c | 5 ++++- arch/arm26/kernel/process.c | 12 +++++------- arch/cris/arch-v32/kernel/smp.c | 1 + arch/cris/kernel/process.c | 2 ++ arch/frv/kernel/process.c | 6 +++++- arch/h8300/kernel/process.c | 28 +++++++++++++++------------- arch/i386/kernel/process.c | 4 +++- arch/i386/kernel/smpboot.c | 1 + arch/ia64/kernel/process.c | 2 ++ arch/ia64/kernel/smpboot.c | 1 + arch/m32r/kernel/process.c | 2 ++ arch/m32r/kernel/smpboot.c | 1 + arch/m68k/kernel/process.c | 2 ++ arch/mips/kernel/process.c | 2 ++ arch/mips/kernel/smp.c | 4 +++- arch/parisc/kernel/process.c | 2 ++ arch/parisc/kernel/smp.c | 1 + arch/powerpc/platforms/iseries/setup.c | 4 ++++ arch/powerpc/platforms/pseries/setup.c | 4 ++++ arch/ppc/kernel/idle.c | 17 ++++++++++++----- arch/ppc/kernel/smp.c | 1 + arch/ppc64/kernel/idle.c | 4 ++++ arch/s390/kernel/process.c | 11 ++++++++--- arch/s390/kernel/smp.c | 1 + arch/sh/kernel/process.c | 2 ++ arch/sh/kernel/smp.c | 5 ++++- arch/sh64/kernel/process.c | 2 ++ arch/sparc/kernel/process.c | 4 ++++ arch/sparc64/kernel/process.c | 4 ++++ arch/sparc64/kernel/smp.c | 3 +++ arch/v850/kernel/process.c | 16 ++++++++++------ arch/x86_64/kernel/process.c | 2 ++ arch/x86_64/kernel/smpboot.c | 1 + arch/xtensa/kernel/process.c | 3 ++- init/main.c | 4 +++- 36 files changed, 125 insertions(+), 43 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c index ba298277bec..93dd92cc12f 100644 --- a/arch/arm/kernel/process.c +++ b/arch/arm/kernel/process.c @@ -116,13 +116,13 @@ void cpu_idle(void) if (!idle) idle = default_idle; - preempt_disable(); leds_event(led_idle_start); while (!need_resched()) idle(); leds_event(led_idle_end); - preempt_enable(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c index 77e2e9ca89f..e55ea952f7a 100644 --- a/arch/arm/kernel/smp.c +++ b/arch/arm/kernel/smp.c @@ -256,7 +256,9 @@ void __cpuexit cpu_die(void) asmlinkage void __cpuinit secondary_start_kernel(void) { struct mm_struct *mm = &init_mm; - unsigned int cpu = smp_processor_id(); + unsigned int cpu; + + cpu = smp_processor_id(); printk("CPU%u: Booted secondary processor\n", cpu); @@ -273,6 +275,7 @@ asmlinkage void __cpuinit secondary_start_kernel(void) local_flush_tlb_all(); cpu_init(); + preempt_disable(); /* * Give the platform a chance to do its own initialisation. diff --git a/arch/arm26/kernel/process.c b/arch/arm26/kernel/process.c index 9eb9964d32a..15833a0057d 100644 --- a/arch/arm26/kernel/process.c +++ b/arch/arm26/kernel/process.c @@ -74,15 +74,13 @@ __setup("hlt", hlt_setup); void cpu_idle(void) { /* endless idle loop with no priority at all */ - preempt_disable(); while (1) { - while (!need_resched()) { - local_irq_disable(); - if (!need_resched() && !hlt_counter) - local_irq_enable(); - } + while (!need_resched()) + cpu_relax(); + preempt_enable_no_resched(); + schedule(); + preempt_disable(); } - schedule(); } static char reboot_mode = 'h'; diff --git a/arch/cris/arch-v32/kernel/smp.c b/arch/cris/arch-v32/kernel/smp.c index 957f551ba5c..13867f4fad1 100644 --- a/arch/cris/arch-v32/kernel/smp.c +++ b/arch/cris/arch-v32/kernel/smp.c @@ -161,6 +161,7 @@ void __init smp_callin(void) REG_WR(intr_vect, irq_regs[cpu], rw_mask, vect_mask); unmask_irq(IPI_INTR_VECT); unmask_irq(TIMER_INTR_VECT); + preempt_disable(); local_irq_enable(); cpu_set(cpu, cpu_online_map); diff --git a/arch/cris/kernel/process.c b/arch/cris/kernel/process.c index 949a0e40e03..7c80afb1046 100644 --- a/arch/cris/kernel/process.c +++ b/arch/cris/kernel/process.c @@ -218,7 +218,9 @@ void cpu_idle (void) idle = default_idle; idle(); } + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/frv/kernel/process.c b/arch/frv/kernel/process.c index 3001b82b151..54a452136f0 100644 --- a/arch/frv/kernel/process.c +++ b/arch/frv/kernel/process.c @@ -77,16 +77,20 @@ void (*idle)(void) = core_sleep_idle; */ void cpu_idle(void) { + int cpu = smp_processor_id(); + /* endless idle loop with no priority at all */ while (1) { while (!need_resched()) { - irq_stat[smp_processor_id()].idle_timestamp = jiffies; + irq_stat[cpu].idle_timestamp = jiffies; if (!frv_dma_inprogress && idle) idle(); } + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/h8300/kernel/process.c b/arch/h8300/kernel/process.c index 27f1fce64ce..fe21adf3e75 100644 --- a/arch/h8300/kernel/process.c +++ b/arch/h8300/kernel/process.c @@ -53,22 +53,18 @@ asmlinkage void ret_from_fork(void); #if !defined(CONFIG_H8300H_SIM) && !defined(CONFIG_H8S_SIM) void default_idle(void) { - while(1) { - if (!need_resched()) { - local_irq_enable(); - __asm__("sleep"); - local_irq_disable(); - } - schedule(); - } + local_irq_disable(); + if (!need_resched()) { + local_irq_enable(); + /* XXX: race here! What if need_resched() gets set now? */ + __asm__("sleep"); + } else + local_irq_enable(); } #else void default_idle(void) { - while(1) { - if (need_resched()) - schedule(); - } + cpu_relax(); } #endif void (*idle)(void) = default_idle; @@ -81,7 +77,13 @@ void (*idle)(void) = default_idle; */ void cpu_idle(void) { - idle(); + while (1) { + while (!need_resched()) + idle(); + preempt_enable_no_resched(); + schedule(); + preempt_disable(); + } } void machine_restart(char * __unused) diff --git a/arch/i386/kernel/process.c b/arch/i386/kernel/process.c index 7a14fdfd3af..5296e284ea3 100644 --- a/arch/i386/kernel/process.c +++ b/arch/i386/kernel/process.c @@ -179,7 +179,7 @@ static inline void play_dead(void) */ void cpu_idle(void) { - int cpu = raw_smp_processor_id(); + int cpu = smp_processor_id(); /* endless idle loop with no priority at all */ while (1) { @@ -201,7 +201,9 @@ void cpu_idle(void) __get_cpu_var(irq_stat).idle_timestamp = jiffies; idle(); } + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/i386/kernel/smpboot.c b/arch/i386/kernel/smpboot.c index 47ec76794d0..bc5a9d97466 100644 --- a/arch/i386/kernel/smpboot.c +++ b/arch/i386/kernel/smpboot.c @@ -485,6 +485,7 @@ static void __devinit start_secondary(void *unused) * things done here to the most necessary things. */ cpu_init(); + preempt_disable(); smp_callin(); while (!cpu_isset(smp_processor_id(), smp_commenced_mask)) rep_nop(); diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c index 051e050359e..4c621fc3c3b 100644 --- a/arch/ia64/kernel/process.c +++ b/arch/ia64/kernel/process.c @@ -292,7 +292,9 @@ cpu_idle (void) #ifdef CONFIG_SMP normal_xtp(); #endif + preempt_enable_no_resched(); schedule(); + preempt_disable(); check_pgt_cache(); if (cpu_is_offline(smp_processor_id())) play_dead(); diff --git a/arch/ia64/kernel/smpboot.c b/arch/ia64/kernel/smpboot.c index 400a4898712..8f44e7d2df6 100644 --- a/arch/ia64/kernel/smpboot.c +++ b/arch/ia64/kernel/smpboot.c @@ -399,6 +399,7 @@ start_secondary (void *unused) Dprintk("start_secondary: starting CPU 0x%x\n", hard_smp_processor_id()); efi_map_pal_code(); cpu_init(); + preempt_disable(); smp_callin(); cpu_idle(); diff --git a/arch/m32r/kernel/process.c b/arch/m32r/kernel/process.c index ea13a8f4d8b..cc4b571e5db 100644 --- a/arch/m32r/kernel/process.c +++ b/arch/m32r/kernel/process.c @@ -104,7 +104,9 @@ void cpu_idle (void) idle(); } + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/m32r/kernel/smpboot.c b/arch/m32r/kernel/smpboot.c index 640d592ea07..b90c54169fa 100644 --- a/arch/m32r/kernel/smpboot.c +++ b/arch/m32r/kernel/smpboot.c @@ -426,6 +426,7 @@ void __init smp_cpus_done(unsigned int max_cpus) int __init start_secondary(void *unused) { cpu_init(); + preempt_disable(); smp_callin(); while (!cpu_isset(smp_processor_id(), smp_commenced_mask)) cpu_relax(); diff --git a/arch/m68k/kernel/process.c b/arch/m68k/kernel/process.c index 11b1b90ba6b..13d109328a4 100644 --- a/arch/m68k/kernel/process.c +++ b/arch/m68k/kernel/process.c @@ -102,7 +102,9 @@ void cpu_idle(void) while (1) { while (!need_resched()) idle(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/mips/kernel/process.c b/arch/mips/kernel/process.c index 4fe3d5715c4..dd725779d91 100644 --- a/arch/mips/kernel/process.c +++ b/arch/mips/kernel/process.c @@ -52,7 +52,9 @@ ATTRIB_NORET void cpu_idle(void) while (!need_resched()) if (cpu_wait) (*cpu_wait)(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/mips/kernel/smp.c b/arch/mips/kernel/smp.c index fcacf1aae98..25472fcaf71 100644 --- a/arch/mips/kernel/smp.c +++ b/arch/mips/kernel/smp.c @@ -82,7 +82,7 @@ extern ATTRIB_NORET void cpu_idle(void); */ asmlinkage void start_secondary(void) { - unsigned int cpu = smp_processor_id(); + unsigned int cpu; cpu_probe(); cpu_report(); @@ -95,6 +95,8 @@ asmlinkage void start_secondary(void) */ calibrate_delay(); + preempt_disable(); + cpu = smp_processor_id(); cpu_data[cpu].udelay_val = loops_per_jiffy; prom_smp_finish(); diff --git a/arch/parisc/kernel/process.c b/arch/parisc/kernel/process.c index 7fdca87ef64..f482f78de43 100644 --- a/arch/parisc/kernel/process.c +++ b/arch/parisc/kernel/process.c @@ -92,7 +92,9 @@ void cpu_idle(void) while (1) { while (!need_resched()) barrier(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); check_pgt_cache(); } } diff --git a/arch/parisc/kernel/smp.c b/arch/parisc/kernel/smp.c index 5db3be4e270..a9ecf646578 100644 --- a/arch/parisc/kernel/smp.c +++ b/arch/parisc/kernel/smp.c @@ -463,6 +463,7 @@ void __init smp_callin(void) #endif smp_cpu_init(slave_id); + preempt_disable(); #if 0 /* NOT WORKING YET - see entry.S */ istack = (void *)__get_free_pages(GFP_KERNEL,ISTACK_ORDER); diff --git a/arch/powerpc/platforms/iseries/setup.c b/arch/powerpc/platforms/iseries/setup.c index d3e4bf756c8..0130f2619da 100644 --- a/arch/powerpc/platforms/iseries/setup.c +++ b/arch/powerpc/platforms/iseries/setup.c @@ -694,7 +694,9 @@ static void iseries_shared_idle(void) if (hvlpevent_is_pending()) process_iSeries_events(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } @@ -726,7 +728,9 @@ static void iseries_dedicated_idle(void) } ppc64_runlatch_on(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/powerpc/platforms/pseries/setup.c b/arch/powerpc/platforms/pseries/setup.c index e78c3936884..4854f5eb5c3 100644 --- a/arch/powerpc/platforms/pseries/setup.c +++ b/arch/powerpc/platforms/pseries/setup.c @@ -539,7 +539,9 @@ static void pseries_dedicated_idle(void) lpaca->lppaca.idle = 0; ppc64_runlatch_on(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING) cpu_die(); @@ -583,7 +585,9 @@ static void pseries_shared_idle(void) lpaca->lppaca.idle = 0; ppc64_runlatch_on(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING) cpu_die(); diff --git a/arch/ppc/kernel/idle.c b/arch/ppc/kernel/idle.c index 11e5b44713f..a6141f05c91 100644 --- a/arch/ppc/kernel/idle.c +++ b/arch/ppc/kernel/idle.c @@ -53,10 +53,6 @@ void default_idle(void) } #endif } - if (need_resched()) - schedule(); - if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING) - cpu_die(); } /* @@ -64,11 +60,22 @@ void default_idle(void) */ void cpu_idle(void) { - for (;;) + int cpu = smp_processor_id(); + + for (;;) { if (ppc_md.idle != NULL) ppc_md.idle(); else default_idle(); + if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING) + cpu_die(); + if (need_resched()) { + preempt_enable_no_resched(); + schedule(); + preempt_disable(); + } + + } } #if defined(CONFIG_SYSCTL) && defined(CONFIG_6xx) diff --git a/arch/ppc/kernel/smp.c b/arch/ppc/kernel/smp.c index bc5bf112483..43b8fc2ca59 100644 --- a/arch/ppc/kernel/smp.c +++ b/arch/ppc/kernel/smp.c @@ -341,6 +341,7 @@ int __devinit start_secondary(void *unused) cpu = smp_processor_id(); smp_store_cpu_info(cpu); set_dec(tb_ticks_per_jiffy); + preempt_disable(); cpu_callin_map[cpu] = 1; printk("CPU %d done callin...\n", cpu); diff --git a/arch/ppc64/kernel/idle.c b/arch/ppc64/kernel/idle.c index 8fec2746980..909ea669af9 100644 --- a/arch/ppc64/kernel/idle.c +++ b/arch/ppc64/kernel/idle.c @@ -61,7 +61,9 @@ void default_idle(void) } ppc64_runlatch_on(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING) cpu_die(); } @@ -77,7 +79,9 @@ void native_idle(void) if (need_resched()) { ppc64_runlatch_on(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); } if (cpu_is_offline(smp_processor_id()) && diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c index 9f3dff6c0b7..66ca5757e36 100644 --- a/arch/s390/kernel/process.c +++ b/arch/s390/kernel/process.c @@ -102,7 +102,6 @@ void default_idle(void) local_irq_disable(); if (need_resched()) { local_irq_enable(); - schedule(); return; } @@ -139,8 +138,14 @@ void default_idle(void) void cpu_idle(void) { - for (;;) - default_idle(); + for (;;) { + while (!need_resched()) + default_idle(); + + preempt_enable_no_resched(); + schedule(); + preempt_disable(); + } } void show_regs(struct pt_regs *regs) diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c index e13c87b446b..5856b3fda6b 100644 --- a/arch/s390/kernel/smp.c +++ b/arch/s390/kernel/smp.c @@ -533,6 +533,7 @@ int __devinit start_secondary(void *cpuvoid) { /* Setup the cpu */ cpu_init(); + preempt_disable(); /* init per CPU timer */ init_cpu_timer(); #ifdef CONFIG_VIRT_TIMER diff --git a/arch/sh/kernel/process.c b/arch/sh/kernel/process.c index 6dce9d0b81f..1cbc26b796a 100644 --- a/arch/sh/kernel/process.c +++ b/arch/sh/kernel/process.c @@ -64,7 +64,9 @@ void default_idle(void) cpu_sleep(); } + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/sh/kernel/smp.c b/arch/sh/kernel/smp.c index 5ecefc02896..59e49b18252 100644 --- a/arch/sh/kernel/smp.c +++ b/arch/sh/kernel/smp.c @@ -112,7 +112,9 @@ int __cpu_up(unsigned int cpu) int start_secondary(void *unused) { - unsigned int cpu = smp_processor_id(); + unsigned int cpu; + + cpu = smp_processor_id(); atomic_inc(&init_mm.mm_count); current->active_mm = &init_mm; @@ -120,6 +122,7 @@ int start_secondary(void *unused) smp_store_cpu_info(cpu); __smp_slave_init(cpu); + preempt_disable(); per_cpu_trap_init(); atomic_inc(&cpus_booted); diff --git a/arch/sh64/kernel/process.c b/arch/sh64/kernel/process.c index efde41c0cd6..0c09537449b 100644 --- a/arch/sh64/kernel/process.c +++ b/arch/sh64/kernel/process.c @@ -334,7 +334,9 @@ void default_idle(void) } local_irq_enable(); } + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/sparc/kernel/process.c b/arch/sparc/kernel/process.c index 29e72b57d4f..c39f4d01096 100644 --- a/arch/sparc/kernel/process.c +++ b/arch/sparc/kernel/process.c @@ -120,7 +120,9 @@ void cpu_idle(void) (*pm_idle)(); } + preempt_enable_no_resched(); schedule(); + preempt_disable(); check_pgt_cache(); } } @@ -133,7 +135,9 @@ void cpu_idle(void) /* endless idle loop with no priority at all */ while(1) { if(need_resched()) { + preempt_enable_no_resched(); schedule(); + preempt_disable(); check_pgt_cache(); } barrier(); /* or else gcc optimizes... */ diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c index 7d10b039709..2f89206e008 100644 --- a/arch/sparc64/kernel/process.c +++ b/arch/sparc64/kernel/process.c @@ -74,7 +74,9 @@ void cpu_idle(void) while (!need_resched()) barrier(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); check_pgt_cache(); } } @@ -93,7 +95,9 @@ void cpu_idle(void) if (need_resched()) { unidle_me(); clear_thread_flag(TIF_POLLING_NRFLAG); + preempt_enable_no_resched(); schedule(); + preempt_disable(); set_thread_flag(TIF_POLLING_NRFLAG); check_pgt_cache(); } diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index 5d90ee9aebf..8aca4b1dc04 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -168,6 +168,9 @@ void __init smp_callin(void) rmb(); cpu_set(cpuid, cpu_online_map); + + /* idle thread is expected to have preempt disabled */ + preempt_disable(); } void cpu_panic(void) diff --git a/arch/v850/kernel/process.c b/arch/v850/kernel/process.c index 9c708c32c1f..39cf247cdae 100644 --- a/arch/v850/kernel/process.c +++ b/arch/v850/kernel/process.c @@ -36,11 +36,8 @@ extern void ret_from_fork (void); /* The idle loop. */ void default_idle (void) { - while (1) { - while (! need_resched ()) - asm ("halt; nop; nop; nop; nop; nop" ::: "cc"); - schedule (); - } + while (! need_resched ()) + asm ("halt; nop; nop; nop; nop; nop" ::: "cc"); } void (*idle)(void) = default_idle; @@ -54,7 +51,14 @@ void (*idle)(void) = default_idle; void cpu_idle (void) { /* endless idle loop with no priority at all */ - (*idle) (); + while (1) { + while (!need_resched()) + (*idle) (); + + preempt_enable_no_resched(); + schedule(); + preempt_disable(); + } } /* diff --git a/arch/x86_64/kernel/process.c b/arch/x86_64/kernel/process.c index b5a89c0bdf5..571f9fe490c 100644 --- a/arch/x86_64/kernel/process.c +++ b/arch/x86_64/kernel/process.c @@ -204,7 +204,9 @@ void cpu_idle (void) idle(); } + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/arch/x86_64/kernel/smpboot.c b/arch/x86_64/kernel/smpboot.c index 4b5b088ec10..c4e59bbdc18 100644 --- a/arch/x86_64/kernel/smpboot.c +++ b/arch/x86_64/kernel/smpboot.c @@ -472,6 +472,7 @@ void __cpuinit start_secondary(void) * things done here to the most necessary things. */ cpu_init(); + preempt_disable(); smp_callin(); /* otherwise gcc will move up the smp_processor_id before the cpu_init */ diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c index 08ef6d82ee5..6a44b54ae81 100644 --- a/arch/xtensa/kernel/process.c +++ b/arch/xtensa/kernel/process.c @@ -96,8 +96,9 @@ void cpu_idle(void) while (1) { while (!need_resched()) platform_idle(); - preempt_enable(); + preempt_enable_no_resched(); schedule(); + preempt_disable(); } } diff --git a/init/main.c b/init/main.c index f142d403534..27f97f9b463 100644 --- a/init/main.c +++ b/init/main.c @@ -394,14 +394,16 @@ static void noinline rest_init(void) kernel_thread(init, NULL, CLONE_FS | CLONE_SIGHAND); numa_default_policy(); unlock_kernel(); - preempt_enable_no_resched(); /* * The boot idle thread must execute schedule() * at least one to get things moving: */ + preempt_enable_no_resched(); schedule(); + preempt_disable(); + /* Call into cpu_idle with preempt disabled */ cpu_idle(); } -- cgit v1.2.3-70-g09d2 From 64c7c8f88559624abdbe12b5da6502e8879f8d28 Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Tue, 8 Nov 2005 21:39:04 -0800 Subject: [PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin Cc: Ingo Molnar Cc: Con Kolivas Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- Documentation/sched-arch.txt | 89 ++++++++++++++++++++++++++++++++++ arch/alpha/kernel/process.c | 10 ++-- arch/arm/kernel/process.c | 14 ++++-- arch/i386/kernel/apm.c | 20 +++++++- arch/i386/kernel/process.c | 64 +++++++++++------------- arch/ia64/kernel/process.c | 32 ++++++------ arch/parisc/kernel/process.c | 2 + arch/powerpc/platforms/iseries/setup.c | 10 +--- arch/powerpc/platforms/pseries/setup.c | 12 ++--- arch/ppc/kernel/idle.c | 20 ++++---- arch/ppc64/kernel/idle.c | 11 +---- arch/s390/kernel/process.c | 13 ++--- arch/sh/kernel/process.c | 12 ++--- arch/sh64/kernel/process.c | 14 ++---- arch/sparc/kernel/process.c | 35 ++++++------- arch/sparc64/kernel/process.c | 20 +++++--- arch/sparc64/kernel/smp.c | 13 +---- arch/x86_64/kernel/process.c | 67 ++++++++++++------------- drivers/acpi/processor_idle.c | 37 ++++++++------ kernel/sched.c | 21 +++++--- 20 files changed, 296 insertions(+), 220 deletions(-) create mode 100644 Documentation/sched-arch.txt (limited to 'arch/sparc64/kernel') diff --git a/Documentation/sched-arch.txt b/Documentation/sched-arch.txt new file mode 100644 index 00000000000..941615a9769 --- /dev/null +++ b/Documentation/sched-arch.txt @@ -0,0 +1,89 @@ + CPU Scheduler implementation hints for architecture specific code + + Nick Piggin, 2005 + +Context switch +============== +1. Runqueue locking +By default, the switch_to arch function is called with the runqueue +locked. This is usually not a problem unless switch_to may need to +take the runqueue lock. This is usually due to a wake up operation in +the context switch. See include/asm-ia64/system.h for an example. + +To request the scheduler call switch_to with the runqueue unlocked, +you must `#define __ARCH_WANT_UNLOCKED_CTXSW` in a header file +(typically the one where switch_to is defined). + +Unlocked context switches introduce only a very minor performance +penalty to the core scheduler implementation in the CONFIG_SMP case. + +2. Interrupt status +By default, the switch_to arch function is called with interrupts +disabled. Interrupts may be enabled over the call if it is likely to +introduce a significant interrupt latency by adding the line +`#define __ARCH_WANT_INTERRUPTS_ON_CTXSW` in the same place as for +unlocked context switches. This define also implies +`__ARCH_WANT_UNLOCKED_CTXSW`. See include/asm-arm/system.h for an +example. + + +CPU idle +======== +Your cpu_idle routines need to obey the following rules: + +1. Preempt should now disabled over idle routines. Should only + be enabled to call schedule() then disabled again. + +2. need_resched/TIF_NEED_RESCHED is only ever set, and will never + be cleared until the running task has called schedule(). Idle + threads need only ever query need_resched, and may never set or + clear it. + +3. When cpu_idle finds (need_resched() == 'true'), it should call + schedule(). It should not call schedule() otherwise. + +4. The only time interrupts need to be disabled when checking + need_resched is if we are about to sleep the processor until + the next interrupt (this doesn't provide any protection of + need_resched, it prevents losing an interrupt). + + 4a. Common problem with this type of sleep appears to be: + local_irq_disable(); + if (!need_resched()) { + local_irq_enable(); + *** resched interrupt arrives here *** + __asm__("sleep until next interrupt"); + } + +5. TIF_POLLING_NRFLAG can be set by idle routines that do not + need an interrupt to wake them up when need_resched goes high. + In other words, they must be periodically polling need_resched, + although it may be reasonable to do some background work or enter + a low CPU priority. + + 5a. If TIF_POLLING_NRFLAG is set, and we do decide to enter + an interrupt sleep, it needs to be cleared then a memory + barrier issued (followed by a test of need_resched with + interrupts disabled, as explained in 3). + +arch/i386/kernel/process.c has examples of both polling and +sleeping idle functions. + + +Possible arch/ problems +======================= + +Possible arch problems I found (and either tried to fix or didn't): + +h8300 - Is such sleeping racy vs interrupts? (See #4a). + The H8/300 manual I found indicates yes, however disabling IRQs + over the sleep mean only NMIs can wake it up, so can't fix easily + without doing spin waiting. + +ia64 - is safe_halt call racy vs interrupts? (does it sleep?) (See #4a) + +sh64 - Is sleeping racy vs interrupts? (See #4a) + +sparc - IRQs on at this point(?), change local_irq_save to _disable. + - TODO: needs secondary CPUs to disable preempt (See #1) + diff --git a/arch/alpha/kernel/process.c b/arch/alpha/kernel/process.c index eb20c3afff5..a8682612abc 100644 --- a/arch/alpha/kernel/process.c +++ b/arch/alpha/kernel/process.c @@ -43,21 +43,17 @@ #include "proto.h" #include "pci_impl.h" -void default_idle(void) -{ - barrier(); -} - void cpu_idle(void) { + set_thread_flag(TIF_POLLING_NRFLAG); + while (1) { - void (*idle)(void) = default_idle; /* FIXME -- EV6 and LCA45 know how to power down the CPU. */ while (!need_resched()) - idle(); + cpu_relax(); schedule(); } } diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c index 93dd92cc12f..c0f6a119de3 100644 --- a/arch/arm/kernel/process.c +++ b/arch/arm/kernel/process.c @@ -86,12 +86,16 @@ EXPORT_SYMBOL(pm_power_off); */ void default_idle(void) { - local_irq_disable(); - if (!need_resched() && !hlt_counter) { - timer_dyn_reprogram(); - arch_idle(); + if (hlt_counter) + cpu_relax(); + else { + local_irq_disable(); + if (!need_resched()) { + timer_dyn_reprogram(); + arch_idle(); + } + local_irq_enable(); } - local_irq_enable(); } /* diff --git a/arch/i386/kernel/apm.c b/arch/i386/kernel/apm.c index 86e80c55147..003548b8735 100644 --- a/arch/i386/kernel/apm.c +++ b/arch/i386/kernel/apm.c @@ -769,8 +769,26 @@ static int set_system_power_state(u_short state) static int apm_do_idle(void) { u32 eax; + u8 ret = 0; + int idled = 0; + int polling; + + polling = test_thread_flag(TIF_POLLING_NRFLAG); + if (polling) { + clear_thread_flag(TIF_POLLING_NRFLAG); + smp_mb__after_clear_bit(); + } + if (!need_resched()) { + idled = 1; + ret = apm_bios_call_simple(APM_FUNC_IDLE, 0, 0, &eax); + } + if (polling) + set_thread_flag(TIF_POLLING_NRFLAG); + + if (!idled) + return 0; - if (apm_bios_call_simple(APM_FUNC_IDLE, 0, 0, &eax)) { + if (ret) { static unsigned long t; /* This always fails on some SMP boards running UP kernels. diff --git a/arch/i386/kernel/process.c b/arch/i386/kernel/process.c index 5296e284ea3..1cb261f225d 100644 --- a/arch/i386/kernel/process.c +++ b/arch/i386/kernel/process.c @@ -99,14 +99,22 @@ EXPORT_SYMBOL(enable_hlt); */ void default_idle(void) { + local_irq_enable(); + if (!hlt_counter && boot_cpu_data.hlt_works_ok) { - local_irq_disable(); - if (!need_resched()) - safe_halt(); - else - local_irq_enable(); + clear_thread_flag(TIF_POLLING_NRFLAG); + smp_mb__after_clear_bit(); + while (!need_resched()) { + local_irq_disable(); + if (!need_resched()) + safe_halt(); + else + local_irq_enable(); + } + set_thread_flag(TIF_POLLING_NRFLAG); } else { - cpu_relax(); + while (!need_resched()) + cpu_relax(); } } #ifdef CONFIG_APM_MODULE @@ -120,29 +128,14 @@ EXPORT_SYMBOL(default_idle); */ static void poll_idle (void) { - int oldval; - local_irq_enable(); - /* - * Deal with another CPU just having chosen a thread to - * run here: - */ - oldval = test_and_clear_thread_flag(TIF_NEED_RESCHED); - - if (!oldval) { - set_thread_flag(TIF_POLLING_NRFLAG); - asm volatile( - "2:" - "testl %0, %1;" - "rep; nop;" - "je 2b;" - : : "i"(_TIF_NEED_RESCHED), "m" (current_thread_info()->flags)); - - clear_thread_flag(TIF_POLLING_NRFLAG); - } else { - set_need_resched(); - } + asm volatile( + "2:" + "testl %0, %1;" + "rep; nop;" + "je 2b;" + : : "i"(_TIF_NEED_RESCHED), "m" (current_thread_info()->flags)); } #ifdef CONFIG_HOTPLUG_CPU @@ -181,6 +174,8 @@ void cpu_idle(void) { int cpu = smp_processor_id(); + set_thread_flag(TIF_POLLING_NRFLAG); + /* endless idle loop with no priority at all */ while (1) { while (!need_resched()) { @@ -246,15 +241,12 @@ static void mwait_idle(void) { local_irq_enable(); - if (!need_resched()) { - set_thread_flag(TIF_POLLING_NRFLAG); - do { - __monitor((void *)¤t_thread_info()->flags, 0, 0); - if (need_resched()) - break; - __mwait(0, 0); - } while (!need_resched()); - clear_thread_flag(TIF_POLLING_NRFLAG); + while (!need_resched()) { + __monitor((void *)¤t_thread_info()->flags, 0, 0); + smp_mb(); + if (need_resched()) + break; + __mwait(0, 0); } } diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c index 4c621fc3c3b..640d6908f8e 100644 --- a/arch/ia64/kernel/process.c +++ b/arch/ia64/kernel/process.c @@ -197,11 +197,15 @@ void default_idle (void) { local_irq_enable(); - while (!need_resched()) - if (can_do_pal_halt) - safe_halt(); - else + while (!need_resched()) { + if (can_do_pal_halt) { + local_irq_disable(); + if (!need_resched()) + safe_halt(); + local_irq_enable(); + } else cpu_relax(); + } } #ifdef CONFIG_HOTPLUG_CPU @@ -263,16 +267,16 @@ void __attribute__((noreturn)) cpu_idle (void) { void (*mark_idle)(int) = ia64_mark_idle; + int cpu = smp_processor_id(); + set_thread_flag(TIF_POLLING_NRFLAG); /* endless idle loop with no priority at all */ while (1) { + if (!need_resched()) { + void (*idle)(void); #ifdef CONFIG_SMP - if (!need_resched()) min_xtp(); #endif - while (!need_resched()) { - void (*idle)(void); - if (__get_cpu_var(cpu_idle_state)) __get_cpu_var(cpu_idle_state) = 0; @@ -284,19 +288,17 @@ cpu_idle (void) if (!idle) idle = default_idle; (*idle)(); - } - - if (mark_idle) - (*mark_idle)(0); - + if (mark_idle) + (*mark_idle)(0); #ifdef CONFIG_SMP - normal_xtp(); + normal_xtp(); #endif + } preempt_enable_no_resched(); schedule(); preempt_disable(); check_pgt_cache(); - if (cpu_is_offline(smp_processor_id())) + if (cpu_is_offline(cpu)) play_dead(); } } diff --git a/arch/parisc/kernel/process.c b/arch/parisc/kernel/process.c index f482f78de43..fee4f1f09ad 100644 --- a/arch/parisc/kernel/process.c +++ b/arch/parisc/kernel/process.c @@ -88,6 +88,8 @@ void default_idle(void) */ void cpu_idle(void) { + set_thread_flag(TIF_POLLING_NRFLAG); + /* endless idle loop with no priority at all */ while (1) { while (!need_resched()) diff --git a/arch/powerpc/platforms/iseries/setup.c b/arch/powerpc/platforms/iseries/setup.c index 0130f2619da..7f8f0cda6a7 100644 --- a/arch/powerpc/platforms/iseries/setup.c +++ b/arch/powerpc/platforms/iseries/setup.c @@ -703,13 +703,10 @@ static void iseries_shared_idle(void) static void iseries_dedicated_idle(void) { long oldval; + set_thread_flag(TIF_POLLING_NRFLAG); while (1) { - oldval = test_and_clear_thread_flag(TIF_NEED_RESCHED); - - if (!oldval) { - set_thread_flag(TIF_POLLING_NRFLAG); - + if (!need_resched()) { while (!need_resched()) { ppc64_runlatch_off(); HMT_low(); @@ -722,9 +719,6 @@ static void iseries_dedicated_idle(void) } HMT_medium(); - clear_thread_flag(TIF_POLLING_NRFLAG); - } else { - set_need_resched(); } ppc64_runlatch_on(); diff --git a/arch/powerpc/platforms/pseries/setup.c b/arch/powerpc/platforms/pseries/setup.c index 4854f5eb5c3..a093a0d4dd6 100644 --- a/arch/powerpc/platforms/pseries/setup.c +++ b/arch/powerpc/platforms/pseries/setup.c @@ -469,6 +469,7 @@ static inline void dedicated_idle_sleep(unsigned int cpu) * more. */ clear_thread_flag(TIF_POLLING_NRFLAG); + smp_mb__after_clear_bit(); /* * SMT dynamic mode. Cede will result in this thread going @@ -481,6 +482,7 @@ static inline void dedicated_idle_sleep(unsigned int cpu) cede_processor(); else local_irq_enable(); + set_thread_flag(TIF_POLLING_NRFLAG); } else { /* * Give the HV an opportunity at the processor, since we are @@ -492,11 +494,11 @@ static inline void dedicated_idle_sleep(unsigned int cpu) static void pseries_dedicated_idle(void) { - long oldval; struct paca_struct *lpaca = get_paca(); unsigned int cpu = smp_processor_id(); unsigned long start_snooze; unsigned long *smt_snooze_delay = &__get_cpu_var(smt_snooze_delay); + set_thread_flag(TIF_POLLING_NRFLAG); while (1) { /* @@ -505,10 +507,7 @@ static void pseries_dedicated_idle(void) */ lpaca->lppaca.idle = 1; - oldval = test_and_clear_thread_flag(TIF_NEED_RESCHED); - if (!oldval) { - set_thread_flag(TIF_POLLING_NRFLAG); - + if (!need_resched()) { start_snooze = __get_tb() + *smt_snooze_delay * tb_ticks_per_usec; @@ -531,9 +530,6 @@ static void pseries_dedicated_idle(void) } HMT_medium(); - clear_thread_flag(TIF_POLLING_NRFLAG); - } else { - set_need_resched(); } lpaca->lppaca.idle = 0; diff --git a/arch/ppc/kernel/idle.c b/arch/ppc/kernel/idle.c index a6141f05c91..3c4e4cb6107 100644 --- a/arch/ppc/kernel/idle.c +++ b/arch/ppc/kernel/idle.c @@ -63,18 +63,18 @@ void cpu_idle(void) int cpu = smp_processor_id(); for (;;) { - if (ppc_md.idle != NULL) - ppc_md.idle(); - else - default_idle(); - if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING) - cpu_die(); - if (need_resched()) { - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + while (need_resched()) { + if (ppc_md.idle != NULL) + ppc_md.idle(); + else + default_idle(); } + if (cpu_is_offline(cpu) && system_state == SYSTEM_RUNNING) + cpu_die(); + preempt_enable_no_resched(); + schedule(); + preempt_disable(); } } diff --git a/arch/ppc64/kernel/idle.c b/arch/ppc64/kernel/idle.c index 909ea669af9..715bc0e71e0 100644 --- a/arch/ppc64/kernel/idle.c +++ b/arch/ppc64/kernel/idle.c @@ -34,15 +34,11 @@ extern void power4_idle(void); void default_idle(void) { - long oldval; unsigned int cpu = smp_processor_id(); + set_thread_flag(TIF_POLLING_NRFLAG); while (1) { - oldval = test_and_clear_thread_flag(TIF_NEED_RESCHED); - - if (!oldval) { - set_thread_flag(TIF_POLLING_NRFLAG); - + if (!need_resched()) { while (!need_resched() && !cpu_is_offline(cpu)) { ppc64_runlatch_off(); @@ -55,9 +51,6 @@ void default_idle(void) } HMT_medium(); - clear_thread_flag(TIF_POLLING_NRFLAG); - } else { - set_need_resched(); } ppc64_runlatch_on(); diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c index 66ca5757e36..78b64fe5e7c 100644 --- a/arch/s390/kernel/process.c +++ b/arch/s390/kernel/process.c @@ -99,14 +99,15 @@ void default_idle(void) { int cpu, rc; + /* CPU is going idle. */ + cpu = smp_processor_id(); + local_irq_disable(); - if (need_resched()) { + if (need_resched()) { local_irq_enable(); - return; - } + return; + } - /* CPU is going idle. */ - cpu = smp_processor_id(); rc = notifier_call_chain(&idle_chain, CPU_IDLE, (void *)(long) cpu); if (rc != NOTIFY_OK && rc != NOTIFY_DONE) BUG(); @@ -119,7 +120,7 @@ void default_idle(void) __ctl_set_bit(8, 15); #ifdef CONFIG_HOTPLUG_CPU - if (cpu_is_offline(smp_processor_id())) + if (cpu_is_offline(cpu)) cpu_die(); #endif diff --git a/arch/sh/kernel/process.c b/arch/sh/kernel/process.c index 1cbc26b796a..fd4f240b833 100644 --- a/arch/sh/kernel/process.c +++ b/arch/sh/kernel/process.c @@ -51,14 +51,13 @@ void enable_hlt(void) EXPORT_SYMBOL(enable_hlt); -void default_idle(void) +void cpu_idle(void) { /* endless idle loop with no priority at all */ while (1) { if (hlt_counter) { - while (1) - if (need_resched()) - break; + while (!need_resched()) + cpu_relax(); } else { while (!need_resched()) cpu_sleep(); @@ -70,11 +69,6 @@ void default_idle(void) } } -void cpu_idle(void) -{ - default_idle(); -} - void machine_restart(char * __unused) { /* SR.BL=1 and invoke address error to let CPU reset (manual reset) */ diff --git a/arch/sh64/kernel/process.c b/arch/sh64/kernel/process.c index 0c09537449b..b95d0414185 100644 --- a/arch/sh64/kernel/process.c +++ b/arch/sh64/kernel/process.c @@ -307,23 +307,19 @@ __setup("hlt", hlt_setup); static inline void hlt(void) { - if (hlt_counter) - return; - __asm__ __volatile__ ("sleep" : : : "memory"); } /* * The idle loop on a uniprocessor SH.. */ -void default_idle(void) +void cpu_idle(void) { /* endless idle loop with no priority at all */ while (1) { if (hlt_counter) { - while (1) - if (need_resched()) - break; + while (!need_resched()) + cpu_relax(); } else { local_irq_disable(); while (!need_resched()) { @@ -338,11 +334,7 @@ void default_idle(void) schedule(); preempt_disable(); } -} -void cpu_idle(void) -{ - default_idle(); } void machine_restart(char * __unused) diff --git a/arch/sparc/kernel/process.c b/arch/sparc/kernel/process.c index c39f4d01096..ea864741146 100644 --- a/arch/sparc/kernel/process.c +++ b/arch/sparc/kernel/process.c @@ -67,13 +67,6 @@ extern void fpsave(unsigned long *, unsigned long *, void *, unsigned long *); struct task_struct *last_task_used_math = NULL; struct thread_info *current_set[NR_CPUS]; -/* - * default_idle is new in 2.5. XXX Review, currently stolen from sparc64. - */ -void default_idle(void) -{ -} - #ifndef CONFIG_SMP #define SUN4C_FAULT_HIGH 100 @@ -92,12 +85,11 @@ void cpu_idle(void) static unsigned long fps; unsigned long now; unsigned long faults; - unsigned long flags; extern unsigned long sun4c_kernel_faults; extern void sun4c_grow_kernel_ring(void); - local_irq_save(flags); + local_irq_disable(); now = jiffies; count -= (now - last_jiffies); last_jiffies = now; @@ -113,13 +105,16 @@ void cpu_idle(void) sun4c_grow_kernel_ring(); } } - local_irq_restore(flags); + local_irq_enable(); } - while((!need_resched()) && pm_idle) { - (*pm_idle)(); + if (pm_idle) { + while (!need_resched()) + (*pm_idle)(); + } else { + while (!need_resched()) + cpu_relax(); } - preempt_enable_no_resched(); schedule(); preempt_disable(); @@ -132,15 +127,15 @@ void cpu_idle(void) /* This is being executed in task 0 'user space'. */ void cpu_idle(void) { + set_thread_flag(TIF_POLLING_NRFLAG); /* endless idle loop with no priority at all */ while(1) { - if(need_resched()) { - preempt_enable_no_resched(); - schedule(); - preempt_disable(); - check_pgt_cache(); - } - barrier(); /* or else gcc optimizes... */ + while (!need_resched()) + cpu_relax(); + preempt_enable_no_resched(); + schedule(); + preempt_disable(); + check_pgt_cache(); } } diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c index 2f89206e008..02f9dec1d45 100644 --- a/arch/sparc64/kernel/process.c +++ b/arch/sparc64/kernel/process.c @@ -85,23 +85,31 @@ void cpu_idle(void) /* * the idle loop on a UltraMultiPenguin... + * + * TIF_POLLING_NRFLAG is set because we do not sleep the cpu + * inside of the idler task, so an interrupt is not needed + * to get a clean fast response. + * + * XXX Reverify this assumption... -DaveM + * + * Addendum: We do want it to do something for the signal + * delivery case, we detect that by just seeing + * if we are trying to send this to an idler or not. */ -#define idle_me_harder() (cpu_data(smp_processor_id()).idle_volume += 1) -#define unidle_me() (cpu_data(smp_processor_id()).idle_volume = 0) void cpu_idle(void) { + cpuinfo_sparc *cpuinfo = &local_cpu_data(); set_thread_flag(TIF_POLLING_NRFLAG); + while(1) { if (need_resched()) { - unidle_me(); - clear_thread_flag(TIF_POLLING_NRFLAG); + cpuinfo->idle_volume = 0; preempt_enable_no_resched(); schedule(); preempt_disable(); - set_thread_flag(TIF_POLLING_NRFLAG); check_pgt_cache(); } - idle_me_harder(); + cpuinfo->idle_volume++; /* The store ordering is so that IRQ handlers on * other cpus see our increasing idleness for the buddy diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index 8aca4b1dc04..797a65493fb 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -1152,20 +1152,9 @@ void __init smp_cpus_done(unsigned int max_cpus) (bogosum/(5000/HZ))%100); } -/* This needn't do anything as we do not sleep the cpu - * inside of the idler task, so an interrupt is not needed - * to get a clean fast response. - * - * XXX Reverify this assumption... -DaveM - * - * Addendum: We do want it to do something for the signal - * delivery case, we detect that by just seeing - * if we are trying to send this to an idler or not. - */ void smp_send_reschedule(int cpu) { - if (cpu_data(cpu).idle_volume == 0) - smp_receive_signal(cpu); + smp_receive_signal(cpu); } /* This is a nop because we capture all other cpus diff --git a/arch/x86_64/kernel/process.c b/arch/x86_64/kernel/process.c index 571f9fe490c..59be85d9a4b 100644 --- a/arch/x86_64/kernel/process.c +++ b/arch/x86_64/kernel/process.c @@ -86,12 +86,22 @@ EXPORT_SYMBOL(enable_hlt); */ void default_idle(void) { + local_irq_enable(); + if (!atomic_read(&hlt_counter)) { - local_irq_disable(); - if (!need_resched()) - safe_halt(); - else - local_irq_enable(); + clear_thread_flag(TIF_POLLING_NRFLAG); + smp_mb__after_clear_bit(); + while (!need_resched()) { + local_irq_disable(); + if (!need_resched()) + safe_halt(); + else + local_irq_enable(); + } + set_thread_flag(TIF_POLLING_NRFLAG); + } else { + while (!need_resched()) + cpu_relax(); } } @@ -102,30 +112,16 @@ void default_idle(void) */ static void poll_idle (void) { - int oldval; - local_irq_enable(); - /* - * Deal with another CPU just having chosen a thread to - * run here: - */ - oldval = test_and_clear_thread_flag(TIF_NEED_RESCHED); - - if (!oldval) { - set_thread_flag(TIF_POLLING_NRFLAG); - asm volatile( - "2:" - "testl %0,%1;" - "rep; nop;" - "je 2b;" - : : - "i" (_TIF_NEED_RESCHED), - "m" (current_thread_info()->flags)); - clear_thread_flag(TIF_POLLING_NRFLAG); - } else { - set_need_resched(); - } + asm volatile( + "2:" + "testl %0,%1;" + "rep; nop;" + "je 2b;" + : : + "i" (_TIF_NEED_RESCHED), + "m" (current_thread_info()->flags)); } void cpu_idle_wait(void) @@ -187,6 +183,8 @@ static inline void play_dead(void) */ void cpu_idle (void) { + set_thread_flag(TIF_POLLING_NRFLAG); + /* endless idle loop with no priority at all */ while (1) { while (!need_resched()) { @@ -221,15 +219,12 @@ static void mwait_idle(void) { local_irq_enable(); - if (!need_resched()) { - set_thread_flag(TIF_POLLING_NRFLAG); - do { - __monitor((void *)¤t_thread_info()->flags, 0, 0); - if (need_resched()) - break; - __mwait(0, 0); - } while (!need_resched()); - clear_thread_flag(TIF_POLLING_NRFLAG); + while (!need_resched()) { + __monitor((void *)¤t_thread_info()->flags, 0, 0); + smp_mb(); + if (need_resched()) + break; + __mwait(0, 0); } } diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c index 161db4acfb9..573b6a97bb1 100644 --- a/drivers/acpi/processor_idle.c +++ b/drivers/acpi/processor_idle.c @@ -167,6 +167,19 @@ acpi_processor_power_activate(struct acpi_processor *pr, return; } +static void acpi_safe_halt(void) +{ + int polling = test_thread_flag(TIF_POLLING_NRFLAG); + if (polling) { + clear_thread_flag(TIF_POLLING_NRFLAG); + smp_mb__after_clear_bit(); + } + if (!need_resched()) + safe_halt(); + if (polling) + set_thread_flag(TIF_POLLING_NRFLAG); +} + static atomic_t c3_cpu_count; static void acpi_processor_idle(void) @@ -177,7 +190,7 @@ static void acpi_processor_idle(void) int sleep_ticks = 0; u32 t1, t2 = 0; - pr = processors[raw_smp_processor_id()]; + pr = processors[smp_processor_id()]; if (!pr) return; @@ -197,8 +210,13 @@ static void acpi_processor_idle(void) } cx = pr->power.state; - if (!cx) - goto easy_out; + if (!cx) { + if (pm_idle_save) + pm_idle_save(); + else + acpi_safe_halt(); + return; + } /* * Check BM Activity @@ -278,7 +296,8 @@ static void acpi_processor_idle(void) if (pm_idle_save) pm_idle_save(); else - safe_halt(); + acpi_safe_halt(); + /* * TBD: Can't get time duration while in C1, as resumes * go to an ISR rather than here. Need to instrument @@ -414,16 +433,6 @@ static void acpi_processor_idle(void) */ if (next_state != pr->power.state) acpi_processor_power_activate(pr, next_state); - - return; - - easy_out: - /* do C1 instead of busy loop */ - if (pm_idle_save) - pm_idle_save(); - else - safe_halt(); - return; } static int acpi_processor_set_power_policy(struct acpi_processor *pr) diff --git a/kernel/sched.c b/kernel/sched.c index 0f2def82229..ac3f5cc3bb5 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -864,21 +864,28 @@ static void deactivate_task(struct task_struct *p, runqueue_t *rq) #ifdef CONFIG_SMP static void resched_task(task_t *p) { - int need_resched, nrpolling; + int cpu; assert_spin_locked(&task_rq(p)->lock); - /* minimise the chance of sending an interrupt to poll_idle() */ - nrpolling = test_tsk_thread_flag(p,TIF_POLLING_NRFLAG); - need_resched = test_and_set_tsk_thread_flag(p,TIF_NEED_RESCHED); - nrpolling |= test_tsk_thread_flag(p,TIF_POLLING_NRFLAG); + if (unlikely(test_tsk_thread_flag(p, TIF_NEED_RESCHED))) + return; + + set_tsk_thread_flag(p, TIF_NEED_RESCHED); + + cpu = task_cpu(p); + if (cpu == smp_processor_id()) + return; - if (!need_resched && !nrpolling && (task_cpu(p) != smp_processor_id())) - smp_send_reschedule(task_cpu(p)); + /* NEED_RESCHED must be visible before we test POLLING_NRFLAG */ + smp_mb(); + if (!test_tsk_thread_flag(p, TIF_POLLING_NRFLAG)) + smp_send_reschedule(cpu); } #else static inline void resched_task(task_t *p) { + assert_spin_locked(&task_rq(p)->lock); set_tsk_need_resched(p); } #endif -- cgit v1.2.3-70-g09d2 From 84c1a13a30f21406f39e546684a78ebe9859e6d7 Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Wed, 9 Nov 2005 12:03:42 -0800 Subject: [SPARC64]: Use ARRAY_SIZE macro Use ARRAY_SIZE macro instead of sizeof(x)/sizeof(x[0]) and remove a duplicate of ARRAY_SIZE which is never used anyways. Signed-off-by: Tobias Klauser Signed-off-by: David S. Miller --- arch/sparc64/kernel/cpu.c | 4 ++-- arch/sparc64/kernel/sbus.c | 2 +- arch/sparc64/mm/fault.c | 2 -- 3 files changed, 3 insertions(+), 5 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/cpu.c b/arch/sparc64/kernel/cpu.c index 77ef5df4e5a..00eed88ef2e 100644 --- a/arch/sparc64/kernel/cpu.c +++ b/arch/sparc64/kernel/cpu.c @@ -43,7 +43,7 @@ struct cpu_fp_info linux_sparc_fpu[] = { { 0x3e, 0x22, 0, "UltraSparc IIIi+ integrated FPU"}, }; -#define NSPARCFPU (sizeof(linux_sparc_fpu)/sizeof(struct cpu_fp_info)) +#define NSPARCFPU ARRAY_SIZE(linux_sparc_fpu) struct cpu_iu_info linux_sparc_chips[] = { { 0x17, 0x10, "TI UltraSparc I (SpitFire)"}, @@ -59,7 +59,7 @@ struct cpu_iu_info linux_sparc_chips[] = { { 0x3e, 0x22, "TI UltraSparc IIIi+ (Serrano)"}, }; -#define NSPARCCHIPS (sizeof(linux_sparc_chips)/sizeof(struct cpu_iu_info)) +#define NSPARCCHIPS ARRAY_SIZE(linux_sparc_chips) char *sparc_cpu_type = "cpu-oops"; char *sparc_fpu_type = "fpu-oops"; diff --git a/arch/sparc64/kernel/sbus.c b/arch/sparc64/kernel/sbus.c index e09ddf92765..96b82505566 100644 --- a/arch/sparc64/kernel/sbus.c +++ b/arch/sparc64/kernel/sbus.c @@ -790,7 +790,7 @@ static unsigned long sysio_irq_offsets[] = { #undef bogon -#define NUM_SYSIO_OFFSETS (sizeof(sysio_irq_offsets) / sizeof(sysio_irq_offsets[0])) +#define NUM_SYSIO_OFFSETS ARRAY_SIZE(sysio_irq_offsets) /* Convert Interrupt Mapping register pointer to associated * Interrupt Clear register pointer, SYSIO specific version. diff --git a/arch/sparc64/mm/fault.c b/arch/sparc64/mm/fault.c index 3be278d916d..6f0539aa44d 100644 --- a/arch/sparc64/mm/fault.c +++ b/arch/sparc64/mm/fault.c @@ -30,8 +30,6 @@ #include #include -#define ELEMENTS(arr) (sizeof (arr)/sizeof (arr[0])) - /* * To debug kernel to catch accesses to certain virtual/physical addresses. * Mode = 0 selects physical watchpoints, mode = 1 selects virtual watchpoints. -- cgit v1.2.3-70-g09d2 From 8ca2bdc7a98b9584ac5f640761501405154171c7 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 9 Nov 2005 12:07:18 -0800 Subject: [SPARC] sbus rtc: implement ->compat_ioctl Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 2 -- drivers/sbus/char/rtc.c | 22 ++++++++++++++++++++++ 2 files changed, 22 insertions(+), 2 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index c2e96daa5ab..e62214354bb 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -114,8 +114,6 @@ COMPATIBLE_IOCTL(FBIOGCURPOS) COMPATIBLE_IOCTL(FBIOGCURMAX) /* Little k */ /* Little v, the video4linux ioctls */ -COMPATIBLE_IOCTL(_IOR('p', 20, int[7])) /* RTCGET */ -COMPATIBLE_IOCTL(_IOW('p', 21, int[7])) /* RTCSET */ /* And these ioctls need translation */ /* Note SIOCRTMSG is no longer, so this is safe and * the user would have seen just an -EINVAL anyways. */ HANDLE_IOCTL(FBIOPUTCMAP32, fbiogetputcmap) diff --git a/drivers/sbus/char/rtc.c b/drivers/sbus/char/rtc.c index 9b988baf0b5..5774bdd0e26 100644 --- a/drivers/sbus/char/rtc.c +++ b/drivers/sbus/char/rtc.c @@ -210,6 +210,27 @@ static int rtc_ioctl(struct inode *inode, struct file *file, unsigned int cmd, } } +static long rtc_compat_ioctl(struct file *file, unsigned int cmd, + unsigned long arg) +{ + int rval = -ENOIOCTLCMD; + + switch (cmd) { + /* + * These two are specific to this driver, the generic rtc ioctls + * are hanlded elsewhere. + */ + case RTCGET: + case RTCSET: + lock_kernel(); + rval = rtc_ioctl(file->f_dentry->d_inode, file, cmd, arg); + unlock_kernel(); + break; + } + + return rval; +} + static int rtc_open(struct inode *inode, struct file *file) { int ret; @@ -237,6 +258,7 @@ static struct file_operations rtc_fops = { .owner = THIS_MODULE, .llseek = no_llseek, .ioctl = rtc_ioctl, + .compat_ioctl = rtc_compat_ioctl, .open = rtc_open, .release = rtc_release, }; -- cgit v1.2.3-70-g09d2 From 4d45cbacb8488ae3172065e4fca608079492231b Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Fri, 11 Nov 2005 12:48:56 -0800 Subject: [SPARC64]: Restore 2.4.x /proc/cpuinfo behavior for "ncpus probed" field. Noticed by Tom 'spot' Callaway. Even on uniprocessor we always reported the number of physical cpus in the system via /proc/cpuinfo. But when this got changed to use num_possible_cpus() it always reads as "1" on uniprocessor. This change was unintentional. So scan the firmware device tree and count the number of cpu nodes, and report that, as we always did. Signed-off-by: David S. Miller --- arch/sparc64/kernel/setup.c | 19 +++++++++++++++---- arch/sparc64/kernel/smp.c | 1 - 2 files changed, 15 insertions(+), 5 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index bf1849dd9c4..48180531562 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -587,6 +587,8 @@ extern void mmu_info(struct seq_file *); unsigned int dcache_parity_tl1_occurred; unsigned int icache_parity_tl1_occurred; +static int ncpus_probed; + static int show_cpuinfo(struct seq_file *m, void *__unused) { seq_printf(m, @@ -595,8 +597,8 @@ static int show_cpuinfo(struct seq_file *m, void *__unused) "promlib\t\t: Version 3 Revision %d\n" "prom\t\t: %d.%d.%d\n" "type\t\t: sun4u\n" - "ncpus probed\t: %ld\n" - "ncpus active\t: %ld\n" + "ncpus probed\t: %d\n" + "ncpus active\t: %d\n" "D$ parity tl1\t: %u\n" "I$ parity tl1\t: %u\n" #ifndef CONFIG_SMP @@ -610,8 +612,8 @@ static int show_cpuinfo(struct seq_file *m, void *__unused) prom_prev >> 16, (prom_prev >> 8) & 0xff, prom_prev & 0xff, - (long)num_possible_cpus(), - (long)num_online_cpus(), + ncpus_probed, + num_online_cpus(), dcache_parity_tl1_occurred, icache_parity_tl1_occurred #ifndef CONFIG_SMP @@ -677,6 +679,15 @@ static int __init topology_init(void) int i, err; err = -ENOMEM; + + /* Count the number of physically present processors in + * the machine, even on uniprocessor, so that /proc/cpuinfo + * output is consistent with 2.4.x + */ + ncpus_probed = 0; + while (!cpu_find_by_instance(ncpus_probed, NULL, NULL)) + ncpus_probed++; + for (i = 0; i < NR_CPUS; i++) { if (cpu_possible(i)) { struct cpu *p = kmalloc(sizeof(*p), GFP_KERNEL); diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index 797a65493fb..6efc03df51c 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -39,7 +39,6 @@ #include #include -extern int linux_num_cpus; extern void calibrate_delay(void); /* Please don't make this stuff initdata!!! --DaveM */ -- cgit v1.2.3-70-g09d2 From 9ffb83bcc5c5337f980dc0576bf13ac9bd4fd33d Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Sat, 12 Nov 2005 12:11:12 -0800 Subject: [SBUSFB]: implement ->compat_ioctl This patch adds a new function, sbusfb_compat_ioctl() to drivers/video/sbuslib.c and uses it as compat_ioctl in all sbus fb drivers This remove the last per-arch compat ioctl bits in arch/sparc64/kernel/ioctl32.c so it would be nice if people could test if this actually copiles and works and if yes apply it :) Signed-off-by: Christoph Hellwig Signed-off-by: David S. Miller --- arch/sparc64/kernel/ioctl32.c | 92 ------------------------------------ drivers/video/bw2.c | 3 ++ drivers/video/cg14.c | 3 ++ drivers/video/cg3.c | 3 ++ drivers/video/cg6.c | 3 ++ drivers/video/ffb.c | 3 ++ drivers/video/leo.c | 3 ++ drivers/video/p9100.c | 3 ++ drivers/video/sbuslib.c | 107 ++++++++++++++++++++++++++++++++++++++++++ drivers/video/sbuslib.h | 2 + drivers/video/tcx.c | 3 ++ 11 files changed, 133 insertions(+), 92 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index e62214354bb..196b208665a 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -12,86 +12,10 @@ #define INCLUDES #include "compat_ioctl.c" #include -#include - -/* Use this to get at 32-bit user passed pointers. - * See sys_sparc32.c for description about it. - */ -#define A(__x) compat_ptr(__x) #define CODE #include "compat_ioctl.c" -struct fbcmap32 { - int index; /* first element (0 origin) */ - int count; - u32 red; - u32 green; - u32 blue; -}; - -#define FBIOPUTCMAP32 _IOW('F', 3, struct fbcmap32) -#define FBIOGETCMAP32 _IOW('F', 4, struct fbcmap32) - -static int fbiogetputcmap(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - struct fbcmap32 __user *argp = (void __user *)arg; - struct fbcmap __user *p = compat_alloc_user_space(sizeof(*p)); - u32 addr; - int ret; - - ret = copy_in_user(p, argp, 2 * sizeof(int)); - ret |= get_user(addr, &argp->red); - ret |= put_user(compat_ptr(addr), &p->red); - ret |= get_user(addr, &argp->green); - ret |= put_user(compat_ptr(addr), &p->green); - ret |= get_user(addr, &argp->blue); - ret |= put_user(compat_ptr(addr), &p->blue); - if (ret) - return -EFAULT; - return sys_ioctl(fd, (cmd == FBIOPUTCMAP32) ? FBIOPUTCMAP_SPARC : FBIOGETCMAP_SPARC, (unsigned long)p); -} - -struct fbcursor32 { - short set; /* what to set, choose from the list above */ - short enable; /* cursor on/off */ - struct fbcurpos pos; /* cursor position */ - struct fbcurpos hot; /* cursor hot spot */ - struct fbcmap32 cmap; /* color map info */ - struct fbcurpos size; /* cursor bit map size */ - u32 image; /* cursor image bits */ - u32 mask; /* cursor mask bits */ -}; - -#define FBIOSCURSOR32 _IOW('F', 24, struct fbcursor32) -#define FBIOGCURSOR32 _IOW('F', 25, struct fbcursor32) - -static int fbiogscursor(unsigned int fd, unsigned int cmd, unsigned long arg) -{ - struct fbcursor __user *p = compat_alloc_user_space(sizeof(*p)); - struct fbcursor32 __user *argp = (void __user *)arg; - compat_uptr_t addr; - int ret; - - ret = copy_in_user(p, argp, - 2 * sizeof (short) + 2 * sizeof(struct fbcurpos)); - ret |= copy_in_user(&p->size, &argp->size, sizeof(struct fbcurpos)); - ret |= copy_in_user(&p->cmap, &argp->cmap, 2 * sizeof(int)); - ret |= get_user(addr, &argp->cmap.red); - ret |= put_user(compat_ptr(addr), &p->cmap.red); - ret |= get_user(addr, &argp->cmap.green); - ret |= put_user(compat_ptr(addr), &p->cmap.green); - ret |= get_user(addr, &argp->cmap.blue); - ret |= put_user(compat_ptr(addr), &p->cmap.blue); - ret |= get_user(addr, &argp->mask); - ret |= put_user(compat_ptr(addr), &p->mask); - ret |= get_user(addr, &argp->image); - ret |= put_user(compat_ptr(addr), &p->image); - if (ret) - return -EFAULT; - return sys_ioctl (fd, FBIOSCURSOR, (unsigned long)p); -} - #define COMPATIBLE_IOCTL(cmd) HANDLE_IOCTL((cmd),sys_ioctl) #define HANDLE_IOCTL(cmd,handler) { (cmd), (ioctl_trans_handler_t)(handler), NULL }, #define IOCTL_TABLE_START \ @@ -103,22 +27,6 @@ IOCTL_TABLE_START #include #define DECLARES #include "compat_ioctl.c" -COMPATIBLE_IOCTL(FBIOGTYPE) -COMPATIBLE_IOCTL(FBIOSATTR) -COMPATIBLE_IOCTL(FBIOGATTR) -COMPATIBLE_IOCTL(FBIOSVIDEO) -COMPATIBLE_IOCTL(FBIOGVIDEO) -COMPATIBLE_IOCTL(FBIOGCURSOR32) /* This is not implemented yet. Later it should be converted... */ -COMPATIBLE_IOCTL(FBIOSCURPOS) -COMPATIBLE_IOCTL(FBIOGCURPOS) -COMPATIBLE_IOCTL(FBIOGCURMAX) -/* Little k */ -/* Little v, the video4linux ioctls */ -/* And these ioctls need translation */ -/* Note SIOCRTMSG is no longer, so this is safe and * the user would have seen just an -EINVAL anyways. */ -HANDLE_IOCTL(FBIOPUTCMAP32, fbiogetputcmap) -HANDLE_IOCTL(FBIOGETCMAP32, fbiogetputcmap) -HANDLE_IOCTL(FBIOSCURSOR32, fbiogscursor) #if 0 HANDLE_IOCTL(RTC32_IRQP_READ, do_rtc_ioctl) HANDLE_IOCTL(RTC32_IRQP_SET, do_rtc_ioctl) diff --git a/drivers/video/bw2.c b/drivers/video/bw2.c index f53bf3ba127..d3728f60961 100644 --- a/drivers/video/bw2.c +++ b/drivers/video/bw2.c @@ -51,6 +51,9 @@ static struct fb_ops bw2_ops = { .fb_imageblit = cfb_imageblit, .fb_mmap = bw2_mmap, .fb_ioctl = bw2_ioctl, +#ifdef CONFIG_COMPAT + .fb_compat_ioctl = sbusfb_compat_ioctl, +#endif }; /* OBio addresses for the bwtwo registers */ diff --git a/drivers/video/cg14.c b/drivers/video/cg14.c index 030d4b13b1c..1bed50f2a27 100644 --- a/drivers/video/cg14.c +++ b/drivers/video/cg14.c @@ -49,6 +49,9 @@ static struct fb_ops cg14_ops = { .fb_imageblit = cfb_imageblit, .fb_mmap = cg14_mmap, .fb_ioctl = cg14_ioctl, +#ifdef CONFIG_COMPAT + .fb_compat_ioctl = sbusfb_compat_ioctl, +#endif }; #define CG14_MCR_INTENABLE_SHIFT 7 diff --git a/drivers/video/cg3.c b/drivers/video/cg3.c index b94eee8c42d..a1354e7e051 100644 --- a/drivers/video/cg3.c +++ b/drivers/video/cg3.c @@ -50,6 +50,9 @@ static struct fb_ops cg3_ops = { .fb_imageblit = cfb_imageblit, .fb_mmap = cg3_mmap, .fb_ioctl = cg3_ioctl, +#ifdef CONFIG_COMPAT + .fb_compat_ioctl = sbusfb_compat_ioctl, +#endif }; diff --git a/drivers/video/cg6.c b/drivers/video/cg6.c index 414c4409e92..9debe642fd2 100644 --- a/drivers/video/cg6.c +++ b/drivers/video/cg6.c @@ -54,6 +54,9 @@ static struct fb_ops cg6_ops = { .fb_sync = cg6_sync, .fb_mmap = cg6_mmap, .fb_ioctl = cg6_ioctl, +#ifdef CONFIG_COMPAT + .fb_compat_ioctl = sbusfb_compat_ioctl, +#endif }; /* Offset of interesting structures in the OBIO space */ diff --git a/drivers/video/ffb.c b/drivers/video/ffb.c index 04417dc16c2..2584daec7bb 100644 --- a/drivers/video/ffb.c +++ b/drivers/video/ffb.c @@ -57,6 +57,9 @@ static struct fb_ops ffb_ops = { .fb_sync = ffb_sync, .fb_mmap = ffb_mmap, .fb_ioctl = ffb_ioctl, +#ifdef CONFIG_COMPAT + .fb_compat_ioctl = sbusfb_compat_ioctl, +#endif }; /* Register layout and definitions */ diff --git a/drivers/video/leo.c b/drivers/video/leo.c index 84a7fe435bb..376d4a171ec 100644 --- a/drivers/video/leo.c +++ b/drivers/video/leo.c @@ -51,6 +51,9 @@ static struct fb_ops leo_ops = { .fb_imageblit = cfb_imageblit, .fb_mmap = leo_mmap, .fb_ioctl = leo_ioctl, +#ifdef CONFIG_COMPAT + .fb_compat_ioctl = sbusfb_compat_ioctl, +#endif }; #define LEO_OFF_LC_SS0_KRN 0x00200000UL diff --git a/drivers/video/p9100.c b/drivers/video/p9100.c index 9aaf65fb623..18bcda23d2c 100644 --- a/drivers/video/p9100.c +++ b/drivers/video/p9100.c @@ -48,6 +48,9 @@ static struct fb_ops p9100_ops = { .fb_imageblit = cfb_imageblit, .fb_mmap = p9100_mmap, .fb_ioctl = p9100_ioctl, +#ifdef CONFIG_COMPAT + .fb_compat_ioctl = sbusfb_compat_ioctl, +#endif }; /* P9100 control registers */ diff --git a/drivers/video/sbuslib.c b/drivers/video/sbuslib.c index 34f72edba82..646c43f921c 100644 --- a/drivers/video/sbuslib.c +++ b/drivers/video/sbuslib.c @@ -3,6 +3,7 @@ * Copyright (C) 2003 David S. Miller (davem@redhat.com) */ +#include #include #include #include @@ -182,3 +183,109 @@ int sbusfb_ioctl_helper(unsigned long cmd, unsigned long arg, }; } EXPORT_SYMBOL(sbusfb_ioctl_helper); + +#ifdef CONFIG_COMPAT +struct fbcmap32 { + int index; /* first element (0 origin) */ + int count; + u32 red; + u32 green; + u32 blue; +}; + +#define FBIOPUTCMAP32 _IOW('F', 3, struct fbcmap32) +#define FBIOGETCMAP32 _IOW('F', 4, struct fbcmap32) + +static int fbiogetputcmap(struct file *file, struct fb_info *info, + unsigned int cmd, unsigned long arg) +{ + struct fbcmap32 __user *argp = (void __user *)arg; + struct fbcmap __user *p = compat_alloc_user_space(sizeof(*p)); + u32 addr; + int ret; + + ret = copy_in_user(p, argp, 2 * sizeof(int)); + ret |= get_user(addr, &argp->red); + ret |= put_user(compat_ptr(addr), &p->red); + ret |= get_user(addr, &argp->green); + ret |= put_user(compat_ptr(addr), &p->green); + ret |= get_user(addr, &argp->blue); + ret |= put_user(compat_ptr(addr), &p->blue); + if (ret) + return -EFAULT; + return info->fbops->fb_ioctl(file->f_dentry->d_inode, file, + (cmd == FBIOPUTCMAP32) ? + FBIOPUTCMAP_SPARC : FBIOGETCMAP_SPARC, + (unsigned long)p, info); +} + +struct fbcursor32 { + short set; /* what to set, choose from the list above */ + short enable; /* cursor on/off */ + struct fbcurpos pos; /* cursor position */ + struct fbcurpos hot; /* cursor hot spot */ + struct fbcmap32 cmap; /* color map info */ + struct fbcurpos size; /* cursor bit map size */ + u32 image; /* cursor image bits */ + u32 mask; /* cursor mask bits */ +}; + +#define FBIOSCURSOR32 _IOW('F', 24, struct fbcursor32) +#define FBIOGCURSOR32 _IOW('F', 25, struct fbcursor32) + +static int fbiogscursor(struct file *file, struct fb_info *info, + unsigned long arg) +{ + struct fbcursor __user *p = compat_alloc_user_space(sizeof(*p)); + struct fbcursor32 __user *argp = (void __user *)arg; + compat_uptr_t addr; + int ret; + + ret = copy_in_user(p, argp, + 2 * sizeof (short) + 2 * sizeof(struct fbcurpos)); + ret |= copy_in_user(&p->size, &argp->size, sizeof(struct fbcurpos)); + ret |= copy_in_user(&p->cmap, &argp->cmap, 2 * sizeof(int)); + ret |= get_user(addr, &argp->cmap.red); + ret |= put_user(compat_ptr(addr), &p->cmap.red); + ret |= get_user(addr, &argp->cmap.green); + ret |= put_user(compat_ptr(addr), &p->cmap.green); + ret |= get_user(addr, &argp->cmap.blue); + ret |= put_user(compat_ptr(addr), &p->cmap.blue); + ret |= get_user(addr, &argp->mask); + ret |= put_user(compat_ptr(addr), &p->mask); + ret |= get_user(addr, &argp->image); + ret |= put_user(compat_ptr(addr), &p->image); + if (ret) + return -EFAULT; + return info->fbops->fb_ioctl(file->f_dentry->d_inode, file, + FBIOSCURSOR, (unsigned long)p, info); +} + +long sbusfb_compat_ioctl(struct file *file, unsigned int cmd, + unsigned long arg, struct fb_info *info) +{ + switch (cmd) { + case FBIOGTYPE: + case FBIOSATTR: + case FBIOGATTR: + case FBIOSVIDEO: + case FBIOGVIDEO: + case FBIOGCURSOR32: /* This is not implemented yet. + Later it should be converted... */ + case FBIOSCURPOS: + case FBIOGCURPOS: + case FBIOGCURMAX: + return info->fbops->fb_ioctl(file->f_dentry->d_inode, + file, cmd, arg, info); + case FBIOPUTCMAP32: + return fbiogetputcmap(file, info, cmd, arg); + case FBIOGETCMAP32: + return fbiogetputcmap(file, info, cmd, arg); + case FBIOSCURSOR32: + return fbiogscursor(file, info, arg); + default: + return -ENOIOCTLCMD; + } +} +EXPORT_SYMBOL(sbusfb_compat_ioctl); +#endif diff --git a/drivers/video/sbuslib.h b/drivers/video/sbuslib.h index a6aa33ba09d..b470e52ce9e 100644 --- a/drivers/video/sbuslib.h +++ b/drivers/video/sbuslib.h @@ -20,5 +20,7 @@ extern int sbusfb_mmap_helper(struct sbus_mmap_map *map, int sbusfb_ioctl_helper(unsigned long cmd, unsigned long arg, struct fb_info *info, int type, int fb_depth, unsigned long fb_size); +long sbusfb_compat_ioctl(struct file *file, unsigned int cmd, + unsigned long arg, struct fb_info *info); #endif /* _SBUSLIB_H */ diff --git a/drivers/video/tcx.c b/drivers/video/tcx.c index 59fff29bc02..fe4f63f3849 100644 --- a/drivers/video/tcx.c +++ b/drivers/video/tcx.c @@ -52,6 +52,9 @@ static struct fb_ops tcx_ops = { .fb_imageblit = cfb_imageblit, .fb_mmap = tcx_mmap, .fb_ioctl = tcx_ioctl, +#ifdef CONFIG_COMPAT + .fb_compat_ioctl = sbusfb_compat_ioctl, +#endif }; /* THC definitions */ -- cgit v1.2.3-70-g09d2 From f3d48f0373c14a6203202f7b1dfc7b0d8aaf6ed2 Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Mon, 21 Nov 2005 21:32:22 -0800 Subject: [PATCH] unpaged: fix sound Bad page states Earlier I unifdefed PageCompound, so that snd_pcm_mmap_control_nopage and others can give out a 0-order component of a higher-order page, which won't be mistakenly freed when zap_pte_range unmaps it. But many Bad page states reported a PG_reserved was freed after all: I had missed that we need to say __GFP_COMP to get compound page behaviour. Some of these higher-order pages are allocated by snd_malloc_pages, some by snd_malloc_dev_pages; or if SBUS, by sbus_alloc_consistent - but that has no gfp arg, so add __GFP_COMP into its sparc32/64 implementations. I'm still rather puzzled that DRM seems not to need a similar change. Signed-off-by: Hugh Dickins Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/sparc/kernel/ioport.c | 2 +- arch/sparc64/kernel/sbus.c | 2 +- sound/core/memalloc.c | 2 ++ 3 files changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/sparc64/kernel') diff --git a/arch/sparc/kernel/ioport.c b/arch/sparc/kernel/ioport.c index d0f2bd227c4..d39c9f20627 100644 --- a/arch/sparc/kernel/ioport.c +++ b/arch/sparc/kernel/ioport.c @@ -252,7 +252,7 @@ void *sbus_alloc_consistent(struct sbus_dev *sdev, long len, u32 *dma_addrp) } order = get_order(len_total); - if ((va = __get_free_pages(GFP_KERNEL, order)) == 0) + if ((va = __get_free_pages(GFP_KERNEL|__GFP_COMP, order)) == 0) goto err_nopages; if ((res = kmalloc(sizeof(struct resource), GFP_KERNEL)) == NULL) diff --git a/arch/sparc64/kernel/sbus.c b/arch/sparc64/kernel/sbus.c index 96b82505566..d95a1bcf163 100644 --- a/arch/sparc64/kernel/sbus.c +++ b/arch/sparc64/kernel/sbus.c @@ -327,7 +327,7 @@ void *sbus_alloc_consistent(struct sbus_dev *sdev, size_t size, dma_addr_t *dvma order = get_order(size); if (order >= 10) return NULL; - first_page = __get_free_pages(GFP_KERNEL, order); + first_page = __get_free_pages(GFP_KERNEL|__GFP_COMP, order); if (first_page == 0UL) return NULL; memset((char *)first_page, 0, PAGE_SIZE << order); diff --git a/sound/core/memalloc.c b/sound/core/memalloc.c index 129abab5ce9..e4b8959dd4b 100644 --- a/sound/core/memalloc.c +++ b/sound/core/memalloc.c @@ -197,6 +197,7 @@ void *snd_malloc_pages(size_t size, gfp_t gfp_flags) snd_assert(size > 0, return NULL); snd_assert(gfp_flags != 0, return NULL); + gfp_flags |= __GFP_COMP; /* compound page lets parts be mapped */ pg = get_order(size); if ((res = (void *) __get_free_pages(gfp_flags, pg)) != NULL) { mark_pages(virt_to_page(res), pg); @@ -241,6 +242,7 @@ static void *snd_malloc_dev_pages(struct device *dev, size_t size, dma_addr_t *d snd_assert(dma != NULL, return NULL); pg = get_order(size); gfp_flags = GFP_KERNEL + | __GFP_COMP /* compound page lets parts be mapped */ | __GFP_NORETRY /* don't trigger OOM-killer */ | __GFP_NOWARN; /* no stack trace print - this call is non-critical */ res = dma_alloc_coherent(dev, PAGE_SIZE << pg, dma, gfp_flags); -- cgit v1.2.3-70-g09d2