From 7e5a69e83ba7a0d5917ad830f417cba8b8d6aa72 Mon Sep 17 00:00:00 2001 From: Nicolas Pitre Date: Mon, 29 Mar 2010 21:46:02 +0100 Subject: ARM: 6007/1: fix highmem with VIPT cache and DMA The VIVT cache of a highmem page is always flushed before the page is unmapped. This cache flush is explicit through flush_cache_kmaps() in flush_all_zero_pkmaps(), or through __cpuc_flush_dcache_area() in kunmap_atomic(). There is also an implicit flush of those highmem pages that were part of a process that just terminated making those pages free as the whole VIVT cache has to be flushed on every task switch. Hence unmapped highmem pages need no cache maintenance in that case. However unmapped pages may still be cached with a VIPT cache because the cache is tagged with physical addresses. There is no need for a whole cache flush during task switching for that reason, and despite the explicit cache flushes in flush_all_zero_pkmaps() and kunmap_atomic(), some highmem pages that were mapped in user space end up still cached even when they become unmapped. So, we do have to perform cache maintenance on those unmapped highmem pages in the context of DMA when using a VIPT cache. Unfortunately, it is not possible to perform that cache maintenance using physical addresses as all the L1 cache maintenance coprocessor functions accept virtual addresses only. Therefore we have no choice but to set up a temporary virtual mapping for that purpose. And of course the explicit cache flushing when unmapping a highmem page on a system with a VIPT cache now can go, which should increase performance. While at it, because the code in __flush_dcache_page() has to be modified anyway, let's also make sure the mapped highmem pages are pinned with kmap_high_get() for the duration of the cache maintenance operation. Because kunmap() does unmap highmem pages lazily, it was reported by Gary King that those pages ended up being unmapped during cache maintenance on SMP causing segmentation faults. Signed-off-by: Nicolas Pitre Signed-off-by: Russell King --- arch/arm/include/asm/highmem.h | 15 ++++++++++++++- arch/arm/include/asm/kmap_types.h | 1 + 2 files changed, 15 insertions(+), 1 deletion(-) (limited to 'arch/arm/include/asm') diff --git a/arch/arm/include/asm/highmem.h b/arch/arm/include/asm/highmem.h index 7f36d00600b..feb988a7ec3 100644 --- a/arch/arm/include/asm/highmem.h +++ b/arch/arm/include/asm/highmem.h @@ -11,7 +11,11 @@ #define kmap_prot PAGE_KERNEL -#define flush_cache_kmaps() flush_cache_all() +#define flush_cache_kmaps() \ + do { \ + if (cache_is_vivt()) \ + flush_cache_all(); \ + } while (0) extern pte_t *pkmap_page_table; @@ -21,11 +25,20 @@ extern void *kmap_high(struct page *page); extern void *kmap_high_get(struct page *page); extern void kunmap_high(struct page *page); +extern void *kmap_high_l1_vipt(struct page *page, pte_t *saved_pte); +extern void kunmap_high_l1_vipt(struct page *page, pte_t saved_pte); + +/* + * The following functions are already defined by + * when CONFIG_HIGHMEM is not set. + */ +#ifdef CONFIG_HIGHMEM extern void *kmap(struct page *page); extern void kunmap(struct page *page); extern void *kmap_atomic(struct page *page, enum km_type type); extern void kunmap_atomic(void *kvaddr, enum km_type type); extern void *kmap_atomic_pfn(unsigned long pfn, enum km_type type); extern struct page *kmap_atomic_to_page(const void *ptr); +#endif #endif diff --git a/arch/arm/include/asm/kmap_types.h b/arch/arm/include/asm/kmap_types.h index c019949a518..c4b2ea3fbe4 100644 --- a/arch/arm/include/asm/kmap_types.h +++ b/arch/arm/include/asm/kmap_types.h @@ -18,6 +18,7 @@ enum km_type { KM_IRQ1, KM_SOFTIRQ0, KM_SOFTIRQ1, + KM_L1_CACHE, KM_L2_CACHE, KM_TYPE_NR }; -- cgit v1.2.3-70-g09d2 From 82c6f5a5b3e91ef4d2fb8725de4b8cf7affd4d61 Mon Sep 17 00:00:00 2001 From: Imre Deak Date: Sun, 11 Apr 2010 15:58:27 +0100 Subject: ARM: 6051/1: VFP: preserve the HW context when calling signal handlers From: Imre Deak Signal handlers can use floating point, so prevent them to corrupt the main thread's VFP context. So far there were two signal stack frame formats defined based on the VFP implementation, but the user struct used for ptrace covers all posibilities, so use it for the signal stack too. Introduce also a new user struct for VFP exception registers. In this too fields not relevant to the current VFP architecture are ignored. Support to save / restore the exception registers was added by Will Deacon. Signed-off-by: Imre Deak Signed-off-by: Will Deacon Signed-off-by: Russell King --- arch/arm/include/asm/ucontext.h | 23 +++++----- arch/arm/include/asm/user.h | 12 +++++- arch/arm/kernel/signal.c | 93 +++++++++++++++++++++++++++++++++++++++-- 3 files changed, 111 insertions(+), 17 deletions(-) (limited to 'arch/arm/include/asm') diff --git a/arch/arm/include/asm/ucontext.h b/arch/arm/include/asm/ucontext.h index bf65e9f4525..47f023aa849 100644 --- a/arch/arm/include/asm/ucontext.h +++ b/arch/arm/include/asm/ucontext.h @@ -59,23 +59,22 @@ struct iwmmxt_sigframe { #endif /* CONFIG_IWMMXT */ #ifdef CONFIG_VFP -#if __LINUX_ARM_ARCH__ < 6 -/* For ARM pre-v6, we use fstmiax and fldmiax. This adds one extra - * word after the registers, and a word of padding at the end for - * alignment. */ #define VFP_MAGIC 0x56465001 -#define VFP_STORAGE_SIZE 152 -#else -#define VFP_MAGIC 0x56465002 -#define VFP_STORAGE_SIZE 144 -#endif struct vfp_sigframe { unsigned long magic; unsigned long size; - union vfp_state storage; -}; + struct user_vfp ufp; + struct user_vfp_exc ufp_exc; +} __attribute__((__aligned__(8))); + +/* + * 8 byte for magic and size, 264 byte for ufp, 12 bytes for ufp_exc, + * 4 bytes padding. + */ +#define VFP_STORAGE_SIZE sizeof(struct vfp_sigframe) + #endif /* CONFIG_VFP */ /* @@ -91,7 +90,7 @@ struct aux_sigframe { #ifdef CONFIG_IWMMXT struct iwmmxt_sigframe iwmmxt; #endif -#if 0 && defined CONFIG_VFP /* Not yet saved. */ +#ifdef CONFIG_VFP struct vfp_sigframe vfp; #endif /* Something that isn't a valid magic number for any coprocessor. */ diff --git a/arch/arm/include/asm/user.h b/arch/arm/include/asm/user.h index df95e050f9d..05ac4b06876 100644 --- a/arch/arm/include/asm/user.h +++ b/arch/arm/include/asm/user.h @@ -83,11 +83,21 @@ struct user{ /* * User specific VFP registers. If only VFPv2 is present, registers 16 to 31 - * are ignored by the ptrace system call. + * are ignored by the ptrace system call and the signal handler. */ struct user_vfp { unsigned long long fpregs[32]; unsigned long fpscr; }; +/* + * VFP exception registers exposed to user space during signal delivery. + * Fields not relavant to the current VFP architecture are ignored. + */ +struct user_vfp_exc { + unsigned long fpexc; + unsigned long fpinst; + unsigned long fpinst2; +}; + #endif /* _ARM_USER_H */ diff --git a/arch/arm/kernel/signal.c b/arch/arm/kernel/signal.c index e7714f367eb..907d5a620bc 100644 --- a/arch/arm/kernel/signal.c +++ b/arch/arm/kernel/signal.c @@ -18,6 +18,7 @@ #include #include #include +#include #include "ptrace.h" #include "signal.h" @@ -175,6 +176,90 @@ static int restore_iwmmxt_context(struct iwmmxt_sigframe *frame) #endif +#ifdef CONFIG_VFP + +static int preserve_vfp_context(struct vfp_sigframe __user *frame) +{ + struct thread_info *thread = current_thread_info(); + struct vfp_hard_struct *h = &thread->vfpstate.hard; + const unsigned long magic = VFP_MAGIC; + const unsigned long size = VFP_STORAGE_SIZE; + int err = 0; + + vfp_sync_hwstate(thread); + __put_user_error(magic, &frame->magic, err); + __put_user_error(size, &frame->size, err); + + /* + * Copy the floating point registers. There can be unused + * registers see asm/hwcap.h for details. + */ + err |= __copy_to_user(&frame->ufp.fpregs, &h->fpregs, + sizeof(h->fpregs)); + /* + * Copy the status and control register. + */ + __put_user_error(h->fpscr, &frame->ufp.fpscr, err); + + /* + * Copy the exception registers. + */ + __put_user_error(h->fpexc, &frame->ufp_exc.fpexc, err); + __put_user_error(h->fpinst, &frame->ufp_exc.fpinst, err); + __put_user_error(h->fpinst2, &frame->ufp_exc.fpinst2, err); + + return err ? -EFAULT : 0; +} + +static int restore_vfp_context(struct vfp_sigframe __user *frame) +{ + struct thread_info *thread = current_thread_info(); + struct vfp_hard_struct *h = &thread->vfpstate.hard; + unsigned long magic; + unsigned long size; + unsigned long fpexc; + int err = 0; + + __get_user_error(magic, &frame->magic, err); + __get_user_error(size, &frame->size, err); + + if (err) + return -EFAULT; + if (magic != VFP_MAGIC || size != VFP_STORAGE_SIZE) + return -EINVAL; + + /* + * Copy the floating point registers. There can be unused + * registers see asm/hwcap.h for details. + */ + err |= __copy_from_user(&h->fpregs, &frame->ufp.fpregs, + sizeof(h->fpregs)); + /* + * Copy the status and control register. + */ + __get_user_error(h->fpscr, &frame->ufp.fpscr, err); + + /* + * Sanitise and restore the exception registers. + */ + __get_user_error(fpexc, &frame->ufp_exc.fpexc, err); + /* Ensure the VFP is enabled. */ + fpexc |= FPEXC_EN; + /* Ensure FPINST2 is invalid and the exception flag is cleared. */ + fpexc &= ~(FPEXC_EX | FPEXC_FP2V); + h->fpexc = fpexc; + + __get_user_error(h->fpinst, &frame->ufp_exc.fpinst, err); + __get_user_error(h->fpinst2, &frame->ufp_exc.fpinst2, err); + + if (!err) + vfp_flush_hwstate(thread); + + return err ? -EFAULT : 0; +} + +#endif + /* * Do a signal return; undo the signal stack. These are aligned to 64-bit. */ @@ -233,8 +318,8 @@ static int restore_sigframe(struct pt_regs *regs, struct sigframe __user *sf) err |= restore_iwmmxt_context(&aux->iwmmxt); #endif #ifdef CONFIG_VFP -// if (err == 0) -// err |= vfp_restore_state(&sf->aux.vfp); + if (err == 0) + err |= restore_vfp_context(&aux->vfp); #endif return err; @@ -348,8 +433,8 @@ setup_sigframe(struct sigframe __user *sf, struct pt_regs *regs, sigset_t *set) err |= preserve_iwmmxt_context(&aux->iwmmxt); #endif #ifdef CONFIG_VFP -// if (err == 0) -// err |= vfp_save_state(&sf->aux.vfp); + if (err == 0) + err |= preserve_vfp_context(&aux->vfp); #endif __put_user_error(0, &aux->end_magic, err); -- cgit v1.2.3-70-g09d2