Blackfin arch: Faster C implementation of no-MPU CPLB handler

This is a mixture ofcMichael McTernan's patch and the existing cplb-mpu code.

We ditch the old cplb-nompu implementation, which is a good example of
why a good algorithm in a HLL is preferrable to a bad algorithm written in
assembly.  Rather than try to construct a table of all posible CPLBs and
search it, we just create a (smaller) table of memory regions and
their attributes.  Some of the data structures are now unified for both
the mpu and nompu cases.  A lot of needless complexity in cplbinit.c is
removed.

Further optimizations:
  * compile cplbmgr.c with a lot of -ffixed-reg options, and omit saving
    these registers on the stack when entering a CPLB exception.
  * lose cli/nop/nop/sti sequences for some workarounds - these don't
  * make
    sense in an exception context

Additional code unification should be possible after this.

[Mike Frysinger <vapier.adi@gmail.com>:
 - convert CPP if statements to C if statements
 - remove redundant statements
 - use a do...while loop rather than a for loop to get slightly better
   optimization and to avoid gcc "may be used uninitialized" warnings ...
   we know that the [id]cplb_nr_bounds variables will never be 0, so this
   is OK
 - the no-mpu code was the last user of MAX_MEM_SIZE and with that rewritten,
   we can punt it
 - add some BUG_ON() checks to make sure we dont overflow the small
   cplb_bounds array
 - add i/d cplb entries for the bootrom because there is functions/data in
   there we want to access
 - we do not need a NULL trailing entry as any time we access the bounds
   arrays, we use the nr_bounds variable
]

Signed-off-by: Michael McTernan <mmcternan@airvana.com>
Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Bernd Schmidt <bernds_cb1@t-online.de>
Signed-off-by: Bryan Wu <cooloney@kernel.org>

19 files changed, 515 insertions, 1442 deletions

diff --git a/arch/blackfin/Kconfig b/arch/blackfin/Kconfig
index b58af2b1bff..bbfeecae640 100644
--- a/arch/blackfin/Kconfig
+++ b/arch/blackfin/Kconfig
@@ -524,14 +524,6 @@ config MEM_SDGCTL
 	default 0x0
 endmenu
 
-config MAX_MEM_SIZE
-	int "Max SDRAM Memory Size in MBytes"
-	depends on !MPU
-	default 512
-	help
-	  This is the max memory size that the kernel will create CPLB
-	  tables for.  Your system will not be able to handle any more.
-
 #
 # Max & Min Speeds for various Chips
 #
diff --git a/arch/blackfin/include/asm/context.S b/arch/blackfin/include/asm/context.S
index 9ce21f68e91..16561ab18b3 100644
--- a/arch/blackfin/include/asm/context.S
+++ b/arch/blackfin/include/asm/context.S
@@ -357,3 +357,42 @@
 	SYSCFG = [sp++];
 	csync;
 .endm
+
+.macro save_context_cplb
+	[--sp] = (R7:0, P5:0);
+	[--sp] = fp;
+
+	[--sp] = a0.x;
+	[--sp] = a0.w;
+	[--sp] = a1.x;
+	[--sp] = a1.w;
+
+	[--sp] = LC0;
+	[--sp] = LC1;
+	[--sp] = LT0;
+	[--sp] = LT1;
+	[--sp] = LB0;
+	[--sp] = LB1;
+
+	[--sp] = RETS;
+.endm
+
+.macro restore_context_cplb
+	RETS = [sp++];
+
+	LB1 = [sp++];
+	LB0 = [sp++];
+	LT1 = [sp++];
+	LT0 = [sp++];
+	LC1 = [sp++];
+	LC0 = [sp++];
+
+	a1.w = [sp++];
+	a1.x = [sp++];
+	a0.w = [sp++];
+	a0.x = [sp++];
+
+	fp = [sp++];
+
+	(R7:0, P5:0) = [SP++];
+.endm
diff --git a/arch/blackfin/include/asm/cplb-mpu.h b/arch/blackfin/include/asm/cplb-mpu.h
deleted file mode 100644
index 80680ad7a37..00000000000
--- a/arch/blackfin/include/asm/cplb-mpu.h
+++ /dev/null
@@ -1,62 +0,0 @@
-/*
- * File:         include/asm-blackfin/cplbinit.h
- * Based on:
- * Author:
- *
- * Created:
- * Description:
- *
- * Modified:
- *               Copyright 2004-2006 Analog Devices Inc.
- *
- * Bugs:         Enter bugs at http://blackfin.uclinux.org/
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, see the file COPYING, or write
- * to the Free Software Foundation, Inc.,
- * 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-#ifndef __ASM_BFIN_CPLB_MPU_H
-#define __ASM_BFIN_CPLB_MPU_H
-#include <linux/threads.h>
-
-struct cplb_entry {
-	unsigned long data, addr;
-};
-
-struct mem_region {
-	unsigned long start, end;
-	unsigned long dcplb_data;
-	unsigned long icplb_data;
-};
-
-extern struct cplb_entry dcplb_tbl[NR_CPUS][MAX_CPLBS];
-extern struct cplb_entry icplb_tbl[NR_CPUS][MAX_CPLBS];
-extern int first_switched_icplb;
-extern int first_mask_dcplb;
-extern int first_switched_dcplb;
-
-extern int nr_dcplb_miss[], nr_icplb_miss[], nr_icplb_supv_miss[];
-extern int nr_dcplb_prot[], nr_cplb_flush[];
-
-extern int page_mask_order;
-extern int page_mask_nelts;
-
-extern unsigned long *current_rwx_mask[NR_CPUS];
-
-extern void flush_switched_cplbs(unsigned int);
-extern void set_mask_dcplbs(unsigned long *, unsigned int);
-
-extern void __noreturn panic_cplb_error(int seqstat, struct pt_regs *);
-
-#endif /* __ASM_BFIN_CPLB_MPU_H */
diff --git a/arch/blackfin/include/asm/cplb.h b/arch/blackfin/include/asm/cplb.h
index 5f7545d0620..ad566ff9ad1 100644
--- a/arch/blackfin/include/asm/cplb.h
+++ b/arch/blackfin/include/asm/cplb.h
@@ -116,4 +116,8 @@
 #define CPLB_INOCACHE   	CPLB_USER_RD | CPLB_VALID
 #define CPLB_IDOCACHE   	CPLB_INOCACHE | CPLB_L1_CHBL
 
+#define FAULT_RW        (1 << 16)
+#define FAULT_USERSUPV  (1 << 17)
+#define FAULT_CPLBBITS  0x0000ffff
+
 #endif				/* _CPLB_H */
diff --git a/arch/blackfin/include/asm/cplbinit.h b/arch/blackfin/include/asm/cplbinit.h
index 2aeec87d24e..05b14a631d0 100644
--- a/arch/blackfin/include/asm/cplbinit.h
+++ b/arch/blackfin/include/asm/cplbinit.h
@@ -32,96 +32,56 @@
 
 #include <asm/blackfin.h>
 #include <asm/cplb.h>
+#include <linux/threads.h>
 
-#ifdef CONFIG_MPU
-
-#include <asm/cplb-mpu.h>
-extern void bfin_icache_init(struct cplb_entry *icplb_tbl);
-extern void bfin_dcache_init(struct cplb_entry *icplb_tbl);
-
+#ifdef CONFIG_CPLB_SWITCH_TAB_L1
+# define PDT_ATTR __attribute__((l1_data))
 #else
+# define PDT_ATTR
+#endif
 
-#define INITIAL_T 0x1
-#define SWITCH_T  0x2
-#define I_CPLB    0x4
-#define D_CPLB    0x8
+struct cplb_entry {
+	unsigned long data, addr;
+};
 
-#define ASYNC_MEMORY_CPLB_COVERAGE  ((ASYNC_BANK0_SIZE + ASYNC_BANK1_SIZE + \
-				ASYNC_BANK2_SIZE + ASYNC_BANK3_SIZE) / SIZE_4M)
+struct cplb_boundary {
+	unsigned long eaddr; /* End of this region.  */
+	unsigned long data; /* CPLB data value.  */
+};
 
-#define CPLB_MEM CONFIG_MAX_MEM_SIZE
+extern struct cplb_boundary dcplb_bounds[];
+extern struct cplb_boundary icplb_bounds[];
+extern int dcplb_nr_bounds, icplb_nr_bounds;
 
-/*
-* Number of required data CPLB switchtable entries
-* MEMSIZE / 4 (we mostly install 4M page size CPLBs
-* approx 16 for smaller 1MB page size CPLBs for allignment purposes
-* 1 for L1 Data Memory
-* possibly 1 for L2 Data Memory
-* 1 for CONFIG_DEBUG_HUNT_FOR_ZERO
-* 1 for ASYNC Memory
-*/
-#define MAX_SWITCH_D_CPLBS (((CPLB_MEM / 4) + 16 + 1 + 1 + 1 \
-				 + ASYNC_MEMORY_CPLB_COVERAGE) * 2)
+extern struct cplb_entry dcplb_tbl[NR_CPUS][MAX_CPLBS];
+extern struct cplb_entry icplb_tbl[NR_CPUS][MAX_CPLBS];
+extern int first_switched_icplb;
+extern int first_switched_dcplb;
 
-/*
-* Number of required instruction CPLB switchtable entries
-* MEMSIZE / 4 (we mostly install 4M page size CPLBs
-* approx 12 for smaller 1MB page size CPLBs for allignment purposes
-* 1 for L1 Instruction Memory
-* possibly 1 for L2 Instruction Memory
-* 1 for CONFIG_DEBUG_HUNT_FOR_ZERO
-*/
-#define MAX_SWITCH_I_CPLBS (((CPLB_MEM / 4) + 12 + 1 + 1 + 1) * 2)
-
-/* Number of CPLB table entries, used for cplb-nompu. */
-#define CPLB_TBL_ENTRIES (16 * 4)
-
-enum {
-	ZERO_P, L1I_MEM, L1D_MEM, L2_MEM, SDRAM_KERN, SDRAM_RAM_MTD, SDRAM_DMAZ,
-	RES_MEM, ASYNC_MEM, OCB_ROM
-};
+extern int nr_dcplb_miss[], nr_icplb_miss[], nr_icplb_supv_miss[];
+extern int nr_dcplb_prot[], nr_cplb_flush[];
 
-struct cplb_desc {
-	u32 start; /* start address */
-	u32 end; /* end address */
-	u32 psize; /* prefered size if any otherwise 1MB or 4MB*/
-	u16 attr;/* attributes */
-	u16 i_conf;/* I-CPLB DATA */
-	u16 d_conf;/* D-CPLB DATA */
-	u16 valid;/* valid */
-	const s8 name[30];/* name */
-};
+#ifdef CONFIG_MPU
 
-struct cplb_tab {
-  u_long *tab;
-	u16 pos;
-	u16 size;
-};
+extern int first_mask_dcplb;
 
-extern u_long icplb_tables[NR_CPUS][CPLB_TBL_ENTRIES+1];
-extern u_long dcplb_tables[NR_CPUS][CPLB_TBL_ENTRIES+1];
+extern int page_mask_order;
+extern int page_mask_nelts;
 
-/* Till here we are discussing about the static memory management model.
- * However, the operating envoronments commonly define more CPLB
- * descriptors to cover the entire addressable memory than will fit into
- * the available on-chip 16 CPLB MMRs. When this happens, the below table
- * will be used which will hold all the potentially required CPLB descriptors
- *
- * This is how Page descriptor Table is implemented in uClinux/Blackfin.
- */
+extern unsigned long *current_rwx_mask[NR_CPUS];
 
-extern u_long ipdt_tables[NR_CPUS][MAX_SWITCH_I_CPLBS+1];
-extern u_long dpdt_tables[NR_CPUS][MAX_SWITCH_D_CPLBS+1];
-#ifdef CONFIG_CPLB_INFO
-extern u_long ipdt_swapcount_tables[NR_CPUS][MAX_SWITCH_I_CPLBS];
-extern u_long dpdt_swapcount_tables[NR_CPUS][MAX_SWITCH_D_CPLBS];
-#endif
-extern void bfin_icache_init(u_long icplbs[]);
-extern void bfin_dcache_init(u_long dcplbs[]);
+extern void flush_switched_cplbs(unsigned int);
+extern void set_mask_dcplbs(unsigned long *, unsigned int);
+
+extern void __noreturn panic_cplb_error(int seqstat, struct pt_regs *);
 
 #endif /* CONFIG_MPU */
 
+extern void bfin_icache_init(struct cplb_entry *icplb_tbl);
+extern void bfin_dcache_init(struct cplb_entry *icplb_tbl);
+
 #if defined(CONFIG_BFIN_DCACHE) || defined(CONFIG_BFIN_ICACHE)
+extern void generate_cplb_tables_all(void);
 extern void generate_cplb_tables_cpu(unsigned int cpu);
 #endif
 #endif
diff --git a/arch/blackfin/include/asm/entry.h b/arch/blackfin/include/asm/entry.h
index c4f721e0d00..d94e4f5139d 100644
--- a/arch/blackfin/include/asm/entry.h
+++ b/arch/blackfin/include/asm/entry.h
@@ -53,9 +53,11 @@
 /* This one pushes RETI without using CLI.  Interrupts are enabled.  */
 #define SAVE_CONTEXT_SYSCALL	save_context_syscall
 #define SAVE_CONTEXT		save_context_with_interrupts
+#define SAVE_CONTEXT_CPLB	save_context_cplb
 
 #define RESTORE_ALL_SYS		restore_context_no_interrupts
 #define RESTORE_CONTEXT		restore_context_with_interrupts
+#define RESTORE_CONTEXT_CPLB	restore_context_cplb
 
 #endif				/* __ASSEMBLY__ */
 #endif				/* __BFIN_ENTRY_H */
diff --git a/arch/blackfin/kernel/cplb-mpu/Makefile b/arch/blackfin/kernel/cplb-mpu/Makefile
index bd92301a704..7d70d3bf321 100644
--- a/arch/blackfin/kernel/cplb-mpu/Makefile
+++ b/arch/blackfin/kernel/cplb-mpu/Makefile
@@ -3,3 +3,8 @@
 #
 
 obj-y := cplbinit.o cacheinit.o cplbmgr.o
+
+CFLAGS_cplbmgr.o := -ffixed-I0 -ffixed-I1 -ffixed-I2 -ffixed-I3 \
+		    -ffixed-L0 -ffixed-L1 -ffixed-L2 -ffixed-L3 \
+		    -ffixed-M0 -ffixed-M1 -ffixed-M2 -ffixed-M3 \
+		    -ffixed-B0 -ffixed-B1 -ffixed-B2 -ffixed-B3
diff --git a/arch/blackfin/kernel/cplb-mpu/cplbinit.c b/arch/blackfin/kernel/cplb-mpu/cplbinit.c
index 1ea7c18435a..bdb958486e7 100644
--- a/arch/blackfin/kernel/cplb-mpu/cplbinit.c
+++ b/arch/blackfin/kernel/cplb-mpu/cplbinit.c
@@ -107,3 +107,7 @@ void __init generate_cplb_tables_cpu(unsigned int cpu)
 	while (i_i < MAX_CPLBS)
 		icplb_tbl[cpu][i_i++].data = 0;
 }
+
+void generate_cplb_tables_all(void)
+{
+}
diff --git a/arch/blackfin/kernel/cplb-mpu/cplbmgr.c b/arch/blackfin/kernel/cplb-mpu/cplbmgr.c
index 76bd99177de..5ef5d1a787f 100644
--- a/arch/blackfin/kernel/cplb-mpu/cplbmgr.c
+++ b/arch/blackfin/kernel/cplb-mpu/cplbmgr.c
@@ -25,8 +25,13 @@
 #include <asm/cplbinit.h>
 #include <asm/mmu_context.h>
 
-#define FAULT_RW	(1 << 16)
-#define FAULT_USERSUPV	(1 << 17)
+/*
+ * WARNING
+ *
+ * This file is compiled with certain -ffixed-reg options.  We have to
+ * make sure not to call any functions here that could clobber these
+ * registers.
+ */
 
 int page_mask_nelts;
 int page_mask_order;
diff --git a/arch/blackfin/kernel/cplb-nompu/Makefile b/arch/blackfin/kernel/cplb-nompu/Makefile
index 4010eca1c6c..7d70d3bf321 100644
--- a/arch/blackfin/kernel/cplb-nompu/Makefile
+++ b/arch/blackfin/kernel/cplb-nompu/Makefile
@@ -2,4 +2,9 @@
 # arch/blackfin/kernel/cplb-nompu/Makefile
 #
 
-obj-y := cplbinit.o cacheinit.o cplbhdlr.o cplbmgr.o
+obj-y := cplbinit.o cacheinit.o cplbmgr.o
+
+CFLAGS_cplbmgr.o := -ffixed-I0 -ffixed-I1 -ffixed-I2 -ffixed-I3 \
+		    -ffixed-L0 -ffixed-L1 -ffixed-L2 -ffixed-L3 \
+		    -ffixed-M0 -ffixed-M1 -ffixed-M2 -ffixed-M3 \
+		    -ffixed-B0 -ffixed-B1 -ffixed-B2 -ffixed-B3
diff --git a/arch/blackfin/kernel/cplb-nompu/cacheinit.c b/arch/blackfin/kernel/cplb-nompu/cacheinit.c
index 3a385aec67d..c6ff947f9d3 100644
--- a/arch/blackfin/kernel/cplb-nompu/cacheinit.c
+++ b/arch/blackfin/kernel/cplb-nompu/cacheinit.c
@@ -25,19 +25,15 @@
 #include <asm/cplbinit.h>
 
 #if defined(CONFIG_BFIN_ICACHE)
-void __cpuinit bfin_icache_init(u_long icplb[])
+void __cpuinit bfin_icache_init(struct cplb_entry *icplb_tbl)
 {
-	unsigned long *table = icplb;
 	unsigned long ctrl;
 	int i;
 
+	SSYNC();
 	for (i = 0; i < MAX_CPLBS; i++) {
-		unsigned long addr = *table++;
-		unsigned long data = *table++;
-		if (addr == (unsigned long)-1)
-			break;
-		bfin_write32(ICPLB_ADDR0 + i * 4, addr);
-		bfin_write32(ICPLB_DATA0 + i * 4, data);
+		bfin_write32(ICPLB_ADDR0 + i * 4, icplb_tbl[i].addr);
+		bfin_write32(ICPLB_DATA0 + i * 4, icplb_tbl[i].data);
 	}
 	ctrl = bfin_read_IMEM_CONTROL();
 	ctrl |= IMC | ENICPLB;
@@ -47,24 +43,20 @@ void __cpuinit bfin_icache_init(u_long icplb[])
 #endif
 
 #if defined(CONFIG_BFIN_DCACHE)
-void __cpuinit bfin_dcache_init(u_long dcplb[])
+void __cpuinit bfin_dcache_init(struct cplb_entry *dcplb_tbl)
 {
-	unsigned long *table = dcplb;
 	unsigned long ctrl;
 	int i;
 
+	SSYNC();
 	for (i = 0; i < MAX_CPLBS; i++) {
-		unsigned long addr = *table++;
-		unsigned long data = *table++;
-		if (addr == (unsigned long)-1)
-			break;
-		bfin_write32(DCPLB_ADDR0 + i * 4, addr);
-		bfin_write32(DCPLB_DATA0 + i * 4, data);
+		bfin_write32(DCPLB_ADDR0 + i * 4, dcplb_tbl[i].addr);
+		bfin_write32(DCPLB_DATA0 + i * 4, dcplb_tbl[i].data);
 	}
+
 	ctrl = bfin_read_DMEM_CONTROL();
 	ctrl |= DMEM_CNTR;
 	bfin_write_DMEM_CONTROL(ctrl);
-
 	SSYNC();
 }
 #endif
diff --git a/arch/blackfin/kernel/cplb-nompu/cplbhdlr.S b/arch/blackfin/kernel/cplb-nompu/cplbhdlr.S
deleted file mode 100644
index ecbabc0a1fe..00000000000
--- a/arch/blackfin/kernel/cplb-nompu/cplbhdlr.S
+++ /dev/null
@@ -1,130 +0,0 @@
-/*
- * File:         arch/blackfin/mach-common/cplbhdlr.S
- * Based on:
- * Author:       LG Soft India
- *
- * Created:      ?
- * Description:  CPLB exception handler
- *
- * Modified:
- *               Copyright 2004-2006 Analog Devices Inc.
- *
- * Bugs:         Enter bugs at http://blackfin.uclinux.org/
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, see the file COPYING, or write
- * to the Free Software Foundation, Inc.,
- * 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-
-#include <linux/linkage.h>
-#include <asm/cplb.h>
-#include <asm/entry.h>
-
-#ifdef CONFIG_EXCPT_IRQ_SYSC_L1
-.section .l1.text
-#else
-.text
-#endif
-
-.type _cplb_mgr, STT_FUNC;
-.type _panic_cplb_error, STT_FUNC;
-
-.align 2
-
-ENTRY(__cplb_hdr)
-	R2 = SEQSTAT;
-
-	/* Mask the contents of SEQSTAT and leave only EXCAUSE in R2 */
-	R2 <<= 26;
-	R2 >>= 26;
-
-	R1 = 0x23; /* Data access CPLB protection violation */
-	CC = R2 == R1;
-	IF !CC JUMP .Lnot_data_write;
-	R0 = 2;		/* is a write to data space*/
-	JUMP .Lis_icplb_miss;
-
-.Lnot_data_write:
-	R1 = 0x2C; /* CPLB miss on an instruction fetch */
-	CC = R2 == R1;
-	R0 = 0;		/* is_data_miss == False*/
-	IF CC JUMP .Lis_icplb_miss;
-
-	R1 = 0x26;
-	CC = R2 == R1;
-	IF !CC JUMP .Lunknown;
-
-	R0 = 1;		/* is_data_miss == True*/
-
-.Lis_icplb_miss:
-
-#if defined(CONFIG_BFIN_ICACHE) || defined(CONFIG_BFIN_DCACHE)
-# if defined(CONFIG_BFIN_ICACHE) && !defined(CONFIG_BFIN_DCACHE)
-	R1 = CPLB_ENABLE_ICACHE;
-# endif
-# if !defined(CONFIG_BFIN_ICACHE) && defined(CONFIG_BFIN_DCACHE)
-	R1 = CPLB_ENABLE_DCACHE;
-# endif
-# if defined(CONFIG_BFIN_ICACHE) && defined(CONFIG_BFIN_DCACHE)
-	R1 = CPLB_ENABLE_DCACHE | CPLB_ENABLE_ICACHE;
-# endif
-#else
-	R1 = 0;
-#endif
-
-	[--SP] = RETS;
-	CALL _cplb_mgr;
-	RETS = [SP++];
-	CC = R0 == 0;
-	IF !CC JUMP .Lnot_replaced;
-	RTS;
-
-/*
- * Diagnostic exception handlers
- */
-.Lunknown:
-	R0 = CPLB_UNKNOWN_ERR;
-	JUMP .Lcplb_error;
-
-.Lnot_replaced:
-	CC = R0 == CPLB_NO_UNLOCKED;
-	IF !CC JUMP .Lnext_check;
-	R0 = CPLB_NO_UNLOCKED;
-	JUMP .Lcplb_error;
-
-.Lnext_check:
-	CC = R0 == CPLB_NO_ADDR_MATCH;
-	IF !CC JUMP .Lnext_check2;
-	R0 = CPLB_NO_ADDR_MATCH;
-	JUMP .Lcplb_error;
-
-.Lnext_check2:
-	CC = R0 == CPLB_PROT_VIOL;
-	IF !CC JUMP .Lstrange_return_from_cplb_mgr;
-	R0 = CPLB_PROT_VIOL;
-	JUMP .Lcplb_error;
-
-.Lstrange_return_from_cplb_mgr:
-	IDLE;
-	CSYNC;
-	JUMP .Lstrange_return_from_cplb_mgr;
-
-.Lcplb_error:
-	R1 = sp;
-	SP += -12;
-	call _panic_cplb_error;
-	SP += 12;
-	JUMP.L _handle_bad_cplb;
-
-ENDPROC(__cplb_hdr)
diff --git a/arch/blackfin/kernel/cplb-nompu/cplbinit.c b/arch/blackfin/kernel/cplb-nompu/cplbinit.c
index 4c010ba50a8..0e28f759573 100644
--- a/arch/blackfin/kernel/cplb-nompu/cplbinit.c
+++ b/arch/blackfin/kernel/cplb-nompu/cplbinit.c
@@ -29,417 +29,143 @@
 #include <asm/cplbinit.h>
 #include <asm/mem_map.h>
 
-u_long icplb_tables[NR_CPUS][CPLB_TBL_ENTRIES+1];
-u_long dcplb_tables[NR_CPUS][CPLB_TBL_ENTRIES+1];
+struct cplb_entry icplb_tbl[NR_CPUS][MAX_CPLBS] PDT_ATTR;
+struct cplb_entry dcplb_tbl[NR_CPUS][MAX_CPLBS] PDT_ATTR;
 
-#ifdef CONFIG_CPLB_SWITCH_TAB_L1
-#define PDT_ATTR __attribute__((l1_data))
-#else
-#define PDT_ATTR
-#endif
-
-u_long ipdt_tables[NR_CPUS][MAX_SWITCH_I_CPLBS+1] PDT_ATTR;
-u_long dpdt_tables[NR_CPUS][MAX_SWITCH_D_CPLBS+1] PDT_ATTR;
-#ifdef CONFIG_CPLB_INFO
-u_long ipdt_swapcount_tables[NR_CPUS][MAX_SWITCH_I_CPLBS] PDT_ATTR;
-u_long dpdt_swapcount_tables[NR_CPUS][MAX_SWITCH_D_CPLBS] PDT_ATTR;
-#endif
+int first_switched_icplb PDT_ATTR;
+int first_switched_dcplb PDT_ATTR;
 
-struct s_cplb {
-	struct cplb_tab init_i;
-	struct cplb_tab init_d;
-	struct cplb_tab switch_i;
-	struct cplb_tab switch_d;
-};
+struct cplb_boundary dcplb_bounds[9] PDT_ATTR;
+struct cplb_boundary icplb_bounds[7] PDT_ATTR;
 
-#if defined(CONFIG_BFIN_DCACHE) || defined(CONFIG_BFIN_ICACHE)
-static struct cplb_desc cplb_data[] = {
-	{
-		.start = 0,
-		.end = SIZE_1K,
-		.psize = SIZE_1K,
-		.attr = INITIAL_T | SWITCH_T | I_CPLB | D_CPLB,
-		.i_conf = SDRAM_OOPS,
-		.d_conf = SDRAM_OOPS,
-#if defined(CONFIG_DEBUG_HUNT_FOR_ZERO)
-		.valid = 1,
-#else
-		.valid = 0,
-#endif
-		.name = "Zero Pointer Guard Page",
-	},
-	{
-		.start = 0,	/* dyanmic */
-		.end = 0,	/* dynamic */
-		.psize = SIZE_4M,
-		.attr = INITIAL_T | SWITCH_T | I_CPLB,
-		.i_conf = L1_IMEMORY,
-		.d_conf = 0,
-		.valid = 1,
-		.name = "L1 I-Memory",
-	},
-	{
-		.start = 0,	/* dynamic */
-		.end = 0,	/* dynamic */
-		.psize = SIZE_4M,
-		.attr = INITIAL_T | SWITCH_T | D_CPLB,
-		.i_conf = 0,
-		.d_conf = L1_DMEMORY,
-#if ((L1_DATA_A_LENGTH > 0) || (L1_DATA_B_LENGTH > 0))
-		.valid = 1,
-#else
-		.valid = 0,
-#endif
-		.name = "L1 D-Memory",
-	},
-	{
-		.start = L2_START,
-		.end = L2_START + L2_LENGTH,
-		.psize = SIZE_1M,
-		.attr = L2_ATTR,
-		.i_conf = L2_IMEMORY,
-		.d_conf = L2_DMEMORY,
-		.valid = (L2_LENGTH > 0),
-		.name = "L2 Memory",
-	},
-	{
-		.start = 0,
-		.end = 0,  /* dynamic */
-		.psize = 0,
-		.attr = INITIAL_T | SWITCH_T | I_CPLB | D_CPLB,
-		.i_conf = SDRAM_IGENERIC,
-		.d_conf = SDRAM_DGENERIC,
-		.valid = 1,
-		.name = "Kernel Memory",
-	},
-	{
-		.start = 0, /* dynamic */
-		.end = 0, /* dynamic */
-		.psize = 0,
-		.attr = INITIAL_T | SWITCH_T | D_CPLB,
-		.i_conf = SDRAM_IGENERIC,
-		.d_conf = SDRAM_DNON_CHBL,
-		.valid = 1,
-		.name = "uClinux MTD Memory",
-	},
-	{
-		.start = 0, /* dynamic */
-		.end = 0,   /* dynamic */
-		.psize = SIZE_1M,
-		.attr = INITIAL_T | SWITCH_T | D_CPLB,
-		.d_conf = SDRAM_DNON_CHBL,
-		.valid = 1,
-		.name = "Uncached DMA Zone",
-	},
-	{
-		.start = 0, /* dynamic */
-		.end = 0, /* dynamic */
-		.psize = 0,
-		.attr = SWITCH_T | D_CPLB,
-		.i_conf = 0, /* dynamic */
-		.d_conf = 0, /* dynamic */
-		.valid = 1,
-		.name = "Reserved Memory",
-	},
-	{
-		.start = ASYNC_BANK0_BASE,
-		.end = ASYNC_BANK3_BASE + ASYNC_BANK3_SIZE,
-		.psize = 0,
-		.attr = SWITCH_T | D_CPLB,
-		.d_conf = SDRAM_EBIU,
-		.valid = 1,
-		.name = "Asynchronous Memory Banks",
-	},
-	{
-		.start = BOOT_ROM_START,
-		.end = BOOT_ROM_START + BOOT_ROM_LENGTH,
-		.psize = SIZE_1M,
-		.attr = SWITCH_T | I_CPLB | D_CPLB,
-		.i_conf = SDRAM_IGENERIC,
-		.d_conf = SDRAM_DGENERIC,
-		.valid = 1,
-		.name = "On-Chip BootROM",
-	},
-};
+int icplb_nr_bounds PDT_ATTR;
+int dcplb_nr_bounds PDT_ATTR;
 
-static bool __init lock_kernel_check(u32 start, u32 end)
+void __init generate_cplb_tables_cpu(unsigned int cpu)
 {
-	if (start >= (u32)__init_begin || end <= (u32)_stext)
-		return false;
-
-	/* This cplb block overlapped with kernel area. */
-	return true;
-}
+	int i_d, i_i;
+	unsigned long addr;
 
-static void __init
-fill_cplbtab(struct cplb_tab *table,
-	     unsigned long start, unsigned long end,
-	     unsigned long block_size, unsigned long cplb_data)
-{
-	int i;
+	struct cplb_entry *d_tbl = dcplb_tbl[cpu];
+	struct cplb_entry *i_tbl = icplb_tbl[cpu];
 
-	switch (block_size) {
-	case SIZE_4M:
-		i = 3;
-		break;
-	case SIZE_1M:
-		i = 2;
-		break;
-	case SIZE_4K:
-		i = 1;
-		break;
-	case SIZE_1K:
-	default:
-		i = 0;
-		break;
-	}
+	printk(KERN_INFO "NOMPU: setting up cplb tables\n");
 
-	cplb_data = (cplb_data & ~(3 << 16)) | (i << 16);
+	i_d = i_i = 0;
 
-	while ((start < end) && (table->pos < table->size)) {
+	/* Set up the zero page.  */
+	d_tbl[i_d].addr = 0;
+	d_tbl[i_d++].data = SDRAM_OOPS | PAGE_SIZE_1KB;
 
-		table->tab[table->pos++] = start;
+	/* Cover kernel memory with 4M pages.  */
+	addr = 0;
 
-		if (lock_kernel_check(start, start + block_size))
-			table->tab[table->pos++] =
-			    cplb_data | CPLB_LOCK | CPLB_DIRTY;
-		else
-			table->tab[table->pos++] = cplb_data;
+	for (; addr < memory_start; addr += 4 * 1024 * 1024) {
+		d_tbl[i_d].addr = addr;
+		d_tbl[i_d++].data = SDRAM_DGENERIC | PAGE_SIZE_4MB;
+		i_tbl[i_i].addr = addr;
+		i_tbl[i_i++].data = SDRAM_IGENERIC | PAGE_SIZE_4MB;
+	}
 
-		start += block_size;
+	/* Cover L1 memory.  One 4M area for code and data each is enough.  */
+	if (L1_DATA_A_LENGTH || L1_DATA_B_LENGTH) {
+		d_tbl[i_d].addr = L1_DATA_A_START;
+		d_tbl[i_d++].data = L1_DMEMORY | PAGE_SIZE_4MB;
 	}
-}
+	i_tbl[i_i].addr = L1_CODE_START;
+	i_tbl[i_i++].data = L1_IMEMORY | PAGE_SIZE_4MB;
 
-static void __init close_cplbtab(struct cplb_tab *table)
-{
-	while (table->pos < table->size)
-		table->tab[table->pos++] = 0;
-}
+	first_switched_dcplb = i_d;
+	first_switched_icplb = i_i;
 
-/* helper function */
-static void __init
-__fill_code_cplbtab(struct cplb_tab *t, int i, u32 a_start, u32 a_end)
-{
-	if (cplb_data[i].psize) {
-		fill_cplbtab(t,
-				cplb_data[i].start,
-				cplb_data[i].end,
-				cplb_data[i].psize,
-				cplb_data[i].i_conf);
-	} else {
-#if defined(CONFIG_BFIN_ICACHE)
-		if (ANOMALY_05000263 && i == SDRAM_KERN) {
-			fill_cplbtab(t,
-					cplb_data[i].start,
-					cplb_data[i].end,
-					SIZE_4M,
-					cplb_data[i].i_conf);
-		} else
-#endif
-		{
-			fill_cplbtab(t,
-					cplb_data[i].start,
-					a_start,
-					SIZE_1M,
-					cplb_data[i].i_conf);
-			fill_cplbtab(t,
-					a_start,
-					a_end,
-					SIZE_4M,
-					cplb_data[i].i_conf);
-			fill_cplbtab(t, a_end,
-					cplb_data[i].end,
-					SIZE_1M,
-					cplb_data[i].i_conf);
-		}
-	}
-}
+	BUG_ON(first_switched_dcplb > MAX_CPLBS);
+	BUG_ON(first_switched_icplb > MAX_CPLBS);
 
-static void __init
-__fill_data_cplbtab(struct cplb_tab *t, int i, u32 a_start, u32 a_end)
-{
-	if (cplb_data[i].psize) {
-		fill_cplbtab(t,
-				cplb_data[i].start,
-				cplb_data[i].end,
-				cplb_data[i].psize,
-				cplb_data[i].d_conf);
-	} else {
-		fill_cplbtab(t,
-				cplb_data[i].start,
-				a_start, SIZE_1M,
-				cplb_data[i].d_conf);
-		fill_cplbtab(t, a_start,
-				a_end, SIZE_4M,
-				cplb_data[i].d_conf);
-		fill_cplbtab(t, a_end,
-				cplb_data[i].end,
-				SIZE_1M,
-				cplb_data[i].d_conf);
-	}
+	while (i_d < MAX_CPLBS)
+		d_tbl[i_d++].data = 0;
+	while (i_i < MAX_CPLBS)
+		i_tbl[i_i++].data = 0;
 }
 
-void __init generate_cplb_tables_cpu(unsigned int cpu)
+void __init generate_cplb_tables_all(void)
 {
+	int i_d, i_i;
 
-	u16 i, j, process;
-	u32 a_start, a_end, as, ae, as_1m;
-
-	struct cplb_tab *t_i = NULL;
-	struct cplb_tab *t_d = NULL;
-	struct s_cplb cplb;
-
-	printk(KERN_INFO "NOMPU: setting up cplb tables for global access\n");
-
-	cplb.init_i.size = CPLB_TBL_ENTRIES;
-	cplb.init_d.size = CPLB_TBL_ENTRIES;
-	cplb.switch_i.size = MAX_SWITCH_I_CPLBS;
-	cplb.switch_d.size = MAX_SWITCH_D_CPLBS;
-
-	cplb.init_i.pos = 0;
-	cplb.init_d.pos = 0;
-	cplb.switch_i.pos = 0;
-	cplb.switch_d.pos = 0;
-
-	cplb.init_i.tab = icplb_tables[cpu];
-	cplb.init_d.tab = dcplb_tables[cpu];
-	cplb.switch_i.tab = ipdt_tables[cpu];
-	cplb.switch_d.tab = dpdt_tables[cpu];
-
-	cplb_data[L1I_MEM].start = get_l1_code_start_cpu(cpu);
-	cplb_data[L1I_MEM].end = cplb_data[L1I_MEM].start + L1_CODE_LENGTH;
-	cplb_data[L1D_MEM].start = get_l1_data_a_start_cpu(cpu);
-	cplb_data[L1D_MEM].end = get_l1_data_b_start_cpu(cpu) + L1_DATA_B_LENGTH;
-	cplb_data[SDRAM_KERN].end = memory_end;
-
+	i_d = 0;
+	/* Normal RAM, including MTD FS.  */
 #ifdef CONFIG_MTD_UCLINUX
-	cplb_data[SDRAM_RAM_MTD].start = memory_mtd_start;
-	cplb_data[SDRAM_RAM_MTD].end = memory_mtd_start + mtd_size;
-	cplb_data[SDRAM_RAM_MTD].valid = mtd_size > 0;
-# if defined(CONFIG_ROMFS_FS)
-	cplb_data[SDRAM_RAM_MTD].attr |= I_CPLB;
-
-	/*
-	 * The ROMFS_FS size is often not multiple of 1MB.
-	 * This can cause multiple CPLB sets covering the same memory area.
-	 * This will then cause multiple CPLB hit exceptions.
-	 * Workaround: We ensure a contiguous memory area by extending the kernel
-	 * memory section over the mtd section.
-	 * For ROMFS_FS memory must be covered with ICPLBs anyways.
-	 * So there is no difference between kernel and mtd memory setup.
-	 */
-
-	cplb_data[SDRAM_KERN].end = memory_mtd_start + mtd_size;;
-	cplb_data[SDRAM_RAM_MTD].valid = 0;
-
-# endif
+	dcplb_bounds[i_d].eaddr = memory_mtd_start + mtd_size;
 #else
-	cplb_data[SDRAM_RAM_MTD].valid = 0;
+	dcplb_bounds[i_d].eaddr = memory_end;
 #endif
+	dcplb_bounds[i_d++].data = SDRAM_DGENERIC;
+	/* DMA uncached region.  */
+	if (DMA_UNCACHED_REGION) {
+		dcplb_bounds[i_d].eaddr = _ramend;
+		dcplb_bounds[i_d++].data = SDRAM_DNON_CHBL;
+	}
+	if (_ramend != physical_mem_end) {
+		/* Reserved memory.  */
+		dcplb_bounds[i_d].eaddr = physical_mem_end;
+		dcplb_bounds[i_d++].data = (reserved_mem_dcache_on ?
+					    SDRAM_DGENERIC : SDRAM_DNON_CHBL);
+	}
+	/* Addressing hole up to the async bank.  */
+	dcplb_bounds[i_d].eaddr = ASYNC_BANK0_BASE;
+	dcplb_bounds[i_d++].data = 0;
+	/* ASYNC banks.  */
+	dcplb_bounds[i_d].eaddr = ASYNC_BANK3_BASE + ASYNC_BANK3_SIZE;
+	dcplb_bounds[i_d++].data = SDRAM_EBIU;
+	/* Addressing hole up to BootROM.  */
+	dcplb_bounds[i_d].eaddr = BOOT_ROM_START;
+	dcplb_bounds[i_d++].data = 0;
+	/* BootROM -- largest one should be less than 1 meg.  */
+	dcplb_bounds[i_d].eaddr = BOOT_ROM_START + (1 * 1024 * 1024);
+	dcplb_bounds[i_d++].data = SDRAM_DGENERIC;
+	if (L2_LENGTH) {
+		/* Addressing hole up to L2 SRAM.  */
+		dcplb_bounds[i_d].eaddr = L2_START;
+		dcplb_bounds[i_d++].data = 0;
+		/* L2 SRAM.  */
+		dcplb_bounds[i_d].eaddr = L2_START + L2_LENGTH;
+		dcplb_bounds[i_d++].data = L2_DMEMORY;
+	}
+	dcplb_nr_bounds = i_d;
+	BUG_ON(dcplb_nr_bounds > ARRAY_SIZE(dcplb_bounds));
 
-	cplb_data[SDRAM_DMAZ].start = _ramend - DMA_UNCACHED_REGION;
-	cplb_data[SDRAM_DMAZ].end = _ramend;
-
-	cplb_data[RES_MEM].start = _ramend;
-	cplb_data[RES_MEM].end = physical_mem_end;
-
-	if (reserved_mem_dcache_on)
-		cplb_data[RES_MEM].d_conf = SDRAM_DGENERIC;
-	else
-		cplb_data[RES_MEM].d_conf = SDRAM_DNON_CHBL;
-
-	if (reserved_mem_icache_on)
-		cplb_data[RES_MEM].i_conf = SDRAM_IGENERIC;
-	else
-		cplb_data[RES_MEM].i_conf = SDRAM_INON_CHBL;
-
-	for (i = ZERO_P; i < ARRAY_SIZE(cplb_data); ++i) {
-		if (!cplb_data[i].valid)
-			continue;
-
-		as_1m = cplb_data[i].start % SIZE_1M;
-
-		/* We need to make sure all sections are properly 1M aligned
-		 * However between Kernel Memory and the Kernel mtd section, depending on the
-		 * rootfs size, there can be overlapping memory areas.
-		 */
-
-		if (as_1m && i != L1I_MEM && i != L1D_MEM) {
+	i_i = 0;
+	/* Normal RAM, including MTD FS.  */
 #ifdef CONFIG_MTD_UCLINUX
-			if (i == SDRAM_RAM_MTD) {
-				if ((cplb_data[SDRAM_KERN].end + 1) > cplb_data[SDRAM_RAM_MTD].start)
-					cplb_data[SDRAM_RAM_MTD].start = (cplb_data[i].start & (-2*SIZE_1M)) + SIZE_1M;
-				else
-					cplb_data[SDRAM_RAM_MTD].start = (cplb_data[i].start & (-2*SIZE_1M));
-			} else
+	icplb_bounds[i_i].eaddr = memory_mtd_start + mtd_size;
+#else
+	icplb_bounds[i_i].eaddr = memory_end;
 #endif
-				printk(KERN_WARNING "Unaligned Start of %s at 0x%X\n",
-				       cplb_data[i].name, cplb_data[i].start);
-		}
-
-		as = cplb_data[i].start % SIZE_4M;
-		ae = cplb_data[i].end % SIZE_4M;
-
-		if (as)
-			a_start = cplb_data[i].start + (SIZE_4M - (as));
-		else
-			a_start = cplb_data[i].start;
-
-		a_end = cplb_data[i].end - ae;
-
-		for (j = INITIAL_T; j <= SWITCH_T; j++) {
-
-			switch (j) {
-			case INITIAL_T:
-				if (cplb_data[i].attr & INITIAL_T) {
-					t_i = &cplb.init_i;
-					t_d = &cplb.init_d;
-					process = 1;
-				} else
-					process = 0;
-				break;
-			case SWITCH_T:
-				if (cplb_data[i].attr & SWITCH_T) {
-					t_i = &cplb.switch_i;
-					t_d = &cplb.switch_d;
-					process = 1;
-				} else
-					process = 0;
-				break;
-			default:
-					process = 0;
-				break;
-			}
-
-			if (!process)
-				continue;
-			if (cplb_data[i].attr & I_CPLB)
-				__fill_code_cplbtab(t_i, i, a_start, a_end);
-
-			if (cplb_data[i].attr & D_CPLB)
-				__fill_data_cplbtab(t_d, i, a_start, a_end);
-		}
+	icplb_bounds[i_i++].data = SDRAM_IGENERIC;
+	/* DMA uncached region.  */
+	if (DMA_UNCACHED_REGION) {
+		icplb_bounds[i_i].eaddr = _ramend;
+		icplb_bounds[i_i++].data = 0;
 	}
-
-	/* make sure we locked the kernel start */
-	BUG_ON(cplb.init_i.pos < 2 + cplb_data[ZERO_P].valid);
-	BUG_ON(cplb.init_d.pos < 1 + cplb_data[ZERO_P].valid + cplb_data[L1D_MEM].valid);
-
-	/* make sure we didnt overflow the table */
-	BUG_ON(cplb.init_i.size < cplb.init_i.pos);
-	BUG_ON(cplb.init_d.size < cplb.init_d.pos);
-	BUG_ON(cplb.switch_i.size < cplb.switch_i.pos);
-	BUG_ON(cplb.switch_d.size < cplb.switch_d.pos);
-
-	/* close tables */
-	close_cplbtab(&cplb.init_i);
-	close_cplbtab(&cplb.init_d);
-
-	cplb.init_i.tab[cplb.init_i.pos] = -1;
-	cplb.init_d.tab[cplb.init_d.pos] = -1;
-	cplb.switch_i.tab[cplb.switch_i.pos] = -1;
-	cplb.switch_d.tab[cplb.switch_d.pos] = -1;
-
+	if (_ramend != physical_mem_end) {
+		/* Reserved memory.  */
+		icplb_bounds[i_i].eaddr = physical_mem_end;
+		icplb_bounds[i_i++].data = (reserved_mem_icache_on ?
+					    SDRAM_IGENERIC : SDRAM_INON_CHBL);
+	}
+	/* Addressing hole up to BootROM.  */
+	icplb_bounds[i_i].eaddr = BOOT_ROM_START;
+	icplb_bounds[i_i++].data = 0;
+	/* BootROM -- largest one should be less than 1 meg.  */
+	icplb_bounds[i_i].eaddr = BOOT_ROM_START + (1 * 1024 * 1024);
+	icplb_bounds[i_i++].data = SDRAM_IGENERIC;
+	if (L2_LENGTH) {
+		/* Addressing hole up to L2 SRAM, including the async bank.  */
+		icplb_bounds[i_i].eaddr = L2_START;
+		icplb_bounds[i_i++].data = 0;
+		/* L2 SRAM.  */
+		icplb_bounds[i_i].eaddr = L2_START + L2_LENGTH;
+		icplb_bounds[i_i++].data = L2_IMEMORY;
+	}
+	icplb_nr_bounds = i_i;
+	BUG_ON(icplb_nr_bounds > ARRAY_SIZE(icplb_bounds));
 }
-
-#endif
diff --git a/arch/blackfin/kernel/cplb-nompu/cplbmgr.S b/arch/blackfin/kernel/cplb-nompu/cplbmgr.S
deleted file mode 100644
index f4ca76c7239..00000000000
--- a/arch/blackfin/kernel/cplb-nompu/cplbmgr.S
+++ /dev/null
@@ -1,648 +0,0 @@
-/*
- * File:         arch/blackfin/mach-common/cplbmgtr.S
- * Based on:
- * Author:       LG Soft India
- *
- * Created:      ?
- * Description:  CPLB replacement routine for CPLB mismatch
- *
- * Modified:
- *               Copyright 2004-2006 Analog Devices Inc.
- *
- * Bugs:         Enter bugs at http://blackfin.uclinux.org/
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, see the file COPYING, or write
- * to the Free Software Foundation, Inc.,
- * 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-
-/* Usage: int _cplb_mgr(is_data_miss,int enable_cache)
- * is_data_miss==2 => Mark as Dirty, write to the clean data page
- * is_data_miss==1 => Replace a data CPLB.
- * is_data_miss==0 => Replace an instruction CPLB.
- *
- * Returns:
- * CPLB_RELOADED	=> Successfully updated CPLB table.
- * CPLB_NO_UNLOCKED	=> All CPLBs are locked, so cannot be evicted.
- *			   This indicates that the CPLBs in the configuration
- *			   tablei are badly configured, as this should never
- *			   occur.
- * CPLB_NO_ADDR_MATCH	=> The address being accessed, that triggered the
- *			   exception, is not covered by any of the CPLBs in
- *			   the configuration table. The application is
- *			   presumably misbehaving.
- * CPLB_PROT_VIOL	=> The address being accessed, that triggered the
- *			   exception, was not a first-write to a clean Write
- *			   Back Data page, and so presumably is a genuine
- *			   violation of the page's protection attributes.
- *			   The application is misbehaving.
- */
-
-#include <linux/linkage.h>
-#include <asm/blackfin.h>
-#include <asm/cplb.h>
-#include <asm/asm-offsets.h>
-
-#ifdef CONFIG_EXCPT_IRQ_SYSC_L1
-.section .l1.text
-#else
-.text
-#endif
-
-.align 2;
-ENTRY(_cplb_mgr)
-
-	[--SP]=( R7:4,P5:3 );
-
-	CC = R0 == 2;
-	IF CC JUMP .Ldcplb_write;
-
-	CC = R0 == 0;
-	IF !CC JUMP .Ldcplb_miss_compare;
-
-	/* ICPLB Miss Exception. We need to choose one of the
-	* currently-installed CPLBs, and replace it with one
-	* from the configuration table.
-	*/
-
-	/* A multi-word instruction can cross a page boundary. This means the
-	 * first part of the instruction can be in a valid page, but the
-	 * second part is not, and hence generates the instruction miss.
-	 * However, the fault address is for the start of the instruction,
-	 * not the part that's in the bad page. Therefore, we have to check
-	 * whether the fault address applies to a page that is already present
-	 * in the table.
-	 */
-
-	P4.L = LO(ICPLB_FAULT_ADDR);
-	P4.H = HI(ICPLB_FAULT_ADDR);
-
-	P1 = 16;
-	P5.L = _page_size_table;
-	P5.H = _page_size_table;
-
-	P0.L = LO(ICPLB_DATA0);
-	P0.H = HI(ICPLB_DATA0);
-	R4 = [P4];		/* Get faulting address*/
-	R6 = 64;		/* Advance past the fault address, which*/
-	R6 = R6 + R4;		/* we'll use if we find a match*/
-	R3 = ((16 << 8) | 2);	/* Extract mask, two bits at posn 16 */
-
-	R5 = 0;
-.Lisearch:
-
-	R1 = [P0-0x100];	/* Address for this CPLB */
-
-	R0 = [P0++];		/* Info for this CPLB*/
-	CC = BITTST(R0,0);	/* Is the CPLB valid?*/
-	IF !CC JUMP .Lnomatch;	/* Skip it, if not.*/
-	CC = R4 < R1(IU);	/* If fault address less than page start*/
-	IF CC JUMP .Lnomatch;	/* then skip this one.*/
-	R2 = EXTRACT(R0,R3.L) (Z);	/* Get page size*/
-	P1 = R2;
-	P1 = P5 + (P1<<2);	/* index into page-size table*/
-	R2 = [P1];		/* Get the page size*/
-	R1 = R1 + R2;		/* and add to page start, to get page end*/
-	CC = R4 < R1(IU);	/* and see whether fault addr is in page.*/
-	IF !CC R4 = R6;		/* If so, advance the address and finish loop.*/
-	IF !CC JUMP .Lisearch_done;
-.Lnomatch:
-	/* Go around again*/
-	R5 += 1;
-	CC = BITTST(R5, 4);	/* i.e CC = R5 >= 16*/
-	IF !CC JUMP .Lisearch;
-
-.Lisearch_done:
-	I0 = R4;		/* Fault address we'll search for*/
-
-	/* set up pointers */
-	P0.L = LO(ICPLB_DATA0);
-	P0.H = HI(ICPLB_DATA0);
-
-	/* The replacement procedure for ICPLBs */
-
-	P4.L = LO(IMEM_CONTROL);
-	P4.H = HI(IMEM_CONTROL);
-
-	/* Turn off CPLBs while we work, necessary according to HRM before
-	 * modifying CPLB descriptors
-	 */
-	R5 = [P4];		/* Control Register*/
-	BITCLR(R5,ENICPLB_P);
-	CLI R1;
-	SSYNC;		/* SSYNC required before writing to IMEM_CONTROL. */
-	.align 8;
-	[P4] = R5;
-	SSYNC;
-	STI R1;
-
-	R1 = -1;		/* end point comparison */
-	R3 = 16;		/* counter */
-
-	/* Search through CPLBs for first non-locked entry */
-	/* Overwrite it by moving everyone else up by 1 */
-.Licheck_lock:
-	R0 = [P0++];
-	R3 = R3 + R1;
-	CC = R3 == R1;
-	IF CC JUMP .Lall_locked;
-	CC = BITTST(R0, 0);		/* an invalid entry is good */
-	IF !CC JUMP .Lifound_victim;
-	CC = BITTST(R0,1);		/* but a locked entry isn't */
-	IF CC JUMP .Licheck_lock;
-
-.Lifound_victim:
-#ifdef CONFIG_CPLB_INFO
-	R7 = [P0 - 0x104];
-	GET_PDA(P2, R2);
-	P3 = [P2 + PDA_IPDT_SWAPCOUNT];
-	P2 = [P2 + PDA_IPDT];
-	P3 += -4;
-.Licount:
-	R2 = [P2];	/* address from config table */
-	P2 += 8;
-	P3 += 8;
-	CC = R2==-1;
-	IF CC JUMP .Licount_done;
-	CC = R7==R2;
-	IF !CC JUMP .Licount;
-	R7 = [P3];
-	R7 += 1;
-	[P3] = R7;
-	CSYNC;
-.Licount_done:
-#endif
-	LC0=R3;
-	LSETUP(.Lis_move,.Lie_move) LC0;
-.Lis_move:
-	R0 = [P0];
-	[P0 - 4] = R0;
-	R0 = [P0 - 0x100];
-	[P0-0x104] = R0;
-.Lie_move:
-	P0+=4;
-
-	/* Clear ICPLB_DATA15, in case we don't find a replacement
-	 * otherwise, we would have a duplicate entry, and will crash
-	 */
-	R0 = 0;
-	[P0 - 4] = R0;
-
-	/* We've made space in the ICPLB table, so that ICPLB15
-	 * is now free to be overwritten. Next, we have to determine
-	 * which CPLB we need to install, from the configuration
-	 * table. This is a matter of getting the start-of-page
-	 * addresses and page-lengths from the config table, and
-	 * determining whether the fault address falls within that
-	 * range.
- 	 */
-
-	GET_PDA(P3, R0);
-	P2 = [P3 + PDA_IPDT];
-#ifdef	CONFIG_CPLB_INFO
-	P3 = [P3 + PDA_IPDT_SWAPCOUNT];
-	P3 += -8;
-#endif
-	P0.L = _page_size_table;
-	P0.H = _page_size_table;
-
-	/* Retrieve our fault address (which may have been advanced
-	 * because the faulting instruction crossed a page boundary).
-	 */
-
-	R0 = I0;
-
-	/* An extraction pattern, to get the page-size bits from
-	 * the CPLB data entry. Bits 16-17, so two bits at posn 16.
-	 */
-
-	R1 = ((16<<8)|2);
-.Linext:	R4 = [P2++];	/* address from config table */
-	R2 = [P2++];	/* data from config table */
-#ifdef	CONFIG_CPLB_INFO
-	P3 += 8;
-#endif
-
-	CC = R4 == -1;	/* End of config table*/
-	IF CC JUMP .Lno_page_in_table;
-
-	/* See if failed address > start address */
-	CC = R4 <= R0(IU);
-	IF !CC JUMP .Linext;
-
-	/* extract page size (17:16)*/
-	R3 = EXTRACT(R2, R1.L) (Z);
-
-	/* add page size to addr to get range */
-
-	P5 = R3;
-	P5 = P0 + (P5 << 2);	/* scaled, for int access*/
-	R3 = [P5];
-	R3 = R3 + R4;
-
-	/* See if failed address < (start address + page size) */
-	CC = R0 < R3(IU);
-	IF !CC JUMP .Linext;
-
-	/* We've found a CPLB in the config table that covers
-	 * the faulting address, so install this CPLB into the
-	 * last entry of the table.
-	 */
-
-	P1.L = LO(ICPLB_DATA15);		/* ICPLB_DATA15 */
-	P1.H = HI(ICPLB_DATA15);
-	[P1] = R2;
-	[P1-0x100] = R4;
-#ifdef	CONFIG_CPLB_INFO
-	R3 = [P3];
-	R3 += 1;
-	[P3] = R3;
-#endif
-
-	/* P4 points to IMEM_CONTROL, and R5 contains its old
-	 * value, after we disabled ICPLBS. Re-enable them.
-	 */
-
-	BITSET(R5,ENICPLB_P);
-	CLI R2;
-	SSYNC;		/* SSYNC required before writing to IMEM_CONTROL. */
-	.align 8;
-	[P4] = R5;
-	SSYNC;
-	STI R2;
-
-	( R7:4,P5:3 ) = [SP++];
-	R0 = CPLB_RELOADED;
-	RTS;
-
-/* FAILED CASES*/
-.Lno_page_in_table:
-	R0 = CPLB_NO_ADDR_MATCH;
-	JUMP .Lfail_ret;
-
-.Lall_locked:
-	R0 = CPLB_NO_UNLOCKED;
-	JUMP .Lfail_ret;
-
-.Lprot_violation:
-	R0 = CPLB_PROT_VIOL;
-
-.Lfail_ret:
-	/* Make sure we turn protection/cache back on, even in the failing case */
-	BITSET(R5,ENICPLB_P);
-	CLI R2;
-	SSYNC;          /* SSYNC required before writing to IMEM_CONTROL. */
-	.align 8;
-	[P4] = R5;
-	SSYNC;
-	STI R2;
-
-	( R7:4,P5:3 ) = [SP++];
-	RTS;
-
-.Ldcplb_write:
-
-	/* if a DCPLB is marked as write-back (CPLB_WT==0), and
-	 * it is clean (CPLB_DIRTY==0), then a write to the
-	 * CPLB's page triggers a protection violation. We have to
-	 * mark the CPLB as dirty, to indicate that there are
-	 * pending writes associated with the CPLB.
-	 */
-
-	P4.L = LO(DCPLB_STATUS);
-	P4.H = HI(DCPLB_STATUS);
-	P3.L = LO(DCPLB_DATA0);
-	P3.H = HI(DCPLB_DATA0);
-	R5 = [P4];
-
-	/* A protection violation can be caused by more than just writes
-	 * to a clean WB page, so we have to ensure that:
-	 * - It's a write
-	 * - to a clean WB page
-	 * - and is allowed in the mode the access occurred.
-	 */
-
-	CC = BITTST(R5, 16);	/* ensure it was a write*/
-	IF !CC JUMP .Lprot_violation;
-
-	/* to check the rest, we have to retrieve the DCPLB.*/
-
-	/* The low half of DCPLB_STATUS is a bit mask*/
-
-	R2 = R5.L (Z);	/* indicating which CPLB triggered the event.*/
-	R3 = 30;	/* so we can use this to determine the offset*/
-	R2.L = SIGNBITS R2;
-	R2 = R2.L (Z);	/* into the DCPLB table.*/
-	R3 = R3 - R2;
-	P4 = R3;
-	P3 = P3 + (P4<<2);
-	R3 = [P3];	/* Retrieve the CPLB*/
-
-	/* Now we can check whether it's a clean WB page*/
-
-	CC = BITTST(R3, 14);	/* 0==WB, 1==WT*/
-	IF CC JUMP .Lprot_violation;
-	CC = BITTST(R3, 7);	/* 0 == clean, 1 == dirty*/
-	IF CC JUMP .Lprot_violation;
-
-	/* Check whether the write is allowed in the mode that was active.*/
-
-	R2 = 1<<3;		/* checking write in user mode*/
-	CC = BITTST(R5, 17);	/* 0==was user, 1==was super*/
-	R5 = CC;
-	R2 <<= R5;		/* if was super, check write in super mode*/
-	R2 = R3 & R2;
-	CC = R2 == 0;
-	IF CC JUMP .Lprot_violation;
-
-	/* It's a genuine write-to-clean-page.*/
-
-	BITSET(R3, 7);		/* mark as dirty*/
-	[P3] = R3;		/* and write back.*/
-	NOP;
-	CSYNC;
-	( R7:4,P5:3 ) = [SP++];
-	R0 = CPLB_RELOADED;
-	RTS;
-
-.Ldcplb_miss_compare:
-
-	/* Data CPLB Miss event. We need to choose a CPLB to
-	 * evict, and then locate a new CPLB to install from the
-	 * config table, that covers the faulting address.
-	 */
-
-	P1.L = LO(DCPLB_DATA15);
-	P1.H = HI(DCPLB_DATA15);
-
-	P4.L = LO(DCPLB_FAULT_ADDR);
-	P4.H = HI(DCPLB_FAULT_ADDR);
-	R4 = [P4];
-	I0 = R4;
-
-	/* The replacement procedure for DCPLBs*/
-
-	R6 = R1;	/* Save for later*/
-
-	/* Turn off CPLBs while we work.*/
-	P4.L = LO(DMEM_CONTROL);
-	P4.H = HI(DMEM_CONTROL);
-	R5 = [P4];
-	BITCLR(R5,ENDCPLB_P);
-	CLI R0;
-	SSYNC;		/* SSYNC required before writing to DMEM_CONTROL. */
-	.align 8;
-	[P4] = R5;
-	SSYNC;
-	STI R0;
-
-	/* Start looking for a CPLB to evict. Our order of preference
-	 * is: invalid CPLBs, clean CPLBs, dirty CPLBs. Locked CPLBs
-	 * are no good.
-	 */
-
-	I1.L = LO(DCPLB_DATA0);
-	I1.H = HI(DCPLB_DATA0);
-	P1 = 2;
-	P2 = 16;
-	I2.L = _dcplb_preference;
-	I2.H = _dcplb_preference;
-	LSETUP(.Lsdsearch1, .Ledsearch1) LC0 = P1;
-.Lsdsearch1:
-	R0 = [I2++];		/* Get the bits we're interested in*/
-	P0 = I1;		/* Go back to start of table*/
-	LSETUP (.Lsdsearch2, .Ledsearch2) LC1 = P2;
-.Lsdsearch2:
-	R1 = [P0++];		/* Fetch each installed CPLB in turn*/
-	R2 = R1 & R0;		/* and test for interesting bits.*/
-	CC = R2 == 0;		/* If none are set, it'll do.*/
-	IF !CC JUMP .Lskip_stack_check;
-
-	R2 = [P0 - 0x104]; 	/* R2 - PageStart */
-	P3.L = _page_size_table; /* retrieve end address */
-	P3.H = _page_size_table; /* retrieve end address */
-	R3 = 0x1002;		/* 16th - position, 2 bits -length */
-#if ANOMALY_05000209
-	nop;			/* Anomaly 05000209 */
-#endif
-	R7 = EXTRACT(R1,R3.l);
-	R7 = R7 << 2;		/* Page size index offset */
-	P5 = R7;
-	P3 = P3 + P5;
-	R7 = [P3];		/* page size in bytes */
-
-	R7 = R2 + R7;		/* R7 - PageEnd */
-	R4 = SP; 		/* Test SP is in range */
-
-	CC = R7 < R4;		/* if PageEnd < SP */
-	IF CC JUMP .Ldfound_victim;
-	R3 = 0x284;		/* stack length from start of trap till
-				 * the point.
-				 * 20 stack locations for future modifications
-				 */
-	R4 = R4 + R3;
-	CC = R4 < R2;		/* if SP + stacklen < PageStart */
-	IF CC JUMP .Ldfound_victim;
-.Lskip_stack_check:
-
-.Ledsearch2: NOP;
-.Ledsearch1: NOP;
-
-	/* If we got here, we didn't find a DCPLB we considered
-	 * replacable, which means all of them were locked.
-	 */
-
-	JUMP .Lall_locked;
-.Ldfound_victim:
-
-#ifdef CONFIG_CPLB_INFO
-	R7 = [P0 - 0x104];
-	GET_PDA(P2, R2);
-	P3 = [P2 + PDA_DPDT_SWAPCOUNT];
-	P2 = [P2 + PDA_DPDT];
-	P3 += -4;
-.Ldicount:
-	R2 = [P2];
-	P2 += 8;
-	P3 += 8;
-	CC = R2==-1;
-	IF CC JUMP .Ldicount_done;
-	CC = R7==R2;
-	IF !CC JUMP .Ldicount;
-	R7 = [P3];
-	R7 += 1;
-	[P3] = R7;
-.Ldicount_done:
-#endif
-
-	/* Clean down the hardware loops*/
-	R2 = 0;
-	LC1 = R2;
-	LC0 = R2;
-
-	/* There's a suitable victim in [P0-4] (because we've
-	 * advanced already).
-	 */
-
-.LDdoverwrite:
-
-	/* [P0-4] is a suitable victim CPLB, so we want to
-	 * overwrite it by moving all the following CPLBs
-	 * one space closer to the start.
-	 */
-
-	R1.L = LO(DCPLB_DATA16);		/* DCPLB_DATA15 + 4 */
-	R1.H = HI(DCPLB_DATA16);
-	R0 = P0;
-
-	/* If the victim happens to be in DCPLB15,
-	 * we don't need to move anything.
-	 */
-
-	CC = R1 == R0;
-	IF CC JUMP .Lde_moved;
-	R1 = R1 - R0;
-	R1 >>= 2;
-	P1 = R1;
-	LSETUP(.Lds_move, .Lde_move) LC0=P1;
-.Lds_move:
-	R0 = [P0++];	/* move data */
-	[P0 - 8] = R0;
-	R0 = [P0-0x104]	/* move address */
-.Lde_move:
-	 [P0-0x108] = R0;
-
-.Lde_moved:
-	NOP;
-
-	/* Clear DCPLB_DATA15, in case we don't find a replacement
-	 * otherwise, we would have a duplicate entry, and will crash
-	 */
-	R0 = 0;
-	[P0 - 0x4] = R0;
-
-	/* We've now made space in DCPLB15 for the new CPLB to be
-	 * installed. The next stage is to locate a CPLB in the
-	 * config table that covers the faulting address.
-	 */
-
-	R0 = I0;		/* Our faulting address */
-
-	GET_PDA(P3, R1);
-	P2 = [P3 + PDA_DPDT];
-#ifdef	CONFIG_CPLB_INFO
-	P3 = [P3 + PDA_DPDT_SWAPCOUNT];
-	P3 += -8;
-#endif
-
-	P1.L = _page_size_table;
-	P1.H = _page_size_table;
-
-	/* An extraction pattern, to retrieve bits 17:16.*/
-
-	R1 = (16<<8)|2;
-.Ldnext:	R4 = [P2++];	/* address */
-	R2 = [P2++];	/* data */
-#ifdef	CONFIG_CPLB_INFO
-	P3 += 8;
-#endif
-
-	CC = R4 == -1;
-	IF CC JUMP .Lno_page_in_table;
-
-	/* See if failed address > start address */
-	CC = R4 <= R0(IU);
-	IF !CC JUMP .Ldnext;
-
-	/* extract page size (17:16)*/
-	R3 = EXTRACT(R2, R1.L) (Z);
-
-	/* add page size to addr to get range */
-
-	P5 = R3;
-	P5 = P1 + (P5 << 2);
-	R3 = [P5];
-	R3 = R3 + R4;
-
-	/* See if failed address < (start address + page size) */
-	CC = R0 < R3(IU);
-	IF !CC JUMP .Ldnext;
-
-	/* We've found the CPLB that should be installed, so
-	 * write it into CPLB15, masking off any caching bits
-	 * if necessary.
-	 */
-
-	P1.L = LO(DCPLB_DATA15);
-	P1.H = HI(DCPLB_DATA15);
-
-	/* If the DCPLB has cache bits set, but caching hasn't
-	 * been enabled, then we want to mask off the cache-in-L1
-	 * bit before installing. Moreover, if caching is off, we
-	 * also want to ensure that the DCPLB has WT mode set, rather
-	 * than WB, since WB pages still trigger first-write exceptions
-	 * even when not caching is off, and the page isn't marked as
-	 * cachable. Finally, we could mark the page as clean, not dirty,
-	 * but we choose to leave that decision to the user; if the user
-	 * chooses to have a CPLB pre-defined as dirty, then they always
-	 * pay the cost of flushing during eviction, but don't pay the
-	 * cost of first-write exceptions to mark the page as dirty.
-	 */
-
-#ifdef CONFIG_BFIN_WT
-	BITSET(R6, 14);		/* Set WT*/
-#endif
-
-	[P1] = R2;
-	[P1-0x100] = R4;
-#ifdef	CONFIG_CPLB_INFO
-	R3 = [P3];
-	R3 += 1;
-	[P3] = R3;
-#endif
-
-	/* We've installed the CPLB, so re-enable CPLBs. P4
-	 * points to DMEM_CONTROL, and R5 is the value we
-	 * last wrote to it, when we were disabling CPLBs.
-	 */
-
-	BITSET(R5,ENDCPLB_P);
-	CLI R2;
-	.align 8;
-	[P4] = R5;
-	SSYNC;
-	STI R2;
-
-	( R7:4,P5:3 ) = [SP++];
-	R0 = CPLB_RELOADED;
-	RTS;
-ENDPROC(_cplb_mgr)
-
-#ifdef CONFIG_CPLB_SWITCH_TAB_L1
-.section .l1.data
-#else
-.data
-#endif
-
-ENTRY(_page_size_table)
-.byte4	0x00000400;	/* 1K */
-.byte4	0x00001000;	/* 4K */
-.byte4	0x00100000;	/* 1M */
-.byte4	0x00400000;	/* 4M */
-END(_page_size_table)
-
-ENTRY(_dcplb_preference)
-.byte4	0x00000001;	/* valid bit */
-.byte4	0x00000002;	/* lock bit */
-END(_dcplb_preference)
diff --git a/arch/blackfin/kernel/cplb-nompu/cplbmgr.c b/arch/blackfin/kernel/cplb-nompu/cplbmgr.c
new file mode 100644
index 00000000000..376249ab269
--- /dev/null
+++ b/arch/blackfin/kernel/cplb-nompu/cplbmgr.c
@@ -0,0 +1,283 @@
+/*
+ * File:         arch/blackfin/kernel/cplb-nompu-c/cplbmgr.c
+ * Based on:     arch/blackfin/kernel/cplb-mpu/cplbmgr.c
+ * Author:       Michael McTernan <mmcternan@airvana.com>
+ *
+ * Created:      01Nov2008
+ * Description:  CPLB miss handler.
+ *
+ * Modified:
+ *               Copyright 2008 Airvana Inc.
+ *               Copyright 2004-2007 Analog Devices Inc.
+ *
+ * Bugs:         Enter bugs at http://blackfin.uclinux.org/
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/kernel.h>
+#include <asm/blackfin.h>
+#include <asm/cplbinit.h>
+#include <asm/cplb.h>
+#include <asm/mmu_context.h>
+
+/*
+ * WARNING
+ *
+ * This file is compiled with certain -ffixed-reg options.  We have to
+ * make sure not to call any functions here that could clobber these
+ * registers.
+ */
+
+int nr_dcplb_miss[NR_CPUS], nr_icplb_miss[NR_CPUS];
+int nr_dcplb_supv_miss[NR_CPUS], nr_icplb_supv_miss[NR_CPUS];
+int nr_cplb_flush[NR_CPUS], nr_dcplb_prot[NR_CPUS];
+
+#ifdef CONFIG_EXCPT_IRQ_SYSC_L1
+#define MGR_ATTR __attribute__((l1_text))
+#else
+#define MGR_ATTR
+#endif
+
+/*
+ * We're in an exception handler.  The normal cli nop nop workaround
+ * isn't going to do very much, as the only thing that can interrupt
+ * us is an NMI, and the cli isn't going to stop that.
+ */
+#define NOWA_SSYNC __asm__ __volatile__ ("ssync;")
+
+/* Anomaly handlers provide SSYNCs, so avoid extra if anomaly is present */
+#if ANOMALY_05000125
+
+#define bfin_write_DMEM_CONTROL_SSYNC(v)    bfin_write_DMEM_CONTROL(v)
+#define bfin_write_IMEM_CONTROL_SSYNC(v)    bfin_write_IMEM_CONTROL(v)
+
+#else
+
+#define bfin_write_DMEM_CONTROL_SSYNC(v) \
+    do { NOWA_SSYNC; bfin_write_DMEM_CONTROL(v); NOWA_SSYNC; } while (0)
+#define bfin_write_IMEM_CONTROL_SSYNC(v) \
+    do { NOWA_SSYNC; bfin_write_IMEM_CONTROL(v); NOWA_SSYNC; } while (0)
+
+#endif
+
+static inline void write_dcplb_data(int cpu, int idx, unsigned long data,
+				    unsigned long addr)
+{
+	unsigned long ctrl = bfin_read_DMEM_CONTROL();
+	bfin_write_DMEM_CONTROL_SSYNC(ctrl & ~ENDCPLB);
+	bfin_write32(DCPLB_DATA0 + idx * 4, data);
+	bfin_write32(DCPLB_ADDR0 + idx * 4, addr);
+	bfin_write_DMEM_CONTROL_SSYNC(ctrl);
+
+#ifdef CONFIG_CPLB_INFO
+	dcplb_tbl[cpu][idx].addr = addr;
+	dcplb_tbl[cpu][idx].data = data;
+#endif
+}
+
+static inline void write_icplb_data(int cpu, int idx, unsigned long data,
+				    unsigned long addr)
+{
+	unsigned long ctrl = bfin_read_IMEM_CONTROL();
+
+	bfin_write_IMEM_CONTROL_SSYNC(ctrl & ~ENICPLB);
+	bfin_write32(ICPLB_DATA0 + idx * 4, data);
+	bfin_write32(ICPLB_ADDR0 + idx * 4, addr);
+	bfin_write_IMEM_CONTROL_SSYNC(ctrl);
+
+#ifdef CONFIG_CPLB_INFO
+	icplb_tbl[cpu][idx].addr = addr;
+	icplb_tbl[cpu][idx].data = data;
+#endif
+}
+
+/*
+ * Given the contents of the status register, return the index of the
+ * CPLB that caused the fault.
+ */
+static inline int faulting_cplb_index(int status)
+{
+	int signbits = __builtin_bfin_norm_fr1x32(status & 0xFFFF);
+	return 30 - signbits;
+}
+
+/*
+ * Given the contents of the status register and the DCPLB_DATA contents,
+ * return true if a write access should be permitted.
+ */
+static inline int write_permitted(int status, unsigned long data)
+{
+	if (status & FAULT_USERSUPV)
+		return !!(data & CPLB_SUPV_WR);
+	else
+		return !!(data & CPLB_USER_WR);
+}
+
+/* Counters to implement round-robin replacement.  */
+static int icplb_rr_index[NR_CPUS] PDT_ATTR;
+static int dcplb_rr_index[NR_CPUS] PDT_ATTR;
+
+/*
+ * Find an ICPLB entry to be evicted and return its index.
+ */
+static int evict_one_icplb(int cpu)
+{
+	int i = first_switched_icplb + icplb_rr_index[cpu];
+	if (i >= MAX_CPLBS) {
+		i -= MAX_CPLBS - first_switched_icplb;
+		icplb_rr_index[cpu] -= MAX_CPLBS - first_switched_icplb;
+	}
+	icplb_rr_index[cpu]++;
+	return i;
+}
+
+static int evict_one_dcplb(int cpu)
+{
+	int i = first_switched_dcplb + dcplb_rr_index[cpu];
+	if (i >= MAX_CPLBS) {
+		i -= MAX_CPLBS - first_switched_dcplb;
+		dcplb_rr_index[cpu] -= MAX_CPLBS - first_switched_dcplb;
+	}
+	dcplb_rr_index[cpu]++;
+	return i;
+}
+
+MGR_ATTR static int icplb_miss(int cpu)
+{
+	unsigned long addr = bfin_read_ICPLB_FAULT_ADDR();
+	int status = bfin_read_ICPLB_STATUS();
+	int idx;
+	unsigned long i_data, base, addr1, eaddr;
+
+	nr_icplb_miss[cpu]++;
+	if (unlikely(status & FAULT_USERSUPV))
+		nr_icplb_supv_miss[cpu]++;
+
+	base = 0;
+	for (idx = 0; idx < icplb_nr_bounds; idx++) {
+		eaddr = icplb_bounds[idx].eaddr;
+		if (addr < eaddr)
+			break;
+		base = eaddr;
+	}
+	if (unlikely(idx == icplb_nr_bounds))
+		return CPLB_NO_ADDR_MATCH;
+
+	i_data = icplb_bounds[idx].data;
+	if (unlikely(i_data == 0))
+		return CPLB_NO_ADDR_MATCH;
+
+	addr1 = addr & ~(SIZE_4M - 1);
+	addr &= ~(SIZE_1M - 1);
+	i_data |= PAGE_SIZE_1MB;
+	if (addr1 >= base && (addr1 + SIZE_4M) <= eaddr) {
+		/*
+		 * This works because
+		 * (PAGE_SIZE_4MB & PAGE_SIZE_1MB) == PAGE_SIZE_1MB.
+		 */
+		i_data |= PAGE_SIZE_4MB;
+		addr = addr1;
+	}
+
+	/* Pick entry to evict */
+	idx = evict_one_icplb(cpu);
+
+	write_icplb_data(cpu, idx, i_data, addr);
+
+	return CPLB_RELOADED;
+}
+
+MGR_ATTR static int dcplb_miss(int cpu)
+{
+	unsigned long addr = bfin_read_DCPLB_FAULT_ADDR();
+	int status = bfin_read_DCPLB_STATUS();
+	int idx;
+	unsigned long d_data, base, addr1, eaddr;
+
+	nr_dcplb_miss[cpu]++;
+	if (unlikely(status & FAULT_USERSUPV))
+		nr_dcplb_supv_miss[cpu]++;
+
+	base = 0;
+	for (idx = 0; idx < dcplb_nr_bounds; idx++) {
+		eaddr = dcplb_bounds[idx].eaddr;
+		if (addr < eaddr)
+			break;
+		base = eaddr;
+	}
+	if (unlikely(idx == dcplb_nr_bounds))
+		return CPLB_NO_ADDR_MATCH;
+
+	d_data = dcplb_bounds[idx].data;
+	if (unlikely(d_data == 0))
+		return CPLB_NO_ADDR_MATCH;
+
+	addr1 = addr & ~(SIZE_4M - 1);
+	addr &= ~(SIZE_1M - 1);
+	d_data |= PAGE_SIZE_1MB;
+	if (addr1 >= base && (addr1 + SIZE_4M) <= eaddr) {
+		/*
+		 * This works because
+		 * (PAGE_SIZE_4MB & PAGE_SIZE_1MB) == PAGE_SIZE_1MB.
+		 */
+		d_data |= PAGE_SIZE_4MB;
+		addr = addr1;
+	}
+
+	/* Pick entry to evict */
+	idx = evict_one_dcplb(cpu);
+
+	write_dcplb_data(cpu, idx, d_data, addr);
+
+	return CPLB_RELOADED;
+}
+
+MGR_ATTR static noinline int dcplb_protection_fault(int cpu)
+{
+	int status = bfin_read_DCPLB_STATUS();
+
+	nr_dcplb_prot[cpu]++;
+
+	if (likely(status & FAULT_RW)) {
+		int idx = faulting_cplb_index(status);
+		unsigned long regaddr = DCPLB_DATA0 + idx * 4;
+		unsigned long data = bfin_read32(regaddr);
+
+		/* Check if fault is to dirty a clean page */
+		if (!(data & CPLB_WT) && !(data & CPLB_DIRTY) &&
+		    write_permitted(status, data)) {
+
+			dcplb_tbl[cpu][idx].data = data;
+			bfin_write32(regaddr, data);
+			return CPLB_RELOADED;
+		}
+	}
+
+	return CPLB_PROT_VIOL;
+}
+
+MGR_ATTR int cplb_hdr(int seqstat, struct pt_regs *regs)
+{
+	int cause = seqstat & 0x3f;
+	unsigned int cpu = smp_processor_id();
+	switch (cause) {
+	case 0x2C:
+		return icplb_miss(cpu);
+	case 0x26:
+		return dcplb_miss(cpu);
+	default:
+		if (unlikely(cause == 0x23))
+			return dcplb_protection_fault(cpu);
+
+		return CPLB_UNKNOWN_ERR;
+	}
+}
diff --git a/arch/blackfin/kernel/cplbinfo.c b/arch/blackfin/kernel/cplbinfo.c
index 1d3bbec1a19..64d78300dd0 100644
--- a/arch/blackfin/kernel/cplbinfo.c
+++ b/arch/blackfin/kernel/cplbinfo.c
@@ -20,8 +20,6 @@ static char const page_strtbl[][3] = { "1K", "4K", "1M", "4M" };
 #define page(flags)    (((flags) & 0x30000) >> 16)
 #define strpage(flags) page_strtbl[page(flags)]
 
-#ifdef CONFIG_MPU
-
 struct cplbinfo_data {
 	loff_t pos;
 	char cplb_type;
@@ -75,88 +73,6 @@ static void cplbinfo_seq_init(struct cplbinfo_data *cdata, unsigned int cpu)
 	}
 }
 
-#else
-
-struct cplbinfo_data {
-	loff_t pos;
-	char cplb_type;
-	u32 mem_control;
-	unsigned long *pdt_tables, *pdt_swapcount;
-	unsigned long cplb_addr, cplb_data;
-};
-
-extern int page_size_table[];
-
-static int cplb_find_entry(unsigned long addr_tbl, unsigned long data_tbl,
-                           unsigned long addr_find, unsigned long data_find)
-{
-	int i;
-
-	for (i = 0; i < 16; ++i) {
-		unsigned long cplb_addr = bfin_read32(addr_tbl + i * 4);
-		unsigned long cplb_data = bfin_read32(data_tbl + i * 4);
-		if (addr_find >= cplb_addr &&
-		    addr_find < cplb_addr + page_size_table[page(cplb_data)] &&
-		    cplb_data == data_find)
-			return i;
-	}
-
-	return -1;
-}
-
-static void cplbinfo_print_header(struct seq_file *m)
-{
-	seq_printf(m, "Address\t\tData\tSize\tValid\tLocked\tSwapin\tiCount\toCount\n");
-}
-
-static int cplbinfo_nomore(struct cplbinfo_data *cdata)
-{
-	return cdata->pdt_tables[cdata->pos * 2] == 0xffffffff;
-}
-
-static int cplbinfo_show(struct seq_file *m, void *p)
-{
-	struct cplbinfo_data *cdata;
-	unsigned long data, addr;
-	int entry;
-	loff_t pos;
-
-	cdata = p;
-	pos = cdata->pos * 2;
-	addr = cdata->pdt_tables[pos];
-	data = cdata->pdt_tables[pos + 1];
-	entry = cplb_find_entry(cdata->cplb_addr, cdata->cplb_data, addr, data);
-
-	seq_printf(m,
-		"0x%08lx\t0x%05lx\t%s\t%c\t%c\t%2d\t%ld\t%ld\n",
-		addr, data, strpage(data),
-		(data & CPLB_VALID) ? 'Y' : 'N',
-		(data & CPLB_LOCK) ? 'Y' : 'N', entry,
-		cdata->pdt_swapcount[pos],
-		cdata->pdt_swapcount[pos + 1]);
-
-	return 0;
-}
-
-static void cplbinfo_seq_init(struct cplbinfo_data *cdata, unsigned int cpu)
-{
-	if (cdata->cplb_type == 'I') {
-		cdata->mem_control = bfin_read_IMEM_CONTROL();
-		cdata->pdt_tables = ipdt_tables[cpu];
-		cdata->pdt_swapcount = ipdt_swapcount_tables[cpu];
-		cdata->cplb_addr = ICPLB_ADDR0;
-		cdata->cplb_data = ICPLB_DATA0;
-	} else {
-		cdata->mem_control = bfin_read_DMEM_CONTROL();
-		cdata->pdt_tables = dpdt_tables[cpu];
-		cdata->pdt_swapcount = dpdt_swapcount_tables[cpu];
-		cdata->cplb_addr = DCPLB_ADDR0;
-		cdata->cplb_data = DCPLB_DATA0;
-	}
-}
-
-#endif
-
 static void *cplbinfo_start(struct seq_file *m, loff_t *pos)
 {
 	struct cplbinfo_data *cdata = m->private;
diff --git a/arch/blackfin/kernel/setup.c b/arch/blackfin/kernel/setup.c
index d3d37e7f465..20d04a1bc86 100644
--- a/arch/blackfin/kernel/setup.c
+++ b/arch/blackfin/kernel/setup.c
@@ -88,6 +88,7 @@ void __init generate_cplb_tables(void)
 {
 	unsigned int cpu;
 
+	generate_cplb_tables_all();
 	/* Generate per-CPU I&D CPLB tables */
 	for (cpu = 0; cpu < num_possible_cpus(); ++cpu)
 		generate_cplb_tables_cpu(cpu);
@@ -97,19 +98,11 @@ void __init generate_cplb_tables(void)
 void __cpuinit bfin_setup_caches(unsigned int cpu)
 {
 #ifdef CONFIG_BFIN_ICACHE
-#ifdef CONFIG_MPU
 	bfin_icache_init(icplb_tbl[cpu]);
-#else
-	bfin_icache_init(icplb_tables[cpu]);
-#endif
 #endif
 
 #ifdef CONFIG_BFIN_DCACHE
-#ifdef CONFIG_MPU
 	bfin_dcache_init(dcplb_tbl[cpu]);
-#else
-	bfin_dcache_init(dcplb_tables[cpu]);
-#endif
 #endif
 
 	/*
diff --git a/arch/blackfin/mach-common/entry.S b/arch/blackfin/mach-common/entry.S
index 0efd5f5b7ba..fae77465137 100644
--- a/arch/blackfin/mach-common/entry.S
+++ b/arch/blackfin/mach-common/entry.S
@@ -112,24 +112,21 @@ ENTRY(_ex_dcplb_viol)
 ENTRY(_ex_dcplb_miss)
 ENTRY(_ex_icplb_miss)
 	(R7:6,P5:4) = [sp++];
-	ASTAT = [sp++];
-	SAVE_ALL_SYS
-#ifdef CONFIG_MPU
+	/* We leave the previously pushed ASTAT on the stack.  */
+	SAVE_CONTEXT_CPLB
+
 	/* We must load R1 here, _before_ DEBUG_HWTRACE_SAVE, since that
 	 * will change the stack pointer.  */
 	R0 = SEQSTAT;
 	R1 = SP;
-#endif
+
 	DEBUG_HWTRACE_SAVE(p5, r7)
-#ifdef CONFIG_MPU
+
 	sp += -12;
 	call _cplb_hdr;
 	sp += 12;
 	CC = R0 == 0;
 	IF !CC JUMP _handle_bad_cplb;
-#else
-	call __cplb_hdr;
-#endif
 
 #ifdef CONFIG_DEBUG_DOUBLEFAULT
 	/* While we were processing this, did we double fault? */
@@ -143,7 +140,8 @@ ENTRY(_ex_icplb_miss)
 #endif
 
 	DEBUG_HWTRACE_RESTORE(p5, r7)
-	RESTORE_ALL_SYS
+	RESTORE_CONTEXT_CPLB
+	ASTAT = [SP++];
 	SP = EX_SCRATCH_REG;
 	rtx;
 ENDPROC(_ex_icplb_miss)
@@ -298,9 +296,8 @@ ENTRY(_handle_bad_cplb)
 	 * the stack to get ready so, we can fall through - we
 	 * need to make a CPLB exception look like a normal exception
 	 */
-
-	RESTORE_ALL_SYS
-	[--sp] = ASTAT;
+	RESTORE_CONTEXT_CPLB
+	/* ASTAT is still on the stack, where it is needed.  */
 	[--sp] = (R7:6,P5:4);
 
 ENTRY(_ex_replaceable)
diff --git a/arch/blackfin/mm/init.c b/arch/blackfin/mm/init.c
index 57d306b9c56..d0532b72bba 100644
--- a/arch/blackfin/mm/init.c
+++ b/arch/blackfin/mm/init.c
@@ -119,16 +119,6 @@ asmlinkage void init_pda(void)
 
 	cpu_pda[cpu].ex_stack = exception_stack[cpu + 1];
 
-#ifdef CONFIG_MPU
-#else
-	cpu_pda[cpu].ipdt = ipdt_tables[cpu];
-	cpu_pda[cpu].dpdt = dpdt_tables[cpu];
-#ifdef CONFIG_CPLB_INFO
-	cpu_pda[cpu].ipdt_swapcount = ipdt_swapcount_tables[cpu];
-	cpu_pda[cpu].dpdt_swapcount = dpdt_swapcount_tables[cpu];
-#endif
-#endif
-
 #ifdef CONFIG_SMP
 	cpu_pda[cpu].imask = 0x1f;
 #endif