4 files changed, 113 insertions, 50 deletions

diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index 1b8229d9c9d5..33ed048502a3 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -372,7 +372,7 @@ config CPU_FEROCEON
         select CPU_PABRT_NOIFAR
         select CPU_CACHE_VIVT
         select CPU_CP15_MMU
-        select CPU_COPY_V4WB if MMU
+        select CPU_COPY_FEROCEON if MMU
         select CPU_TLB_V4WBI if MMU
 
 config CPU_FEROCEON_OLD_ID
@@ -523,6 +523,9 @@ config CPU_COPY_V4WT
 config CPU_COPY_V4WB
         bool
 
+config CPU_COPY_FEROCEON
+        bool
+
 config CPU_COPY_V6
         bool
 
@@ -658,7 +661,7 @@ config CPU_DCACHE_SIZE
 
 config CPU_DCACHE_WRITETHROUGH
         bool "Force write through D-cache"
-        depends on (CPU_ARM740T || CPU_ARM920T || CPU_ARM922T || CPU_ARM925T || CPU_ARM926T || CPU_ARM940T || CPU_ARM946E || CPU_ARM1020 || CPU_FEROCEON) && !CPU_DCACHE_DISABLE
+        depends on (CPU_ARM740T || CPU_ARM920T || CPU_ARM922T || CPU_ARM925T || CPU_ARM926T || CPU_ARM940T || CPU_ARM946E || CPU_ARM1020) && !CPU_DCACHE_DISABLE
         default y if CPU_ARM925T
         help
           Say Y here to use the data cache in writethrough mode. Unless you
diff --git a/arch/arm/mm/Makefile b/arch/arm/mm/Makefile
index 44536a0b995a..32b2d2d213a6 100644
--- a/arch/arm/mm/Makefile
+++ b/arch/arm/mm/Makefile
@@ -36,6 +36,7 @@ obj-$(CONFIG_CPU_CACHE_V7)	+= cache-v7.o
 obj-$(CONFIG_CPU_COPY_V3)       += copypage-v3.o
 obj-$(CONFIG_CPU_COPY_V4WT)     += copypage-v4wt.o
 obj-$(CONFIG_CPU_COPY_V4WB)     += copypage-v4wb.o
+obj-$(CONFIG_CPU_COPY_FEROCEON) += copypage-feroceon.o
 obj-$(CONFIG_CPU_COPY_V6)       += copypage-v6.o context.o
 obj-$(CONFIG_CPU_SA1100)        += copypage-v4mc.o
 obj-$(CONFIG_CPU_XSCALE)        += copypage-xscale.o
diff --git a/arch/arm/mm/copypage-feroceon.S b/arch/arm/mm/copypage-feroceon.S
new file mode 100644
index 000000000000..7eb0d320d240
--- /dev/null
+++ b/arch/arm/mm/copypage-feroceon.S
@@ -0,0 +1,95 @@
+/*
+ *  linux/arch/arm/lib/copypage-feroceon.S
+ *
+ *  Copyright (C) 2008 Marvell Semiconductors
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This handles copy_user_page and clear_user_page on Feroceon
+ * more optimally than the generic implementations.
+ */
+#include <linux/linkage.h>
+#include <linux/init.h>
+#include <asm/asm-offsets.h>
+
+        .text
+        .align  5
+
+ENTRY(feroceon_copy_user_page)
+        stmfd   sp!, {r4-r9, lr}
+        mov     ip, #PAGE_SZ
+1:      mov     lr, r1
+        ldmia   r1!, {r2 - r9}
+        pld     [lr, #32]
+        pld     [lr, #64]
+        pld     [lr, #96]
+        pld     [lr, #128]
+        pld     [lr, #160]
+        pld     [lr, #192]
+        pld     [lr, #224]
+        stmia   r0, {r2 - r9}
+        ldmia   r1!, {r2 - r9}
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        stmia   r0, {r2 - r9}
+        ldmia   r1!, {r2 - r9}
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        stmia   r0, {r2 - r9}
+        ldmia   r1!, {r2 - r9}
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        stmia   r0, {r2 - r9}
+        ldmia   r1!, {r2 - r9}
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        stmia   r0, {r2 - r9}
+        ldmia   r1!, {r2 - r9}
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        stmia   r0, {r2 - r9}
+        ldmia   r1!, {r2 - r9}
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        stmia   r0, {r2 - r9}
+        ldmia   r1!, {r2 - r9}
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        stmia   r0, {r2 - r9}
+        subs    ip, ip, #(32 * 8)
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        bne     1b
+        mcr     p15, 0, ip, c7, c10, 4          @ drain WB
+        ldmfd   sp!, {r4-r9, pc}
+
+        .align  5
+
+ENTRY(feroceon_clear_user_page)
+        stmfd   sp!, {r4-r7, lr}
+        mov     r1, #PAGE_SZ/32
+        mov     r2, #0
+        mov     r3, #0
+        mov     r4, #0
+        mov     r5, #0
+        mov     r6, #0
+        mov     r7, #0
+        mov     ip, #0
+        mov     lr, #0
+1:      stmia   r0, {r2-r7, ip, lr}
+        subs    r1, r1, #1
+        mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D line
+        add     r0, r0, #32
+        bne     1b
+        mcr     p15, 0, r1, c7, c10, 4          @ drain WB
+        ldmfd   sp!, {r4-r7, pc}
+
+        __INITDATA
+
+        .type   feroceon_user_fns, #object
+ENTRY(feroceon_user_fns)
+        .long   feroceon_clear_user_page
+        .long   feroceon_copy_user_page
+        .size   feroceon_user_fns, . - feroceon_user_fns
diff --git a/arch/arm/mm/proc-feroceon.S b/arch/arm/mm/proc-feroceon.S
index 90e7594e29b1..a02c1712b52d 100644
--- a/arch/arm/mm/proc-feroceon.S
+++ b/arch/arm/mm/proc-feroceon.S
@@ -93,7 +93,7 @@ ENTRY(cpu_feroceon_reset)
  *
  * Called with IRQs disabled
  */
-        .align  10
+        .align  5
 ENTRY(cpu_feroceon_do_idle)
         mov     r0, #0
         mcr     p15, 0, r0, c7, c10, 4          @ Drain write buffer
@@ -106,6 +106,7 @@ ENTRY(cpu_feroceon_do_idle)
  *      Clean and invalidate all cache entries in a particular
  *      address space.
  */
+        .align  5
 ENTRY(feroceon_flush_user_cache_all)
         /* FALLTHROUGH */
 
@@ -118,12 +119,8 @@ ENTRY(feroceon_flush_kern_cache_all)
         mov     r2, #VM_EXEC
         mov     ip, #0
 __flush_whole_cache:
-#ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
-        mcr     p15, 0, ip, c7, c6, 0           @ invalidate D cache
-#else
 1:      mrc     p15, 0, r15, c7, c14, 3         @ test,clean,invalidate
         bne     1b
-#endif
         tst     r2, #VM_EXEC
         mcrne   p15, 0, ip, c7, c5, 0           @ invalidate I cache
         mcrne   p15, 0, ip, c7, c10, 4          @ drain WB
@@ -139,27 +136,19 @@ __flush_whole_cache:
  *      - end   - end address (exclusive)
  *      - flags - vm_flags describing address space
  */
+        .align  5
 ENTRY(feroceon_flush_user_cache_range)
         mov     ip, #0
         sub     r3, r1, r0                      @ calculate total size
         cmp     r3, #CACHE_DLIMIT
         bgt     __flush_whole_cache
 1:      tst     r2, #VM_EXEC
-#ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
-        mcr     p15, 0, r0, c7, c6, 1           @ invalidate D entry
-        mcrne   p15, 0, r0, c7, c5, 1           @ invalidate I entry
-        add     r0, r0, #CACHE_DLINESIZE
-        mcr     p15, 0, r0, c7, c6, 1           @ invalidate D entry
-        mcrne   p15, 0, r0, c7, c5, 1           @ invalidate I entry
-        add     r0, r0, #CACHE_DLINESIZE
-#else
         mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D entry
         mcrne   p15, 0, r0, c7, c5, 1           @ invalidate I entry
         add     r0, r0, #CACHE_DLINESIZE
         mcr     p15, 0, r0, c7, c14, 1          @ clean and invalidate D entry
         mcrne   p15, 0, r0, c7, c5, 1           @ invalidate I entry
         add     r0, r0, #CACHE_DLINESIZE
-#endif
         cmp     r0, r1
         blo     1b
         tst     r2, #VM_EXEC
@@ -176,6 +165,7 @@ ENTRY(feroceon_flush_user_cache_range)
  *      - start - virtual start address
  *      - end   - virtual end address
  */
+        .align  5
 ENTRY(feroceon_coherent_kern_range)
         /* FALLTHROUGH */
 
@@ -207,6 +197,7 @@ ENTRY(feroceon_coherent_user_range)
  *
  *      - addr  - page aligned address
  */
+        .align  5
 ENTRY(feroceon_flush_kern_dcache_page)
         add     r1, r0, #PAGE_SZ
 1:      mcr     p15, 0, r0, c7, c14, 1          @ clean+invalidate D entry
@@ -231,13 +222,12 @@ ENTRY(feroceon_flush_kern_dcache_page)
  *
  * (same as v4wb)
  */
+        .align  5
 ENTRY(feroceon_dma_inv_range)
-#ifndef CONFIG_CPU_DCACHE_WRITETHROUGH
         tst     r0, #CACHE_DLINESIZE - 1
         mcrne   p15, 0, r0, c7, c10, 1          @ clean D entry
         tst     r1, #CACHE_DLINESIZE - 1
         mcrne   p15, 0, r1, c7, c10, 1          @ clean D entry
-#endif
         bic     r0, r0, #CACHE_DLINESIZE - 1
 1:      mcr     p15, 0, r0, c7, c6, 1           @ invalidate D entry
         add     r0, r0, #CACHE_DLINESIZE
@@ -256,14 +246,13 @@ ENTRY(feroceon_dma_inv_range)
  *
  * (same as v4wb)
  */
+        .align  5
 ENTRY(feroceon_dma_clean_range)
-#ifndef CONFIG_CPU_DCACHE_WRITETHROUGH
         bic     r0, r0, #CACHE_DLINESIZE - 1
 1:      mcr     p15, 0, r0, c7, c10, 1          @ clean D entry
         add     r0, r0, #CACHE_DLINESIZE
         cmp     r0, r1
         blo     1b
-#endif
         mcr     p15, 0, r0, c7, c10, 4          @ drain WB
         mov     pc, lr
 
@@ -275,14 +264,10 @@ ENTRY(feroceon_dma_clean_range)
  *      - start - virtual start address
  *      - end   - virtual end address
  */
+        .align  5
 ENTRY(feroceon_dma_flush_range)
         bic     r0, r0, #CACHE_DLINESIZE - 1
-1:
-#ifndef CONFIG_CPU_DCACHE_WRITETHROUGH
-        mcr     p15, 0, r0, c7, c14, 1          @ clean+invalidate D entry
-#else
-        mcr     p15, 0, r0, c7, c10, 1          @ clean D entry
-#endif
+1:      mcr     p15, 0, r0, c7, c14, 1          @ clean+invalidate D entry
         add     r0, r0, #CACHE_DLINESIZE
         cmp     r0, r1
         blo     1b
@@ -300,13 +285,12 @@ ENTRY(feroceon_cache_fns)
         .long   feroceon_dma_clean_range
         .long   feroceon_dma_flush_range
 
+        .align  5
 ENTRY(cpu_feroceon_dcache_clean_area)
-#ifndef CONFIG_CPU_DCACHE_WRITETHROUGH
 1:      mcr     p15, 0, r0, c7, c10, 1          @ clean D entry
         add     r0, r0, #CACHE_DLINESIZE
         subs    r1, r1, #CACHE_DLINESIZE
         bhi     1b
-#endif
         mcr     p15, 0, r0, c7, c10, 4          @ drain WB
         mov     pc, lr
 
@@ -323,13 +307,9 @@ ENTRY(cpu_feroceon_dcache_clean_area)
 ENTRY(cpu_feroceon_switch_mm)
 #ifdef CONFIG_MMU
         mov     ip, #0
-#ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
-        mcr     p15, 0, ip, c7, c6, 0           @ invalidate D cache
-#else
 @ && 'Clean & Invalidate whole DCache'
 1:      mrc     p15, 0, r15, c7, c14, 3         @ test,clean,invalidate
         bne     1b
-#endif
         mcr     p15, 0, ip, c7, c5, 0           @ invalidate I cache
         mcr     p15, 0, ip, c7, c10, 4          @ drain WB
         mcr     p15, 0, r0, c2, c0, 0           @ load page table pointer
@@ -362,16 +342,9 @@ ENTRY(cpu_feroceon_set_pte_ext)
         tst     r1, #L_PTE_PRESENT | L_PTE_YOUNG        @ Present and Young?
         movne   r2, #0
 
-#ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
-        eor     r3, r2, #0x0a                   @ C & small page?
-        tst     r3, #0x0b
-        biceq   r2, r2, #4
-#endif
         str     r2, [r0]                        @ hardware version
         mov     r0, r0
-#ifndef CONFIG_CPU_DCACHE_WRITETHROUGH
         mcr     p15, 0, r0, c7, c10, 1          @ clean D entry
-#endif
         mcr     p15, 0, r0, c7, c10, 4          @ drain WB
 #endif
         mov     pc, lr
@@ -387,20 +360,11 @@ __feroceon_setup:
         mcr     p15, 0, r0, c8, c7              @ invalidate I,D TLBs on v4
 #endif
 
-
-#ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
-        mov     r0, #4                          @ disable write-back on caches explicitly
-        mcr     p15, 7, r0, c15, c0, 0
-#endif
-
         adr     r5, feroceon_crval
         ldmia   r5, {r5, r6}
         mrc     p15, 0, r0, c1, c0              @ get control register v4
         bic     r0, r0, r5
         orr     r0, r0, r6
-#ifdef CONFIG_CPU_CACHE_ROUND_ROBIN
-        orr     r0, r0, #0x4000                 @ .1.. .... .... ....
-#endif
         mov     pc, lr
         .size   __feroceon_setup, . - __feroceon_setup
 
@@ -476,7 +440,7 @@ __feroceon_old_id_proc_info:
         .long   cpu_feroceon_name
         .long   feroceon_processor_functions
         .long   v4wbi_tlb_fns
-        .long   v4wb_user_fns
+        .long   feroceon_user_fns
         .long   feroceon_cache_fns
         .size   __feroceon_old_id_proc_info, . - __feroceon_old_id_proc_info
 #endif
@@ -502,6 +466,6 @@ __feroceon_proc_info:
         .long   cpu_feroceon_name
         .long   feroceon_processor_functions
         .long   v4wbi_tlb_fns
-        .long   v4wb_user_fns
+        .long   feroceon_user_fns
         .long   feroceon_cache_fns
         .size   __feroceon_proc_info, . - __feroceon_proc_info