- Update to 2.6.25-rc3.

[linux-flexiantxendom0-3.2.10.git] / arch / mips / lib / memset.S

diff --git a/arch/mips/lib/memset.S b/arch/mips/lib/memset.S
index 3f8b8b3..77dc3b2 100644 (file)
--- a/arch/mips/lib/memset.S
+++ b/arch/mips/lib/memset.S
@@ -5,6 +5,7 @@
  *
  * Copyright (C) 1998, 1999, 2000 by Ralf Baechle
  * Copyright (C) 1999, 2000 Silicon Graphics, Inc.
+ * Copyright (C) 2007  Maciej W. Rozycki
  */
 #include <asm/asm.h>
 #include <asm/asm-offsets.h>
@@ -71,34 +72,45 @@ LEAF(memset)
 
 FEXPORT(__bzero)
        sltiu           t0, a2, LONGSIZE        /* very small region? */
-       bnez            t0, small_memset
+       bnez            t0, .Lsmall_memset
         andi           t0, a0, LONGMASK        /* aligned? */
 
+#ifndef CONFIG_CPU_DADDI_WORKAROUNDS
        beqz            t0, 1f
         PTR_SUBU       t0, LONGSIZE            /* alignment in bytes */
+#else
+       .set            noat
+       li              AT, LONGSIZE
+       beqz            t0, 1f
+        PTR_SUBU       t0, AT                  /* alignment in bytes */
+       .set            at
+#endif
 
+       R10KCBARRIER(0(ra))
 #ifdef __MIPSEB__
-       EX(LONG_S_L, a1, (a0), first_fixup)     /* make word/dword aligned */
+       EX(LONG_S_L, a1, (a0), .Lfirst_fixup)   /* make word/dword aligned */
 #endif
 #ifdef __MIPSEL__
-       EX(LONG_S_R, a1, (a0), first_fixup)     /* make word/dword aligned */
+       EX(LONG_S_R, a1, (a0), .Lfirst_fixup)   /* make word/dword aligned */
 #endif
        PTR_SUBU        a0, t0                  /* long align ptr */
        PTR_ADDU        a2, t0                  /* correct size */
 
 1:     ori             t1, a2, 0x3f            /* # of full blocks */
        xori            t1, 0x3f
-       beqz            t1, memset_partial      /* no block to fill */
+       beqz            t1, .Lmemset_partial    /* no block to fill */
         andi           t0, a2, 0x40-LONGSIZE
 
        PTR_ADDU        t1, a0                  /* end address */
        .set            reorder
 1:     PTR_ADDIU       a0, 64
-       f_fill64 a0, -64, a1, fwd_fixup
+       R10KCBARRIER(0(ra))
+       f_fill64 a0, -64, a1, .Lfwd_fixup
        bne             t1, a0, 1b
        .set            noreorder
 
-memset_partial:
+.Lmemset_partial:
+       R10KCBARRIER(0(ra))
        PTR_LA          t1, 2f                  /* where to start */
 #if LONGSIZE == 4
        PTR_SUBU        t1, t0
@@ -106,7 +118,7 @@ memset_partial:
        .set            noat
        LONG_SRL                AT, t0, 1
        PTR_SUBU        t1, AT
-       .set            noat
+       .set            at
 #endif
        jr              t1
         PTR_ADDU       a0, t0                  /* dest ptr */
@@ -114,26 +126,28 @@ memset_partial:
        .set            push
        .set            noreorder
        .set            nomacro
-       f_fill64 a0, -64, a1, partial_fixup     /* ... but first do longs ... */
+       f_fill64 a0, -64, a1, .Lpartial_fixup   /* ... but first do longs ... */
 2:     .set            pop
        andi            a2, LONGMASK            /* At most one long to go */
 
        beqz            a2, 1f
         PTR_ADDU       a0, a2                  /* What's left */
+       R10KCBARRIER(0(ra))
 #ifdef __MIPSEB__
-       EX(LONG_S_R, a1, -1(a0), last_fixup)
+       EX(LONG_S_R, a1, -1(a0), .Llast_fixup)
 #endif
 #ifdef __MIPSEL__
-       EX(LONG_S_L, a1, -1(a0), last_fixup)
+       EX(LONG_S_L, a1, -1(a0), .Llast_fixup)
 #endif
 1:     jr              ra
         move           a2, zero
 
-small_memset:
+.Lsmall_memset:
        beqz            a2, 2f
         PTR_ADDU       t1, a0, a2
 
 1:     PTR_ADDIU       a0, 1                   /* fill bytewise */
+       R10KCBARRIER(0(ra))
        bne             t1, a0, 1b
         sb             a1, -1(a0)
 
@@ -141,11 +155,11 @@ small_memset:
         move           a2, zero
        END(memset)
 
-first_fixup:
+.Lfirst_fixup:
        jr      ra
         nop
 
-fwd_fixup:
+.Lfwd_fixup:
        PTR_L           t0, TI_TASK($28)
        LONG_L          t0, THREAD_BUADDR(t0)
        andi            a2, 0x3f
@@ -153,7 +167,7 @@ fwd_fixup:
        jr              ra
         LONG_SUBU      a2, t0
 
-partial_fixup:
+.Lpartial_fixup:
        PTR_L           t0, TI_TASK($28)
        LONG_L          t0, THREAD_BUADDR(t0)
        andi            a2, LONGMASK
@@ -161,6 +175,6 @@ partial_fixup:
        jr              ra
         LONG_SUBU      a2, t0
 
-last_fixup:
+.Llast_fixup:
        jr              ra
         andi           v1, a2, LONGMASK