[MIPS] Unify memset.S

The 32-bit version and 64-bit version are almost equal. Unify them. This makes further improvements (for example, supporting CDEX, etc.) easier. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
git-mirror · Feb 6, 2007 · a583158 · a583158
1 parent c44e8d5
commit a583158
Show file tree

Hide file tree

Showing 5 changed files with 31 additions and 152 deletions.
diff --git a/arch/mips/lib-32/Makefile b/arch/mips/lib-32/Makefile
@@ -2,7 +2,7 @@
 # Makefile for MIPS-specific library files..
 #
 
-lib-y	+= memset.o watch.o
+lib-y	+= watch.o
 
 obj-$(CONFIG_CPU_MIPS32)	+= dump_tlb.o
 obj-$(CONFIG_CPU_MIPS64)	+= dump_tlb.o

diff --git a/arch/mips/lib-64/Makefile b/arch/mips/lib-64/Makefile
@@ -2,7 +2,7 @@
 # Makefile for MIPS-specific library files..
 #
 
-lib-y	+= memset.o watch.o
+lib-y	+= watch.o
 
 obj-$(CONFIG_CPU_MIPS32)	+= dump_tlb.o
 obj-$(CONFIG_CPU_MIPS64)	+= dump_tlb.o

diff --git a/arch/mips/lib-64/memset.S b/arch/mips/lib-64/memset.S
diff --git a/arch/mips/lib/Makefile b/arch/mips/lib/Makefile
@@ -2,7 +2,7 @@
 # Makefile for MIPS-specific library files..
 #
 
-lib-y	+= csum_partial.o memcpy.o promlib.o \
+lib-y	+= csum_partial.o memcpy.o memset.o promlib.o \
 	   strlen_user.o strncpy_user.o strnlen_user.o uncached.o
 
 obj-y	+= iomap.o

diff --git a/arch/mips/lib-32/memset.S → arch/mips/lib/memset.S b/arch/mips/lib-32/memset.S → arch/mips/lib/memset.S
@@ -10,6 +10,14 @@
 #include <asm/asm-offsets.h>
 #include <asm/regdef.h>
 
+#if LONGSIZE == 4
+#define LONG_S_L swl
+#define LONG_S_R swr
+#else
+#define LONG_S_L sdl
+#define LONG_S_R sdr
+#endif
+
 #define EX(insn,reg,addr,handler)			\
 9:	insn	reg, addr;				\
 	.section __ex_table,"a"; 			\
@@ -25,6 +33,7 @@
 	EX(LONG_S, \val, (\offset +  5 * LONGSIZE)(\dst), \fixup)
 	EX(LONG_S, \val, (\offset +  6 * LONGSIZE)(\dst), \fixup)
 	EX(LONG_S, \val, (\offset +  7 * LONGSIZE)(\dst), \fixup)
+#if LONGSIZE == 4
 	EX(LONG_S, \val, (\offset +  8 * LONGSIZE)(\dst), \fixup)
 	EX(LONG_S, \val, (\offset +  9 * LONGSIZE)(\dst), \fixup)
 	EX(LONG_S, \val, (\offset + 10 * LONGSIZE)(\dst), \fixup)
@@ -33,6 +42,7 @@
 	EX(LONG_S, \val, (\offset + 13 * LONGSIZE)(\dst), \fixup)
 	EX(LONG_S, \val, (\offset + 14 * LONGSIZE)(\dst), \fixup)
 	EX(LONG_S, \val, (\offset + 15 * LONGSIZE)(\dst), \fixup)
+#endif
 	.endm
 
 /*
@@ -49,9 +59,13 @@ LEAF(memset)
 	 move		v0, a0			/* result */
 
 	andi		a1, 0xff		/* spread fillword */
-	sll		t1, a1, 8
+	LONG_SLL		t1, a1, 8
 	or		a1, t1
-	sll		t1, a1, 16
+	LONG_SLL		t1, a1, 16
+#if LONGSIZE == 8
+	or		a1, t1
+	LONG_SLL		t1, a1, 32
+#endif
 	or		a1, t1
 1:
 
@@ -64,18 +78,18 @@ FEXPORT(__bzero)
 	 PTR_SUBU	t0, LONGSIZE		/* alignment in bytes */
 
 #ifdef __MIPSEB__
-	EX(swl, a1, (a0), first_fixup)		/* make word aligned */
+	EX(LONG_S_L, a1, (a0), first_fixup)	/* make word/dword aligned */
 #endif
 #ifdef __MIPSEL__
-	EX(swr, a1, (a0), first_fixup)		/* make word aligned */
+	EX(LONG_S_R, a1, (a0), first_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 */
-	 andi		t0, a2, 0x3c
+	 andi		t0, a2, 0x40-LONGSIZE
 
 	PTR_ADDU	t1, a0			/* end address */
 	.set		reorder
@@ -86,7 +100,14 @@ FEXPORT(__bzero)
 
 memset_partial:
 	PTR_LA		t1, 2f			/* where to start */
+#if LONGSIZE == 4
 	PTR_SUBU	t1, t0
+#else
+	.set		noat
+	LONG_SRL		AT, t0, 1
+	PTR_SUBU	t1, AT
+	.set		noat
+#endif
 	jr		t1
 	 PTR_ADDU	a0, t0			/* dest ptr */
 
@@ -100,10 +121,10 @@ memset_partial:
 	beqz		a2, 1f
 	 PTR_ADDU	a0, a2			/* What's left */
 #ifdef __MIPSEB__
-	EX(swr, a1, -1(a0), last_fixup)
+	EX(LONG_S_R, a1, -1(a0), last_fixup)
 #endif
 #ifdef __MIPSEL__
-	EX(swl, a1, -1(a0), last_fixup)
+	EX(LONG_S_L, a1, -1(a0), last_fixup)
 #endif
 1:	jr		ra
 	 move		a2, zero