linux/arch/mips/lib/memset.S

303 lines
6.6 KiB
ArmAsm
Raw Normal View History

/*
* This file is subject to the terms and conditions of the GNU General Public
* License. See the file "COPYING" in the main directory of this archive
* for more details.
*
* Copyright (C) 1998, 1999, 2000 by Ralf Baechle
* Copyright (C) 1999, 2000 Silicon Graphics, Inc.
* Copyright (C) 2007 by Maciej W. Rozycki
* Copyright (C) 2011, 2012 MIPS Technologies, Inc.
*/
#include <asm/asm.h>
#include <asm/asm-offsets.h>
#include <asm/export.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
#ifdef CONFIG_CPU_MICROMIPS
#define STORSIZE (LONGSIZE * 2)
#define STORMASK (STORSIZE - 1)
#define FILL64RG t8
#define FILLPTRG t7
#undef LONG_S
#define LONG_S LONG_SP
#else
#define STORSIZE LONGSIZE
#define STORMASK LONGMASK
#define FILL64RG a1
#define FILLPTRG t0
#endif
#define LEGACY_MODE 1
#define EVA_MODE 2
/*
* No need to protect it with EVA #ifdefery. The generated block of code
* will never be assembled if EVA is not enabled.
*/
#define __EVAFY(insn, reg, addr) __BUILD_EVA_INSN(insn##e, reg, addr)
#define ___BUILD_EVA_INSN(insn, reg, addr) __EVAFY(insn, reg, addr)
#define EX(insn,reg,addr,handler) \
.if \mode == LEGACY_MODE; \
9: insn reg, addr; \
.else; \
9: ___BUILD_EVA_INSN(insn, reg, addr); \
.endif; \
.section __ex_table,"a"; \
PTR 9b, handler; \
.previous
.macro f_fill64 dst, offset, val, fixup, mode
EX(LONG_S, \val, (\offset + 0 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 1 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 2 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 3 * STORSIZE)(\dst), \fixup)
#if ((defined(CONFIG_CPU_MICROMIPS) && (LONGSIZE == 4)) || !defined(CONFIG_CPU_MICROMIPS))
EX(LONG_S, \val, (\offset + 4 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 5 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 6 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 7 * STORSIZE)(\dst), \fixup)
#endif
#if (!defined(CONFIG_CPU_MICROMIPS) && (LONGSIZE == 4))
EX(LONG_S, \val, (\offset + 8 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 9 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 10 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 11 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 12 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 13 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 14 * STORSIZE)(\dst), \fixup)
EX(LONG_S, \val, (\offset + 15 * STORSIZE)(\dst), \fixup)
#endif
.endm
.set noreorder
.align 5
/*
* Macro to generate the __bzero{,_user} symbol
* Arguments:
* mode: LEGACY_MODE or EVA_MODE
*/
.macro __BUILD_BZERO mode
/* Initialize __memset if this is the first time we call this macro */
.ifnotdef __memset
.set __memset, 1
.hidden __memset /* Make sure it does not leak */
.endif
sltiu t0, a2, STORSIZE /* very small region? */
bnez t0, .Lsmall_memset\@
andi t0, a0, STORMASK /* aligned? */
#ifdef CONFIG_CPU_MICROMIPS
move t8, a1 /* used by 'swp' instruction */
move t9, a1
#endif
[MIPS] R4000/R4400 daddiu erratum workaround This complements the generic R4000/R4400 errata workaround code and adds bits for the daddiu problem. In most places it just modifies handwritten assembly code so that the assembler is allowed to use a temporary register as daddiu may now be treated as a macro that expands to a sequence of li and daddu. It is the AT register or, where AT is unavailable or used explicitly for another purpose, an explicitly-named register is selected, using the .set at=<reg> feature added recently to gas. This feature is only used if CONFIG_CPU_DADDI_WORKAROUNDS has been set, so if the workaround remains disabled, the required version of binutils stays unchanged. Similarly, daddiu instructions put in branch delay slots in noreorder fragments are now taken out of them and the assembler is allowed to reorder them itself as possible (which it does making the whole idea of scheduling them into delay slots manually questionable). Also in the very few places where such a simple conversion was not possible, a handcoded longer sequence is implemented. Other than that there are changes to code responsible for building the TLB fault and page clear/copy handlers to avoid daddiu as appropriate. These are only effective if the erratum is verified to be present at the run time. Finally there is a trivial update to __delay(), because it uses daddiu in a branch delay slot. Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2007-10-23 19:43:25 +08:00
#ifndef CONFIG_CPU_DADDI_WORKAROUNDS
beqz t0, 1f
PTR_SUBU t0, STORSIZE /* alignment in bytes */
[MIPS] R4000/R4400 daddiu erratum workaround This complements the generic R4000/R4400 errata workaround code and adds bits for the daddiu problem. In most places it just modifies handwritten assembly code so that the assembler is allowed to use a temporary register as daddiu may now be treated as a macro that expands to a sequence of li and daddu. It is the AT register or, where AT is unavailable or used explicitly for another purpose, an explicitly-named register is selected, using the .set at=<reg> feature added recently to gas. This feature is only used if CONFIG_CPU_DADDI_WORKAROUNDS has been set, so if the workaround remains disabled, the required version of binutils stays unchanged. Similarly, daddiu instructions put in branch delay slots in noreorder fragments are now taken out of them and the assembler is allowed to reorder them itself as possible (which it does making the whole idea of scheduling them into delay slots manually questionable). Also in the very few places where such a simple conversion was not possible, a handcoded longer sequence is implemented. Other than that there are changes to code responsible for building the TLB fault and page clear/copy handlers to avoid daddiu as appropriate. These are only effective if the erratum is verified to be present at the run time. Finally there is a trivial update to __delay(), because it uses daddiu in a branch delay slot. Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2007-10-23 19:43:25 +08:00
#else
.set noat
li AT, STORSIZE
[MIPS] R4000/R4400 daddiu erratum workaround This complements the generic R4000/R4400 errata workaround code and adds bits for the daddiu problem. In most places it just modifies handwritten assembly code so that the assembler is allowed to use a temporary register as daddiu may now be treated as a macro that expands to a sequence of li and daddu. It is the AT register or, where AT is unavailable or used explicitly for another purpose, an explicitly-named register is selected, using the .set at=<reg> feature added recently to gas. This feature is only used if CONFIG_CPU_DADDI_WORKAROUNDS has been set, so if the workaround remains disabled, the required version of binutils stays unchanged. Similarly, daddiu instructions put in branch delay slots in noreorder fragments are now taken out of them and the assembler is allowed to reorder them itself as possible (which it does making the whole idea of scheduling them into delay slots manually questionable). Also in the very few places where such a simple conversion was not possible, a handcoded longer sequence is implemented. Other than that there are changes to code responsible for building the TLB fault and page clear/copy handlers to avoid daddiu as appropriate. These are only effective if the erratum is verified to be present at the run time. Finally there is a trivial update to __delay(), because it uses daddiu in a branch delay slot. Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2007-10-23 19:43:25 +08:00
beqz t0, 1f
PTR_SUBU t0, AT /* alignment in bytes */
[MIPS] R4000/R4400 daddiu erratum workaround This complements the generic R4000/R4400 errata workaround code and adds bits for the daddiu problem. In most places it just modifies handwritten assembly code so that the assembler is allowed to use a temporary register as daddiu may now be treated as a macro that expands to a sequence of li and daddu. It is the AT register or, where AT is unavailable or used explicitly for another purpose, an explicitly-named register is selected, using the .set at=<reg> feature added recently to gas. This feature is only used if CONFIG_CPU_DADDI_WORKAROUNDS has been set, so if the workaround remains disabled, the required version of binutils stays unchanged. Similarly, daddiu instructions put in branch delay slots in noreorder fragments are now taken out of them and the assembler is allowed to reorder them itself as possible (which it does making the whole idea of scheduling them into delay slots manually questionable). Also in the very few places where such a simple conversion was not possible, a handcoded longer sequence is implemented. Other than that there are changes to code responsible for building the TLB fault and page clear/copy handlers to avoid daddiu as appropriate. These are only effective if the erratum is verified to be present at the run time. Finally there is a trivial update to __delay(), because it uses daddiu in a branch delay slot. Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2007-10-23 19:43:25 +08:00
.set at
#endif
#ifndef CONFIG_CPU_MIPSR6
R10KCBARRIER(0(ra))
#ifdef __MIPSEB__
EX(LONG_S_L, a1, (a0), .Lfirst_fixup\@) /* make word/dword aligned */
#else
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 */
#else /* CONFIG_CPU_MIPSR6 */
#define STORE_BYTE(N) \
EX(sb, a1, N(a0), .Lbyte_fixup\@); \
beqz t0, 0f; \
PTR_ADDU t0, 1;
PTR_ADDU a2, t0 /* correct size */
PTR_ADDU t0, 1
STORE_BYTE(0)
STORE_BYTE(1)
#if LONGSIZE == 4
EX(sb, a1, 2(a0), .Lbyte_fixup\@)
#else
STORE_BYTE(2)
STORE_BYTE(3)
STORE_BYTE(4)
STORE_BYTE(5)
EX(sb, a1, 6(a0), .Lbyte_fixup\@)
#endif
0:
ori a0, STORMASK
xori a0, STORMASK
PTR_ADDIU a0, STORSIZE
#endif /* CONFIG_CPU_MIPSR6 */
1: ori t1, a2, 0x3f /* # of full blocks */
xori t1, 0x3f
beqz t1, .Lmemset_partial\@ /* no block to fill */
andi t0, a2, 0x40-STORSIZE
PTR_ADDU t1, a0 /* end address */
.set reorder
1: PTR_ADDIU a0, 64
R10KCBARRIER(0(ra))
f_fill64 a0, -64, FILL64RG, .Lfwd_fixup\@, \mode
bne t1, a0, 1b
.set noreorder
.Lmemset_partial\@:
R10KCBARRIER(0(ra))
PTR_LA t1, 2f /* where to start */
#ifdef CONFIG_CPU_MICROMIPS
LONG_SRL t7, t0, 1
#endif
#if LONGSIZE == 4
PTR_SUBU t1, FILLPTRG
#else
.set noat
LONG_SRL AT, FILLPTRG, 1
PTR_SUBU t1, AT
[MIPS] R4000/R4400 daddiu erratum workaround This complements the generic R4000/R4400 errata workaround code and adds bits for the daddiu problem. In most places it just modifies handwritten assembly code so that the assembler is allowed to use a temporary register as daddiu may now be treated as a macro that expands to a sequence of li and daddu. It is the AT register or, where AT is unavailable or used explicitly for another purpose, an explicitly-named register is selected, using the .set at=<reg> feature added recently to gas. This feature is only used if CONFIG_CPU_DADDI_WORKAROUNDS has been set, so if the workaround remains disabled, the required version of binutils stays unchanged. Similarly, daddiu instructions put in branch delay slots in noreorder fragments are now taken out of them and the assembler is allowed to reorder them itself as possible (which it does making the whole idea of scheduling them into delay slots manually questionable). Also in the very few places where such a simple conversion was not possible, a handcoded longer sequence is implemented. Other than that there are changes to code responsible for building the TLB fault and page clear/copy handlers to avoid daddiu as appropriate. These are only effective if the erratum is verified to be present at the run time. Finally there is a trivial update to __delay(), because it uses daddiu in a branch delay slot. Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2007-10-23 19:43:25 +08:00
.set at
#endif
jr t1
PTR_ADDU a0, t0 /* dest ptr */
.set push
.set noreorder
.set nomacro
/* ... but first do longs ... */
f_fill64 a0, -64, FILL64RG, .Lpartial_fixup\@, \mode
2: .set pop
andi a2, STORMASK /* At most one long to go */
beqz a2, 1f
#ifndef CONFIG_CPU_MIPSR6
PTR_ADDU a0, a2 /* What's left */
R10KCBARRIER(0(ra))
#ifdef __MIPSEB__
EX(LONG_S_R, a1, -1(a0), .Llast_fixup\@)
#else
EX(LONG_S_L, a1, -1(a0), .Llast_fixup\@)
#endif
#else
PTR_SUBU t0, $0, a2
PTR_ADDIU t0, 1
STORE_BYTE(0)
STORE_BYTE(1)
#if LONGSIZE == 4
EX(sb, a1, 2(a0), .Lbyte_fixup\@)
#else
STORE_BYTE(2)
STORE_BYTE(3)
STORE_BYTE(4)
STORE_BYTE(5)
EX(sb, a1, 6(a0), .Lbyte_fixup\@)
#endif
0:
#endif
1: jr ra
move a2, zero
.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)
2: jr ra /* done */
move a2, zero
.if __memset == 1
END(memset)
.set __memset, 0
.hidden __memset
.endif
#ifdef CONFIG_CPU_MIPSR6
.Lbyte_fixup\@:
PTR_SUBU a2, $0, t0
jr ra
PTR_ADDIU a2, 1
#endif /* CONFIG_CPU_MIPSR6 */
.Lfirst_fixup\@:
jr ra
nop
.Lfwd_fixup\@:
PTR_L t0, TI_TASK($28)
andi a2, 0x3f
LONG_L t0, THREAD_BUADDR(t0)
LONG_ADDU a2, t1
jr ra
LONG_SUBU a2, t0
.Lpartial_fixup\@:
PTR_L t0, TI_TASK($28)
andi a2, STORMASK
LONG_L t0, THREAD_BUADDR(t0)
LONG_ADDU a2, t1
jr ra
LONG_SUBU a2, t0
.Llast_fixup\@:
jr ra
andi v1, a2, STORMASK
.endm
/*
* memset(void *s, int c, size_t n)
*
* a0: start of area to clear
* a1: char to fill with
* a2: size of area to clear
*/
LEAF(memset)
EXPORT_SYMBOL(memset)
beqz a1, 1f
move v0, a0 /* result */
andi a1, 0xff /* spread fillword */
LONG_SLL t1, a1, 8
or a1, t1
LONG_SLL t1, a1, 16
#if LONGSIZE == 8
or a1, t1
LONG_SLL t1, a1, 32
#endif
or a1, t1
1:
#ifndef CONFIG_EVA
FEXPORT(__bzero)
EXPORT_SYMBOL(__bzero)
#else
FEXPORT(__bzero_kernel)
EXPORT_SYMBOL(__bzero_kernel)
#endif
__BUILD_BZERO LEGACY_MODE
#ifdef CONFIG_EVA
LEAF(__bzero)
EXPORT_SYMBOL(__bzero)
__BUILD_BZERO EVA_MODE
END(__bzero)
#endif