Gabe Black (gabeblack@chromium.org) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/3762
-gerrit
commit 84427fdbdfa4c6df288e305fccb7302ad6114a8b Author: Gabe Black gabeblack@google.com Date: Sun Jul 7 14:08:30 2013 -0700
x86: Add and enable an arch verson of memmove.
This is from memcpy_32.c in the Linux kernel. There was no copyright header in the original file either.
Change-Id: Ifd259cb8a87615dce79ed1e551cc4bacb0414b4f Signed-off-by: Gabe Black gabeblack@chromium.org --- src/arch/x86/Kconfig | 1 + src/arch/x86/lib/Makefile.inc | 4 + src/arch/x86/lib/memmove.c | 187 ++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 192 insertions(+)
diff --git a/src/arch/x86/Kconfig b/src/arch/x86/Kconfig index 287e2bc..581a048 100644 --- a/src/arch/x86/Kconfig +++ b/src/arch/x86/Kconfig @@ -5,6 +5,7 @@ config X86_ARCH_OPTIONS default y select HAVE_ARCH_MEMSET select HAVE_ARCH_MEMCPY + select HAVE_ARCH_MEMMOVE
config MARK_GRAPHICS_MEM_WRCOMB bool "Mark graphics memory as write-combining." diff --git a/src/arch/x86/lib/Makefile.inc b/src/arch/x86/lib/Makefile.inc index 56e4171..8588c76 100644 --- a/src/arch/x86/lib/Makefile.inc +++ b/src/arch/x86/lib/Makefile.inc @@ -6,6 +6,7 @@ ramstage-y += exception.c ramstage-$(CONFIG_IOAPIC) += ioapic.c ramstage-y += memset.c ramstage-y += memcpy.c +ramstage-y += memmove.c ramstage-y += ebda.c ramstage-y += rom_media.c ramstage-$(CONFIG_COOP_MULTITASKING) += thread.c @@ -15,13 +16,16 @@ romstage-$(CONFIG_EARLY_CONSOLE) += romstage_console.c romstage-y += cbfs_and_run.c romstage-y += memset.c romstage-y += memcpy.c +romstage-y += memmove.c romstage-y += rom_media.c
smm-y += memset.c smm-y += memcpy.c +smm-y += memmove.c smm-y += rom_media.c
rmodules-y += memset.c rmodules-y += memcpy.c +rmodules-y += memmove.c
$(obj)/arch/x86/lib/console.ramstage.o :: $(obj)/build.h diff --git a/src/arch/x86/lib/memmove.c b/src/arch/x86/lib/memmove.c new file mode 100644 index 0000000..ba12127 --- /dev/null +++ b/src/arch/x86/lib/memmove.c @@ -0,0 +1,187 @@ +#include <string.h> + +void *memmove(void *dest, const void *src, size_t n) +{ + int d0,d1,d2,d3,d4,d5; + char *ret = dest; + + __asm__ __volatile__( + /* Handle more 16bytes in loop */ + "cmp $0x10, %0\n\t" + "jb 1f\n\t" + + /* Decide forward/backward copy mode */ + "cmp %2, %1\n\t" + "jb 2f\n\t" + + /* + * movs instruction have many startup latency + * so we handle small size by general register. + */ + "cmp $680, %0\n\t" + "jb 3f\n\t" + /* + * movs instruction is only good for aligned case. + */ + "mov %1, %3\n\t" + "xor %2, %3\n\t" + "and $0xff, %3\n\t" + "jz 4f\n\t" + "3:\n\t" + "sub $0x10, %0\n\t" + + /* + * We gobble 16byts forward in each loop. + */ + "3:\n\t" + "sub $0x10, %0\n\t" + "mov 0*4(%1), %3\n\t" + "mov 1*4(%1), %4\n\t" + "mov %3, 0*4(%2)\n\t" + "mov %4, 1*4(%2)\n\t" + "mov 2*4(%1), %3\n\t" + "mov 3*4(%1), %4\n\t" + "mov %3, 2*4(%2)\n\t" + "mov %4, 3*4(%2)\n\t" + "lea 0x10(%1), %1\n\t" + "lea 0x10(%2), %2\n\t" + "jae 3b\n\t" + "add $0x10, %0\n\t" + "jmp 1f\n\t" + + /* + * Handle data forward by movs. + */ + ".p2align 4\n\t" + "4:\n\t" + "mov -4(%1, %0), %3\n\t" + "lea -4(%2, %0), %4\n\t" + "shr $2, %0\n\t" + "rep movsl\n\t" + "mov %3, (%4)\n\t" + "jmp 11f\n\t" + /* + * Handle data backward by movs. + */ + ".p2align 4\n\t" + "6:\n\t" + "mov (%1), %3\n\t" + "mov %2, %4\n\t" + "lea -4(%1, %0), %1\n\t" + "lea -4(%2, %0), %2\n\t" + "shr $2, %0\n\t" + "std\n\t" + "rep movsl\n\t" + "mov %3,(%4)\n\t" + "cld\n\t" + "jmp 11f\n\t" + + /* + * Start to prepare for backward copy. + */ + ".p2align 4\n\t" + "2:\n\t" + "cmp $680, %0\n\t" + "jb 5f\n\t" + "mov %1, %3\n\t" + "xor %2, %3\n\t" + "and $0xff, %3\n\t" + "jz 6b\n\t" + + /* + * Calculate copy position to tail. + */ + "5:\n\t" + "add %0, %1\n\t" + "add %0, %2\n\t" + "sub $0x10, %0\n\t" + + /* + * We gobble 16byts backward in each loop. + */ + "7:\n\t" + "sub $0x10, %0\n\t" + + "mov -1*4(%1), %3\n\t" + "mov -2*4(%1), %4\n\t" + "mov %3, -1*4(%2)\n\t" + "mov %4, -2*4(%2)\n\t" + "mov -3*4(%1), %3\n\t" + "mov -4*4(%1), %4\n\t" + "mov %3, -3*4(%2)\n\t" + "mov %4, -4*4(%2)\n\t" + "lea -0x10(%1), %1\n\t" + "lea -0x10(%2), %2\n\t" + "jae 7b\n\t" + /* + * Calculate copy position to head. + */ + "add $0x10, %0\n\t" + "sub %0, %1\n\t" + "sub %0, %2\n\t" + + /* + * Move data from 8 bytes to 15 bytes. + */ + ".p2align 4\n\t" + "1:\n\t" + "cmp $8, %0\n\t" + "jb 8f\n\t" + "mov 0*4(%1), %3\n\t" + "mov 1*4(%1), %4\n\t" + "mov -2*4(%1, %0), %5\n\t" + "mov -1*4(%1, %0), %1\n\t" + + "mov %3, 0*4(%2)\n\t" + "mov %4, 1*4(%2)\n\t" + "mov %5, -2*4(%2, %0)\n\t" + "mov %1, -1*4(%2, %0)\n\t" + "jmp 11f\n\t" + + /* + * Move data from 4 bytes to 7 bytes. + */ + ".p2align 4\n\t" + "8:\n\t" + "cmp $4, %0\n\t" + "jb 9f\n\t" + "mov 0*4(%1), %3\n\t" + "mov -1*4(%1, %0), %4\n\t" + "mov %3, 0*4(%2)\n\t" + "mov %4, -1*4(%2, %0)\n\t" + "jmp 11f\n\t" + + /* + * Move data from 2 bytes to 3 bytes. + */ + ".p2align 4\n\t" + "9:\n\t" + "cmp $2, %0\n\t" + "jb 10f\n\t" + "movw 0*2(%1), %%dx\n\t" + "movw -1*2(%1, %0), %%bx\n\t" + "movw %%dx, 0*2(%2)\n\t" + "movw %%bx, -1*2(%2, %0)\n\t" + "jmp 11f\n\t" + + /* + * Move data for 1 byte. + */ + ".p2align 4\n\t" + "10:\n\t" + "cmp $1, %0\n\t" + "jb 11f\n\t" + "movb (%1), %%cl\n\t" + "movb %%cl, (%2)\n\t" + ".p2align 4\n\t" + "11:" + : "=&c" (d0), "=&S" (d1), "=&D" (d2), + "=r" (d3),"=r" (d4), "=r"(d5) + :"0" (n), + "1" (src), + "2" (dest) + :"memory"); + + return ret; + +}