mapi: Improve the x86 tsd stubs performance.

This skips touching %ebx most times and it shows that glGetString performance
increased from 114M/s to 120M/s on my desktop.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Lepton Wu <lepton@chromium.org>

diff --git a/src/mapi/entry_x86_tsd.h b/src/mapi/entry_x86_tsd.h
index bd8db7b19f9748e21e34046cd9157e87b1b85154..9526cdcef1a83ef3440f1cfcd7adab6e530b3dd5 100644 (file)
--- a/src/mapi/entry_x86_tsd.h
+++ b/src/mapi/entry_x86_tsd.h
@@ -44,18 +44,19 @@ __asm__(".text\n"
    func ":"
 
 #define STUB_ASM_CODE(slot)         \
-   "push %ebx\n\t"                  \
    "call 1f\n\t"                    \
    "1:\n\t"                         \
-   "popl %ebx\n\t"                  \
-   "addl $_GLOBAL_OFFSET_TABLE_+[.-1b], %ebx\n\t" \
-   "movl " ENTRY_CURRENT_TABLE "@GOT(%ebx), %eax\n\t" \
+   "popl %ecx\n\t"                  \
+   "addl $_GLOBAL_OFFSET_TABLE_+[.-1b], %ecx\n\t" \
+   "movl " ENTRY_CURRENT_TABLE "@GOT(%ecx), %eax\n\t" \
    "mov (%eax), %eax\n\t"           \
    "testl %eax, %eax\n\t"           \
    "jne 1f\n\t"                     \
+   "push %ebx\n\t"                  \
+   "movl %ecx, %ebx\n\t"            \
    "call " ENTRY_CURRENT_TABLE_GET "@PLT\n\t" \
+   "popl %ebx\n\t"                  \
    "1:\n\t"                         \
-   "pop %ebx\n\t"                   \
    "jmp *(4 * " slot ")(%eax)"
 
 #define MAPI_TMP_STUB_ASM_GCC