Hi,
recent bugreport from Chris Rankin pointed out to me
that last jump in both __read_lock_failed and __write_lock_failed
uses 4 byte offset, although 1 byte is enough. It happens
because of __write_lock_failed is .globl, and so gas emits
4 byte offset + R_386_PC32 relocation (should it? I always thought
that only .weak symbols may be overwritten from other file).
Simple patch below makes SMP kernel 8 bytes shorter.
Best regards,
Petr Vandrovec
vandrove@vc.cvut.cz
diff -urdN linux/arch/i386/kernel/semaphore.c linux/arch/i386/kernel/semaphore.c
--- linux/arch/i386/kernel/semaphore.c 2002-11-26 19:55:59.000000000 +0000
+++ linux/arch/i386/kernel/semaphore.c 2002-11-29 23:24:50.000000000 +0000
@@ -275,12 +275,13 @@
".align 4\n"
".globl __write_lock_failed\n"
"__write_lock_failed:\n\t"
+"0:\n\t"
LOCK "addl $" RW_LOCK_BIAS_STR ",(%eax)\n"
"1: rep; nop\n\t"
"cmpl $" RW_LOCK_BIAS_STR ",(%eax)\n\t"
"jne 1b\n\t"
LOCK "subl $" RW_LOCK_BIAS_STR ",(%eax)\n\t"
- "jnz __write_lock_failed\n\t"
+ "jnz 0b\n\t"
"ret"
);
@@ -289,12 +290,13 @@
".align 4\n"
".globl __read_lock_failed\n"
"__read_lock_failed:\n\t"
+"0:\n\t"
LOCK "incl (%eax)\n"
"1: rep; nop\n\t"
"cmpl $1,(%eax)\n\t"
"js 1b\n\t"
LOCK "decl (%eax)\n\t"
- "js __read_lock_failed\n\t"
+ "js 0b\n\t"
"ret"
);
#endif
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Nov 30 2002 - 22:00:23 EST