Date: Sat, 30 Nov 2002 00:40:35 +0100
From: Petr Vandrovec <vandrove@vc.cvut.cz>
To: alan@lxorguk.ukuu.org.uk, torvalds@transmeta.com
Cc: linux-kernel@vger.kernel.org
Subject: [PATCH] Save 8 bytes in __{read,write}_lock_failed...
Message-ID: <20021129234035.GA22666@vana>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.4i
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1577
Lines: 51

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/