[uClibc] Re: Kernel 2.4 and gcc 3.4

Peter S. Mazinger ps.m at gmx.net
Mon Feb 28 15:43:39 UTC 2005


Attached a patch that I used, it seems that -Os in gcc-3.4 optimizes away 
memcmp

Disclaimer: don't blaim me, if it is incorrect, if you loose data, etc ;)
I haven't gotten any feedback on this patch.

Peter

-- 
Peter S. Mazinger <ps dot m at gmx dot net>           ID: 0xA5F059F2
Key fingerprint = 92A4 31E1 56BC 3D5A 2D08  BB6E C389 975E A5F0 59F2
-------------- next part --------------
--- lib/string.c.mps	Tue Nov 30 14:22:32 2004
+++ lib/string.c	Tue Nov 30 14:24:03 2004
@@ -459,6 +459,7 @@
 			break;
 	return res;
 }
+EXPORT_SYMBOL(memcmp);
 #endif
 
 #ifndef __HAVE_ARCH_MEMSCAN
--- lib/Makefile.mps	Tue Nov 30 14:24:07 2004
+++ lib/Makefile	Tue Nov 30 14:25:35 2004
@@ -9,7 +9,7 @@
 L_TARGET := lib.a
 
 export-objs := cmdline.o dec_and_lock.o rwsem-spinlock.o rwsem.o \
-	       rbtree.o crc32.o firmware_class.o
+	       rbtree.o crc32.o firmware_class.o string.o
 
 obj-y := errno.o ctype.o string.o vsprintf.o brlock.o cmdline.o \
 	 bust_spinlocks.o rbtree.o dump_stack.o
--- include/asm-i386/string.h.mps	2004-11-30 15:36:30.000000000 +0100
+++ include/asm-i386/string.h	2004-11-30 15:06:17.000000000 +0100
@@ -332,7 +332,9 @@
 return dest;
 }
 
+#ifdef __HAVE_ARCH_MEMCMP
 #define memcmp __builtin_memcmp
+#endif
 
 #define __HAVE_ARCH_MEMCHR
 static inline void * memchr(const void * cs,int c,size_t count)


More information about the uClibc mailing list