This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[cs] patch to revert reset of funcdef_no


This reverts my 10-29 patch to reset the funcdef_no counter
when the compile server starts a new output file.  The
intention was to get identical assembly output with or
without the server (for testing/QA purposes).  Alas, this
doesn't work because of saved (inline) instructions are
reused, and so the numbers may clash.

For now I guess we'll aim for identical .o files, rather than
identical .s files.  (Local labels in .s files normally don't
show up in the .o files.)

Checked into the compile server branch.
--
	--Per Bothner
per@bothner.com   http://per.bothner.com/


2003-12-12  Per Bothner  <pbothner@apple.com>

	* function.c (init_function_once):  Don't reset funcdef_no.
	It doesn't work when RTL is reused across translation units.
	This reverts a 10-29 change.

Index: function.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/function.c,v
retrieving revision 1.448.2.3
diff -u -p -r1.448.2.3 function.c
--- function.c	11 Dec 2003 22:19:22 -0000	1.448.2.3
+++ function.c	13 Dec 2003 00:29:22 -0000
@@ -8025,8 +8025,6 @@ init_function_once (void)
   VARRAY_INT_INIT (prologue, 0, "prologue");
   VARRAY_INT_INIT (epilogue, 0, "epilogue");
   VARRAY_INT_INIT (sibcall_epilogue, 0, "sibcall_epilogue");
-
-  funcdef_no = 0;
 }
 
 #include "gt-function.h"

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]