On Tue, 18 May 2010, Jeffrey Lee wrote:
The easiest fix is likely to be to amend the NULL check that was added to
__gcc_alloca_restore() for bug #174 so that it also returns if
(chunk->block != block). I'd have tested this myself by now, but something
is preventing my current source tree from building, so I'll have to try again
in a day or two when I have some more time available to get my tree
working again.
After fixing my build issues, I can confirm that the attached patch fixes
the problems I've been seeing. Want me to enter it as a bug report so it
doesn't get forgotten?
Cheers,
- Jeffrey
Index: gcc4/recipe/files/gcc/libunixlib/gcccompat/galloca.c
===================================================================
--- gcc4/recipe/files/gcc/libunixlib/gcccompat/galloca.c (revision 4622)
+++ gcc4/recipe/files/gcc/libunixlib/gcccompat/galloca.c (working copy)
@@ -193,11 +193,13 @@
/* Bug 174: If GCC determines that a call to __gcc_alloca can never be
reached,
it will optimise it away. However, the subsequent call to
__gcc_alloca_restore
is left intact and, as a block was never allocated, crashes because
- chunk == NULL . */
- if (chunk == NULL)
+ chunk == NULL .
+ We also need to protect against chunk->block != block, for the situation
where
+ __gcc_alloca has been skipped for this stack frame but the chunk list is
+ nonempty due to use of __gcc_alloca in a parent stack frame */
+ if ((chunk == NULL) || (chunk->block != block))
return;
- assert (chunk->block == block);
assert (chunk->fp == callee_fp);
/* Delete chunk with id block. */
_______________________________________________
GCCSDK mailing list [email protected]
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK
