From: SourceForge.net <no...@so...> - 2004-03-26 19:06:09
|
Bugs item #923803, was opened at 2004-03-26 10:58 Message generated for change (Comment added) made by richi You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=101627&aid=923803&group_id=1627 Category: Build Process Group: None Status: Closed Resolution: Fixed Priority: 5 Submitted By: Volker Stolz (volkersf) Assigned to: Richard Guenther (richi) Summary: varargs on FreeBSD 5 Initial Comment: Somehow FreeBSD 5.x's va_copy is not detected and memcpy breaks on other architectures than i386, e.g. alpha. Currently, the attached patch helps. ---------------------------------------------------------------------- >Comment By: Richard Guenther (richi) Date: 2004-03-26 19:06 Message: Logged In: YES user_id=7575 As always, testing ia32 is mood - the patch fails on ppc because I missed some spot - and viola - va_copy is back :( Added a check for it (but no workaround for a missing one) to unstable CVS, too. See attached patch "p2" for a working version (for systems with va_copy). ---------------------------------------------------------------------- Comment By: Richard Guenther (richi) Date: 2004-03-26 17:22 Message: Logged In: YES user_id=7575 I checked it on debian unstable on ia32. But it should work in theory because now we avoid the fragile va_copy stuff. Checked in into unstable, will backport soon. Thanks for the report, Richard. ---------------------------------------------------------------------- Comment By: Volker Stolz (volkersf) Date: 2004-03-26 17:04 Message: Logged In: YES user_id=126328 Yes, glame builds fine with your patch as well. Thanks! I didn't check if it's doing what it's supposed to do though :) ---------------------------------------------------------------------- Comment By: Richard Guenther (richi) Date: 2004-03-26 12:59 Message: Logged In: YES user_id=7575 Ok, can you try the attached patch which just removes the need for va_copy? It's untested, but maybe you get the idea and can fix up any simple errors I made. Thanks, Richard. ---------------------------------------------------------------------- Comment By: Richard Guenther (richi) Date: 2004-03-26 11:19 Message: Logged In: YES user_id=7575 I guess it's time for a configure check. I'll go over this piece again. Don't exactly remember why I put the memcpy fallback in, but I think I should drop it in favor of a configure failure. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=101627&aid=923803&group_id=1627 |