From a fresh *STABLE* stage3, build log as follows... The culprit is the 02_all_gcc-4.3-alpha-PR42113.patch patch, excluding it it compiles successfully.
Created attachment 223429 [details] build.log
The patch is used in the 4.3, 4.4, and 4.5 branches and is included in 4.4.3 already. The problem is, apparently, that we're missing some other necessary patch from 4.3 SVN. I'll see if I can track it down.
@toolchain: could we exclude the patch meanwhile? Thanks
(In reply to comment #3) > @toolchain: could we exclude the patch meanwhile? > > Thanks The problem with that is that the patch fixes internal compiler errors itself. So gcc will compile, but rsync and libperl won't. (Sorry I haven't had time to figure out what patches we need to make this compile.)
yes, i can vouch for the rsync ICE. i saw the same problem with 4.4.2 and upgrading to 4.4.3 fixed rsync building.
Created attachment 224705 [details, diff] alpha.patch I tracked the commits done to the gcc-4_3-branch until the patch from PR42113. This patch replaces 02_all_gcc-4.3-alpha-PR42113.patch , since that patch is included in this one as well. Thanks
i would want to extract some info from each svn commit to put into the patch header ...
Created attachment 224941 [details] svn.log Ah yes, sorry. Commits are: 150466 <- PR40577 150735 <- PR8603 151160 151710 152344 <- PR22903 154465 <- PR42113 Attached is the svn log for each commit.
that's perfect, thanks. any feedback Matt before i toss this in ?
(In reply to comment #9) > that's perfect, thanks. any feedback Matt before i toss this in ? Nope. Looks great to me. Thanks a _ton_ armin76! I owe you a beer, whenever I come to Spain. :D
Ugh. Apparently there is another fix to this: http://gcc.gnu.org/ml/gcc-patches/2010-03/msg01146.html Please include that as well after we test it.
Excluded in the ebuild until we have a full patch.
Reducing priority as this isn't a problem now. I'd vote for waiting for a 4.3.5 release, to be honest...at least i'm not going to bother to fix this, with the small improvement it gives(fixing -O3).
Since 4.4.3 is supposed to go to stable, can we forget about this bug?
there is a 4.3.5 release which i'm working on, so i'll close this once i bump
gcc-4.3.5 is now in the tree