gcc-4.1.1 upstream bugfixes

[Matthew Burgess]

As reported by Alexander Patrakov via Archaic:

echo | /lfs-livecd/packages/gcc/gcc-build/./gcc/xgcc
-B/lfs-livecd/packages/gcc/gcc-build/./gcc/ -B/usr/i486-pc-linux-gnu/bin/
-B/usr/i486-pc-linux-gnu/lib/ -isystem /usr/i486-pc-linux-gnu/include -isystem
/usr/i486-pc-linux-gnu/sys-include -E -dM - | \
          sed -n 's/^#define \([^_][a-zA-Z0-9_]*\).*/\1/p ; \
                s/^#define \(_[^_A-Z][a-zA-Z0-9_]*\).*/\1/p' | \
          sort -u > tmp-macro_list
sed: -e expression #1, char 88: unterminated address regex

[alexander@…]

Adds more backoprted bugfixes

[alexander@…]

Please use the new -fixes-1 patch instead. It also fixes some cases of silent miscompilation.

It would be also nice to backport ICE fixes, but I did not do that because the LiveCD build doesn't hit any ICE.

[Matthew Burgess]

Am I right in thinking that the upstream-fixes patch should be applied to all 3 passes of GCC? The host may well be running Make-3.81 after all.

[alexander@…]

It is certainly the safest option. However, the Make-3.81 bug affects only installation of the fixincludes script, something we don't care about. Other bugfixes are vital if one adds -Os to CFLAGS.

[alexander@…]

FIXME: the patch is incomplete, e.g., it doesn't include the fix for PR 26719.

[Matthew Burgess]

OK, the patch for PR26719 (and also PR27795 & PR27639) is at ​http://gcc.gnu.org/ml/gcc-patches/2006-07/msg00898.html. Another patch (for PR27937) at ​http://gcc.gnu.org/ml/gcc-patches/2006-07/msg00229.html was committed following discussions related to that original patch but needs an additional trivial fix mentioned at ​http://gcc.gnu.org/ml/gcc-patches/2006-07/msg01057.html. I'll put a consolidated patch together tonight, and test it out. No doubt once this is all done, GCC-4.1.2 will be released rendering all this effort meaningless :-)

[Matthew Burgess]

Well, those patches have rejections all over the place. As it is, GCC-4.1.2-RC1 is due out anytime, meaning the final release won't be too far away. I'm tempted to just wait until then to get these fixes. As this bug is 2 months old anyway, is it unreasonable to wait for another couple of weeks? If not, I'd appreciate someone else working on getting the patches mentioned above into a state where they'll apply cleanly.

[alexander@…]

Not fixing this bug (maybe even without PR 26719, as it is too complex) would lead to the word "Official" being dropped from the unstable LiveCD name.

[Matthew Burgess]

Why would you have to drop the "Official" designation of the LiveCD? You already build Linux with patches that aren't in the LFS book (reiser4 and tulip-is-not-a-proper-driver-for-dm910x.patch), so why would using an additional patch in GCC cause problems with the officiality of the LiveCD?

That said, I'll apply the fixes patch to the book, as it does remove the incompatibility with Make. I'll leave the fixes for PR26719 et al until GCC-4.1.2 comes out, which, if all goes well, will be in about a week: ​http://gcc.gnu.org/ml/gcc/2007-01/msg01223.html.

[alexander@…]

Then it surely makes sense to wait a bit.

[Matthew Burgess]

OK, I'll close this as WONTFIX then. We'll pick the bug fixes up in 4.1.2.

[Jeremy Huntwork]

Milestone 6.3 deleted