When building dev-db/mysql with gcc 4.5, the build fails with the following message: make[2]: *** [do_abi_check] Error 1 make[2]: Leaving directory `/var/tmp/portage/dev-db/mysql-5.1.45-r1/work/mysql' make[1]: *** [abi_check] Error 2 make[1]: Leaving directory `/var/tmp/portage/dev-db/mysql-5.1.45-r1/work/mysql' make: *** [all-recursive] Error 1 * ERROR: dev-db/mysql-5.1.45-r1 failed: * emake failed * * Call stack: * ebuild.sh, line 48: Called src_compile * environment, line 4415: Called mysql_src_compile * environment, line 3878: Called die * The specific snippet of code: * emake || die "emake failed"; The failure occurs during an ABI-change check designed for internal MySQL developers. The failure results from a change in behavior of GCC's preprocessor. Previously, when the preprocessor was unable to #include a header, it would leave the #include intact and issue an error on stderr, but continue to process the file. With GCC 4.5, the preprocessor immediately aborts and does not complete the preprocessing. Here's a comparison: cd /var/tmp/portage/dev-db/mysql-5.1.45-r1/work/mysql /usr/x86_64-pc-linux-gnu/gcc-bin/4.4.3/gcc -E -nostdinc -dI -I./include -I./include/mysql -I./sql -I./include -I./include/mysql -I./sql include/mysql.h include/mysql.h:47:23: error: sys/types.h: No such file or directory <much valid preprocessed output, snipped> /usr/x86_64-pc-linux-gnu/gcc-bin/4.5.0-pre9999/gcc -E -nostdinc -dI -I./include -I./include/mysql -I./sql -I./include -I./include/mysql -I./sql include/mysql.h include/mysql.h:47:23: fatal error: sys/types.h: No such file or directory compilation terminated. # 1 "include/mysql.h" # 1 "<built-in>" # 1 "<command-line>" # 1 "include/mysql.h" # 47 "include/mysql.h" #include <sys/types.h> Since do_abi_check is for internal debugging use, it can be safely eliminated for end-user builds.
Created attachment 225971 [details] build.log
Created attachment 225973 [details, diff] Patch to skip do_abi_check
Raised upstream: http://bugs.mysql.com/bug.php?id=52514 I'll wait for their response before including that patch, because it's meant as a safety value to detect changes due to user systems too.
Upstream has confirmed, but has no fix yet. I'd expect to see it in 5.1.47 probably.
do you mind if we leave this open until upstream releases a fixed version so it shows up on the 4.5 tracker?
*** Bug 317681 has been marked as a duplicate of this bug. ***
(In reply to comment #5) > do you mind if we leave this open until upstream releases a fixed version so it > shows up on the 4.5 tracker? > a patch is ready, please put it into the package: http://bugs.mysql.com/file.php?id=14421 http://bugs.mysql.com/bug.php?id=52514
(In reply to comment #7) > (In reply to comment #5) > a patch is ready, please put it into the package: > > http://bugs.mysql.com/file.php?id=14421 > http://bugs.mysql.com/bug.php?id=52514 Have you noticed there hasn't been any change since Robin's last comment? He's the reporter of the bug (in case you missed it).
*** Bug 319655 has been marked as a duplicate of this bug. ***
what's the holdup here? now mysql wont even start after the openssl upgrade and there's a patch from 4 months ago that will fix this?
(In reply to comment #10) > what's the holdup here? now mysql wont even start after the openssl upgrade and > there's a patch from 4 months ago that will fix this? > That was just a temporary workaround. There was a patch (or 3?) committed upstream yesterday. The referenced bug, mysql 52514, now has the status 'documenting' and the fix should be included in 5.1.49. You can switch to gcc 4.4, build mysql, then switch back to gcc 4.5 if you're in need of mysql quickly.
(In reply to comment #11) > That was just a temporary workaround. There was a patch (or 3?) committed > upstream yesterday. The referenced bug, mysql 52514, now has the status > 'documenting' and the fix should be included in 5.1.49. As I don't really use the ssl feature currently I cheated with: ln -s libssl.so.1.0.0 libssl.so.0.9.8 ln -s libcrypto.so.1.0.0 libcrypto.so.0.9.8 Hopefully we'll see 5.1.49 soon.
mysql 5.1.49 was released: http://forums.mysql.com/read.php?3,377551 A version bump would not only fix this but, but this one as well: http://bugs.gentoo.org/show_bug.cgi?id=321791 A version bump in ~arch would be nice.
5.1.49 added to overlay, testing it now. http://git.overlays.gentoo.org/gitweb/?p=proj/mysql.git;a=commit;h=11aa801ac65212e678ed8da7715c56a39bd5077b Hopefully it passes the tests where 5.1.48 failed.
The bug was fixed before 5.1.49 was released, but it will not be included until 5.1.50. Using the overlay doesn't help any. http://dev.mysql.com/doc/refman/5.1/en/news-5-1-50.html
We've got it backported for 5.1.49, but there are other problems causing it to not pass the testsuite, so it's still only in the overlay.
Please retest w/ 5.1.49 now in the tree.
(In reply to comment #17) > Please retest w/ 5.1.49 now in the tree. Still fails here: make[2]: *** [do_abi_check] Error 1 make[2]: Leaving directory `/var/tmp/portage/dev-db/mysql-5.1.49/work/mysql' make[1]: *** [abi_check] Error 2 make[1]: Leaving directory `/var/tmp/portage/dev-db/mysql-5.1.49/work/mysql' make: *** [all-recursive] Error 1
retest w/ 5.1.49-r1 now.
(In reply to comment #19) > retest w/ 5.1.49-r1 now. Compiled fine with -r1.
