=== kylem [n=kyle@cabal.ca] has joined #ubuntu-toolchain === __keybuk [n=scott@quest.netsplit.com] has joined #ubuntu-toolchain === mdz [n=mdz@studiocity-motorola-bsr1-70-36-194-85.vnnyca.adelphia.net] has joined #ubuntu-toolchain === mdz [n=mdz@studiocity-motorola-bsr1-70-36-194-85.vnnyca.adelphia.net] has joined #ubuntu-toolchain === doko [n=doko@dslb-088-073-105-251.pools.arcor-ip.net] has joined #ubuntu-toolchain === rodarvus [n=rodarvus@ubuntu/member/rodarvus] has joined #ubuntu-toolchain === jbailey [n=jbailey@209.217.74.66] has joined #ubuntu-toolchain [03:32] doko: Ran my box out of drive space last night, restarted the build this morning. === rodarvus [n=rodarvus@ubuntu/member/rodarvus] has joined #ubuntu-toolchain [03:33] jbailey: ouch [03:34] rodarvus: Is there an #ubuntu-x channel where I should lurk before filing random bugs? =) [03:34] yes, #ubuntu-x :) [03:34] Whee! [04:27] doko: Going into the testuite with 900MB Free this time. SHould be much better. =) [04:48] doko: Remind me, are you doing gcj-4.2 for edgy, or sticking with 4.1? [04:50] jbailey: didn't look too careful ... FC seems to have a backport of the gcjwebplugin patches to the 4.1 branch. Maybe I'll look at that one. [04:51] Ah, interesting. [04:51] I know that gcjwebplugin is now integrated into classpath itself upstream. === mdz [n=mdz@studiocity-motorola-bsr1-70-36-194-85.vnnyca.adelphia.net] has joined #ubuntu-toolchain [07:45] doko: Starting glibc build with rebuilt gcc [08:41] doko: The solves the glibc build failure on ppc. [08:50] jbailey: so that was the a rebuilt gcc with ssp turned off? [08:50] Yeah, lemme debdiff for you. [08:53] hmm, strange === jbailey [n=jbailey@209.217.74.66] has joined #ubuntu-toolchain === pitti [n=pitti@ubuntu/member/pitti] has joined #ubuntu-toolchain [11:53] hi [11:53] hi [11:53] one minute, please [11:53] Heya Martin! [11:53] glibc on powerpc doesn't build anymore with the ssp enabled gcc [11:54] argh argh [11:54] hm, it built before, didn't it? [11:54] pitti: Haven't had a chance to go through and really debug why, but it's an interesting problem. [11:54] glibc itself does -fstack-protector when available. [11:55] right [11:56] so gcc falls over if you specify -fs-p explicitly? [11:56] glibc, but yes. [11:56] It seems to cause some sort of symbol leak resulting in a link error. [11:57] oh, ok, so glibc must be fixed, not gcc? [11:57] jbailey: interesting, I saw a similar bug when building postgresql on sid, remember? === pitti actually looks at build log [11:58] " edgy powerpc Successfully built" [11:58] hm, that won't help me [11:58] pitti: I'm not sure who's bug it is. glibc is correct for an upstream gcc and DTRT for stack protection for itself. [11:58] is there any build log somewhere? [11:59] jbailey: when calling gcc -E, I currently do not turn on -fstack-protector, resulting in a missing __SSP__ define. but that should be consistent across architectures [11:59] ouch, SSP influences -E? [12:00] I thought inline functions and such were only decorated with -fstack-protector-all or oso [12:01] I only say, that a macro is defined. [12:01] ah [12:02] jbailey: so, it fails for you locally? [12:02] pitti: Yes. [12:02] jbailey: amd64 and i386 builds fail as well [12:02] doko: Thanks. [12:03] I didn't have much time to research it with the sprint last week. [12:03] I can try to look through it this weekend, I guess. I'm behind on so many things right now. [12:03] gcc-4.1 -nostdlib -nostartfiles -r -o /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/elf/librtld.map.o '-Wl,-(' /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/elf/dl-allobjs.os /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/libc_pic.a -lgcc '-Wl,-)' -Wl,-Map,/home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/elf/librtld.mapT [12:03] /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/libc_pic.a(init-first.os):(.data+0x0): multiple definition of `__libc_multiple_libcs' [12:03] /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/elf/dl-allobjs.os:(.bss+0xec): first defined here [12:03] /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/libc_pic.a(_itoa.os): In function `_itoa': [12:03] _itoa.c:(.text+0xb0): multiple definition of `_itoa' [12:03] /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/elf/dl-allobjs.os:(.text+0x13960): first defined here [12:03] /usr/bin/ld: Warning: size of symbol `_itoa' changed from 120 in /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/elf/dl-allobjs.os to 432 in /home/packages/glibc/u/glibc-2.4/build-tree/amd64-libc/libc_pic.a(_itoa.os) [12:03] jbailey: curious why it works on the buildds [12:03] collect2: ld returned 1 exit status [12:03] that is amd64 ... [12:03] pitti: It didn't. [12:03] pitti: we didn't try yet :) [12:04] pitti: The last glibc build on the buildds was before the forced ssp change. [12:04] oh [12:04] gcc-4.1 -nostdlib -nostartfiles -shared -o /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/ld.so \ [12:04] -Wl,-z,combreloc -Wl,-z,relro -Wl,-z,defs \ [12:04] /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/librtld.os -Wl,--version-script=/home/packages/tmp/glibc-2.4/build-tree/i386-libc/ld.map \ [12:04] -Wl,-soname=ld-linux.so.2 -T /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/ld.so.lds [12:04] /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/librtld.os: In function `process_dl_debug': [12:04] /home/packages/tmp/glibc-2.4/build-tree/glibc-2.4/elf/rtld.c:2452: undefined reference to `__stack_chk_fail_local' [12:04] /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/librtld.os: In function `process_envvars': [12:04] /home/packages/tmp/glibc-2.4/build-tree/glibc-2.4/elf/rtld.c:2711: undefined reference to `__stack_chk_fail_local' [12:05] /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/librtld.os: In function `dl_main': [12:05] /home/packages/tmp/glibc-2.4/build-tree/glibc-2.4/elf/rtld.c:2332: undefined reference to `__stack_chk_fail_local' [12:05] /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/librtld.os: In function `print_search_path': [12:05] /home/packages/tmp/glibc-2.4/build-tree/glibc-2.4/elf/dl-load.c:1548: undefined reference to `__stack_chk_fail_local' [12:05] /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/librtld.os: In function `open_verify': [12:05] /home/packages/tmp/glibc-2.4/build-tree/glibc-2.4/elf/dl-load.c:1762: undefined reference to `__stack_chk_fail_local' [12:05] /home/packages/tmp/glibc-2.4/build-tree/i386-libc/elf/librtld.os:/home/packages/tmp/glibc-2.4/build-tree/glibc-2.4/elf/dl-load.c:1917: more undefined references to `__stack_chk_fail_local' follow [12:05] collect2: ld returned 1 exit status [12:05] hm, that's strange -- glibc itself is shipping __stack_chk_fail() [12:05] so it should have _local as well... [12:06] Right, so I'm confused by this. [12:06] Also by the need to patch in libssp into gcc itself. [12:06] vacuum coming through, bbias.