| *** jude_ has quit IRC | 01:59 | |
| *** gtristan has joined #baserock | 03:00 | |
| *** noisecell has joined #baserock | 07:54 | |
| *** rdale has joined #baserock | 08:02 | |
| *** jonathanmaw has joined #baserock | 08:08 | |
| *** paulw has joined #baserock | 08:15 | |
| *** toscalix has joined #baserock | 08:20 | |
| *** ssam2 has joined #baserock | 09:01 | |
| *** ChanServ sets mode: +v ssam2 | 09:01 | |
| *** jude_ has joined #baserock | 09:28 | |
| *** locallycompact has joined #baserock | 09:35 | |
| *** locallycompact has quit IRC | 09:44 | |
| *** locallycompact has joined #baserock | 10:37 | |
| ssam2 | jonathanmaw, having merged some stuff from the sam/buildstream branch into my cleaned-up one, I'm now hitting the same ld.so issue even for x86_64 builds | 10:48 |
|---|---|---|
| gitlab-br-bot | ybd: merge request (benbrown/lockfiles->master: Fix file locking errors with higher number of instances) #371 changed state ("opened"): https://gitlab.com/baserock/ybd/merge_requests/371 | 11:38 |
| ssam2 | jonathanmaw, am just trying out this to see if it fixes the symlink issue ... https://pastebin.com/N6RtJTuV | 11:39 |
| jonathanmaw | ssam2: ditto | 11:39 |
| ssam2 | great minds :-) | 11:39 |
| jjardon | Hi | 11:41 |
| ssam2 | Hello!! | 11:41 |
| *** paulsherwood has joined #baserock | 11:42 | |
| jjardon | :) Any idea if network access can be disabled in the ybd sandbox ? | 11:42 |
| ssam2 | YBD has multiple sandbox backends | 11:42 |
| ssam2 | which one are you using? | 11:42 |
| jjardon | good question, one sec | 11:42 |
| paulsherwood | jjardon: it is already disabled? | 11:42 |
| ssam2 | I think linux-user-chroot makes this easy | 11:42 |
| ssam2 | but the chroot backend in sandbox doesn't have any support for controlling network access | 11:43 |
| ssam2 | that said, it doesn't copy in /etc/resolv.conf, so DNS won't work anyway | 11:43 |
| jjardon | ssam2: chroot | 11:53 |
| ssam2 | ok | 11:53 |
| ssam2 | i don't really know the best way to disable network access then | 11:53 |
| jjardon | paulsherwood: ok, so its always disable, its not a configuration option or something | 11:53 |
| paulsherwood | jjardon: i had thought so, but i may be wrong | 11:53 |
| ssam2 | if it's using the chroot backend, I don't think there's any code that does anything to do with network access ? it'll just do whatever the host system allows | 11:53 |
| jjardon | seems we tried linux-user-root before but it didnt work for some reason, maybe we should try again | 11:54 |
| ssam2 | but probably without DNS, since /etc/resolv.conf will be missing in the chroot | 11:54 |
| ssam2 | so in practice most network access would break | 11:54 |
| ssam2 | if something pulls from a hardcoded IP that would probably work though | 11:54 |
| jjardon | well, seems good enough :) | 11:54 |
| paulsherwood | jjardon: we had 'too many mounts' for cloud-basesd ybd running with l-u-c iirc | 11:56 |
| ssam2 | jonathanmaw, i've ended up with this patch, which at least builds stage2-glibc ... https://pastebin.com/US0Bq0c0 | 12:08 |
| ssam2 | waiting for further results | 12:08 |
| ssam2 | my previous one was broken in various ways :-) | 12:08 |
| gitlab-br-bot | ybd: issue #265 ("RPM deployment creates too many sandboxes") changed state ("opened") https://gitlab.com/baserock/ybd/issues/265 | 12:14 |
| gitlab-br-bot | definitions: merge request (sam/fixes-for-buildstream->master: Some fixes that came out of my BuildStream conversion branch) #51 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/51 | 12:47 |
| gitlab-br-bot | definitions: merge request (sam/fixes-for-buildstream->master: Some fixes that came out of my BuildStream conversion branch) #51 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/51 | 12:55 |
| *** lchlan has quit IRC | 12:56 | |
| *** lchlan has joined #baserock | 12:58 | |
| gitlab-br-bot | definitions: merge request (sam/fixes-for-buildstream->master: Some fixes that came out of my BuildStream conversion branch) #51 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/51 | 13:15 |
| jonathanmaw | ssam2: make's still failing for me, did you get past it on your run? | 13:32 |
| ssam2 | hmm | 13:49 |
| ssam2 | yes, but i wasn't doing a cross bootstrap | 13:50 |
| ssam2 | the `make` thing is weird though ... for me the failure was in linux-api-headers | 13:50 |
| ssam2 | which is earlier on.. | 13:50 |
| jonathanmaw | ssam2: still failing to find cc1 for me. should cc1 be in /usr/libexec, or is it going wrong somewhere else? | 13:56 |
| ssam2 | i think the location is corrent | 13:59 |
| ssam2 | *correct | 13:59 |
| ssam2 | on my machine cc1 is in /usr/libexec/gcc/x86_64-redhat-linux/7/cc1 for example | 13:59 |
| ssam2 | where is our GCC looking for it again ? | 13:59 |
| jonathanmaw | ssam2: my cc1 is at /usr/libexec/gcc/powerpc64le-unknown-linux-gnu/7.1.0/cc1, not sure how to figure out where gcc looks for it | 14:00 |
| ssam2 | so the location looks correct | 14:01 |
| ssam2 | but finding out what path GCC looks for it in is tricky. .. | 14:01 |
| ssam2 | not sure we can get strace in there, unless we can produce a statically compiled binary of it | 14:01 |
| ssam2 | what happens if you run `gcc -X/usr/libexec/gcc/powerpc64le-unknown-linux-gnu/7.1.0` to build a C program ? | 14:02 |
| ssam2 | oh, I mean -B I think | 14:02 |
| ssam2 | yeah -B | 14:02 |
| jonathanmaw | aha, gcc -print-search-dirs tells me some paths | 14:02 |
| ssam2 | aha! | 14:02 |
| jonathanmaw | looking for programs in =/tools/bin/../libexec/gcc/powerpc64le-unknown-linux-gnu/7.1.0/:/tools/bin/../libexec/gcc/:/tools/bin/../lib/gcc/powerpc64le-unknown-linux-gnu/7.1.0/../../../../powerpc64le-unknown-linux-gnu/bin/powerpc64le-unknown-linux-gnu/7.1.0/:/tools/bin/../lib/gcc/powerpc64le-unknown-linux-gnu/7.1.0/../../../../powerpc64le-unknown-linux-gnu/bin/ | 14:03 |
| jonathanmaw | so, problem is it's looking in /tools | 14:03 |
| ssam2 | right yeah | 14:03 |
| *** juergbi has quit IRC | 14:03 | |
| ssam2 | the final GCC should be looking in /usr | 14:03 |
| jonathanmaw | I'm assuming that PATH needs to include /tools/bin and /tools/sbin at this point | 14:06 |
| ssam2 | yeah | 14:10 |
| ssam2 | but PATH is separate from the GCC search dirs ... | 14:11 |
| ssam2 | --prefix is relevant, but I'm guessing we set --prefix=/usr | 14:11 |
| *** juergbi has joined #baserock | 14:12 | |
| ssam2 | nothing has set COMPILER_PATH in the environment, right ? | 14:15 |
| jonathanmaw | ssam2: no COMPILER_PATH | 14:18 |
| ssam2 | ok | 14:18 |
| jonathanmaw | prefix is not set, so AIUI it'll default to /usr | 14:18 |
| ssam2 | hmm, weird that it's not set though | 14:18 |
| ssam2 | in fact it usually defaults to /usr/local ... | 14:19 |
| jonathanmaw | ssam2: it's not set in the definitions | 14:19 |
| jonathanmaw | there's --prefix="%{prefix}" in the configure line, though | 14:19 |
| ssam2 | oh right | 14:19 |
| ssam2 | yes that's fine | 14:19 |
| ssam2 | although check that `bst source-bundle` has generated what we expect | 14:19 |
| ssam2 | maybe it expanded %{prefix} wrongly... (although then all the files wouldn't make it into /usr) | 14:20 |
| jonathanmaw | it does --prefix="/usr" | 14:20 |
| ssam2 | nice, ok | 14:20 |
| jonathanmaw | though there's an environment variable PREFIX, that it set to /buildstream/install | 14:20 |
| jonathanmaw | not sure what's going on there | 14:21 |
| ssam2 | no, that's a bit weird | 14:21 |
| ssam2 | this is relevant I think: http://git.baserock.org/cgit/delta/gcc-tarball.git/tree/gcc/Makefile.in?h=baserock/gcc-7.1.0#n2116 | 14:22 |
| ssam2 | the STANDARD_EXEC_PREFIX and STANDARD_LIBEXEC_PREFIX appear to go into `gcc -print-search-dirs` | 14:22 |
| ssam2 | and appear to be based on libdir and libexecdir ... so if those ended up with weird values, it might explain this | 14:22 |
| ssam2 | gcc/config.log from the GCC build tree should tell you what libdir and libexecdir got set to | 14:23 |
| jonathanmaw | ssam2: hrm, gcc got cleaned away by the bootstrap script. I'll try rebuilding gcc and hope it's guilty of the same problem. | 14:25 |
| gitlab-br-bot | definitions: merge request (staging/gnome-apps-legacy->master: gnome apps legacy) #32 changed state ("closed"): https://gitlab.com/baserock/definitions/merge_requests/32 | 14:30 |
| jonathanmaw | worse; new gcc has replaced old gcc, so I can no longer build gcc. that's going to be a nuisance to retry :/ | 14:30 |
| ssam2 | start from scratch I guess.. and prevent the script from deleting stuff | 14:30 |
| *** CTtpollard has quit IRC | 16:01 | |
| *** paulw has joined #baserock | 16:08 | |
| jonathanmaw | \o/ finally have the config.log from the gcc | 16:35 |
| jonathanmaw | unfortunately, it's 7000 lines long, so I'm going to be grepping heavily | 16:36 |
| *** jude_ has quit IRC | 16:37 | |
| ssam2 | the list of variables is usually the very last thing | 16:37 |
| *** jude_ has joined #baserock | 16:37 | |
| jonathanmaw | ssam2: about 500 lines of it, slightly more manageable | 16:38 |
| *** toscalix has quit IRC | 16:46 | |
| jonathanmaw | hrm, considering it's looking for cc1 in /tools, surprisingly the only output variable in that chunk that has /tools in it is "LD='/tools/lib/gcc/powerpc64le-bootstrap-linux-gnu/7.1.0/../../../../powerpc64le-bootstrap-linux-gnu/bin/ld'" | 16:47 |
| ssam2 | right | 16:48 |
| ssam2 | what value does it get for STANDARD_EXEC_PREFIX and STANDARD_LIBEXEC_PREFIX ? I guess /usr/lib and /usr/libexec ? | 16:48 |
| jonathanmaw | STANDARD_EXEC_PREFIX doesn't appear in the log, though there's exec_prefix, which is the value of ${prefix}, which is /usr | 16:50 |
| jonathanmaw | no libexec_prefix, but libexecdir is /usr/libexec, after unpacking all the substitutions | 16:50 |
| jonathanmaw | libdir is /usr/lib | 16:51 |
| ssam2 | OK | 16:52 |
| ssam2 | oh yeah, it's the gcc/Makefile.in that actually sets STANDARD_EXEC_PREFIX, and uses ${libdir} for that | 16:52 |
| ssam2 | ok, this is pretty weird then | 16:52 |
| ssam2 | it seems like the build system is doing exactly what it should | 16:52 |
| ssam2 | so it could be that we've triggered a bug in gcc ... | 16:52 |
| jonathanmaw | hrm, it also seems to be including from /tools, I've seen it include /tools/include/unistd.h. not sure if that means anything yet | 16:54 |
| ssam2 | that's fine, that's where the stage2 glibc is | 16:54 |
| jonathanmaw | apart from that, mentions of /tools are that I'm using an ld that comes from /tools, and that the gcc it was using to compile gcc used /tools | 16:55 |
| jonathanmaw | so that doesn't look out of the ordinary :S | 16:55 |
| ssam2 | yeah | 16:55 |
| ssam2 | maybe the thing to do will be to attach a debugger to `gcc -print-search-dirs` and trace through it, to work out how it gets so confused | 16:56 |
| ssam2 | i guess that means building gdb, but that should be possible to do manually | 16:56 |
| jonathanmaw | If I do it before the broken gcc is built, probably. | 16:59 |
| ssam2 | yeah | 16:59 |
| jonathanmaw | none of the other binaries seemed to be funted by it | 16:59 |
| *** locallycompact has quit IRC | 17:03 | |
| *** gtristan has quit IRC | 17:26 | |
| *** jonathanmaw has quit IRC | 17:29 | |
| *** noisecell has quit IRC | 17:36 | |
| ssam2 | gitlab is really messing me around tonight | 17:48 |
| ssam2 | this is the least fun | 17:50 |
| ssam2 | i just want to have a build running overnight | 17:50 |
| ssam2 | but instead it just ignores my branch, even when i push new commits | 17:50 |
| ssam2 | i give up, you win Gitlab, thanks for wasting the last hour | 17:51 |
| *** ssam2 has quit IRC | 17:51 | |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 17:51 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:00 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:04 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:04 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:05 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:05 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:06 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:06 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:09 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:09 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 18:14 |
| *** rdale has quit IRC | 18:18 | |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 19:05 |
| *** gtristan has joined #baserock | 19:07 | |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 19:19 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 19:33 |
| gitlab-br-bot | definitions: merge request (sam/buildstream-autoconvert->master: Add BuildStream conversion script and do ongoing BuildStream convert + build as part of the CI) #52 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/52 | 20:05 |
| *** gtristan has quit IRC | 23:59 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!