| *** gtristan has joined #baserock | 01:16 | |
| *** rdale_ct has joined #baserock | 02:31 | |
| *** rdale has quit IRC | 02:34 | |
| *** rdale_ct has quit IRC | 03:38 | |
| *** jude_ has joined #baserock | 06:58 | |
| *** ctbruce has joined #baserock | 07:27 | |
| *** paulw has joined #baserock | 07:56 | |
| *** toscalix has joined #baserock | 08:03 | |
| *** jonathanmaw has joined #baserock | 08:04 | |
| *** rdale has joined #baserock | 08:10 | |
| *** rdale_ct has joined #baserock | 08:42 | |
| *** rdale has quit IRC | 08:45 | |
| *** tiagogomes has joined #baserock | 08:49 | |
| *** ssam2 has joined #baserock | 08:51 | |
| *** ChanServ sets mode: +v ssam2 | 08:51 | |
| gitlab-br-bot | ybd: merge request (ps-kbas-cull->master: Add culling functionality to KBAS) #384 changed state ("opened"): https://gitlab.com/baserock/ybd/merge_requests/384 | 09:53 |
|---|---|---|
| *** noisecell has joined #baserock | 09:56 | |
| gitlab-br-bot | ybd: merge request (ps-kbas-cull->master: Add culling functionality to KBAS) #384 changed state ("opened"): https://gitlab.com/baserock/ybd/merge_requests/384 | 10:09 |
| *** locallycompact has joined #baserock | 10:13 | |
| *** locallycompact has quit IRC | 10:14 | |
| *** locallycompact has joined #baserock | 10:22 | |
| *** gtristan has quit IRC | 12:51 | |
| *** gtristan has joined #baserock | 13:36 | |
| ssam2 | jonathanmaw, the ld.so stuff is kinda fragile because... | 14:17 |
| ssam2 | glibc decides what it's going to call the ld.so file | 14:17 |
| ssam2 | gcc works out what its going to be called based on some hardcoded logic in gcc that duplicates the logic in glibc | 14:17 |
| ssam2 | and then we faff with it as well in the build instructions | 14:17 |
| ssam2 | here in particular: https://gitlab.com/baserock/definitions/blob/sam/buildstream/elements/gnu-toolchain/stage2-glibc.bst#L115 | 14:19 |
| ssam2 | that appeared to be working for me, but maybe something has changed somewhere | 14:20 |
| jonathanmaw | ssam2: ok, first red flag seems to be that there's no /lib64 in the sysroot | 14:22 |
| ssam2 | seems bad | 14:22 |
| jonathanmaw | no /ld64.so.1 at att :S | 14:23 |
| jonathanmaw | *all | 14:23 |
| ssam2 | i remember having to add something to fhs-dirs to create a /lib64 symlink matching the one we have for /lib | 14:23 |
| ssam2 | maybe that got lost | 14:23 |
| ssam2 | oh seems like it did | 14:24 |
| jonathanmaw | ssam2: ok, I'll give that a go | 14:26 |
| ssam2 | i've just pushed a patch to sam/buildstream that should fix things | 14:26 |
| ssam2 | found it in my `git reflog` | 14:26 |
| jonathanmaw | ssam2: ah, cool | 14:26 |
| ssam2 | not really sure how it got dropped from the branch .. probably a careless rebase | 14:26 |
| ssam2 | oh that will need extending to cover ppc64l and ppc64b | 14:27 |
| jonathanmaw | ssam2: does it take the same ARCH value in both cases? | 14:27 |
| ssam2 | i made it use arch conditionals | 14:27 |
| ssam2 | and no, the arch is different for LE and BE so there needs to be even more duplication | 14:28 |
| jonathanmaw | ssam2: ah, ok. | 14:28 |
| ssam2 | we should probably move this into the create-fhs-dirs script in fhs-dirs.git instead, which would allows us to use a glob pattern for the architecture... | 14:28 |
| persia | In this context, "arch" is really proxy for "ABI", so things like x86 vs x32 x 86_64 are also different. Better to have a big list than a composable field anyway. | 14:29 |
| ssam2 | i guess | 14:30 |
| ssam2 | it'd still be nice to avoid duplicating four identical lines for each 64-bit ABI, though | 14:30 |
| persia | Maybe a big conditional to set states, and then processing stanzas for each state? | 14:31 |
| persia | Some years ago I was involed in a project to create a new architecture that was x86 except without the CMOV instruction, for an example of the strange sorts of arhitectures people sometimes want. | 14:31 |
| persia | That aside, for architectures supporting multiple personalities (so needing /lib and /lib64), it ought be sensible to separate them. | 14:32 |
| persia | (as contrasted with native 64-bit architectures (e.g. Alpha), where there is no /lib64 because /lib is correct (as there is no non-64 option, ever). | 14:33 |
| jonathanmaw | ssam2: looking at linux-api-headers, ARCH should match one of the subdirs of "arch", and there's just "powerpc", no option for endianness | 14:47 |
| ssam2 | I think linux-api-headers is fine | 14:48 |
| ssam2 | the patch I pushed touches fhs-dirs and stage2-fhs-dirs | 14:48 |
| ssam2 | but only handles armv8[lb]64 and x86_64 | 14:48 |
| ssam2 | i guess i misunderstood your question befoire | 14:48 |
| ssam2 | *before | 14:48 |
| *** ctbruce has quit IRC | 16:22 | |
| *** jude_ has quit IRC | 16:34 | |
| *** toscalix has quit IRC | 16:50 | |
| *** locallycompact has quit IRC | 16:53 | |
| *** noisecell has quit IRC | 16:56 | |
| *** jonathanmaw has quit IRC | 17:01 | |
| *** jude_ has joined #baserock | 17:15 | |
| *** ssam2 has quit IRC | 17:27 | |
| gitlab-br-bot | ybd: merge request (ps-kbas-cull->master: Add culling functionality to KBAS) #384 changed state ("opened"): https://gitlab.com/baserock/ybd/merge_requests/384 | 18:24 |
| *** rdale_ct has quit IRC | 18:42 | |
| *** gtristan has quit IRC | 21:06 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!