| ahasenack_ | sudip: hi, quick question about yhour MPs | 15:13 |
|---|---|---|
| ahasenack_ | they are showing up in our server-only report at https://code.launchpad.net/~canonical-server-reporter/+activereviews | 15:13 |
| ahasenack_ | have you requested those server slots, or was it git ubuntu submit, or something else? | 15:13 |
| ahasenack_ | for example, https://code.launchpad.net/~sudipmuk/ubuntu/+source/python-django-compressor/+git/python-django-compressor/+merge/475376 has slots for canonical-server-reporter, and git-ubuntu import | 15:13 |
| sudip | ahasenack_: no, I have not requested anything specifically, I noticed sponsors were not added, so just added them | 15:14 |
| ahasenack_ | sudip: how did you create the mps, do you remember? | 15:14 |
| ahasenack_ | was it via git-ubuntu submit, or manually? | 15:14 |
| ahasenack_ | as in, you just accepted the normal default reviewers? | 15:14 |
| sudip | https://github.com/canonical/ubuntu-maintainers-handbook/blob/main/PackageMerging.md | 15:14 |
| sudip | "git ubuntu submit --target-branch debian/sid" | 15:15 |
| ahasenack_ | cool, thanks | 15:15 |
| ahasenack_ | mkukri: hi, still around? | 18:07 |
| ahasenack_ | mkukri: I think the s390x optimization patches you added to zlib in noble introduced a bug: https://bugs.launchpad.net/ubuntu/+source/rsyslog/+bug/2083700 | 18:08 |
| -ubottu:#ubuntu-devel- Launchpad bug 2083700 in zlib (Ubuntu Plucky) "rsyslog FTBFS (s390x only) against zlib 1:1.3.dfsg+really1.3.1-1ubuntu1" [Undecided, Confirmed] | 18:08 | |
| mkukri | that's weird cause they were in a previous version of zlib, i just updated the layout of them | 18:09 |
| mkukri | the previous zlib was an ugly mess, but also had a large set of s390x and ppc64el optimizations | 18:10 |
| ahasenack_ | I did see good rsyslog builds with that version of zlib, but now they are failing | 18:10 |
| ahasenack_ | I wonder if the optimization never kicked in before because of lack of cpu support? | 18:10 |
| mkukri | hmm im not sure, but this zlib upload has been in noble for a really long time | 18:10 |
| ahasenack_ | I noticed that in a PPA the test always fails, but in a canonistack s390x vm it passes, and the cpu flags are different | 18:10 |
| ahasenack_ | -features : esan3 zarch stfle msa ldisp eimm dfp edat etf3eh highgprs te vx | 18:10 |
| ahasenack_ | +features : esan3 zarch stfle msa ldisp eimm dfp edat etf3eh highgprs te vx vxd vxe gs vxe2 vxp sort dflt | 18:10 |
| ahasenack_ | cpus with the longer list is where the test fails | 18:10 |
| ahasenack_ | the smaller list is what I get in a canonistack s390x vm, for example | 18:11 |
| mkukri | maybe the old zlib would fail as well on the new cpus? | 18:11 |
| ahasenack_ | and if we drop the patches, the test passes (there are two: can't tell yet if both need to be dropped or just one) | 18:11 |
| ahasenack_ | mkukri: maybe, but this is being caught by an rsyslog test, btw | 18:11 |
| ahasenack_ | how old are we talking? | 18:11 |
| mkukri | but the patches were cherry picked from an IBM developers github (links should be in the dep-3 metadata), so maybe we need to tell IBM | 18:12 |
| ahasenack_ | doing that in parallel | 18:12 |
| mkukri | https://launchpad.net/ubuntu/+source/zlib/1:1.3.dfsg-3ubuntu1 | 18:12 |
| ahasenack_ | my interest here is because of rsyslog, I need to update it in noble with some fixes, but it fails to build on s390x because of this | 18:12 |
| mkukri | i think this is my first zlib upload with the patches | 18:12 |
| mkukri | so nov 2023 | 18:12 |
| ahasenack_ | right, 1:1.3.dfsg-3ubuntu1 is when they are first mentioned. You said the patches were there before? | 18:13 |
| ahasenack_ | ah, in different files | 18:13 |
| mkukri | under different names yeah | 18:13 |
| ahasenack_ | 410.patch, 478.patch, etc | 18:13 |
| mkukri | the changelog is a bit confusing in retrospect | 18:13 |
| ahasenack_ | ok, I thought they were new in noble | 18:13 |
| ahasenack_ | the reorg is what happened in noble then | 18:14 |
| mkukri | yes. and i also updated them from the author's github, but they looked more or less identical | 18:14 |
| mkukri | and it was also rebased on a newer zlib at the same time, so there is potential for breakage | 18:14 |
| ahasenack_ | ok, so a) the reorg changed something and introduced the bug; or b) the cpus used by these builders were old and now are new, actually supporting what the patch add | 18:15 |
| ahasenack_ | adds* | 18:15 |
| mkukri | yeah it's one of those. unfortunately i dont have easy access to s390x machines to debug this | 18:15 |
| ahasenack_ | I have to a vm that does NOT show the bug :) | 18:16 |
| ahasenack_ | maybe frank/ibm can help | 18:16 |
| mkukri | the patches come from IBM, so yeah that's probably the best thing to do | 18:16 |
| mkukri | and i think i was told at some point that we were asked by IBM to add them originally | 18:17 |
| ahasenack_ | from what I read, the performance improvement is quite good, so that makes sense | 18:17 |
| ahasenack_ | too bad they are not upstream | 18:18 |
| mkukri | i think they have a PR trying to upstream (some) of them at least | 18:18 |
| mkukri | https://github.com/madler/zlib/pull/410 | 18:18 |
| mkukri | https://github.com/madler/zlib/pull/478 | 18:18 |
| -ubottu:#ubuntu-devel- Pull 410 in madler/zlib "Add support for IBM Z hardware-accelerated deflate" [Open] | 18:18 | |
| ahasenack_ | let me see if they changed in since nov2023 in that iii-i github repo | 18:18 |
| -ubottu:#ubuntu-devel- Pull 478 in madler/zlib "Add optimized crc32 for Power 8+ processors" [Open] | 18:18 | |
| ahasenack_ | mkukri: I got confirmation that the builders' s390x hardware indeed change | 18:32 |
| ahasenack_ | the issue is being forwarded to ibm | 18:33 |
| === pushkarnk1 is now known as pushkarnk | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!