[00:42] <mitchdz> Anyone seen this error before on arm64? Seeing this on optee-os and thought the newer binutils would fix it, but it's still happening with the newer binutils. aarch64-linux-gnu-ld.bfd: Error: unable to disambiguate: -specs=/usr/share/dpkg/elf-package-metadata.specs (did you mean --specs=/usr/share/dpkg/elf-package-metadata.specs ?)
[00:44] <sarnold> mitchdz: forgive me for asking, but is the - vs -- difference important?
[00:45] <mitchdz> Is the what important? sorry didn't follow
[00:46] <sarnold> mitchdz: the first one has a single hyphen and the second one has two hyphens
[00:46] <mitchdz> Yeah, not sure why that flag is becoming that
[00:46] <mitchdz> ah might need a new gcc too let me check if that's in proposed
[00:48] <mitchdz> Okay I'm on the newest gcc
[00:49] <mitchdz> that -specs=/usr/share/dpkg/elf-package-metadata.specs is showing up in the LDFLAGS of our build, but in the debian build system I'm seeing LDFLAGS=''
[00:52] <mitchdz> just not too sure where that's even coming from
[00:55] <sarnold> gcc is a challenge to grok :( heh
[01:02] <mitchdz> looking at gcc build failures makes me feel some sort of way
[01:02] <sarnold> lol yes
[01:02] <mitchdz> but on good news I see multipath-tools just migrated :)
[01:02] <sarnold> yay
[01:03] <sarnold> yeah i'm out of ideas on this one, sorry :(
[01:11] <mitchdz> maybe if I sleep it will come to me in the form of a dream
[01:11] <sarnold> or you'll invent benzine! either way
[01:37] <mwhudson> mitchdz: got a link to a build?
[01:38] <mwhudson> it sounds related to bdrung's recent changes for https://systemd.io/ELF_PACKAGE_METADATA/
[01:38] <mwhudson> possibly in combination with some evil hackes in d/rules, at a guess
[01:39] <mwhudson> "Strip -Wl, from LDFLAGS as optee_os calls ld directly." oh.
[01:42] <sarnold> or, if you sleep, mwhudson might show up with a pretty good idea :)
[01:45] <mwhudson> i guess you _could_ put the --package-metadata=JSON thingy into LDFLAGS directly but i'm sure shell quoting gremlins will eat your brain if you try that
[01:45] <sarnold> base64 encode all the things
[01:47] <mwhudson> sarnold: did you know base64 is bad actually? https://www.opswat.com/blog/how-base64-encoding-opens-the-door-for-malware
[01:47] <mwhudson> (do not read this page)
[01:48] <mwhudson> actually maybe that's not the really bad one i was looking for
[01:54] <mitchdz> no build link, this is a local build. I could upload to pastebin
[01:58] <mwhudson> well i think i can see what is going on by inspection
[01:58] <mwhudson> did the words above make any sense?
[01:59] <mwhudson> basically the code that munges LDFLAGS should probably strip out -specs= too
[02:00] <mwhudson> (i see it already filters out -f*, maybe it should _only_ keep things that start -Wl,? i don't nor do i want to speak GNU make very well)
[02:07] <mwhudson> https://paste.ubuntu.com/p/5kdw5szmN9/ maybe?
[02:07] <mwhudson> mitchdz: ^^
[02:07] <mwhudson> (completely untested)
[02:07] <mwhudson> afk for a bit now
[02:08] <mitchdz> I'll try that later tonight after I'm back :)
[03:57] <ginggs> @pilot in
=== ChanServ changed the topic of #ubuntu-devel to: Archive: open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Focal-Noble | Patch Pilots: ginggs
[04:30] <mitchdz> mwhudson: tried that and the substitution failed, might need a bit more of a tweak
[04:46] <mitchdz> hmm got it to build locally with this ugly line https://git.launchpad.net/~mitchdz/ubuntu/+source/optee-os/commit/?id=d71fa06f0f5e7ce3279f221e56eabfd7f091a844
[04:46] -ubottu:#ubuntu-devel- Commit d71fa06 in ~mitchdz/ubuntu/+source/optee-os "d/rules: also remove elf-package-metadata.specs from LDFLAGS"
[04:47] <mitchdz> Honestly not familiar with this /usr/share/dpkg/elf-package-metadata.specs file and if it's safe to just not link against
[05:08] <mitchdz> Cool it built. If someone more knowledgeable could give their $0.02 on if dropping the elf-package-metadata.specs is fine that'd be great
[05:09] <mitchdz> from what I'm reading it seems that it's _fine_ but could shoot ourselves in the foot when trying to debug a coredump
[05:13] <mwhudson> mitchdz: eh i guess that works but seems a bit specific
[05:13] <mwhudson> otoh GNU make is not a good programming language so
[05:15] <mwhudson> it's not create to skip the metadata but not the end of the world (aiui anyway)
[05:15] <mwhudson> uh
[05:15] <mwhudson> it's not *great to ...
[05:17] <mitchdz> That's my understanding too. It's nice to have a ready solution though :)
[05:17] <mitchdz> (even if it's not the best)
[09:26] <ginggs> @pilot out
=== ChanServ changed the topic of #ubuntu-devel to: Archive: open | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Focal-Noble | Patch Pilots: N/A
[09:26] <ginggs> Thank you for flying Ubuntu Air!
=== schopin_ is now known as schopin
[10:32] <sudip> LocutusOfBorg: is it ok if I do the gnuradio and synfig transitions later today.. just wanted to sync with you before you start with them..
[10:45] <LocutusOfBorg> sudip, ask also jbicha :)
[10:47] <LocutusOfBorg> I did slurm and miniupnpc
[10:47] <LocutusOfBorg> and fixing optee
[10:48] <sudip> I looked at miniupnpc, but that had one dependency in main so did not ask
[10:49]  * sudip will wait for an ack from jbicha before doing gnuradio and synfig today evening (after they are ready)
[10:56] <ginggs> sudip: i think j.bicha is on vac
[10:57] <LocutusOfBorg> sudip, just do them :D
[10:58] <sudip> ack, will do. thanks ginggs LocutusOfBorg
[11:06] <LocutusOfBorg> sudip, nobody forbids you to do transition and then ask just to rebuild main packages to some core-dev :)
[11:07] <LocutusOfBorg> at least I won't
[11:09] <sudip> LocutusOfBorg: its not about forbid. its about wasting time and effort if 2 person is doing the same. imho, just sending a line to you and j.bicha when I want to look at some transition will save that wasted effort :)
[12:19] <LocutusOfBorg> sudip, I watch transitions based on the html page, so if you do something, I just see in red what is missing
[12:20] <LocutusOfBorg> there is not huge risk of duplicating work
[13:28] <sudip> ok :)
=== kerbero is now known as Kerbero
=== Kerbero is now known as kerbero
[14:12] <flag> LocutusOfBorg: the last upload of bpfcc is completly missing on ppc64
[14:12] <mkukri_> flag LocutusOfBorg, bpftrace runtime_tests still failing on your latest upload (and blocking linux-meta)
[14:14] <mkukri_> and likely the same for llvm-toolchain-18
[14:28] <flag> mkukri_: ??? - bpftrace 0.21.0-1ubuntu6 passed all tests, but is blocked on bpfcc, not on testing
[14:29] <mkukri_> bpftrace/0.21.0-1ubuntu6 x linux-meta 6.10.0-15.15 is marked as a regression for me
[14:29] <mkukri_> https://objectstorage.prodstack5.canonical.com/swift/v1/AUTH_0f9aae918d5b4744bf7b827671c86842/autopkgtest-oracular/oracular/amd64/b/bpftrace/20240724_093853_fe7f9@/log.gz
[14:29] <mkukri_> there is a test that is only broken by the new kernel i believe
[14:30] <flag> mkukri_: flaky test, running runtime-tests manually against 6.10 everything passed
[14:38] <mkukri_> then please retry it until it passes
[14:42] <flag> mkukri_: already restarted
=== rkratky__ is now known as rkratky
[15:05] <mitchdz> LocutusOfBorg: You mentioned earlier that you are working on optee, I may have a fix for optee-os FTBFS
[15:11] <mitchdz> Hmmm
[15:12] <mitchdz> I was bothered about the new flags that we see in the optee-os build (-specs=/usr/share/dpkg/elf-package-metadata.specs), I'm not sure why/where they are from. I decided to build the current version in release (4.2.0) and it FTBFS for the same reason as the proposed package
[15:12] <mitchdz> so it looks like something on our end changed to cause this
[15:12] <mitchdz> any idea on what package change would include that flag?
[15:36] <LocutusOfBorg> mitchdz, I already changed and fixed?
[15:36] <LocutusOfBorg> dpkg is injecting them by default now
[15:36] <mitchdz> ah
[15:36] <LocutusOfBorg> and I removed the feature for it
[15:36] <LocutusOfBorg> check proposed :)
[15:36] <mitchdz> which package in proposed? dpkg
[15:37] <mitchdz> oh optee-os
[15:37] <mitchdz> ah darn I had this bug up for it :( https://bugs.launchpad.net/ubuntu/+source/optee-os/+bug/2074012
[15:37] -ubottu:#ubuntu-devel- Launchpad bug 2074012 in optee-os (Ubuntu) "arm64 build failure" [Undecided, New]
=== pushkarnk1 is now known as pushkarnk
[15:46] <mitchdz> LocutusOfBorg: did you see my bug linked in excuses page before you started optee?
[16:04] <LocutusOfBorg> mitchdz, no...
[16:04] <LocutusOfBorg> let me open
[16:05] <LocutusOfBorg> I'm uploading your change
[16:05] <mitchdz> Just curious what I can do to help make sure we don't do parallel work :)
[16:05] <LocutusOfBorg> mitchdz, I usually check excuses, but since this package was in an already open chrome tab, I didn't check again
[16:05] <LocutusOfBorg> my fault
[16:05] <mitchdz> no worries
[16:05] <LocutusOfBorg> I have like 30 tabs already open of packages
[16:05] <LocutusOfBorg> but I'm uploading your fix
[16:05] <mitchdz> hehe I feel that
[16:06] <mitchdz> Is that a good way to document I'm working on it though?
[16:06] <mitchdz> I realize we don't really have great documentation on how to handle that (at least IMO)
[16:06] <LocutusOfBorg> yes it should be indeed
[16:06] <LocutusOfBorg> but that update-excuses sometimes shows really old bugs, so people might not notice it
[16:06] <LocutusOfBorg> including me
[16:07] <mitchdz> Yeah, it definitely is a double edge sword
[16:07] <mitchdz> No hard feelings, glad to see we came to the same conclusion :)
[16:08] <LocutusOfBorg> maybe you can forward to Debian
[16:09] <LocutusOfBorg> basically this link script adds some metadata in the binary itself
[16:09] <LocutusOfBorg> https://bugs.launchpad.net/ubuntu/+source/doxygen/+bug/2071468
[16:09] -ubottu:#ubuntu-devel- Launchpad bug 2071468 in texinfo (Ubuntu) "ELF package metadata failure: environment variable ‘DEB_HOST_ARCH’ not defined" [High, Triaged]
[16:10] <mitchdz> Yeah I'll make a debian report once I get back
[16:10] <mitchdz> Even though they don't suffer from it right now
[16:11] <LocutusOfBorg> DEB_BUILD_OS_RELEASE_ID, DEB_HOST_ARCH,
[16:11] <LocutusOfBorg>     DEB_SOURCE, and DEB_VERSION
[17:41] <mitchdz> https://www.irccloud.com/pastebin/wmqNAQac/
[17:41] <mitchdz> whoops weird pasting
[17:42] <mitchdz> I'll just paste raw
[17:42] <mitchdz> before I go down this rabbit hole I'll ask quickly here, were there recent changes to case Cmake to do this?
[17:42] <mitchdz> -- Detecting C compiler ABI info - failed
[17:45] <mitchdz> (seeing this in pico-sdk)
[18:20] <jbicha> sudip: I'm returning from GUADEC today with very limited connectivity so I won't be duplicating any of your work :)
[18:21] <tarzeau> jbicha: hope you enjoyed, looking forward for things :)
[18:24] <sudip> those are still in pending publication stage, so please consider that an advance ping for tomorrow :)
[18:49] <sudip> if the builds of the dependencies fails in a transition then whats the usual process followed? do we wait for Debian to complete the transition or is it ok to add a delta now and sync/merge from Debian later ?
[19:07] <sarnold> mwhudson: ahh, img src=data:image/png;base64, my old friend! I once did a very hasty "copy image link" from firefox and pasted it into irssi over ssh (or mosh?) and then watched the next hour as a line of infinite length was sent to the remote server with no way to kill it except *killing* it
=== arif-ali_ is now known as arif-ali