Date: Mon, 29 Jan 2024 08:53:54 -0700 From: Warner Losh <imp@bsdimp.com> To: Guido Falsi <mad@madpilot.net> Cc: Nathan Reilly-list <lists@nreilly.com>, emulation@freebsd.org, "freebsd-arm@freebsd.org" <freebsd-arm@freebsd.org>, freebsd-pkg@freebsd.org Subject: Re: qemu-user-static aarch64 lockup/race? (was Re: Python failure in poudriere on arm64 (via qemu-user-static cross compiling)) Message-ID: <CANCZdfpUFYHeqkw7RdeyO=394PVfENfr8RbsG-SrrhAr8_2=Zw@mail.gmail.com> In-Reply-To: <990427ae-0491-463e-92c7-c74700deb6fa@madpilot.net> References: <6a33726b-eb6f-418e-9fbd-6d0b9b4bfaa8@madpilot.net> <0fc7f929-6e5b-4a33-97d2-8a9c0c07d524@madpilot.net> <79a5eb0f-d04e-4c1a-9d8a-185e1fb4e4a2@madpilot.net> <CANCZdfr-0w2EMa=_hFT3p4gFSDO-P1Yf8Vb-1eLiwRVomo1Jfg@mail.gmail.com> <a1845758-3535-4aa0-9274-d3b13dd3801b@madpilot.net> <5ef2ab66-25ef-45f1-aa5a-4b614eab2f40@madpilot.net> <D2DD631F-8AED-48B7-8FB3-86F93BA707F2@nreilly.com> <CANCZdfqELPcaCj-d%2BLj_qocR6gMiHp1RL1Y92myq=TnR-W6Y1w@mail.gmail.com> <e434f5a7-5730-498e-b54d-b01310f95f7a@madpilot.net> <990427ae-0491-463e-92c7-c74700deb6fa@madpilot.net>
index | next in thread | previous in thread | raw e-mail
[-- Attachment #1 --] On Mon, Jan 29, 2024, 8:48 AM Guido Falsi <mad@madpilot.net> wrote: > On 29/01/24 09:26, Guido Falsi wrote: > > On 29/01/24 02:10, Warner Losh wrote: > >> > >> > >> On Sun, Jan 28, 2024 at 4:45 PM Nathan Reilly-list <lists@nreilly.com > >> <mailto:lists@nreilly.com>> wrote: > >> > >> > >> > >>> On 29 Jan 2024, at 8:43 am, Guido Falsi <mad@madpilot.net > >>> <mailto:mad@madpilot.net>> wrote: > >>> On 28/01/24 22:34, Guido Falsi wrote: > >>>> On 28/01/24 22:23, Warner Losh wrote: > >>>>> On Sun, Jan 28, 2024, 12:38 PM Guido Falsi <mad@madpilot.net > >>>>> <mailto:mad@madpilot.net> <mailto:mad@madpilot.net > >>>>> <mailto:mad@madpilot.net>>> wrote: > >>>>> > >>>>> On 28/01/24 15:15, Guido Falsi wrote: > >>>>> [snip] > >>>>> > Creating repository in /tmp/packages: 0% > >>>>> > > >>>>> > >>>>> BTW, forgot to mention last time this worked without issue > >>>>> was around > >>>>> 20th December. > >>>>> > >>>>> > >>>>> I think this is a bsd-user issue. There is a race somewhere in > >>>>> that code that causes the hangs. I'd love a reproducible test > >>>>> case that is somewhat smaller than python... there are bigger > >>>>> races with the newer stuff and I've not had the time to chase it > >>>>> there either. 😞 > >>>> First of all thanks for your feedback. It encourages me having > >>>> someone else with better knowledge about this confirm that a race > >>>> condition is actually a possible cause! > >>>> Strange this has not been happening up to mid December. > >>>> My main and fully reproducible use case is actually mostly with > >>>> pkg. > >>>> at the end of the run poudriere runs `pkg repo` to create the > >>>> meta files and sign the repo. It forks itself (ncpus + 2 I guess, > >>>> even forcing it to 1 worker I see three processes), and then > >>>> locks up, with all the processes stopping using CPU (ps output is > >>>> in my message) > >>>> I guess this can be reproduced with any poudriere repo with at > >>>> least more than ncpus packages in it. can also be reproduced > >>>> using `poudriere pkgclean -u <etc>` > >>>> If that does not work I'm not sure how to reproduce it in other > >>>> ways, but I can try writing some code mocking what pkg seems to > >>>> be doing, not an expert at such things, though. > >>> > >>> In case it helps further norrow doen things, It looks like the > >>> lockup is happening somewhere around here: > >>> > >>> > >>> > https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L778 > < > https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L778 > > > >>> > >>> and/or in the pkg_create_repo_worker() function here: > >>> > >>> > >>> > https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L341 > < > https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L341 > > > >>> > >>> > >>> (I'm trying to spare you the time needed to find the actual code > >>> being executed, I guess you would have identified this in a few > >>> minutes yourself, but I'm trying to make myself useful) > >> > >> > >> There appears to be a GitHub issue for poudriere with this, but > >> seems to be looking in another direction. > >> > >> https://github.com/freebsd/poudriere/issues/1009 > >> <https://github.com/freebsd/poudriere/issues/1009>; > >> > > > > This one looks quite similar. > > > > In my case the ports/pkg are aligned between host and jail, in fact I > > have built them from the exact same git checkout. > > > > I noticed pkg head has been converted to using pthreads instead of fork, > > maybe that could help. I will make time to perform some testing. > > Thanks for pointing me here, it looks like this was "it", in that by > fixing this issue it uses native pkg-static, and sidesteps the issue. > > > Unluckily there ARE qemu races and lockups that prevent arm64 pkg-static > binary to be correctly emulated by qemu-user-static. such conditions > also cause sporadic failures in some ports being built. > > I filed a PR with a fix for that issue: > > https://github.com/freebsd/poudriere/pull/1115 Ok. This dodges the problem. But it papers over things. Any chance you could give me the state of pkg before + the package added as a test case for qemu? Warner > > -- > Guido Falsi <mad@madpilot.net> > > [-- Attachment #2 --] <div dir="auto"><div><br><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Jan 29, 2024, 8:48 AM Guido Falsi <<a href="mailto:mad@madpilot.net">mad@madpilot.net</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On 29/01/24 09:26, Guido Falsi wrote:<br> > On 29/01/24 02:10, Warner Losh wrote:<br> >><br> >><br> >> On Sun, Jan 28, 2024 at 4:45 PM Nathan Reilly-list <<a href="mailto:lists@nreilly.com" target="_blank" rel="noreferrer">lists@nreilly.com</a> <br> >> <mailto:<a href="mailto:lists@nreilly.com" target="_blank" rel="noreferrer">lists@nreilly.com</a>>> wrote:<br> >><br> >><br> >><br> >>> On 29 Jan 2024, at 8:43 am, Guido Falsi <<a href="mailto:mad@madpilot.net" target="_blank" rel="noreferrer">mad@madpilot.net</a><br> >>> <mailto:<a href="mailto:mad@madpilot.net" target="_blank" rel="noreferrer">mad@madpilot.net</a>>> wrote:<br> >>> On 28/01/24 22:34, Guido Falsi wrote:<br> >>>> On 28/01/24 22:23, Warner Losh wrote:<br> >>>>> On Sun, Jan 28, 2024, 12:38 PM Guido Falsi <<a href="mailto:mad@madpilot.net" target="_blank" rel="noreferrer">mad@madpilot.net</a><br> >>>>> <mailto:<a href="mailto:mad@madpilot.net" target="_blank" rel="noreferrer">mad@madpilot.net</a>> <mailto:<a href="mailto:mad@madpilot.net" target="_blank" rel="noreferrer">mad@madpilot.net</a><br> >>>>> <mailto:<a href="mailto:mad@madpilot.net" target="_blank" rel="noreferrer">mad@madpilot.net</a>>>> wrote:<br> >>>>><br> >>>>> On 28/01/24 15:15, Guido Falsi wrote:<br> >>>>> [snip]<br> >>>>> > Creating repository in /tmp/packages: 0%<br> >>>>> ><br> >>>>><br> >>>>> BTW, forgot to mention last time this worked without issue<br> >>>>> was around<br> >>>>> 20th December.<br> >>>>><br> >>>>><br> >>>>> I think this is a bsd-user issue. There is a race somewhere in<br> >>>>> that code that causes the hangs. I'd love a reproducible test<br> >>>>> case that is somewhat smaller than python... there are bigger<br> >>>>> races with the newer stuff and I've not had the time to chase it<br> >>>>> there either. 😞<br> >>>> First of all thanks for your feedback. It encourages me having<br> >>>> someone else with better knowledge about this confirm that a race<br> >>>> condition is actually a possible cause!<br> >>>> Strange this has not been happening up to mid December.<br> >>>> My main and fully reproducible use case is actually mostly with <br> >>>> pkg.<br> >>>> at the end of the run poudriere runs `pkg repo` to create the<br> >>>> meta files and sign the repo. It forks itself (ncpus + 2 I guess,<br> >>>> even forcing it to 1 worker I see three processes), and then<br> >>>> locks up, with all the processes stopping using CPU (ps output is<br> >>>> in my message)<br> >>>> I guess this can be reproduced with any poudriere repo with at<br> >>>> least more than ncpus packages in it. can also be reproduced<br> >>>> using `poudriere pkgclean -u <etc>`<br> >>>> If that does not work I'm not sure how to reproduce it in other<br> >>>> ways, but I can try writing some code mocking what pkg seems to<br> >>>> be doing, not an expert at such things, though.<br> >>><br> >>> In case it helps further norrow doen things, It looks like the<br> >>> lockup is happening somewhere around here:<br> >>><br> >>> <br> >>> <a href="https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L778" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L778</a>; <<a href="https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L778" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L778</a>><br>; >>><br> >>> and/or in the pkg_create_repo_worker() function here:<br> >>><br> >>> <br> >>> <a href="https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L341" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L341</a>; <<a href="https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L341" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/pkg/blob/56fa3f87d9d9644348b89680dfd8af47a860ee82/libpkg/pkg_repo_create.c#L341</a>><br>; >>><br> >>><br> >>> (I'm trying to spare you the time needed to find the actual code<br> >>> being executed, I guess you would have identified this in a few<br> >>> minutes yourself, but I'm trying to make myself useful)<br> >><br> >><br> >> There appears to be a GitHub issue for poudriere with this, but<br> >> seems to be looking in another direction.<br> >><br> >> <a href="https://github.com/freebsd/poudriere/issues/1009" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/poudriere/issues/1009</a><br>; >> <<a href="https://github.com/freebsd/poudriere/issues/1009" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/poudriere/issues/1009</a>><br>; >><br> > <br> > This one looks quite similar.<br> > <br> > In my case the ports/pkg are aligned between host and jail, in fact I <br> > have built them from the exact same git checkout.<br> > <br> > I noticed pkg head has been converted to using pthreads instead of fork, <br> > maybe that could help. I will make time to perform some testing.<br> <br> Thanks for pointing me here, it looks like this was "it", in that by <br> fixing this issue it uses native pkg-static, and sidesteps the issue.<br> <br> <br> Unluckily there ARE qemu races and lockups that prevent arm64 pkg-static <br> binary to be correctly emulated by qemu-user-static. such conditions <br> also cause sporadic failures in some ports being built.<br> <br> I filed a PR with a fix for that issue:<br> <br> <a href="https://github.com/freebsd/poudriere/pull/1115" rel="noreferrer noreferrer" target="_blank">https://github.com/freebsd/poudriere/pull/1115</a></blockquote></div></div><div dir="auto"><br></div><div dir="auto">Ok. This dodges the problem. But it papers over things.</div><div dir="auto"><br></div><div dir="auto">Any chance you could give me the state of pkg before + the package added as a test case for qemu?</div><div dir="auto"><br></div><div dir="auto">Warner</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br> <br> -- <br> Guido Falsi <<a href="mailto:mad@madpilot.net" target="_blank" rel="noreferrer">mad@madpilot.net</a>><br> <br> </blockquote></div></div></div>home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfpUFYHeqkw7RdeyO=394PVfENfr8RbsG-SrrhAr8_2=Zw>