Bug 1903578
| Summary: | kernnel-rt-debug: do not call blocking ops when !TASK_RUNNING; state=1 set at [<0000000050e86018>] handle_userfault+0x530/0x1820 | |||
|---|---|---|---|---|
| Product: | Red Hat Enterprise Linux 8 | Reporter: | Chunyu Hu <chuhu> | |
| Component: | kernel-rt | Assignee: | Andrew Halaney <ahalaney> | |
| kernel-rt sub component: | Memory Management | QA Contact: | Chunyu Hu <chuhu> | |
| Status: | CLOSED ERRATA | Docs Contact: | ||
| Severity: | low | |||
| Priority: | low | CC: | ahalaney, bhu, jlelli, kcarcia, liwan, mm-maint, mstowell, pifang, rt-maint, rt-qe, williams | |
| Version: | 8.4 | Keywords: | Triaged, ZStream | |
| Target Milestone: | rc | |||
| Target Release: | 8.4 | |||
| Hardware: | x86_64 | |||
| OS: | Linux | |||
| Whiteboard: | ||||
| Fixed In Version: | kernel-rt-4.18.0-343.rt7.125.el8 | Doc Type: | If docs needed, set a value | |
| Doc Text: | Story Points: | --- | ||
| Clone Of: | ||||
| : | 2029420 2029421 2029422 (view as bug list) | Environment: | ||
| Last Closed: | 2022-05-10 14:41:26 UTC | Type: | Bug | |
| Regression: | --- | Mount Type: | --- | |
| Documentation: | --- | CRM: | ||
| Verified Versions: | Category: | --- | ||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | ||
| Cloudforms Team: | --- | Target Upstream Version: | ||
| Embargoed: | ||||
| Bug Depends On: | ||||
| Bug Blocks: | 2020013, 2029420, 2029421, 2029422 | |||
| Deadline: | 2021-09-13 | |||
|
Description
Chunyu Hu
2020-12-02 12:23:18 UTC
This is only seen on kernel-rt. This can be reproduced on upstream as well (will report). Also, it can be reproduced on all 8.x RT versions. Guess we never noticed until now because the test is actually a PASS. FYI, I think this is a false warning and I've posted a patch upstream to address it: https://lore.kernel.org/lkml/20210406221952.50399-1-ahalaney@redhat.com/ tglx said this would land in the next RT release, will wait until I see it there to backport: https://lore.kernel.org/lkml/877dkoud19.ffs@nanos.tec.linutronix.de/ Sorry for the delay -- I've been slowly trying to get this to land somewhere upstream since the last comment. linux-rt-devel overhauled their rwsem implementation for RT and thus tglx never picked up the patch after that above thread. I've been poking the stable maintainers offline to get it landed there, and Steven has included it in a RC today. Once that officially releases I'll go ahead and post a patch. Thanks for being patient. https://lore.kernel.org/linux-rt-users/20210820234737.244832083@goodmis.org/ Landed upstream in the 5.10 RT stable branch: https://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-stable-rt.git/commit/?h=v5.10-rt&id=b2ed0a4302faf2bb09e97529dd274233c082689b Too close to 8.5 beta to target that. Setting ITR to 8.6 for that reason. It sounds like 8.6 might use the new RT patchset, in which case this won't ever land on 8.6 and will have to target 8.5 directly. Until that happens though I'm assuming this needs to land in 8.6. Also, this affects 8.2-rt - 8.5-rt as well. I don't think I'm supposed to set ZTR until this is verified though, so holding off on that. Premature, but here's the brew build for the debug kernel: http://brew-task-repos.usersys.redhat.com/repos/scratch/ahalaney/kernel-rt/4.18.0/337.rt7.118.el8bz1903578v1/ I've posted the MR but I'm going to leave this in the ASSIGNED state until 8.5 branches off. Once main-rt targets the 8.6 release I'll change it to MODIFIED. Please let me know if anyone has objections to that strategy. MR: https://gitlab.com/redhat/rhel/src/kernel/rhel-8/-/merge_requests/1249 (In reply to Andrew Halaney from comment #9) > Premature, but here's the brew build for the debug kernel: > http://brew-task-repos.usersys.redhat.com/repos/scratch/ahalaney/kernel-rt/4. > 18.0/337.rt7.118.el8bz1903578v1/ This is tested, result is good. :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: /kernel/general/memory/function/userfaultfd2 :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: [ 21:54:45 ] :: [ LOG ] :: JOURNAL XML: /var/tmp/beakerlib-REi9TkB/journal.xml :: [ 21:54:45 ] :: [ LOG ] :: JOURNAL TXT: /var/tmp/beakerlib-REi9TkB/journal.txt :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :: Duration: 213s :: Phases: 3 good, 0 bad :: OVERALL RESULT: PASS (/kernel/general/memory/function/userfaultfd2) [root@ibm-x3650m4-05 userfaultfd2]# [root@ibm-x3650m4-05 userfaultfd2]# dmesg [ 507.204710] runtest.sh (1898): drop_caches: 3 [root@ibm-x3650m4-05 userfaultfd2]# dmesg [ 507.204710] runtest.sh (1898): drop_caches: 3 [root@ibm-x3650m4-05 userfaultfd2]# uname -r 4.18.0-337.rt7.118.el8bz1903578v1.x86_64+debug [root@ibm-x3650m4-05 userfaultfd2]# dmesg [ 507.204710] runtest.sh (1898): drop_caches: 3 > > I've posted the MR but I'm going to leave this in the ASSIGNED state until > 8.5 branches off. Once main-rt targets the 8.6 release I'll change it to > MODIFIED. Please let me know if anyone has objections to that strategy. The ‘Devel target Milestone' is '2', not sure if we got enough reviews/acks for the MR, is time to move to MODIFIED? as it's ITM2 already. > > MR: https://gitlab.com/redhat/rhel/src/kernel/rhel-8/-/merge_requests/1249 (In reply to Chunyu Hu from comment #10) > (In reply to Andrew Halaney from comment #9) > > I've posted the MR but I'm going to leave this in the ASSIGNED state until > > 8.5 branches off. Once main-rt targets the 8.6 release I'll change it to > > MODIFIED. Please let me know if anyone has objections to that strategy. > > The ‘Devel target Milestone' is '2', not sure if we got enough reviews/acks > for the > MR, is time to move to MODIFIED? as it's ITM2 already. Ah, my apologies @chuhu, the rhel-8 kernel doesn't have a place to land 8.6-rt changes yet but that should happen today. I didn't realize that would be the case (I assumed by DTM 2 there would be somewhere to land the patch). I'm moving to DTM 3 to give buffer room for review after the branching happens today, can you please confirm that ITM 5 is ok and if not adjust? Thanks and sorry! (In reply to Andrew Halaney from comment #11) > (In reply to Chunyu Hu from comment #10) > > (In reply to Andrew Halaney from comment #9) > > > I've posted the MR but I'm going to leave this in the ASSIGNED state until > > > 8.5 branches off. Once main-rt targets the 8.6 release I'll change it to > > > MODIFIED. Please let me know if anyone has objections to that strategy. > > > > The ‘Devel target Milestone' is '2', not sure if we got enough reviews/acks > > for the > > MR, is time to move to MODIFIED? as it's ITM2 already. > > Ah, my apologies @chuhu, the rhel-8 kernel doesn't have a place to > land 8.6-rt changes yet but that should happen today. > > I didn't realize that would be the case (I assumed by DTM 2 there would be > somewhere to land the patch). > I'm moving to DTM 3 to give buffer room for review after the branching > happens today, > can you please confirm that ITM 5 is ok and if not adjust? Thanks and sorry! It works for me for ITM-5. Thanks! The MR is there and bot has already attached the build, and if the workflow-bot can notify when the MR gets enough reviews/acks, that would be great. You are right, the MR is now targeting the correct branch for 8.6-rt release. The MR has enough reviews/acks now (I guess the bot doesn't post that), so all that should be left is adding "Verified: Tested" when appropriate (not sure if you want to take the MR artifacts for a spin or if you trust my brew artifacts) and a maintainer to merge. Thanks! (In reply to Andrew Halaney from comment #14) > You are right, the MR is now targeting the correct branch for 8.6-rt release. > > The MR has enough reviews/acks now (I guess the bot doesn't post that), > so all that should be left is adding "Verified: Tested" when > appropriate (not sure if you want to take the MR artifacts for > a spin or if you trust my brew artifacts) and a maintainer to merge. > Thanks! I trust the brew build, set 'Tested', there's no debug kernel build in the MR. Thanks！ Hi Andrew, When are we going to add this into candidate/official kernel-rt build? The target ITM for this is ITM-5, which is from Sep-28 to Oct 4. Maybe we need to defer this for several ITMs? if that's the case, I'll adjust the ITM field of the bz. Thanks! Regards, Chunyu Hu Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Important: kernel-rt security and bug fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2022:1975 |