Bug 1886109
Summary: | BUG: using smp_processor_id() in preemptible [00000000] code: handler106/3082 [rhel-rt-8.4.0] | |||
---|---|---|---|---|
Product: | Red Hat Enterprise Linux 8 | Reporter: | Juri Lelli <jlelli> | |
Component: | kernel-rt | Assignee: | Juri Lelli <jlelli> | |
kernel-rt sub component: | Networking | QA Contact: | Rona Gugliemino <rgugliem> | |
Status: | CLOSED ERRATA | Docs Contact: | ||
Severity: | urgent | |||
Priority: | unspecified | CC: | ailan, atheurer, atragler, bhu, brault, ctrautma, daolivei, dblack, dcain, echaudro, fbaudin, fhallal, fiezzi, fromani, jiji, kcarcia, kholtz, krister, kzhang, lcapitulino, mcornea, mniranja, msecaur, msimmons, mstowell, nilal, pezhang, rgugliem, trwilson, william.caban, williams, yjoseph | |
Version: | 8.4 | Keywords: | TestOnly, Triaged | |
Target Milestone: | rc | |||
Target Release: | 8.0 | |||
Hardware: | Unspecified | |||
OS: | Unspecified | |||
Whiteboard: | ||||
Fixed In Version: | kernel-rt-4.18.0-255.rt7.20.el8 | Doc Type: | If docs needed, set a value | |
Doc Text: | Story Points: | --- | ||
Clone Of: | ||||
: | 1888237 (view as bug list) | Environment: | ||
Last Closed: | 2021-05-18 15:12:49 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: | 1888237 | |||
Bug Blocks: | 1883636, 1898189 |
Description
Juri Lelli
2020-10-07 16:47:10 UTC
This is currently a blocker for the OCP Telco effort. It causes extreme latency spikes on our 4.6 test cluster. > Potentially introduced by upstream commit eac87c413bf97 ("net: openvswitch:
> reorder masks array based on usage")
This is a very important performance enhancement for OVS. If this is a problem, we need to find a way to keep this optimization if we can.
(In reply to Andrew Theurer from comment #2) > > Potentially introduced by upstream commit eac87c413bf97 ("net: openvswitch: > > reorder masks array based on usage") > > This is a very important performance enhancement for OVS. If this is a > problem, we need to find a way to keep this optimization if we can. I'm thinking this below might be acceptable --->8--- net: openvswitch: Fix using smp_processor_id() in preemptible code Signed-off-by: Juri Lelli <juri.lelli> --- net/openvswitch/flow_table.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c index aff660fe3a141..88b03cfdec9ec 100644 --- a/net/openvswitch/flow_table.c +++ b/net/openvswitch/flow_table.c @@ -706,11 +706,14 @@ static struct sw_flow *flow_lookup(struct flow_table *tbl, u32 *n_mask_hit, u32 *index) { - u64 *usage_counters = this_cpu_ptr(ma->masks_usage_cntr); + u64 *usage_counters; struct sw_flow *flow; struct sw_flow_mask *mask; int i; + get_cpu_light(); + usage_counters = this_cpu_ptr(ma->masks_usage_cntr); + if (likely(*index < ma->max)) { mask = rcu_dereference_ovsl(ma->masks[*index]); if (mask) { @@ -719,6 +722,7 @@ static struct sw_flow *flow_lookup(struct flow_table *tbl, u64_stats_update_begin(&ma->syncp); usage_counters[*index]++; u64_stats_update_end(&ma->syncp); + put_cpu_light(); return flow; } } @@ -739,10 +743,12 @@ static struct sw_flow *flow_lookup(struct flow_table *tbl, u64_stats_update_begin(&ma->syncp); usage_counters[*index]++; u64_stats_update_end(&ma->syncp); + put_cpu_light(); return flow; } } + put_cpu_light(); return NULL; } (In reply to Juri Lelli from comment #3) > (In reply to Andrew Theurer from comment #2) > > > Potentially introduced by upstream commit eac87c413bf97 ("net: openvswitch: > > > reorder masks array based on usage") > > > > This is a very important performance enhancement for OVS. If this is a > > problem, we need to find a way to keep this optimization if we can. > > I'm thinking this below might be acceptable > > --->8--- > net: openvswitch: Fix using smp_processor_id() in preemptible code > > Signed-off-by: Juri Lelli <juri.lelli> > --- > net/openvswitch/flow_table.c | 8 +++++++- > 1 file changed, 7 insertions(+), 1 deletion(-) > > diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c > index aff660fe3a141..88b03cfdec9ec 100644 > --- a/net/openvswitch/flow_table.c > +++ b/net/openvswitch/flow_table.c > @@ -706,11 +706,14 @@ static struct sw_flow *flow_lookup(struct flow_table > *tbl, > u32 *n_mask_hit, > u32 *index) > { > - u64 *usage_counters = this_cpu_ptr(ma->masks_usage_cntr); > + u64 *usage_counters; > struct sw_flow *flow; > struct sw_flow_mask *mask; > int i; > > + get_cpu_light(); > + usage_counters = this_cpu_ptr(ma->masks_usage_cntr); > + > if (likely(*index < ma->max)) { > mask = rcu_dereference_ovsl(ma->masks[*index]); > if (mask) { > @@ -719,6 +722,7 @@ static struct sw_flow *flow_lookup(struct flow_table > *tbl, > u64_stats_update_begin(&ma->syncp); > usage_counters[*index]++; > u64_stats_update_end(&ma->syncp); > + put_cpu_light(); > return flow; > } > } > @@ -739,10 +743,12 @@ static struct sw_flow *flow_lookup(struct flow_table > *tbl, > u64_stats_update_begin(&ma->syncp); > usage_counters[*index]++; > u64_stats_update_end(&ma->syncp); > + put_cpu_light(); > return flow; > } > } > > + put_cpu_light(); > return NULL; > } I have no experience with the RT kernel, but reading a bit into it, I guess the suggested change is ok. Thanks for having a look. Patch posted upstream: https://marc.info/?l=linux-rt-users&m=160224770202585&w=2 (In reply to Eelco Chaudron from comment #4) [...] > > I have no experience with the RT kernel, but reading a bit into it, I guess > the suggested change is ok. Interesting. Sebastian (on the upstream discussion) identified additional potential problems with the commit identified in comment#0: https://lore.kernel.org/lkml/20201009154116.a4fcrrm7flxonidd@linutronix.de/ Eelco, would you be up for following up on this? I will have a look as well, but I guess you'll be quicker to respond, as most probably more familiar with the code and its associated locking scheme. Thanks, Juri (In reply to Juri Lelli from comment #6) Replied upstream: https://lore.kernel.org/lkml/65BBD0B4-2A74-421A-BF81-357CD5F84747@redhat.com/ Looks like we could fix this with a general fix, see the following post: https://lore.kernel.org/netdev/160259304349.181017.7492443293310262978.stgit@ebuild/T/#u Juri can you build an RT kernel with this patch to make sure no more splats happen? RT 8.2 kernel build started: https://brewweb.engineering.redhat.com/brew/taskinfo?taskID=31946895 I had to slightly modify your change as 8.2 doesn't seem to have n_cache_hit variable. Also, I noticed that Sebastian had additional questions upstream (but I started the build anyway to try to buy us some time). BTW, as the eventual fix should apply to kernel (and reach RHEL-RT via RHEL), should I be changing the component to networking and assign the BZ to you? I will of course continue monitoring and helping with builds and testing. *** Bug 1887805 has been marked as a duplicate of this bug. *** (In reply to Juri Lelli from comment #9) > Also, I noticed that Sebastian had additional questions upstream (but I > started the > build anyway to try to buy us some time). Need some additional fix to solve the u64 sync issue, which is only related to 32-bit architectures. I will work on this today and send an updated patch upstream. > BTW, as the eventual fix should apply to kernel (and reach RHEL-RT via > RHEL), should > I be changing the component to networking and assign the BZ to you? I will > of course > continue monitoring and helping with builds and testing. Not sure how the RHEL-RT procedure goes. I think the easiest way is to create a clone of this BZ for RHEL where I will fix the issue and that this BZ depends on it. So you get notified when RT changes are needed? If this is all not necessary then assign the BZ to me. (In reply to Eelco Chaudron from comment #11) > (In reply to Juri Lelli from comment #9) > > Also, I noticed that Sebastian had additional questions upstream (but I > > started the > > build anyway to try to buy us some time). > > Need some additional fix to solve the u64 sync issue, which is only related > to 32-bit architectures. I will work on this today and send an updated patch > upstream. > > > BTW, as the eventual fix should apply to kernel (and reach RHEL-RT via > > RHEL), should > > I be changing the component to networking and assign the BZ to you? I will > > of course > > continue monitoring and helping with builds and testing. > > Not sure how the RHEL-RT procedure goes. I think the easiest way is to > create a clone of this BZ for RHEL where I will fix the issue and that this > BZ depends on it. So you get notified when RT changes are needed? If this is > all not necessary then assign the BZ to me. OK, thanks for driving this! I cloned the bug: https://bugzilla.redhat.com/show_bug.cgi?id=1888237 Unfortunately we can't use the kernel packages from brew in comment #9 because it has install deps which are not possible in RHCOS. The only packages installed for RHCOS for RT are: kernel-rt-core kernel-rt-modules kernel-rt-modules-extra kernel-rt-kvm These normally require other RPMs like: Resolving dependencies... done error: Could not depsolve transaction; 7 problems detected: Problem 1: conflicting requests - nothing provides hdparm needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-configobj needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-dbus needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-decorator needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-gobject-base needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-linux-procfs needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-perf needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-pyudev needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-schedutils needed by tuned-2.14.0-4.el8.noarch - nothing provides python3-syspurpose needed by tuned-2.14.0-4.el8.noarch - nothing provides virt-what needed by tuned-2.14.0-4.el8.noarch Problem 2: conflicting requests - nothing provides squashfs-tools needed by dracut-squash-049-53.git20191001.el8.x86_64 Problem 3: conflicting requests - nothing provides python3-ethtool needed by tuna-0.14-4.el8.noarch - nothing provides python3-linux-procfs >= 0.6 needed by tuna-0.14-4.el8.noarch - nothing provides python3-schedutils >= 0.6 needed by tuna-0.14-4.el8.noarch Problem 4: package tuned-profiles-realtime-2.14.0-4.el8.noarch requires tuna, but none of the providers can be installed - conflicting requests - nothing provides python3-ethtool needed by tuna-0.14-4.el8.noarch - nothing provides python3-linux-procfs >= 0.6 needed by tuna-0.14-4.el8.noarch - nothing provides python3-schedutils >= 0.6 needed by tuna-0.14-4.el8.noarch Problem 5: package rt-setup-2.1-2.el8.x86_64 requires tuna, but none of the providers can be installed - conflicting requests - nothing provides python3-ethtool needed by tuna-0.14-4.el8.noarch - nothing provides python3-linux-procfs >= 0.6 needed by tuna-0.14-4.el8.noarch - nothing provides python3-schedutils >= 0.6 needed by tuna-0.14-4.el8.noarch Problem 6: package kexec-tools-2.0.20-20.el8.x86_64 requires dracut-squash >= 049, but none of the providers can be installed - conflicting requests - nothing provides squashfs-tools needed by dracut-squash-049-53.git20191001.el8.x86_64 Problem 7: package kernel-rt-4.18.0-193.24.1.rt13.74.el8_2.1886109.test.cki.kt0.x86_64 requires rt-setup, but none of the providers can be installed - package rt-setup-2.1-2.el8.x86_64 requires tuna, but none of the providers can be installed - conflicting requests - nothing provides python3-ethtool needed by tuna-0.14-4.el8.noarch - nothing provides python3-linux-procfs >= 0.6 needed by tuna-0.14-4.el8.noarch - nothing provides python3-schedutils >= 0.6 needed by tuna-0.14-4.el8.noarch I suspect there is a slightly different build process for kernel-rt for RHCOS, which ensures packages like tuned, tuna, tuned-profiles-resltime, etc are not deps. Is anyone aware of this? (In reply to Andrew Theurer from comment #13) > Unfortunately we can't use the kernel packages from brew in comment #9 > because it has install deps which are not possible in RHCOS. The only > packages installed for RHCOS for RT are: > > kernel-rt-core > kernel-rt-modules > kernel-rt-modules-extra > kernel-rt-kvm > > These normally require other RPMs like: > > Resolving dependencies... done > error: Could not depsolve transaction; 7 problems detected: > Problem 1: conflicting requests > - nothing provides hdparm needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-configobj needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-dbus needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-decorator needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-gobject-base needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-linux-procfs needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-perf needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-pyudev needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-schedutils needed by tuned-2.14.0-4.el8.noarch > - nothing provides python3-syspurpose needed by tuned-2.14.0-4.el8.noarch > - nothing provides virt-what needed by tuned-2.14.0-4.el8.noarch > Problem 2: conflicting requests > - nothing provides squashfs-tools needed by > dracut-squash-049-53.git20191001.el8.x86_64 > Problem 3: conflicting requests > - nothing provides python3-ethtool needed by tuna-0.14-4.el8.noarch > - nothing provides python3-linux-procfs >= 0.6 needed by > tuna-0.14-4.el8.noarch > - nothing provides python3-schedutils >= 0.6 needed by > tuna-0.14-4.el8.noarch > Problem 4: package tuned-profiles-realtime-2.14.0-4.el8.noarch requires > tuna, but none of the providers can be installed > - conflicting requests > - nothing provides python3-ethtool needed by tuna-0.14-4.el8.noarch > - nothing provides python3-linux-procfs >= 0.6 needed by > tuna-0.14-4.el8.noarch > - nothing provides python3-schedutils >= 0.6 needed by > tuna-0.14-4.el8.noarch > Problem 5: package rt-setup-2.1-2.el8.x86_64 requires tuna, but none of the > providers can be installed > - conflicting requests > - nothing provides python3-ethtool needed by tuna-0.14-4.el8.noarch > - nothing provides python3-linux-procfs >= 0.6 needed by > tuna-0.14-4.el8.noarch > - nothing provides python3-schedutils >= 0.6 needed by > tuna-0.14-4.el8.noarch > Problem 6: package kexec-tools-2.0.20-20.el8.x86_64 requires dracut-squash > >= 049, but none of the providers can be installed > - conflicting requests > - nothing provides squashfs-tools needed by > dracut-squash-049-53.git20191001.el8.x86_64 > Problem 7: package > kernel-rt-4.18.0-193.24.1.rt13.74.el8_2.1886109.test.cki.kt0.x86_64 requires > rt-setup, but none of the providers can be installed > - package rt-setup-2.1-2.el8.x86_64 requires tuna, but none of the > providers can be installed > - conflicting requests > - nothing provides python3-ethtool needed by tuna-0.14-4.el8.noarch > - nothing provides python3-linux-procfs >= 0.6 needed by > tuna-0.14-4.el8.noarch > - nothing provides python3-schedutils >= 0.6 needed by > tuna-0.14-4.el8.noarch > > I suspect there is a slightly different build process for kernel-rt for > RHCOS, which ensures packages like tuned, tuna, tuned-profiles-resltime, etc > are not deps. Is anyone aware of this? I believe I successfully installed a scratch kernel-rt by doing something like the following in a worker node: $ sudo rpm-ostree reset $ sudo rpm-ostree override remove kernel{,-core,-modules,-modules-extra} --install http://brew-task-repos.usersys.redhat.com/repos/scratch/jlelli/kernel-rt/4.18.0/193.24.1.rt13.74.el8_2.1886109.test.cki.kt0/x86_64/kernel-rt-core-4.18.0-193.24.1.rt13.74.el8_2.1886109.test.cki.kt0.x86_64.rpm --install ... $ sudo systemctl reboot Not sure though if the other packages were installed already though. Also, I now see > kernel-rt-4.18.0-193.24.1.rt13.74.el8_2.1886109.test.cki.kt0.x86_64 requires ... Are trying to install kernel-rt-4.18.0-193.24.1.rt13.74.el8_2.1886109.test.cki.kt0.x86_64 as well (the "main" package)? We don't need that to test a new kernel and that one usually brings in the other deps. Using this kernel on OCP 4.6.0-rc.2 with downstream PAO, and stalld running with an interval of 30, oslat still has unacceptable latency (~47 us max) in even a short 10-minute test, but that is a huge improvement from the 17k us (stock kernel from PAO, stalld enabled) and 3k us (stock kernel from PAO, stalld disabled) latencies we had been seeing. uid: 0 **** allowed cpu list: 4-22 oslat 4.18.0-193.24.1.rt13.74.el8_2.1886109.test.cki.kt0.x86_64 Cloning into 'oslat'... cc -O2 -Wall -c -o main.o main.c cc -O2 -Wall -c -o rt-utils.o rt-utils.c cc -O2 -Wall -c -o error.o error.c cc -O2 -Wall -c -o trace.o trace.c cc -o oslat -lpthread -lnuma -lm main.o rt-utils.o error.o trace.o new cpu list: 5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22 cmd to run: oslat --runtime 600 --rtprio 1 --cpu-list 5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22 --cpu-main-thread 4 Version: v0.1.7 Total runtime: 600 seconds Thread priority: SCHED_FIFO:1 CPU list: 5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22 CPU for main thread: 4 Workload: no Workload mem: 0 (KiB) Preheat cores: 18 Pre-heat for 1 seconds... Test starts... Test completed. Core: 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 CPU Freq: 2590 2590 2590 2590 2590 2597 2590 2590 2590 2597 2590 2590 2590 2597 2590 2590 2590 2597 (Mhz) 001 (us): 23035708776 23032836891 23038885569 23040713165 23043529386 23052157548 23043740675 23041749321 23035458992 23052772087 23036229957 23055929040 23033220514 22754539032 23042830347 23040469655 23041587872 23040229132 002 (us): 557607 592739 590590 597392 594439 596796 594056 597734 594503 595716 595112 596917 592570 597233 594496 595650 595140 592852 003 (us): 43752 8379 10528 4120 6640 4314 6883 3392 7187 5551 5827 4617 8680 3843 7009 6077 6158 8480 004 (us): 518 287 252 347 299 352 314 327 341 350 349 335 351 389 91 349 104 154 005 (us): 150 89 92 75 80 86 93 92 76 90 76 87 94 80 49 74 38 62 006 (us): 41 30 32 32 38 24 40 23 35 28 39 26 27 22 23 37 18 29 007 (us): 10 6 11 5 18 7 13 3 17 5 16 7 5 4 7 13 7 5 008 (us): 1 0 8 1 1 2 3 4 6 1 1 2 2 1 3 1 1 2 009 (us): 1 0 1 1 3 5 2 1 0 1 0 0 0 0 0 1 0 1 010 (us): 0 1 4 4 5 1 1 0 0 0 0 0 0 0 1 0 0 0 011 (us): 3 1 1 4 0 0 0 1 0 0 0 0 0 0 0 0 0 2 012 (us): 1 4 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 013 (us): 3 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 014 (us): 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 015 (us): 0 1 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 016 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 017 (us): 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 018 (us): 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 019 (us): 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 020 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 021 (us): 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 022 (us): 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 023 (us): 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 024 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 025 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 026 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 027 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 1 1 028 (us): 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 029 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 030 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 031 (us): 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 032 (us): 0 0 0 0 0 0 0 0 0 0 0 1 0 0 10 0 10 1 (including overflows) Minimum: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 (us) Average: 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 (us) Maximum: 13 21 28 11 23 10 10 11 8 9 13 34 8 8 47 9 45 34 (us) Max-Min: 12 20 27 10 22 9 9 10 7 8 12 33 7 7 46 8 44 33 (us) Duration: 600.902 600.902 600.902 600.902 600.902 599.283 600.902 600.902 600.902 599.283 600.902 600.902 600.902 599.283 600.902 600.902 600.902 599.283 (sec) of_rules test passed after re-running 7 failed tests. First run: https://beaker.engineering.redhat.com/jobs/4891591 Re-run https://beaker.engineering.redhat.com/jobs/4893133 "nightly" vsperf gating ci test passed job: https://beaker.engineering.redhat.com/jobs/4891593 results: https://docs.google.com/spreadsheets/d/1J0xiaF9YxQTuJxh6z9FEQmDaIQrVZPmZgPD92sgxroU/edit#gid=984973794 A few issues that are being investigated: 1. no results seen for sr-iov part of test 2. rt-kernel is not installed on guest-vm. We are working on solving that problem currently. Running "gating" vsperf ci test over the weekend. Marking bz Verified because it's not certain "gating" and not solely "nightly" vsperf test is needed in order to verify fix. Hi, Is this fix going to be backported to RHEL 8.2? I have a customer who is requesting it. If there is no plan for that, I will make the request. Thanks! (In reply to Matthew Secaur from comment #42) > Hi, > > Is this fix going to be backported to RHEL 8.2? I have a customer who is > requesting it. If there is no plan for that, I will make the request. > > Thanks! Hi Matthew, Thank you for asking, because I don't think it was on our radar. I set the ZTR to 8.2.0, so we can evaluate it. Thanks, Beth Hi, Hasn't this been already backported to 8.2.z via bz1893282 and thus fixed in kernel-4.18.0-193.36.1.el8_2 and corresponding kernel-rt-4.18.0-193.36.1.rt13.86.el8_2 ? 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-2021:1739 |