On Wed, Feb 10, 2021 at 12:25:31PM -0500, Mathieu Desnoyers wrote:
----- On Feb 10, 2021, at 12:09 PM, Greg Kroah-Hartman gregkh@xxxxxxxxxxxxxxxxxxx wrote:
On Wed, Feb 10, 2021 at 11:04:19AM -0500, Mathieu Desnoyers wrote:
Hi,
While reconciling the lttng-modules writeback instrumentation with its
counterpart
within the upstream Linux kernel, I notice that the following commit introduced
in
5.6 is present in stable branches 5.4 and 5.5, but is missing from LTS stable
branches
for 4.4, 4.9, 4.14, 4.19:
commit 68f23b89067fdf187763e75a56087550624fdbee
("memcg: fix a crash in wb_workfn when a device disappears")
Considering that this fix was CC'd to the stable mailing list, is there any
reason why it has not been integrated into those LTS branches ?
Yes, it doesn't apply at all. If you think this is needed, I will
gladly take backported and tested patches.
But why do you think this is needed in older kernels? Have you hit
this in real-life?
No, I have not hit this in real-life. Looking at the patch commit message,
the conditions needed to trigger this issue are very specific: memcg must
be enabled, and a device must be hotremoved while writeback is going on,
with writeback tracing active.
AFAIU memcg was present in those LTS releases and devices can be hotremoved
(please correct me if I'm wrong here), so all the preconditions appear to be
met.
Considering that I don't have the setup ready to reproduce this issue, I will
have to defer to the original patch authors for a properly tested backport,
if it happens to be relevant at all.
I just though reporting what appears to be a missing fix in LTS branches
would be the right thing to do.
Looks like it doesn't apply due to churn with tracepoints, I think it's
fixable. Let me try and get something for <=4.19.
--
Thanks,
Sasha
