Re: [PATCH] x86/hyper-v: guard against cpu mask changes in hyperv_flush_tlb_others()

From
Wei Liu <wei.liu@kernel.org>
Date
2021-08-22 17:32:32
Also in
lkml
Reply to
David Mozes, 2021-08-22
On Sun, Aug 22, 2021 at 04:25:19PM +0000, David Mozes wrote:
This is not visible since we need a very high load to reproduce. 
We have tried a lot but can't achieve the desired load 
On our kernel with less load, it is not reproducible as well.
There isn't much upstream can do if there is no way to reproduce the
issue with an upstream kernel.

You can check all the code paths which may modify cpumask and analyze
them. KCSAN may be useful too, but that's only available in 5.8 and
later.

Thanks,
Wei.
-----Original Message-----
From: Wei Liu <wei.liu@kernel.org> 
Sent: Sunday, August 22, 2021 6:25 PM
To: David Mozes <redacted>
Cc: David Moses <redacted>; Wei Liu <wei.liu@kernel.org>; Michael Kelley <redacted>; תומר אבוטבול <redacted>; <redacted>; <redacted>
Subject: Re: [PATCH] x86/hyper-v: guard against cpu mask changes in hyperv_flush_tlb_others()

On Thu, Aug 19, 2021 at 07:55:06AM +0000, David Mozes wrote:
quoted
Hi Wei ,
I move the print cpumask to other two places after the treatment on the empty mask see below
And I got the folwing:


Aug 19 02:01:51 c-node05 kernel: [25936.562674] Hyper-V: ERROR_HYPERV2: cpu_last=
Aug 19 02:01:51 c-node05 kernel: [25936.562686] WARNING: CPU: 11 PID: 56432 at arch/x86/include/asm/mshyperv.h:301 hyperv_flush_tlb_others+0x23f/0x7b0

So we got empty on different place on the code .
Let me know if you need further information from us.
How you sagest to handle this situation?
Please find a way to reproduce this issue with upstream kernels.

Thanks,
Wei.
Thread (3 messages) 3 messages, 2 authors, last reply 2021-08-22
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help