Re: [qubes-devel] Kernel regression affecting VM network interface attach/detach operations

2018-03-04 Thread M. Vefa Bicakci

On 03/02/2018 01:51 PM, Simon Gaiser wrote:

Reverting this commit appears to resolve the regression for the time being. I
haven't had the time to formulate a proper correction.


Did you find out yet why this change is broken?


Unfortunately, no, I haven't had the time to investigate further.

Vefa

--
You received this message because you are subscribed to the Google Groups 
"qubes-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to qubes-devel+unsubscr...@googlegroups.com.
To post to this group, send email to qubes-devel@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/qubes-devel/1b7edd67-6038-6b5a-8363-49f2bbe3add4%40runbox.com.
For more options, visit https://groups.google.com/d/optout.


Re: [qubes-devel] Kernel regression affecting VM network interface attach/detach operations

2018-03-02 Thread Simon Gaiser
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

M. Vefa Bicakci:
> Hello all,
> 
> I noticed that the following commit causes a regression in QubesOS when
> a VM's network interface is detached and re-attached a few times in a row:
> 
>   5b5971df3bc2775107ddad164018a8a8db633b81
>   ("xen-netfront: remove warning when unloading module")
> 
> The regression exhibits itself via VM unresponsiveness to qrexec commands
> and heavier than usual CPU utilization involving the xenbus and xenwatch
> kernel threads in the VM whose network interface had been 
> detached/re-attached.
> 
> This commit has been back-ported to the kernel trees for versions 4.9.80 and
> 4.14.17, which is why QubesOS's testing kernel (based on 4.14.18) is also
> affected.

Thanks, for tracking this down! And sorry that I missed your report
until now (Otherwise I would already have included it in the latest
kernel PR).

> I am currently using QubesOS R3.2, but I believe this issue affects R4.0 as 
> well.

I'm observing this behavior on R4.0.

Will try if reverting fixes the problem for me.

> Reverting this commit appears to resolve the regression for the time being. I
> haven't had the time to formulate a proper correction.

Did you find out yet why this change is broken?
-BEGIN PGP SIGNATURE-
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=az44
-END PGP SIGNATURE-

-- 
You received this message because you are subscribed to the Google Groups 
"qubes-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to qubes-devel+unsubscr...@googlegroups.com.
To post to this group, send email to qubes-devel@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/qubes-devel/efaec423-a5da-68f7-7a16-8aed217193be%40invisiblethingslab.com.
For more options, visit https://groups.google.com/d/optout.