Re: [PATCH for-4.19 0/9] x86/irq: fixes for CPU hot{,un}plug

[Jan Beulich <jbeulich@xxxxxxxx>]

On 29.05.2024 11:37, Oleksii K. wrote:
> On Wed, 2024-05-29 at 11:01 +0200, Roger Pau Monne wrote:
>> Hello,
>>
>> The following series aim to fix interrupt handling when doing CPU
>> plug/unplug operations.  Without this series running:
>>
>> cpus=`xl info max_cpu_id`
>> while [ 1 ]; do
>>     for i in `seq 1 $cpus`; do
>>         xen-hptool cpu-offline $i;
>>         xen-hptool cpu-online $i;
>>     done
>> done
>>
>> Quite quickly results in interrupts getting lost and "No irq handler
>> for
>> vector" messages on the Xen console.  Drivers in dom0 also start
>> getting
>> interrupt timeouts and the system becomes unusable.
>>
>> After applying the series running the loop over night still result in
>> a
>> fully usable system, no  "No irq handler for vector" messages at all,
>> no
>> interrupt loses reported by dom0.  Test with
>> x2apic-mode={mixed,cluster}.
>>
>> I'm tagging this for 4.19 as it's IMO bugfixes, but the series has
>> grown
>> quite bigger than expected, and hence we need to be careful to not
>> introduce breakages late in the release cycle.  I've attempted to
>> document all code as good as I could, interrupt handling has some
>> unexpected corner cases that are hard to diagnose and reason about.
> Despite of the fact that it can be considered as bugfixes, it seems to
> me that this patch series can be risky. Let's wait for maintainers
> opinion...

Working my way through v2 of this series, I think I'd be okay with
including stuff there up to patch 5. Patch 6, which I just finished
taking a first look at, is likely correct (and it's just me missing some
aspects to fully grok the changes done there), but at the same time
looks to be more intrusive than we would like to have it at this point
of the release cycle. That said, I'd be pretty okay to be overridden in
this regard by Roger and/or Andrew.

Jan