Re: [PATCH v3 1/8] tests/functional: Add remote interrupts test for PowerNV

["Cédric Le Goater" <clg@redhat.com>]

On 6/30/26 13:04, Aditya Gupta wrote:
> 
> On 29/06/26 20:13, Cédric Le Goater wrote:
>> On 6/29/26 13:39, Aditya Gupta wrote:
>>> On 26/06/29 01:20PM, Cédric Le Goater wrote:
>>>> On 6/29/26 13:04, Aditya Gupta wrote:
>>>>> In the past there have been hard to recreate issues where XIVE changes
>>>>> cause qemu crashes due to multi-socket interrupts such as in [1].
>>>>>
>>>>> Add a functional test explicitly to test whether remote interrupts work.
>>>>>
>>>>> The test can also work as additional boot test for multi-socket boot,
>>>>> initrd boot test, as well as a check for e1000e to be working in powernv,
>>>>
>>>> you should consider moving to a more modern nic model: igb, which
>>>> has sr-iov support too.
>>>
>>> yes, can do. with igb i see multiple irqs, so test will have to take sum
>>> of all rx irqs, and compare:
>>>
>>>     2026-06-29 17:00:46,288: cat /proc/interrupts | grep eth0
>>>     2026-06-29 17:00:46,391: 135:          0          0 0 PNV-PCI-MSI 2416443392 Edge      eth0
>>>     2026-06-29 17:00:46,393: 136:          0          6 0 PNV-PCI-MSI 2416443393 Edge      eth0-rx-0
>>>     2026-06-29 17:00:46,394: 137:          0         14 0 PNV-PCI-MSI 2416443394 Edge      eth0-rx-1
>>>     2026-06-29 17:00:46,395: 138:          0          4 0 PNV-PCI-MSI 2416443395 Edge      eth0-rx-2
>>>     2026-06-29 17:00:46,397: 139:          0          4 0 PNV-PCI-MSI 2416443396 Edge      eth0-rx-3
>>>     2026-06-29 17:00:46,398: 140:          0         16 0 PNV-PCI-MSI 2416443397 Edge      eth0-tx-0
>>>     2026-06-29 17:00:46,399: 141:          0          6 0 PNV-PCI-MSI 2416443398 Edge      eth0-tx-1
>>>     2026-06-29 17:00:46,399: 142:          0          6 0 PNV-PCI-MSI 2416443399 Edge      eth0-tx-2
>>>     2026-06-29 17:00:46,400: 143:          0          4 0 PNV-PCI-MSI 2416443400 Edge      eth0-tx-3
>>>
>>> I feel e1000e suffices for this test, should i stay with e1000e or use
>>> igb here ?
>>
>> or use a single cpu per socket config. Which seems better for the test.
> 
> 
> The test uses single cpu per socket, since i have `-smp 4,sockets=4,threads=1`. Was this what you meant ?

yes. or cores=1, as you wish.

Thanks,

C.