Re: XenVif div by zero on Tx path after resume.

["Durrant, Paul" <xadimgnik@xxxxxxxxx>]

On 22/04/2022 13:21, Martin Harvey wrote:
[snip]
> So we are definitely in PdoSuspendCallbackLate, and we don't exit it, and we 
> don't get round to the VifCallbackLate.
>
> Even a cursory examination of the logs shows that xeniface is executing at a 
> lower Irql, amongst other things.
>
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539037:xen_platform_log xen platform: 
> xenvif|PdoSuspendCallbackLate: (PdoSuspendCallbackLate) MCHTRACE ====>
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539141:xen_platform_log xen platform: xenvif|__PdoD0ToD3: 
> (__PdoD0ToD3) MCHTRACE ====>
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539267:xen_platform_log xen platform: xenvif|FrontendSetState: 
> FrontendSetState:device/vif/0: MCHTRACE ====> 'CLOSED' -> 'CLOSED'
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539383:xen_platform_log xen platform: xenvif|FrontendSetState: 
> FrontendSetState:device/vif/0: MCHTRACE <=====
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539486:xen_platform_log xen platform: xenvif|__PdoD0ToD3: 
> (__PdoD0ToD3) MCHTRACE <====
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539591:xen_platform_log xen platform: xenvif|__PdoD3ToD0: 
> (__PdoD3ToD0) MCHTRACE ====>
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539717:xen_platform_log xen platform: xenvif|FrontendSetState: 
> FrontendSetState:device/vif/0: MCHTRACE ====> 'CLOSED' -> 'CONNECTED'
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539817:xen_platform_log xen platform: xeniface|IoctlLog: USER: 
> RTC is in local time
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.539933:xen_platform_log xen platform: 
> xeniface|FdoInitialiseXSRegistryEntries: About to convert unicode string
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.540125:xen_platform_log xen platform: 
> xeniface|FdoInitialiseXSRegistryEntries: About to write unicode string
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.540241:xen_platform_log xen platform: xeniface|IoctlLog: USER: 
> Time Now = 2022/4/20 13:26:07.483
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.540345:xen_platform_log xen platform: xeniface|IoctlLog: USER: 
> New Time = 2022/4/20 13:31:39.14
> Apr 20 13:34:24 xrtmia-13-06 qemu-dm-46[29574]: 
> 29574@1650461664.540449:xen_platform_log xen platform: 
> xenvif|TransmitterQueuePacket: BUG: !NT_SUCCESS(status)
>
> I'll start looking at IRQL violations (blocking from DISPATCH), as well as how 
> on earth the corral is supposed to work.
Yeah, something is screwed up there. It would be useful to annotate the 
log lines with which CPU they happen on. There's no way user-space 
should be running before all the suspend callbacks have completed.
  Paul