On Mon, Aug 03, 2026 at 05:25:04PM -0700, Alison Schofield wrote:
> On Sat, Aug 01, 2026 at 02:30:24AM -0700, Anisa Su wrote:
> > On Fri, Jul 31, 2026 at 09:04:43AM +0000, [email protected] wrote:
> > > Thank you for your contribution! Sashiko AI review found 3 potential
> > > issue(s) to consider:
> > >
> > > New issues:
> > > - [High] Infinite IRQ Thread Loop / CPU Spin when handling DCD events
> > >
> > False Positive. DCD events are ignored because DCD is turned off
> > (mds->dcd_supported = false).
>
> Hi Anisa,
>
> I have related review feedback -
>
>
> snip
>
> > >
> > > [Severity: High]
> > > Since mask now includes CXLDEV_EVENT_STATUS_DCD, status can have this bit
> > > set.
> > > However, cxl_mem_get_event_records() does not appear to handle the DCD
> > > event:
> > >
> > The mask never includes CXLDEV_EVENT_STATUS_DCD.
> > Above in cxl_event_drain_mask:
> >
> > if (cxl_dcd_supported(mds))
> > mask |= CXLDEV_EVENT_STATUS_DCD;
> >
> > mds->dcd_supported is set to false in Patch 1 so the DCD bit is never
> > set.
> >
> > So status &= mask becomes zero and we break from the loop.
> >
> > - Anisa
>
> I understand that dcd_supported being forced false makes this unreachable
> today. My concern is that this patch adds the DCD bit to the drain path
> before there is code to consume and clear that log.
>
> As soon as a later patch enables dcd_supported, the bit can enter status
> and the handler can loop without clearing it. Could the DCD bit be added
> to the mask in the same patch that adds the DCD drain handling?
>
> -- Alison
Sure, that makes sense.
cxl_event_drain_mask() no longer sets the DCD bit, and it
loses the mds argument it needed for checking cxl_dcd_supported() to set it:
/* Event logs the driver drains: standard logs when native_cxl */
static u32 cxl_event_drain_mask(struct pci_host_bridge *host_bridge)
{
if (host_bridge->native_cxl_error)
return CXLDEV_EVENT_STATUS_ALL &
~CXLDEV_EVENT_STATUS_DCD;
return 0;
}
The bit is added back by "cxl/mem: Set up framework for handling DC
Events", the patch that adds the DCD case to
cxl_mem_get_event_records(), so the mask and its consumer arrive
together.
The commit message is updated to make it clear that the DCD event log
is not drained in this patch:
The DCD event log is not drained here. cxl_event_drain_mask() reports
only the logs the driver can service, and the DCD bit is added by
the patch introducing DCD event handling. Until then a DCD interrupt
wakes the event thread, which finds no log it owns and returns.
Thanks,
Anisa
