Re: Recent tpm_tis IRQ handling changes are causing kernel backtraces

[Hans de Goede <hdegoede@redhat.com>]

Hi,

On 2/11/21 2:09 PM, Hans de Goede wrote:
> Hi Jerry,
> 
> It looks like there still is an issue with the recent changes to the tpm_tis IRQ
> handling. At least I think those are the cause I did not dive any deeper,
> I just noticed that we (Fedora) have been receiving an aweful lot of
> kernel tpm_tis_send_data backtraces with most starting with tpm_tis_probe_irq_...
> 
> See for example:
> https://bugzilla.redhat.com/show_bug.cgi?id=1912167
> https://bugzilla.redhat.com/show_bug.cgi?id=1927610
> 
> Those are just the 3 which landed in my inbox today, for much more see:
> https://bugzilla.redhat.com/buglist.cgi?quicksearch=tpm_tis_send_data
> (this shows 18 bugs atm).
> 
> These were reported through the Fedora ABRT tools which automatically
> collects backtraces, the bugs have links to the ABRT reports, e.g. :
> https://retrace.fedoraproject.org/faf/reports/28155/
> https://retrace.fedoraproject.org/faf/reports/37107/
> 
> The 28155 report says that so far there have been 308,412 (ouch) automatic
> uploads of that particular variant of these backtraces
> 
> Note the second (37107) retrace report is about this happening
> on resume, rather then on probe/tpm_tis_probe_irq_... time.
> 
> Did your work on this work land in 5.10 ? Or could it be that the
> issue is an incomplete backport to the 5.10.y stable series ?

Ping ?

It is raining bug-reports about this:

https://bugzilla.redhat.com/buglist.cgi?quicksearch=tpm_tis_send_data

Currently lists 25 bugs and that is excluding bugs which have already
been marked as a duplicate.

Can someone involved in the patch-series which is causing this regression
please take a look at these kernel backtraces ?

Regards,

Hans