Re: [PATCH] drm/kms-helper: disable hpd_irq handling when poll=0

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, Apr 8, 2013 at 8:43 AM, Daniel Vetter <daniel.vetter@xxxxxxxx> wrote:
> On Mon, Apr 8, 2013 at 2:28 PM, Alex Deucher <alexdeucher@xxxxxxxxx> wrote:
>> On Sat, Apr 6, 2013 at 12:05 PM, Daniel Vetter <daniel.vetter@xxxxxxxx> wrote:
>>> When inlining the actual hpd output probing in
>>>
>>> commit 69787f7da6b2adc4054357a661aaa1701a9ca76f
>>> Author: Daniel Vetter <daniel.vetter@xxxxxxxx>
>>> Date:   Tue Oct 23 18:23:34 2012 +0000
>>>
>>>     drm: run the hpd irq event code directly
>>>
>>> the check for the drm_kms_hlper.poll module option was lost. This
>>> regressed systems where this option is used to work-around output
>>> probing issues (like irq storms). Restore the old behaviour.
>>
>> NAK.  It's been a while since I looked at it, but the whole point of
>> this patch set was to be able to disabling polling independently of
>> hpd.  If you add this check back, setting poll=0 disables both polling
>> and hpd.  I'd suggest we add a separate hpd option to disable hpd.
>
> The point for me was that the _driver_ can separate hpd handling from
> polling. Which it still can with this patch applied.
>
> The issue at hand is that the old poll=0 also managed to paper over
> some hpd handling irq storms and so breaking that is a regression. To
> fix that we should imo apply this patch.
>
> We can revert this again once i915 has the hpd irq storm rate-limiting
> stuff applied (presuming there's no other bug report for other
> drivers). Also in 3.9 we have the reworked kms locking, so the usual
> reason that polling causes cursor/pageflip stalls to set poll=0
> evaporated. So imo the only people who should still set poll=0 on 3.9
> are those with irq storms. And we've just broken that part ...

There are also a lot of people that have non-HPD capable ports
(usually old analog stuff like VGA) that they aren't using and they
don't want to waste the CPU cycles polling them, but they still want
hpd.  I expect we'll get regression reports about the poll option
getting broken again (hpd stops working with polling disabled, or
monitor detection no longer works) if this patch is applied.

Alex
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [Linux Kernel]     [Kernel Development Newbies]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite Hiking]     [Linux Kernel]     [Linux SCSI]