[Intel-gfx] How does i915 CRT hotplug detection work?

[Bryan Freed]

Hi intel-gfx folks.

This issue is coming up for me again.  Can anyone comment on where I can go
for an answer?

bryan.

On Fri, Dec 10, 2010 at 10:15 AM, Bryan Freed <bfreed at chromium.org> wrote:

> The only additional information I have is that the irq occurs on a regular
> 6.47s interval.
>
> Can anyone comment on the mechanism?
>
> bryan.
>
>
> On Tue, Dec 7, 2010 at 5:08 PM, Bryan Freed <bfreed at chromium.org> wrote:
>
>> I have a pineview system with VGA output that takes anywhere from 1 to 5
>> seconds to detect an external monitor plug in/out, and I am trying to reduce
>> that latency.  But I do not know how hotplug detection works.
>>
>> My first sign of detection is in i915_driver_irq_handler() where we
>> read PORT_HOTPLUG_STAT (0x61114).  We quickly get
>> to i915_hotplug_work_func() and drm_sysfs_hotplug_event() after that, but
>> just getting to i915_driver_irq_handler() takes 1 to 5 seconds.
>>
>> My questions are 1) what takes so long to trigger the irq, and 2) what is
>> the irq mechanism?
>>
>> I first thought the i915 was polling over ddc to detect a monitor status
>> change.  But I do not see any activity when I probe DDC_DATA or DDC_CLK.
>> I have not probed while plugging in an actual monitor since that removes
>> my easy-to-probe point.
>>
>> Does i915 look for a glitch on one of the signals before it probes DDC?
>>
>> bryan.
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20110107/f8a9963c/attachment.html>