[PATCH v6 0/8] drm: zynqmp_dp: IRQ cleanups and debugfs support

[Sean Anderson]

On 10/2/24 10:50, Tomi Valkeinen wrote:
> Hi,
> 
> On 01/10/2024 21:31, Sean Anderson wrote:
>> On 8/9/24 15:35, Sean Anderson wrote:
>>> This series cleans up the zyqnmp_dp IRQ and locking situation. Once
>>> that's done, it adds debugfs support. The intent is to enable compliance
>>> testing or to help debug signal-integrity issues.
> 
> I think the patches 1-7 look fine, and I think I can pick those already to drm-misc if you're ok with that.
> 
> I'm a bit unsure about patch 8, probably mainly because I don't have experience with the compliance testing.
> 
> How have you tested this? With some DP analyzer/tester, I presume?

For my test setup I used an oscilloscope hooked up to the displayport
output using a fixture that broke the signals out to SMA. Since the
oscilloscope cannot emulate a sink, I first had the output connected to
a monitor. Then I disabled HPD and reconnected the output to my fixture.
This process is described in more detail in the documentation.

> I think none of this (patch 8) is needed by almost anybody.

Well, I found it very useful for debugging a signal integrity issue I
was having. Once I could have a look at the signals it was very clear
what the problem was.

> Even among zynqmp_dp developers I assume it's very rare to have the
> hardware for this. I wonder if it would make sense to have the debugfs
> and related code behind a compile option (which would be nice as the
> code wouldn't even compiled in), or maybe a module parameter (which
> would be nice as then "anyone" can easily enable it for compliance
> testing). What do you think?

Other drivers with these features just enabled it unconditionally, so I
didn't bother with any special config.

> I also somehow recall that there was some discussion earlier about
> how/if other drivers support compliance testing. But I can't find the
> discussion. Do you remember if there was such discussion, and what was
> the conclusion? With a quick look, everything in the debugfs looks
> generic, not xilinx specific.

The last it got discussed was back in [1], but I never got any further
response. I agree that some of this is generic, and could probably be
reworked into some internal helpers. But I don't have the bandwidth at
the moment to do that work.

--Sean

[1] http://lore.kernel.org/dri-devel/cda22b0c-8d7c-4ce2-9a7c-3b5ab540fa1f@linux.dev