[Intel-gfx] [PATCH] drm: Only warn about an invalid EDID block prior to rejection

[Ville Syrjälä]

On Mon, Mar 09, 2015 at 03:39:01PM +0000, Chris Wilson wrote:
> On Mon, Mar 09, 2015 at 05:29:44PM +0200, Ville Syrjälä wrote:
> > On Mon, Mar 09, 2015 at 11:44:05AM +0000, Chris Wilson wrote:
> > > On a noisy link, we may retry the EDID reads multiple times per block
> > > and still succeed. In such cases, we don't want to flood the kernel logs
> > > with *ERROR* messages as we then succeed. Refactor the EDID dumping and
> > > push it into the caller rather than the validator so we can suppress the
> > > warnings from transient failures. In the process, we want to refactor
> > > drm_load_edid_firmware() to use the common drm_do_get_edid() to share
> > > the same logic (since it already duplicates it).
> > > 
> > > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > > References: https://bugs.freedesktop.org/show_bug.cgi?id=89454
> > 
> > Could we get it to dump the number of retries it had to make to get a
> > valid EDID? The retries could be due to driver bugs so we may want to
> > know that they in fact occured. Otherwise we might end up with eg.
> > large delays during resume without any indication what caused them.
> 
> We also log the number of times the edid was bad and the edid was zero.

I don't see where we log those. We do count them though.

> Knowing how many times we retried would be nice as well. Seems like it
> should be possible to output that to debugfs/.../connector/edid_info ?
> 
> What interface do you envisage being useful? I'm not convinced adding it
> to the user dmesg would be useful, but can a debugfs be integrated into
> whatever test you have in mind?

Well I just figure having it in the dmesg would be OK. That way when we
get a bug titled 'Resume takes forever' we can just ask for the dmesg as
usual, and we'd be able to see that something fishy was going on with
the EDID reads.

-- 
Ville Syrjälä
Intel OTC