[Intel-gfx] [PATCH] drm/i915: Retry on every aux read.

[Daniel Vetter]

On Tue, Oct 20, 2015 at 10:02:21AM +0300, Jani Nikula wrote:
> On Tue, 20 Oct 2015, Rodrigo Vivi <rodrigo.vivi at intel.com> wrote:
> > We have an inconsistency on our code on using intel_dp_dpcd_read_wake with
> > retries and when using drm_dp_dpcd_read helper without retry.
> 
> We're supposed to do the retries when the sink may be in a power down
> state. We're not very good at tracking that, and we've cargo culted the
> retry variants all over the place without thinking. This is why we're
> inconsistent.
> 
> > Since the retries help in many cases let's be consistent and be on
> > the safe side retrying on every aux read, including i2c ones.
> 
> Please let's not add superfluous retries at different levels of the
> stack just to be safe. It's like a variant of the C programmer's
> disease.
> 
> If the retries really help [citation needed] we need to figure out what
> we're doing wrong and where, and preferrably fix this at the DP helper
> level for all drivers, if possible.

Yeah, same comment as I've done last time around. If we need this, we need
to do this in core dp helpers.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch