Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1457333 > unrolled thread

[PATCH 0/7] Minor DP aux transaction fixes

Started byLyude <cpaul@redhat.com>
First post2016-08-06 22:30 +0200
Last post2016-08-09 08:20 +0200
Articles 3 — 3 participants

Back to article view | Back to linux.kernel


Contents

  [PATCH 0/7] Minor DP aux transaction fixes Lyude <cpaul@redhat.com> - 2016-08-06 22:30 +0200
    Re: [PATCH 0/7] Minor DP aux transaction fixes Alex Deucher <alexdeucher@gmail.com> - 2016-08-08 19:40 +0200
      Re: [PATCH 0/7] Minor DP aux transaction fixes Daniel Vetter <daniel@ffwll.ch> - 2016-08-09 08:20 +0200

#1457333 — [PATCH 0/7] Minor DP aux transaction fixes

FromLyude <cpaul@redhat.com>
Date2016-08-06 22:30 +0200
Subject[PATCH 0/7] Minor DP aux transaction fixes
Message-ID<s3goh-263-5@gated-at.bofh.it>
While I was investigating an unrelated bug on the radeon driver, I noticed that
it's become rather difficult to actually read through dmesg with drm.debug
turned on, on account of the huge number of messages we end up printing from
failed DP aux transactions that happen every time we reprobe each connector.

Timed out transactions are relatively normal, and as well there's a lot of
places in radeon/amdgpu where we're printing redundant debugging information
dozens of times each time we attempt a DP aux transactions.

Additionally, I've removed some of the retry loops in amdgpu/radeon. These were
definitely useful at one point, but since we now retry any failed aux
transaction unconditionally in DRM's dp helpers they don't serve much purpose
other then to make failing aux transactions take a lot more time then they need
to.

Lyude (7):
  drm/dp_helper: Print first error received on failure in
    drm_dp_dpcd_access()
  drm/radeon: Don't print error on aux transaction timeouts
  drm/radeon: Don't retry 7 times in radeon_dp_dpcd()
  drm/amdgpu: Don't print error on aux transaction timeouts
  drm/amdgpu: Don't retry 7 times in amdgpu_atombios_dp_get_dpcd()
  drm: Add ratelimited versions of the DRM_DEBUG* macros
  drm/dp_helper: Rate limit timeout errors from drm_dp_i2c_do_msg()

 drivers/gpu/drm/amd/amdgpu/atombios_dp.c | 22 ++++++++++------------
 drivers/gpu/drm/drm_dp_helper.c          | 14 ++++++++++++--
 drivers/gpu/drm/radeon/atombios_dp.c     | 21 ++++++++++-----------
 drivers/gpu/drm/radeon/radeon_dp_auxch.c |  1 -
 include/drm/drmP.h                       | 30 ++++++++++++++++++++++++++++++
 5 files changed, 62 insertions(+), 26 deletions(-)

-- 
2.7.4

[toc] | [next] | [standalone]


#1457947

FromAlex Deucher <alexdeucher@gmail.com>
Date2016-08-08 19:40 +0200
Message-ID<s3WGR-4qo-7@gated-at.bofh.it>
In reply to#1457333
On Fri, Aug 5, 2016 at 8:30 PM, Lyude <cpaul@redhat.com> wrote:
> While I was investigating an unrelated bug on the radeon driver, I noticed that
> it's become rather difficult to actually read through dmesg with drm.debug
> turned on, on account of the huge number of messages we end up printing from
> failed DP aux transactions that happen every time we reprobe each connector.
>
> Timed out transactions are relatively normal, and as well there's a lot of
> places in radeon/amdgpu where we're printing redundant debugging information
> dozens of times each time we attempt a DP aux transactions.
>
> Additionally, I've removed some of the retry loops in amdgpu/radeon. These were
> definitely useful at one point, but since we now retry any failed aux
> transaction unconditionally in DRM's dp helpers they don't serve much purpose
> other then to make failing aux transactions take a lot more time then they need
> to.

I've applied the amdgpu and radeon patches.  For the drm patches, I
can either take them through my tree or via drm-misc.

Alex

>
> Lyude (7):
>   drm/dp_helper: Print first error received on failure in
>     drm_dp_dpcd_access()
>   drm/radeon: Don't print error on aux transaction timeouts
>   drm/radeon: Don't retry 7 times in radeon_dp_dpcd()
>   drm/amdgpu: Don't print error on aux transaction timeouts
>   drm/amdgpu: Don't retry 7 times in amdgpu_atombios_dp_get_dpcd()
>   drm: Add ratelimited versions of the DRM_DEBUG* macros
>   drm/dp_helper: Rate limit timeout errors from drm_dp_i2c_do_msg()
>
>  drivers/gpu/drm/amd/amdgpu/atombios_dp.c | 22 ++++++++++------------
>  drivers/gpu/drm/drm_dp_helper.c          | 14 ++++++++++++--
>  drivers/gpu/drm/radeon/atombios_dp.c     | 21 ++++++++++-----------
>  drivers/gpu/drm/radeon/radeon_dp_auxch.c |  1 -
>  include/drm/drmP.h                       | 30 ++++++++++++++++++++++++++++++
>  5 files changed, 62 insertions(+), 26 deletions(-)
>
> --
> 2.7.4
>
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

[toc] | [prev] | [next] | [standalone]


#1458429

FromDaniel Vetter <daniel@ffwll.ch>
Date2016-08-09 08:20 +0200
Message-ID<s48yl-3Q0-9@gated-at.bofh.it>
In reply to#1457947
On Mon, Aug 08, 2016 at 01:30:37PM -0400, Alex Deucher wrote:
> On Fri, Aug 5, 2016 at 8:30 PM, Lyude <cpaul@redhat.com> wrote:
> > While I was investigating an unrelated bug on the radeon driver, I noticed that
> > it's become rather difficult to actually read through dmesg with drm.debug
> > turned on, on account of the huge number of messages we end up printing from
> > failed DP aux transactions that happen every time we reprobe each connector.
> >
> > Timed out transactions are relatively normal, and as well there's a lot of
> > places in radeon/amdgpu where we're printing redundant debugging information
> > dozens of times each time we attempt a DP aux transactions.
> >
> > Additionally, I've removed some of the retry loops in amdgpu/radeon. These were
> > definitely useful at one point, but since we now retry any failed aux
> > transaction unconditionally in DRM's dp helpers they don't serve much purpose
> > other then to make failing aux transactions take a lot more time then they need
> > to.
> 
> I've applied the amdgpu and radeon patches.  For the drm patches, I
> can either take them through my tree or via drm-misc.

I applied the 2 core patches to drm-misc, thanks.
-Daniel

> 
> Alex
> 
> >
> > Lyude (7):
> >   drm/dp_helper: Print first error received on failure in
> >     drm_dp_dpcd_access()
> >   drm/radeon: Don't print error on aux transaction timeouts
> >   drm/radeon: Don't retry 7 times in radeon_dp_dpcd()
> >   drm/amdgpu: Don't print error on aux transaction timeouts
> >   drm/amdgpu: Don't retry 7 times in amdgpu_atombios_dp_get_dpcd()
> >   drm: Add ratelimited versions of the DRM_DEBUG* macros
> >   drm/dp_helper: Rate limit timeout errors from drm_dp_i2c_do_msg()
> >
> >  drivers/gpu/drm/amd/amdgpu/atombios_dp.c | 22 ++++++++++------------
> >  drivers/gpu/drm/drm_dp_helper.c          | 14 ++++++++++++--
> >  drivers/gpu/drm/radeon/atombios_dp.c     | 21 ++++++++++-----------
> >  drivers/gpu/drm/radeon/radeon_dp_auxch.c |  1 -
> >  include/drm/drmP.h                       | 30 ++++++++++++++++++++++++++++++
> >  5 files changed, 62 insertions(+), 26 deletions(-)
> >
> > --
> > 2.7.4
> >
> > _______________________________________________
> > dri-devel mailing list
> > dri-devel@lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/dri-devel
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web