Skip to contentSkip to conversation

Increase LSPCON timeout

PATCH Superseded Fredrik Schön 7 participants 16 messages Aug 13, 2018 – Aug 16, 2018

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Increase LSPCON timeout

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

---
 drivers/gpu/drm/i915/intel_lspcon.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_lspcon.c +1 -1
index 8ae8f42f430a..be1b08f589a4 100644
@@ -74,7 +74,7 @@ static enum drm_lspcon_mode lspcon_wait_mode(struct intel_lspcon *lspcon,
 	DRM_DEBUG_KMS("Waiting for LSPCON mode %s to settle\n",
 		      lspcon_mode_name(mode));
 
-	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 100);
+	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 1000);
 	if (current_mode != mode)
 		DRM_ERROR("LSPCON mode hasn't settled\n");
 
Signature · 6 lines
2.17.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

✓ Fi.CI.BAT: success for Increase LSPCON timeout

== Series Details ==

== Series Details ==

Series: Increase LSPCON timeout
URL : https://patchwork.freedesktop.org/series/48183/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4665 -> Patchwork_9937 =

== Summary - SUCCESS ==

No regressions found.

External URL: https://patchwork.freedesktop.org/api/1.0/series/48183/revisions/1/mbox/

== Known issues ==

Here are the changes found in Patchwork_9937 that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@debugfs_test@read_all_entries:
      {fi-icl-u}:         PASS -> DMESG-FAIL (fdo#107411)

    igt@drv_selftest@live_hangcheck:
      fi-kbl-guc:         PASS -> INCOMPLETE (fdo#106693, fdo#107207)

    igt@gem_exec_reloc@basic-gtt-read-noreloc:
      {fi-icl-u}:         PASS -> DMESG-WARN (fdo#107411) +77

    igt@kms_frontbuffer_tracking@basic:
      {fi-byt-clapper}:   PASS -> FAIL (fdo#103167)

    igt@kms_pipe_crc_basic@nonblocking-crc-pipe-b:
      {fi-byt-clapper}:   PASS -> FAIL (fdo#107362)

    
    ==== Possible fixes ====

    igt@debugfs_test@read_all_entries:
      fi-snb-2520m:       INCOMPLETE (fdo#103713) -> PASS

    igt@drv_selftest@live_hangcheck:
      fi-skl-guc:         DMESG-FAIL (fdo#107174) -> PASS

    igt@drv_selftest@live_workarounds:
      fi-cnl-psr:         DMESG-FAIL (fdo#107292) -> PASS
      fi-skl-6700k2:      DMESG-FAIL (fdo#107292) -> PASS

    igt@kms_chamelium@dp-crc-fast:
      fi-kbl-7500u:       DMESG-FAIL (fdo#103841) -> PASS

    igt@kms_pipe_crc_basic@suspend-read-crc-pipe-c:
      fi-bxt-dsi:         INCOMPLETE (fdo#103927) -> PASS

    {igt@kms_psr@primary_mmap_gtt}:
      fi-cnl-psr:         DMESG-WARN (fdo#107372) -> PASS

{name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).

fdo#103167 https://bugs.freedesktop.org/show_bug.cgi?id=103167
fdo#103713 https://bugs.freedesktop.org/show_bug.cgi?id=103713
fdo#103841 https://bugs.freedesktop.org/show_bug.cgi?id=103841
fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927
fdo#106693 https://bugs.freedesktop.org/show_bug.cgi?id=106693
fdo#107174 https://bugs.freedesktop.org/show_bug.cgi?id=107174
fdo#107207 https://bugs.freedesktop.org/show_bug.cgi?id=107207
fdo#107292 https://bugs.freedesktop.org/show_bug.cgi?id=107292
fdo#107362 https://bugs.freedesktop.org/show_bug.cgi?id=107362
fdo#107372 https://bugs.freedesktop.org/show_bug.cgi?id=107372
fdo#107411 https://bugs.freedesktop.org/show_bug.cgi?id=107411

== Participating hosts (54 -> 49) ==

Missing (5): fi-ctg-p8600 fi-ilk-m540 fi-byt-squawks fi-bsw-cyan fi-hsw-4200u

== Build changes ==

* Linux: CI_DRM_4665 -> Patchwork_9937

CI_DRM_4665: 3a43bf5efa33be9f88d70427dbfba4db4377eaf6 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4594: b0263e5d0563a81a42cf66e7d3b84662d3222862 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_9937: 977c85e40247ce7d9dc570748f6bbea3473bad75 @ git://anongit.freedesktop.org/gfx-ci/linux

== Linux commits ==

977c85e40247 Increase LSPCON timeout

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_9937/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#2 lore raw

✓ Fi.CI.IGT: success for Increase LSPCON timeout

In reply to Fredrik Schön
== Series Details ==

== Series Details ==

Series: Increase LSPCON timeout
URL : https://patchwork.freedesktop.org/series/48183/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4665_full -> Patchwork_9937_full =

== Summary - SUCCESS ==

No regressions found.

== Known issues ==

Here are the changes found in Patchwork_9937_full that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@gem_workarounds@suspend-resume:
      shard-glk:          NOTRUN -> FAIL (fdo#103375)

    igt@kms_setmode@basic:
      shard-apl:          PASS -> FAIL (fdo#99912)

    
    ==== Possible fixes ====

    igt@drv_suspend@shrink:
      shard-glk:          INCOMPLETE (fdo#106886, k.org#198133, fdo#103359) -> PASS

    igt@gem_workarounds@suspend-resume:
      shard-apl:          FAIL (fdo#103375) -> PASS

fdo#103359 https://bugs.freedesktop.org/show_bug.cgi?id=103359
fdo#103375 https://bugs.freedesktop.org/show_bug.cgi?id=103375
fdo#106886 https://bugs.freedesktop.org/show_bug.cgi?id=106886
fdo#99912 https://bugs.freedesktop.org/show_bug.cgi?id=99912
k.org#198133 https://bugzilla.kernel.org/show_bug.cgi?id=198133

== Participating hosts (5 -> 5) ==

No changes in participating hosts

== Build changes ==

* Linux: CI_DRM_4665 -> Patchwork_9937

CI_DRM_4665: 3a43bf5efa33be9f88d70427dbfba4db4377eaf6 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4594: b0263e5d0563a81a42cf66e7d3b84662d3222862 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_9937: 977c85e40247ce7d9dc570748f6bbea3473bad75 @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4509: fdc5a4ca11124ab8413c7988896eec4c97336694 @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_9937/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

maintainer Reviewed-by on v1 1/1
#3 lore raw
In reply to Fredrik Schön

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 6 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode to settle")
Cc Shashank Sharma <[email protected]>
Cc Imre Deak <[email protected]>
Cc Jani Nikula <[email protected]>
Cc <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

Quoted 25 lines
---
 drivers/gpu/drm/i915/intel_lspcon.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_lspcon.c +1 -1
index 8ae8f42f430a..be1b08f589a4 100644
@@ -74,7 +74,7 @@ static enum drm_lspcon_mode lspcon_wait_mode(struct intel_lspcon *lspcon,
 	DRM_DEBUG_KMS("Waiting for LSPCON mode %s to settle\n",
 		      lspcon_mode_name(mode));
 
-	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 100);
+	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 1000);
 	if (current_mode != mode)
 		DRM_ERROR("LSPCON mode hasn't settled\n");
 
Signature · 6 lines
2.17.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

maintainer
#4 lore raw
Quoted 30 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 6 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode to settle")
Cc Shashank Sharma <[email protected]>
Cc Imre Deak <[email protected]>
Cc Jani Nikula <[email protected]>
Cc <[email protected]> # v4.11+
Quoted 7 lines

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

now, really cc'ing them...

Quoted 22 lines
Quoted 22 lines
---
 drivers/gpu/drm/i915/intel_lspcon.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_lspcon.c +1 -1
index 8ae8f42f430a..be1b08f589a4 100644
@@ -74,7 +74,7 @@ static enum drm_lspcon_mode lspcon_wait_mode(struct intel_lspcon *lspcon,
 	DRM_DEBUG_KMS("Waiting for LSPCON mode %s to settle\n",
 		      lspcon_mode_name(mode));
 
-	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 100);
+	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 1000);
 	if (current_mode != mode)
 		DRM_ERROR("LSPCON mode hasn't settled\n");
 
Signature · 1 line
2.17.1
Quoted 8 lines
Quoted 4 lines

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#5 lore raw
Quoted 34 lines · Rodrigo Vivi
Quoted 33 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 6 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode to settle")
Cc Shashank Sharma <[email protected]>
Cc Imre Deak <[email protected]>
Cc Jani Nikula <[email protected]>
Cc <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

Quoted 6 lines
Quoted 4 lines

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

BR,
Jani.

Quoted 31 lines
Quoted 31 lines
Quoted 27 lines
---
 drivers/gpu/drm/i915/intel_lspcon.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_lspcon.c +1 -1
index 8ae8f42f430a..be1b08f589a4 100644
@@ -74,7 +74,7 @@ static enum drm_lspcon_mode lspcon_wait_mode(struct intel_lspcon *lspcon,
 	DRM_DEBUG_KMS("Waiting for LSPCON mode %s to settle\n",
 		      lspcon_mode_name(mode));
 
-	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 100);
+	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 1000);
 	if (current_mode != mode)
 		DRM_ERROR("LSPCON mode hasn't settled\n");
 
Signature · 6 lines
2.17.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Signature · 5 lines
Jani Nikula, Intel Open Source Graphics Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#6 lore raw

Regards

Shashank

Quoted 34 lines · Jani Nikula
Quoted 33 lines · Rodrigo Vivi
Quoted 32 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 6 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode to settle")
Cc Shashank Sharma <[email protected]>
Cc Imre Deak <[email protected]>
Cc Jani Nikula <[email protected]>
Cc <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

Quoted 6 lines
Quoted 5 lines
Quoted 4 lines

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

This will slow down the lspcon probing and resume part, but both of them happen only when LSPCON device is found. So to answer your question, this will not slow down the non-lspcon path, but will slow down the LSPCON connector resume and probe time. but I would recommend, instead of increasing it to 1000 ms in a single shot, we might want to gradually pick this up, on a wake-and-check way.

- Shashank

Quoted 34 lines

BR,
Jani.

Quoted 29 lines
Quoted 29 lines
Quoted 25 lines
---
  drivers/gpu/drm/i915/intel_lspcon.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_lspcon.c +1 -1
index 8ae8f42f430a..be1b08f589a4 100644
@@ -74,7 +74,7 @@ static enum drm_lspcon_mode lspcon_wait_mode(struct intel_lspcon *lspcon,
  	DRM_DEBUG_KMS("Waiting for LSPCON mode %s to settle\n",
  		      lspcon_mode_name(mode));
 
-	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 100);
+	wait_for((current_mode = lspcon_get_current_mode(lspcon)) == mode, 1000);
  	if (current_mode != mode)
  		DRM_ERROR("LSPCON mode hasn't settled\n");
 
Signature · 6 lines
2.17.1

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#7 lore raw

Quoting Sharma, Shashank (2018-08-16 08:33:36)

Quoted 44 lines

Regards

Shashank

Quoted 38 lines · Jani Nikula
Quoted 37 lines · Rodrigo Vivi
Quoted 36 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 6 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode to settle")
Cc Shashank Sharma <[email protected]>
Cc Imre Deak <[email protected]>
Cc Jani Nikula <[email protected]>
Cc <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

Quoted 8 lines
Quoted 2 lines
Quoted 1 line

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

This will slow down the lspcon probing and resume part, but both of them happen only when LSPCON device is found. So to answer your question, this will not slow down the non-lspcon path, but will slow down the LSPCON connector resume and probe time. but I would recommend, instead of increasing it to 1000 ms in a single shot, we might want to gradually pick this up, on a wake-and-check way.

wait_for() checks every [10us, 1ms] until the condition is met, or it times out. So, so long as we don't enter this path for !LPSCON where we know that it will timeout, the wait_for() will only take as long as is required for the connector to settle.

Can we do other connectors at the same time, or does probing LSPCON block the system?
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#8 lore raw

Hey Chris,

Quoted 53 lines · Chris Wilson

Quoting Sharma, Shashank (2018-08-16 08:33:36)

Quoted 44 lines

Regards

Shashank

Quoted 38 lines · Jani Nikula
Quoted 37 lines · Rodrigo Vivi
Quoted 36 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 6 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode to settle")
Cc Shashank Sharma <[email protected]>
Cc Imre Deak <[email protected]>
Cc Jani Nikula <[email protected]>
Cc <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

Quoted 8 lines
Quoted 2 lines
Quoted 1 line

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

This will slow down the lspcon probing and resume part, but both of them happen only when LSPCON device is found. So to answer your question, this will not slow down the non-lspcon path, but will slow down the LSPCON connector resume and probe time. but I would recommend, instead of increasing it to 1000 ms in a single shot, we might want to gradually pick this up, on a wake-and-check way.

Quoted 4 lines

wait_for() checks every [10us, 1ms] until the condition is met, or it times out. So, so long as we don't enter this path for !LPSCON where we know that it will timeout, the wait_for() will only take as long as is required for the connector to settle.

We wont hit !LSPCON timeout case here, as we have already read the dongle signature successfully by now. But I was thinking that, if the spec recommends max wait time as 100ms (which is of course doesn't seem enough), if we can't detect i2c-over-aux after first 500ms, I guess we wont be able to do that in next 500ms too. So is it really ok to wait this long in the resume sequence ?

I guess Fredrik can provide some inputs here on if there are some experiments behind this number of 1000ms, or this is just a safe bet ?

Quoted 3 lines

Can we do other connectors at the same time, or does probing LSPCON block the system?

We can do other connectors at the same time in DRM layer at-least, LSPCON blocks only this connector. I was curious if are we doing this during the resume scenario or is this in the sequential get_connector() type of call ?
- Shashank

Quoted 1 line

-Chris

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#9 lore raw

Shashank,

Den tors 16 aug. 2018 kl 10:15 skrev Sharma, Shashank <[email protected]>:

Quoted 69 lines

Hey Chris,

Quoted 53 lines · Chris Wilson

Quoting Sharma, Shashank (2018-08-16 08:33:36)

Quoted 44 lines

Regards

Shashank

Quoted 38 lines · Jani Nikula
Quoted 37 lines · Rodrigo Vivi
Quoted 36 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 6 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode to settle")
Cc Shashank Sharma <[email protected]>
Cc Imre Deak <[email protected]>
Cc Jani Nikula <[email protected]>
Cc <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

Quoted 8 lines
Quoted 2 lines
Quoted 1 line

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

This will slow down the lspcon probing and resume part, but both of them happen only when LSPCON device is found. So to answer your question, this will not slow down the non-lspcon path, but will slow down the LSPCON connector resume and probe time. but I would recommend, instead of increasing it to 1000 ms in a single shot, we might want to gradually pick this up, on a wake-and-check way.

Quoted 4 lines

wait_for() checks every [10us, 1ms] until the condition is met, or it times out. So, so long as we don't enter this path for !LPSCON where we know that it will timeout, the wait_for() will only take as long as is required for the connector to settle.

We wont hit !LSPCON timeout case here, as we have already read the dongle signature successfully by now. But I was thinking that, if the spec recommends max wait time as 100ms (which is of course doesn't seem enough), if we can't detect i2c-over-aux after first 500ms, I guess we wont be able to do that in next 500ms too. So is it really ok to wait this long in the resume sequence ?

Quoted 3 lines

I guess Fredrik can provide some inputs here on if there are some experiments behind this number of 1000ms, or this is just a safe bet ?

Quoted 1 line

My first patch attempt - which is attached to the Redhat and FDO Bugzilla bugs - added a retry loop around the original 100 ms timeout. The retry loop did trigger, but never more than once in a row in my testing.

So possibly 200 ms would be a sufficient timeout, but as the wait_for() loop terminates early on success I suggested a conservative value of 1000 ms.

Quoted 8 lines
Quoted 2 lines

Can we do other connectors at the same time, or does probing LSPCON block the system?

We can do other connectors at the same time in DRM layer at-least, LSPCON blocks only this connector. I was curious if are we doing this during the resume scenario or is this in the sequential get_connector() type of call ?
- Shashank

Quoted 1 line

-Chris

/F
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

maintainer
#10 lore raw
Quoted 77 lines · Fredrik Schön

Shashank,

Den tors 16 aug. 2018 kl 10:15 skrev Sharma, Shashank <[email protected]>:

Quoted 69 lines

Hey Chris,

Quoted 53 lines · Chris Wilson

Quoting Sharma, Shashank (2018-08-16 08:33:36)

Quoted 44 lines

Regards

Shashank

Quoted 38 lines · Jani Nikula
Quoted 37 lines · Rodrigo Vivi
Quoted 36 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 6 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver initialisation.

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode to settle")
Cc Shashank Sharma <[email protected]>
Cc Imre Deak <[email protected]>
Cc Jani Nikula <[email protected]>
Cc <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

Quoted 8 lines
Quoted 2 lines
Quoted 1 line

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

This will slow down the lspcon probing and resume part, but both of them happen only when LSPCON device is found. So to answer your question, this will not slow down the non-lspcon path, but will slow down the LSPCON connector resume and probe time. but I would recommend, instead of increasing it to 1000 ms in a single shot, we might want to gradually pick this up, on a wake-and-check way.

Quoted 4 lines

wait_for() checks every [10us, 1ms] until the condition is met, or it times out. So, so long as we don't enter this path for !LPSCON where we know that it will timeout, the wait_for() will only take as long as is required for the connector to settle.

We wont hit !LSPCON timeout case here, as we have already read the dongle signature successfully by now. But I was thinking that, if the spec recommends max wait time as 100ms (which is of course doesn't seem enough), if we can't detect i2c-over-aux after first 500ms, I guess we wont be able to do that in next 500ms too. So is it really ok to wait this long in the resume sequence ?

Quoted 3 lines

I guess Fredrik can provide some inputs here on if there are some experiments behind this number of 1000ms, or this is just a safe bet ?

Quoted 1 line
Quoted 6 lines

My first patch attempt - which is attached to the Redhat and FDO Bugzilla bugs - added a retry loop around the original 100 ms timeout. The retry loop did trigger, but never more than once in a row in my testing.

So possibly 200 ms would be a sufficient timeout, but as the wait_for() loop terminates early on success I suggested a conservative value of 1000 ms.

Since Shashank mentioned 100us came from some spec, maybe the double is already a conservative value.

Since there is the concerns of delaying something when LSPCON fails and we are possibly looping on connectors somewhere/somehow I believe we need to have a balanced approach here.

could you please try the 200 ms approach on your case there for a while and see how it goes?

Quoted 8 lines
Quoted 8 lines
Quoted 3 lines

Can we do other connectors at the same time, or does probing LSPCON block the system?

We can do other connectors at the same time in DRM layer at-least, LSPCON blocks only this connector. I was curious if are we doing this during the resume scenario or is this in the sequential get_connector() type of call ?
- Shashank

Quoted 6 lines
Quoted 1 line
Quoted 1 line

-Chris

/F
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#11 lore raw

tor 2018-08-16 klockan 11:23 -0700 skrev Rodrigo Vivi:

Quoted 35 lines · Fredrik Schön
Quoted 34 lines

Shashank,

Den tors 16 aug. 2018 kl 10:15 skrev Sharma, Shashank <[email protected]>:

Quoted 30 lines

Hey Chris,

Quoted 25 lines · Chris Wilson

Quoting Sharma, Shashank (2018-08-16 08:33:36)

Quoted 24 lines

Regards

Shashank

Quoted 18 lines · Jani Nikula
Quoted 16 lines · Rodrigo Vivi
Quoted 14 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 5 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to
settle. This causes dropped display modes at driver initialisation.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 4 lines

Increase timeout to 1000 ms.

Fixes:
https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

Quoted 80 lines
Quoted 71 lines
Quoted 60 lines
Quoted 40 lines
Quoted 28 lines
Quoted 19 lines
Quoted 19 lines
Quoted 19 lines

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla:
https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for

expected LSPCON mode to settle")
Cc: Shashank Sharma <[email protected]>
Cc: Imre Deak <[email protected]>
Cc: Jani Nikula <[email protected]>
Cc: <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to
help other cases I'm in favor of this move so

Quoted 8 lines
Quoted 6 lines
Quoted 4 lines

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

Quoted 11 lines

This will slow down the lspcon probing and resume part, but both of them
happen only when LSPCON device is found. So to answer your question,
this will not slow down the non-lspcon path, but will slow down the
LSPCON connector resume and probe time. but I would recommend, instead
of increasing it to 1000 ms in a single shot, we might want to gradually
pick this up, on a wake-and-check way.

Quoted 7 lines

wait_for() checks every [10us, 1ms] until the condition is met, or it
times out. So, so long as we don't enter this path for !LPSCON where we
know that it will timeout, the wait_for() will only take as long as is
required for the connector to settle.

We wont hit !LSPCON timeout case here, as we have already read the
dongle signature successfully by now. But I was thinking that, if the
spec recommends max wait time as 100ms (which is of course doesn't seem
enough), if we can't detect i2c-over-aux after first 500ms, I guess we
wont be able to do that in next 500ms too. So is it really ok to wait
this long in the resume sequence ?

Quoted 4 lines

I guess Fredrik can provide some inputs here on if there are some experiments behind this number of 1000ms, or this is just a safe bet ?

Quoted 1 line

My first patch attempt - which is attached to the Redhat and FDO Bugzilla
bugs - added a retry loop around the original 100 ms timeout. The retry loop
did trigger, but never more than once in a row in my testing.

Quoted 4 lines

So possibly 200 ms would be a sufficient timeout, but as the wait_for() loop
terminates early on success I suggested a conservative value of 1000 ms.

Since Shashank mentioned 100us came from some spec, maybe the double is already
a conservative value.

Quoted 9 lines

Since there is the concerns of delaying something when LSPCON fails and we are possibly looping on connectors somewhere/somehow I believe we need
to have a balanced approach here.

could you please try the 200 ms approach on your case there for a while and
see how it goes?

I ran a few stress tests using Nicholas test case from [1]. I can quickly reproduce the failure with timeouts 100 ms, 110 ms, 130 ms, 150 ms and 170 ms. I am unable to reproduce any failures with timeouts 190 ms (n=18) and 200 ms (n=20+16).

So while 200 ms appears to work on my hardware with reasonable confidence, I wouldn't call 200 conservative. But then again, I do not know the specifications. I'm just being empirical.

[1] https://bugs.freedesktop.org/show_bug.cgi?id=107503#c15

Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
Quoted 4 lines

Can we do other connectors at the same time, or does probing LSPCON
block the system?

We can do other connectors at the same time in DRM layer at- least,
LSPCON blocks only this connector. I was curious if are we doing this
during the resume scenario or is this in the sequential get_connector()
type of call ?
- Shashank

Quoted 7 lines
Quoted 7 lines
Quoted 1 line
Quoted 1 line

-Chris

/F
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

maintainer
#12 lore raw
Quoted 135 lines · On Thu, Aug 16, 2018 at 11:35:26PM +0200, [email protected] wrote:

tor 2018-08-16 klockan 11:23 -0700 skrev Rodrigo Vivi:

Quoted 35 lines · Fredrik Schön
Quoted 34 lines

Shashank,

Den tors 16 aug. 2018 kl 10:15 skrev Sharma, Shashank <[email protected]>:

Quoted 30 lines

Hey Chris,

Quoted 25 lines · Chris Wilson

Quoting Sharma, Shashank (2018-08-16 08:33:36)

Quoted 24 lines

Regards

Shashank

Quoted 18 lines · Jani Nikula
Quoted 16 lines · Rodrigo Vivi
Quoted 14 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 5 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to
settle. This causes dropped display modes at driver initialisation.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 4 lines

Increase timeout to 1000 ms.

Fixes:
https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

Quoted 80 lines
Quoted 71 lines
Quoted 60 lines
Quoted 40 lines
Quoted 28 lines
Quoted 19 lines
Quoted 19 lines
Quoted 19 lines

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla:
https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for

expected LSPCON mode to settle")
Cc: Shashank Sharma <[email protected]>
Cc: Imre Deak <[email protected]>
Cc: Jani Nikula <[email protected]>
Cc: <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to
help other cases I'm in favor of this move so

Quoted 8 lines
Quoted 6 lines
Quoted 4 lines

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

Quoted 11 lines

This will slow down the lspcon probing and resume part, but both of them
happen only when LSPCON device is found. So to answer your question,
this will not slow down the non-lspcon path, but will slow down the
LSPCON connector resume and probe time. but I would recommend, instead
of increasing it to 1000 ms in a single shot, we might want to gradually
pick this up, on a wake-and-check way.

Quoted 7 lines

wait_for() checks every [10us, 1ms] until the condition is met, or it
times out. So, so long as we don't enter this path for !LPSCON where we
know that it will timeout, the wait_for() will only take as long as is
required for the connector to settle.

We wont hit !LSPCON timeout case here, as we have already read the
dongle signature successfully by now. But I was thinking that, if the
spec recommends max wait time as 100ms (which is of course doesn't seem
enough), if we can't detect i2c-over-aux after first 500ms, I guess we
wont be able to do that in next 500ms too. So is it really ok to wait
this long in the resume sequence ?

Quoted 4 lines

I guess Fredrik can provide some inputs here on if there are some experiments behind this number of 1000ms, or this is just a safe bet ?

Quoted 1 line

My first patch attempt - which is attached to the Redhat and FDO Bugzilla
bugs - added a retry loop around the original 100 ms timeout. The retry loop
did trigger, but never more than once in a row in my testing.

Quoted 4 lines

So possibly 200 ms would be a sufficient timeout, but as the wait_for() loop
terminates early on success I suggested a conservative value of 1000 ms.

Since Shashank mentioned 100us came from some spec, maybe the double is already
a conservative value.

Quoted 9 lines

Since there is the concerns of delaying something when LSPCON fails and we are possibly looping on connectors somewhere/somehow I believe we need
to have a balanced approach here.

could you please try the 200 ms approach on your case there for a while and
see how it goes?

Quoted 8 lines

I ran a few stress tests using Nicholas test case from [1]. I can quickly reproduce the failure with timeouts 100 ms, 110 ms, 130 ms, 150 ms and 170 ms. I am unable to reproduce any failures with timeouts 190 ms (n=18) and 200 ms (n=20+16).

So while 200 ms appears to work on my hardware with reasonable confidence, I wouldn't call 200 conservative. But then again, I do not know the specifications. I'm just being empirical.

I don't know this specification either and if that exists the empirical shows that it is wrong or we have another bug somewhere else.

So... let's call 400 safe enough for now then?!

Quoted 24 lines

[1] https://bugs.freedesktop.org/show_bug.cgi?id=107503#c15

Quoted 20 lines
Quoted 20 lines
Quoted 5 lines
Quoted 4 lines

Can we do other connectors at the same time, or does probing LSPCON
block the system?

Quoted 9 lines

We can do other connectors at the same time in DRM layer at- least,
LSPCON blocks only this connector. I was curious if are we doing this
during the resume scenario or is this in the sequential get_connector()
type of call ?
- Shashank

Quoted 1 line

-Chris

/F
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#13 lore raw

tor 2018-08-16 klockan 14:43 -0700 skrev Rodrigo Vivi:

Quoted 43 lines
Quoted 41 lines · On Thu, Aug 16, 2018 at 11:35:26PM +0200, [email protected] wrote:

tor 2018-08-16 klockan 11:23 -0700 skrev Rodrigo Vivi:

Quoted 40 lines · Fredrik Schön
Quoted 39 lines

Shashank,

Den tors 16 aug. 2018 kl 10:15 skrev Sharma, Shashank <[email protected]>:

Quoted 35 lines

Hey Chris,

Quoted 30 lines · Chris Wilson

Quoting Sharma, Shashank (2018-08-16 08:33:36)

Quoted 29 lines

Regards

Shashank

Quoted 23 lines · Jani Nikula
Quoted 20 lines · Rodrigo Vivi
Quoted 17 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön
wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 7 lines

100 ms is not enough time for the LSPCON adapter
on
Intel NUC devices to
settle. This causes dropped display modes at
driver
initialisation.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Increase timeout to 1000 ms.

Fixes:

https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

Quoted 120 lines
Quoted 115 lines
Quoted 103 lines
Quoted 86 lines
Quoted 71 lines
Quoted 46 lines
Quoted 30 lines
Quoted 22 lines
Quoted 22 lines
Quoted 22 lines

But also no need for resending... could be fixed when
merging

The right one would be:

Bugzilla:
https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for

expected LSPCON mode to settle")
Cc: Shashank Sharma <[email protected]>
Cc: Imre Deak <[email protected]>
Cc: Jani Nikula <[email protected]>
Cc: <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to
help other cases I'm in favor of this move so

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

Quoted 15 lines

This will slow down the lspcon probing and resume part, but
both of them
happen only when LSPCON device is found. So to answer your
question,
this will not slow down the non-lspcon path, but will slow
down the
LSPCON connector resume and probe time. but I would recommend, instead
of increasing it to 1000 ms in a single shot, we might want
to gradually
pick this up, on a wake-and-check way.

Quoted 9 lines

wait_for() checks every [10us, 1ms] until the condition is met,
or it
times out. So, so long as we don't enter this path for !LPSCON
where we
know that it will timeout, the wait_for() will only take as long as is
required for the connector to settle.

We wont hit !LSPCON timeout case here, as we have already read
the
dongle signature successfully by now. But I was thinking that,
if the
spec recommends max wait time as 100ms (which is of course doesn't seem
enough), if we can't detect i2c-over-aux after first 500ms, I guess we
wont be able to do that in next 500ms too. So is it really ok to
wait
this long in the resume sequence ?

Quoted 6 lines

I guess Fredrik can provide some inputs here on if there are some
experiments behind this number of 1000ms, or this is just a safe
bet ?

Quoted 1 line

My first patch attempt - which is attached to the Redhat and FDO
Bugzilla
bugs - added a retry loop around the original 100 ms timeout. The
retry loop
did trigger, but never more than once in a row in my testing.

Quoted 4 lines

So possibly 200 ms would be a sufficient timeout, but as the wait_for() loop
terminates early on success I suggested a conservative value of 1000 ms.

Since Shashank mentioned 100us came from some spec, maybe the double
is already
a conservative value.

Since there is the concerns of delaying something when LSPCON fails
and we are possibly looping on connectors somewhere/somehow I believe
we need
to have a balanced approach here.

Quoted 4 lines

could you please try the 200 ms approach on your case there for a while and
see how it goes?

I ran a few stress tests using Nicholas test case from [1]. I can quickly reproduce the failure with timeouts 100 ms, 110 ms, 130 ms, 150
ms and 170 ms. I am unable to reproduce any failures with timeouts 190
ms (n=18) and 200 ms (n=20+16).

Quoted 4 lines

So while 200 ms appears to work on my hardware with reasonable confidence, I wouldn't call 200 conservative. But then again, I do not
know the specifications. I'm just being empirical.

Quoted 6 lines

I don't know this specification either and if that exists the empirical
shows that it is wrong or we have another bug somewhere else.

So... let's call 400 safe enough for now then?!

Sound reasonable. Do you want me to respin the patch?

Quoted 18 lines
Quoted 18 lines

[1] https://bugs.freedesktop.org/show_bug.cgi?id=107503#c15

Quoted 15 lines
Quoted 15 lines
Quoted 15 lines
Quoted 5 lines

Can we do other connectors at the same time, or does probing
LSPCON
block the system?

We can do other connectors at the same time in DRM layer at- least,
LSPCON blocks only this connector. I was curious if are we doing
this
during the resume scenario or is this in the sequential get_connector()
type of call ?
- Shashank

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 1 line
Quoted 1 line

-Chris

/F
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

maintainer
#14 lore raw
Quoted 176 lines · Fredrik Schön

tor 2018-08-16 klockan 14:43 -0700 skrev Rodrigo Vivi:

Quoted 43 lines
Quoted 41 lines · On Thu, Aug 16, 2018 at 11:35:26PM +0200, [email protected] wrote:

tor 2018-08-16 klockan 11:23 -0700 skrev Rodrigo Vivi:

Quoted 40 lines · Fredrik Schön
Quoted 39 lines

Shashank,

Den tors 16 aug. 2018 kl 10:15 skrev Sharma, Shashank <[email protected]>:

Quoted 35 lines

Hey Chris,

Quoted 30 lines · Chris Wilson

Quoting Sharma, Shashank (2018-08-16 08:33:36)

Quoted 29 lines

Regards

Shashank

Quoted 23 lines · Jani Nikula
Quoted 20 lines · Rodrigo Vivi
Quoted 17 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön
wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 7 lines

100 ms is not enough time for the LSPCON adapter
on
Intel NUC devices to
settle. This causes dropped display modes at
driver
initialisation.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Increase timeout to 1000 ms.

Fixes:

https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html#fixes

Quoted 120 lines
Quoted 119 lines
Quoted 103 lines
Quoted 86 lines
Quoted 71 lines
Quoted 46 lines
Quoted 30 lines
Quoted 22 lines
Quoted 22 lines
Quoted 22 lines

But also no need for resending... could be fixed when
merging

The right one would be:

Bugzilla:
https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for

expected LSPCON mode to settle")
Cc: Shashank Sharma <[email protected]>
Cc: Imre Deak <[email protected]>
Cc: Jani Nikula <[email protected]>
Cc: <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to
help other cases I'm in favor of this move so

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

Quoted 15 lines

This will slow down the lspcon probing and resume part, but
both of them
happen only when LSPCON device is found. So to answer your
question,
this will not slow down the non-lspcon path, but will slow
down the
LSPCON connector resume and probe time. but I would recommend, instead
of increasing it to 1000 ms in a single shot, we might want
to gradually
pick this up, on a wake-and-check way.

Quoted 9 lines

wait_for() checks every [10us, 1ms] until the condition is met,
or it
times out. So, so long as we don't enter this path for !LPSCON
where we
know that it will timeout, the wait_for() will only take as long as is
required for the connector to settle.

We wont hit !LSPCON timeout case here, as we have already read
the
dongle signature successfully by now. But I was thinking that,
if the
spec recommends max wait time as 100ms (which is of course doesn't seem
enough), if we can't detect i2c-over-aux after first 500ms, I guess we
wont be able to do that in next 500ms too. So is it really ok to
wait
this long in the resume sequence ?

Quoted 6 lines

I guess Fredrik can provide some inputs here on if there are some
experiments behind this number of 1000ms, or this is just a safe
bet ?

Quoted 1 line

My first patch attempt - which is attached to the Redhat and FDO
Bugzilla
bugs - added a retry loop around the original 100 ms timeout. The
retry loop
did trigger, but never more than once in a row in my testing.

Quoted 4 lines

So possibly 200 ms would be a sufficient timeout, but as the wait_for() loop
terminates early on success I suggested a conservative value of 1000 ms.

Since Shashank mentioned 100us came from some spec, maybe the double
is already
a conservative value.

Since there is the concerns of delaying something when LSPCON fails
and we are possibly looping on connectors somewhere/somehow I believe
we need
to have a balanced approach here.

Quoted 4 lines

could you please try the 200 ms approach on your case there for a while and
see how it goes?

I ran a few stress tests using Nicholas test case from [1]. I can quickly reproduce the failure with timeouts 100 ms, 110 ms, 130 ms, 150
ms and 170 ms. I am unable to reproduce any failures with timeouts 190
ms (n=18) and 200 ms (n=20+16).

So while 200 ms appears to work on my hardware with reasonable confidence, I wouldn't call 200 conservative. But then again, I do not
know the specifications. I'm just being empirical.

Quoted 8 lines
Quoted 6 lines

I don't know this specification either and if that exists the empirical
shows that it is wrong or we have another bug somewhere else.

So... let's call 400 safe enough for now then?!

Sound reasonable. Do you want me to respin the patch?

we can give a day for the guys on other timezone to have a chance to ack, nack or give another suggestions and after that yes, please send a v2.

Thanks,
Rodrigo.

Quoted 26 lines
Quoted 18 lines
Quoted 18 lines

[1] https://bugs.freedesktop.org/show_bug.cgi?id=107503#c15

Quoted 15 lines
Quoted 15 lines
Quoted 15 lines
Quoted 5 lines

Can we do other connectors at the same time, or does probing
LSPCON
block the system?

We can do other connectors at the same time in DRM layer at- least,
LSPCON blocks only this connector. I was curious if are we doing
this
during the resume scenario or is this in the sequential get_connector()
type of call ?
- Shashank

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 1 line
Quoted 1 line

-Chris

/F
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Quoted 4 lines

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#15 lore raw
In reply to Jani Nikula

Hi,

Quoted 48 lines

-----Original Message-----
From: Intel-gfx [mailto:[email protected]] On Behalf Of Jani Nikula
Sent: torstai 16. elokuuta 2018 10.18
To: Vivi, Rodrigo <[email protected]>; Fredrik Schön <[email protected]>
Cc: [email protected]; Fredrik Schön <[email protected]>
Subject: Re: [Intel-gfx] [PATCH] [drm][i915] Increase LSPCON timeout

Quoted 11 lines · Rodrigo Vivi
Quoted 10 lines · Rodrigo Vivi

On Mon, Aug 13, 2018 at 07:51:33PM +0200, Fredrik Schön wrote:

First of all we need to fix the commit subject:

drm/i915: Increase LSPCON timeout

(this can be done when merging, no need to resend)

Quoted 2 lines

100 ms is not enough time for the LSPCON adapter on Intel NUC devices to settle. This causes dropped display modes at driver

initialisation.

Quoted 25 lines
Quoted 25 lines
Quoted 4 lines

Increase timeout to 1000 ms.

Missusage of "Fixes:" tag, please read

https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-intel.html
#fixes

But also no need for resending... could be fixed when merging

The right one would be:

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1570392

Fixes 357c0ae9198a ("drm/i915/lspcon: Wait for expected LSPCON mode

to settle")
Cc: Shashank Sharma <[email protected]>
Cc: Imre Deak <[email protected]>
Cc: Jani Nikula <[email protected]>
Cc: <[email protected]> # v4.11+

Since initial 100 seemed to be empirical and this increase seems to help other cases I'm in favor of this move so

Quoted 8 lines
Quoted 6 lines
Quoted 4 lines

However I will wait a bit before merging it so Imre, Shashank, and/or Jani can take a look here...

now, really cc'ing them...

Shashank? Does this slow down non-LSPCON paths?

+ Shashank for real

Quoted 41 lines

BR,
Jani.

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
---
 drivers/gpu/drm/i915/intel_lspcon.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 6 lines
Quoted 6 lines

drivers/gpu/drm/i915/intel_lspcon.c static enum drm_lspcon_mode

Quoted 6 lines
drivers/gpu/drm/i915/intel_lspcon.c +0 -0
index 8ae8f42f430a..be1b08f589a4 100644
@@ -74,7 +74,7 @@ static enum drm_lspcon_mode

lspcon_wait_mode(struct intel_lspcon *lspcon,

Quoted 4 lines
Quoted 4 lines

drivers/gpu/drm/i915/intel_lspcon.c static enum drm_lspcon_mode lspcon_wait_mode(struct intel_lspcon *lspcon,

Quoted 4 lines
drivers/gpu/drm/i915/intel_lspcon.c +0 -1
 	DRM_DEBUG_KMS("Waiting for LSPCON mode %s to settle\n",
 		      lspcon_mode_name(mode));
 
-	wait_for((current_mode = lspcon_get_current_mode(lspcon)) ==

mode, 100);

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
drivers/gpu/drm/i915/intel_lspcon.c +2 -1
+	wait_for((current_mode = lspcon_get_current_mode(lspcon)) ==
+mode, 1000);
 	if (current_mode != mode)
 		DRM_ERROR("LSPCON mode hasn't settled\n");
 
--

2.17.1

Quoted 8 lines
Quoted 8 lines
Quoted 4 lines

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Quoted 6 lines

--
Jani Nikula, Intel Open Source Graphics Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx