drm/i915/gen9: Fix PCODE polling during CDCLK change notification
Subsystems (3)
PATCH v8 1/1 drm/i915/gen9: Fix PCODE polling during CDCLK change notification
commit 848496e5902833600f7992f4faa82dc1546051ba
Author: Ville Syrjälä <[email protected]> Date: Wed Jul 13 16:32:03 2016 +0300
drm/i915: Wait up to 3ms for the pcu to ack the cdclk change request on SKL
increased the timeout to match the spec, but we still see a timeout on at least one SKL. A CDCLK change request following the failed one will succeed nevertheless.
I could reproduce this problem easily by running kms_pipe_crc_basic in a loop. In all failure cases _wait_for() was pre-empted for >3ms and so in the worst case - when the pre-emption happened right after calculating timeout__ in _wait_for() - we called skl_cdclk_wait_for_pcu_ready() only once which failed and so _wait_for() timed out. As opposed to this the spec says to keep retrying the request for at most a 3ms period.
To fix this send the first request explicitly to guarantee that there is 3ms between the first and last request. Though this matches the spec, I noticed that in rare cases this can still time out if we sent only a few requests (in the worst case 2) _and_ PCODE is busy for some reason even after a previous request and a 3ms delay. To work around this retry the polling with pre-emption disabled to maximize the number of requests. Also increase the timeout to 10ms to account for interrupts that could reduce the number of requests. With this change I couldn't trigger the problem.
v2:
- Use 1ms poll period instead of 10us. (Chris)
v3:
- Poll with pre-emption disabled to increase the number of request attempts. (Ville, Chris)
- Factor out a helper to poll, it's also needed by the next patch. v4:
- Pass reply_mask, reply to skl_pcode_request(), instead of assuming the reply is generic. (Ville)
v5:
- List the request specific timeout values as code comment. (Ville) v6:
- Try the poll first with preemption enabled.
- Add code comment about first request being queued by PCODE. (Art)
- Add timeout_base_ms argument. (Ville)
v7:
- Clarify code comment about first queued request. (Chris) v8:
- Rebased on 4.9.2
Reference: https://bugs.freedesktop.org/show_bug.cgi?id=97929
Testcase: igt/kms_pipe_crc_basic/suspend-read-crc-pipe-B
--- drivers/gpu/drm/i915/i915_drv.h | 2 + drivers/gpu/drm/i915/intel_display.c | 31 +++++---------- drivers/gpu/drm/i915/intel_pm.c | 75 ++++++++++++++++++++++++++++++++++++ 3 files changed, 87 insertions(+), 21 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +2 -0
@@ -3684,6 +3684,8 @@ extern void intel_display_print_error_state(struct drm_i915_error_state_buf *e,int sandybridge_pcode_read (struct drm_i915_private *dev_priv,u32 mbox,u32 *val);int sandybridge_pcode_write (struct drm_i915_private *dev_priv,u32 mbox,u32 val); +int skl_pcode_request (struct drm_i915_private *dev_priv,u32 mbox,u32 request, +u32 reply_mask,u32 reply,int timeout_base_ms);/* intel_sideband.c */ u32 vlv_punit_read (struct drm_i915_private *dev_priv,u32 addr);
drivers/gpu/drm/i915/intel_display.c +10 -21
@@ -6262,36 +6262,25 @@ skl_dpll0_disable(struct drm_i915_private *dev_priv) dev_priv->cdclk_pll.vco =0 ; } -static bool skl_cdclk_pcu_ready (struct drm_i915_private *dev_priv) -{ -int ret; -u32 val; - -/* inform PCU we want to change CDCLK */ - val =SKL_CDCLK_PREPARE_FOR_CHANGE ; -mutex_lock (&dev_priv->rps.hw_lock); - ret =sandybridge_pcode_read (dev_priv,SKL_PCODE_CDCLK_CONTROL , &val); -mutex_unlock (&dev_priv->rps.hw_lock); - -return ret ==0 && (val &SKL_CDCLK_READY_FOR_CHANGE ); -} - -static bool skl_cdclk_wait_for_pcu_ready (struct drm_i915_private *dev_priv) -{ -return _wait_for (skl_cdclk_pcu_ready (dev_priv),3000 ,10 ) ==0 ; -} -static void skl_set_cdclk (struct drm_i915_private *dev_priv,int cdclk,int vco) {struct drm_device *dev = &dev_priv->drm;u32 freq_select, pcu_ack; +int ret;WARN_ON ((cdclk ==24000 ) != (vco ==0 ));DRM_DEBUG_DRIVER ("Changing CDCLK to %d kHz (VCO %d kHz)\n" , cdclk, vco); -if (!skl_cdclk_wait_for_pcu_ready (dev_priv)) { -DRM_ERROR ("failed to inform PCU about cdclk change\n" ); +mutex_lock (&dev_priv->rps.hw_lock); + ret =skl_pcode_request (dev_priv,SKL_PCODE_CDCLK_CONTROL , +SKL_CDCLK_PREPARE_FOR_CHANGE , +SKL_CDCLK_READY_FOR_CHANGE , +SKL_CDCLK_READY_FOR_CHANGE ,3 ); +mutex_unlock (&dev_priv->rps.hw_lock); +if (ret) { +DRM_ERROR ("Failed to inform PCU about cdclk change (%d)\n" , + ret);return ; }
drivers/gpu/drm/i915/intel_pm.c +75 -0
81 lines not on this page · raw message
Signature · 1 line
2.5.0
Greg Kroah-Hartman
maintainerQuoted 60 lines · Imre Deak
commit
848496e5902833600f7992f4faa82dc1546051baAuthor: Ville Syrj�l� <[email protected]> Date: Wed Jul 13 16:32:03 2016 +0300drm/i915: Wait up to 3ms for the pcu to ack the cdclk change request on SKL
increased the timeout to match the spec, but we still see a timeout on at least one SKL. A CDCLK change request following the failed one will succeed nevertheless.
I could reproduce this problem easily by running
kms_pipe_crc_basicin a loop. In all failure cases_wait_for()was pre-empted for >3ms and so in the worst case - when the pre-emption happened right after calculatingtimeout__in_wait_for()- we calledskl_cdclk_wait_for_pcu_ready()only once which failed and so_wait_for()timed out. As opposed to this the spec says to keep retrying the request for at most a 3ms period.To fix this send the first request explicitly to guarantee that there is 3ms between the first and last request. Though this matches the spec, I noticed that in rare cases this can still time out if we sent only a few requests (in the worst case 2) _and_ PCODE is busy for some reason even after a previous request and a 3ms delay. To work around this retry the polling with pre-emption disabled to maximize the number of requests. Also increase the timeout to 10ms to account for interrupts that could reduce the number of requests. With this change I couldn't trigger the problem.
v2:
- Use 1ms poll period instead of 10us. (Chris)
v3:
- Poll with pre-emption disabled to increase the number of request attempts. (Ville, Chris)
- Factor out a helper to poll, it's also needed by the next patch. v4:
- Passreply_mask, reply toskl_pcode_request(), instead of assuming the reply is generic. (Ville)
v5:
- List the request specific timeout values as code comment. (Ville) v6:
- Try the poll first with preemption enabled.
- Add code comment about first request being queued by PCODE. (Art)
- Addtimeout_base_msargument. (Ville)
v7:
- Clarify code comment about first queued request. (Chris) v8:
- Rebased on 4.9.2Cc Ville Syrj�l� <[email protected]>Cc Chris Wilson <[email protected]>Cc Art Runyan <[email protected]>Cc <[email protected]> # v4.2- : 3b2c171 : drm/i915: Wait up to 3msCc <[email protected]> # v4.2-Fixes 5d96d8afcfbb ("drm/i915/skl: Deinit/init the display at suspend/resume")Reference: https://bugs.freedesktop.org/show_bug.cgi?id=97929
Testcase: igt/kms_pipe_crc_basic/suspend-read-crc-pipe-BSigned-off-by Imre Deak <[email protected]>Reviewed-by Chris Wilson <[email protected]>(cherry picked from commit a0b8a1fe34430c3a82258e8cb45f5968bdf31afd)Signed-off-by Jani Nikula <[email protected]>
Quoted 5 lines
--- drivers/gpu/drm/i915/i915_drv.h | 2 + drivers/gpu/drm/i915/intel_display.c | 31 +++++---------- drivers/gpu/drm/i915/intel_pm.c | 75 ++++++++++++++++++++++++++++++++++++ 3 files changed, 87 insertions(+), 21 deletions(-)
What tree is this patch for? Please give us a hint, we don't like guessing...
thanks,
greg k-h
Quoted 67 lines · Greg Kroah-Hartman
Quoted 60 lines · Imre Deak
commit
848496e5902833600f7992f4faa82dc1546051ba
Author: Ville Syrjälä <[email protected]> Date: Wed Jul 13 16:32:03 2016 +0300drm/i915: Wait up to 3ms for the pcu to ack the cdclk change request on SKL
increased the timeout to match the spec, but we still see a timeout on at least one SKL. A CDCLK change request following the failed one will succeed nevertheless.
I could reproduce this problem easily by running
kms_pipe_crc_basicin a loop. In all failure cases_wait_for()was pre-empted for >3ms and so in the worst case - when the pre-emption happened right after calculatingtimeout__in_wait_for()- we calledskl_cdclk_wait_for_pcu_ready()only once which failed and so_wait_for()timed out. As opposed to this the spec says to keep retrying the request for at most a 3ms period.To fix this send the first request explicitly to guarantee that there is 3ms between the first and last request. Though this matches the spec, I noticed that in rare cases this can still time out if we sent only a few requests (in the worst case 2) _and_ PCODE is busy for some reason even after a previous request and a 3ms delay. To work around this retry the polling with pre-emption disabled to maximize the number of requests. Also increase the timeout to 10ms to account for interrupts that could reduce the number of requests. With this change I couldn't trigger the problem.
v2:
- Use 1ms poll period instead of 10us. (Chris)
v3:
- Poll with pre-emption disabled to increase the number of request attempts. (Ville, Chris)
- Factor out a helper to poll, it's also needed by the next patch. v4:
- Passreply_mask, reply toskl_pcode_request(), instead of assuming the reply is generic. (Ville)
v5:
- List the request specific timeout values as code comment. (Ville) v6:
- Try the poll first with preemption enabled.
- Add code comment about first request being queued by PCODE. (Art)
- Addtimeout_base_msargument. (Ville)
v7:
- Clarify code comment about first queued request. (Chris) v8:
- Rebased on 4.9.2Cc Ville Syrjälä <[email protected]>Cc Chris Wilson <[email protected]>Cc Art Runyan <[email protected]>Cc <[email protected]> # v4.2- : 3b2c171 : drm/i915: Wait up to 3msCc <[email protected]> # v4.2-Fixes 5d96d8afcfbb ("drm/i915/skl: Deinit/init the display at suspend/resume")Reference: https://bugs.freedesktop.org/show_bug.cgi?id=97929
Testcase: igt/kms_pipe_crc_basic/suspend-read-crc-pipe-BSigned-off-by Imre Deak <[email protected]>Reviewed-by Chris Wilson <[email protected]>(cherry picked from commit a0b8a1fe34430c3a82258e8cb45f5968bdf31afd)Signed-off-by Jani Nikula <[email protected]>Quoted 5 lines
---
drivers/gpu/drm/i915/i915_drv.h | 2 +drivers/gpu/drm/i915/intel_display.c| 31 +++++---------- drivers/gpu/drm/i915/intel_pm.c | 75 ++++++++++++++++++++++++++++++++++++3 files changed, 87 insertions(+), 21 deletions(-)
Quoted 2 lines
What tree is this patch for? Please give us a hint, we don't like guessing...
It's for the 4.9 stable tree and for other stable trees starting from 4.2. It's the upstream commit a0b8a1fe34430c3a82258e8cb45f5968bdf31afd rebased on top of 4.9.2 according to our discussion at
http://www.spinics.net/lists/stable/msg155406.html
--Imre
Greg Kroah-Hartman
maintainerQuoted 68 lines · Imre Deak
Quoted 67 lines · Greg Kroah-Hartman
Quoted 65 lines · Imre Deak
commit
848496e5902833600f7992f4faa82dc1546051ba
Author: Ville Syrj�l� <[email protected]> Date:���Wed Jul 13 16:32:03 2016 +0300����drm/i915: Wait up to 3ms for the pcu to ack the cdclk change request on SKL
increased the timeout to match the spec, but we still see a timeout on at least one SKL. A CDCLK change request following the failed one will succeed nevertheless.
I could reproduce this problem easily by running
kms_pipe_crc_basicin a loop. In all failure cases_wait_for()was pre-empted for >3ms and so in the worst case - when the pre-emption happened right after calculatingtimeout__in_wait_for()- we calledskl_cdclk_wait_for_pcu_ready()only once which failed and so_wait_for()timed out. As opposed to this the spec says to keep retrying the request for at most a 3ms period.To fix this send the first request explicitly to guarantee that there is 3ms between the first and last request. Though this matches the spec, I noticed that in rare cases this can still time out if we sent only a few requests (in the worst case 2) _and_ PCODE is busy for some reason even after a previous request and a 3ms delay. To work around this retry the polling with pre-emption disabled to maximize the number of requests. Also increase the timeout to 10ms to account for interrupts that could reduce the number of requests. With this change I couldn't trigger the problem.
v2:
- Use 1ms poll period instead of 10us. (Chris)
v3:
- Poll with pre-emption disabled to increase the number of request � attempts. (Ville, Chris)
- Factor out a helper to poll, it's also needed by the next patch. v4:
- Passreply_mask, reply toskl_pcode_request(), instead of assuming the � reply is generic. (Ville)
v5:
- List the request specific timeout values as code comment. (Ville) v6:
- Try the poll first with preemption enabled.
- Add code comment about first request being queued by PCODE. (Art)
- Addtimeout_base_msargument. (Ville)
v7:
- Clarify code comment about first queued request. (Chris) v8:
- Rebased on 4.9.2Cc Ville Syrj�l� <[email protected]>Cc Chris Wilson <[email protected]>Cc Art Runyan <[email protected]>Cc <[email protected]> # v4.2- : 3b2c171 : drm/i915: Wait up to 3msCc <[email protected]> # v4.2-Fixes 5d96d8afcfbb ("drm/i915/skl: Deinit/init the display at suspend/resume")Reference: https://bugs.freedesktop.org/show_bug.cgi?id=97929
Testcase: igt/kms_pipe_crc_basic/suspend-read-crc-pipe-BSigned-off-by Imre Deak <[email protected]>Reviewed-by Chris Wilson <[email protected]>(cherry picked from commit a0b8a1fe34430c3a82258e8cb45f5968bdf31afd)Signed-off-by Jani Nikula <[email protected]>---
�drivers/gpu/drm/i915/i915_drv.h������|��2 + �drivers/gpu/drm/i915/intel_display.c | 31 +++++---------- �drivers/gpu/drm/i915/intel_pm.c������| 75 ++++++++++++++++++++++++++++++++++++ �3 files changed, 87 insertions(+), 21 deletions(-)
Quoted 5 lines
Quoted 2 lines
What tree is this patch for?��Please give us a hint, we don't like guessing...
It's for the 4.9 stable tree and for other stable trees starting from 4.2. It's the upstream commit
a0b8a1fe34430c3a82258e8cb45f5968bdf31afd
There is no such git commit id in Linus's tree :(
Do you mean 2c7d0602c815277f7cb7c932b091288710d8aba7?
Quoted 2 lines
rebased on top of 4.9.2 according to our discussion at
http://www.spinics.net/lists/stable/msg155406.html
Please always tell me stuff like this, I have the short-term memory of a squirrel...
thanks,
greg k-h
Quoted 69 lines · Greg Kroah-Hartman
Quoted 68 lines · Imre Deak
Quoted 67 lines · Greg Kroah-Hartman
Quoted 65 lines · Imre Deak
commit
848496e5902833600f7992f4faa82dc1546051ba
Author: Ville Syrjälä <[email protected]> Date: Wed Jul 13 16:32:03 2016 +0300drm/i915: Wait up to 3ms for the pcu to ack the cdclk change request on SKL
increased the timeout to match the spec, but we still see a timeout on at least one SKL. A CDCLK change request following the failed one will succeed nevertheless.
I could reproduce this problem easily by running
kms_pipe_crc_basicin a loop. In all failure cases_wait_for()was pre-empted for >3ms and so in the worst case - when the pre-emption happened right after calculatingtimeout__in_wait_for()- we calledskl_cdclk_wait_for_pcu_ready()only once which failed and so_wait_for()timed out. As opposed to this the spec says to keep retrying the request for at most a 3ms period.To fix this send the first request explicitly to guarantee that there is 3ms between the first and last request. Though this matches the spec, I noticed that in rare cases this can still time out if we sent only a few requests (in the worst case 2) _and_ PCODE is busy for some reason even after a previous request and a 3ms delay. To work around this retry the polling with pre-emption disabled to maximize the number of requests. Also increase the timeout to 10ms to account for interrupts that could reduce the number of requests. With this change I couldn't trigger the problem.
v2:
- Use 1ms poll period instead of 10us. (Chris)
v3:
- Poll with pre-emption disabled to increase the number of request attempts. (Ville, Chris)
- Factor out a helper to poll, it's also needed by the next patch. v4:
- Passreply_mask, reply toskl_pcode_request(), instead of assuming the reply is generic. (Ville)
v5:
- List the request specific timeout values as code comment. (Ville) v6:
- Try the poll first with preemption enabled.
- Add code comment about first request being queued by PCODE. (Art)
- Addtimeout_base_msargument. (Ville)
v7:
- Clarify code comment about first queued request. (Chris) v8:
- Rebased on 4.9.2Cc Ville Syrjälä <[email protected]>Cc Chris Wilson <[email protected]>Cc Art Runyan <[email protected]>Cc <[email protected]> # v4.2- : 3b2c171 : drm/i915: Wait up to 3msCc <[email protected]> # v4.2-Fixes 5d96d8afcfbb ("drm/i915/skl: Deinit/init the display at suspend/resume")Reference: https://bugs.freedesktop.org/show_bug.cgi?id=97929
Testcase: igt/kms_pipe_crc_basic/suspend-read-crc-pipe-BSigned-off-by Imre Deak <[email protected]>Reviewed-by Chris Wilson <[email protected]>(cherry picked from commit a0b8a1fe34430c3a82258e8cb45f5968bdf31afd)Signed-off-by Jani Nikula <[email protected]>---
drivers/gpu/drm/i915/i915_drv.h | 2 +drivers/gpu/drm/i915/intel_display.c| 31 +++++---------- drivers/gpu/drm/i915/intel_pm.c | 75 ++++++++++++++++++++++++++++++++++++3 files changed, 87 insertions(+), 21 deletions(-)
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines
What tree is this patch for? Please give us a hint, we don't like guessing...
It's for the 4.9 stable tree and for other stable trees starting from 4.2. It's the upstream commit
a0b8a1fe34430c3a82258e8cb45f5968bdf31afdThere is no such git commit id in Linus's tree :(
Ah, sorry didn't notice that this SHA is specific to the drm-tip tree.
Quoted 1 line
Do you mean
2c7d0602c815277f7cb7c932b091288710d8aba7?
Yes.
Quoted 5 lines
Quoted 2 lines
rebased on top of 4.9.2 according to our discussion at
http://www.spinics.net/lists/stable/msg155406.htmlPlease always tell me stuff like this, I have the short-term memory of a squirrel...
Ok, will do.
--Imre