releases/4.10.11/drm-i915-gen9-increase-pcode-request-timeout-to-50ms.patch

   1 From d253371c4c2f5fc2d884ef25f64decd7549aff5a Mon Sep 17 00:00:00 2001
   2 From: Imre Deak <imre.deak@intel.com>
   3 Date: Fri, 24 Feb 2017 16:32:10 +0200
   4 Subject: drm/i915/gen9: Increase PCODE request timeout to 50ms
   5 MIME-Version: 1.0
   6 Content-Type: text/plain; charset=UTF-8
   7 Content-Transfer-Encoding: 8bit
   8
   9 From: Imre Deak <imre.deak@intel.com>
  10
  11 commit d253371c4c2f5fc2d884ef25f64decd7549aff5a upstream.
  12
  13 After
  14 commit 2c7d0602c815277f7cb7c932b091288710d8aba7
  15 Author: Imre Deak <imre.deak@intel.com>
  16 Date:   Mon Dec 5 18:27:37 2016 +0200
  17
  18     drm/i915/gen9: Fix PCODE polling during CDCLK change notification
  19
  20 there is still one report of the CDCLK-change request timing out on a
  21 KBL machine, see the Reference link. On that machine the maximum time
  22 the request took to succeed was 34ms, so increase the timeout to 50ms.
  23
  24 v2:
  25 - Change timeout from 100 to 50 ms to maintain the current 50 ms limit
  26   for atomic waits in the driver. (Chris, Tvrtko)
  27
  28 Reference: https://bugs.freedesktop.org/show_bug.cgi?id=99345
  29 Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
  30 Cc: Chris Wilson <chris@chris-wilson.co.uk>
  31 Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
  32 Signed-off-by: Imre Deak <imre.deak@intel.com>
  33 Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
  34 Link: http://patchwork.freedesktop.org/patch/msgid/1487946730-17162-1-git-send-email-imre.deak@intel.com
  35 (cherry picked from commit 0129936ddda26afd5d9d207c4e86b2425952579f)
  36 Signed-off-by: Jani Nikula <jani.nikula@intel.com>
  37 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
  38
  39 ---
  40  drivers/gpu/drm/i915/intel_pm.c |   11 ++++++-----
  41  1 file changed, 6 insertions(+), 5 deletions(-)
  42
  43 --- a/drivers/gpu/drm/i915/intel_pm.c
  44 +++ b/drivers/gpu/drm/i915/intel_pm.c
  45 @@ -7895,10 +7895,10 @@ static bool skl_pcode_try_request(struct
  46   * @timeout_base_ms: timeout for polling with preemption enabled
  47   *
  48   * Keep resending the @request to @mbox until PCODE acknowledges it, PCODE
  49 - * reports an error or an overall timeout of @timeout_base_ms+10 ms expires.
  50 + * reports an error or an overall timeout of @timeout_base_ms+50 ms expires.
  51   * The request is acknowledged once the PCODE reply dword equals @reply after
  52   * applying @reply_mask. Polling is first attempted with preemption enabled
  53 - * for @timeout_base_ms and if this times out for another 10 ms with
  54 + * for @timeout_base_ms and if this times out for another 50 ms with
  55   * preemption disabled.
  56   *
  57   * Returns 0 on success, %-ETIMEDOUT in case of a timeout, <0 in case of some
  58 @@ -7934,14 +7934,15 @@ int skl_pcode_request(struct drm_i915_pr
  59          * worst case) _and_ PCODE was busy for some reason even after a
  60          * (queued) request and @timeout_base_ms delay. As a workaround retry
  61          * the poll with preemption disabled to maximize the number of
  62 -        * requests. Increase the timeout from @timeout_base_ms to 10ms to
  63 +        * requests. Increase the timeout from @timeout_base_ms to 50ms to
  64          * account for interrupts that could reduce the number of these
  65 -        * requests.
  66 +        * requests, and for any quirks of the PCODE firmware that delays
  67 +        * the request completion.
  68          */
  69         DRM_DEBUG_KMS("PCODE timeout, retrying with preemption disabled\n");
  70         WARN_ON_ONCE(timeout_base_ms > 3);
  71         preempt_disable();
  72 -       ret = wait_for_atomic(COND, 10);
  73 +       ret = wait_for_atomic(COND, 50);
  74         preempt_enable();
  75
  76  out: