]> git.ipfire.org Git - thirdparty/kernel/stable-queue.git/commitdiff
Drop drm-i915-implement-waedplinkratedatareload.patch from 5.4
authorSasha Levin <sashal@kernel.org>
Tue, 20 Sep 2022 13:42:03 +0000 (09:42 -0400)
committerSasha Levin <sashal@kernel.org>
Tue, 20 Sep 2022 13:42:13 +0000 (09:42 -0400)
queue-5.4/drm-i915-implement-waedplinkratedatareload.patch [deleted file]
queue-5.4/series

diff --git a/queue-5.4/drm-i915-implement-waedplinkratedatareload.patch b/queue-5.4/drm-i915-implement-waedplinkratedatareload.patch
deleted file mode 100644 (file)
index 2f5b846..0000000
+++ /dev/null
@@ -1,95 +0,0 @@
-From 36c9ca7e117d715e270e5151e4c9ecf97f565e6e Mon Sep 17 00:00:00 2001
-From: Sasha Levin <sashal@kernel.org>
-Date: Fri, 2 Sep 2022 10:03:18 +0300
-Subject: drm/i915: Implement WaEdpLinkRateDataReload
-MIME-Version: 1.0
-Content-Type: text/plain; charset=UTF-8
-Content-Transfer-Encoding: 8bit
-
-From: Ville Syrjälä <ville.syrjala@linux.intel.com>
-
-[ Upstream commit 672d6ca758651f0ec12cd0d59787067a5bde1c96 ]
-
-A lot of modern laptops use the Parade PS8461E MUX for eDP
-switching. The MUX can operate in jitter cleaning mode or
-redriver mode, the first one resulting in higher link
-quality. The jitter cleaning mode needs to know the link
-rate used and the MUX achieves this by snooping the
-LINK_BW_SET, LINK_RATE_SELECT and SUPPORTED_LINK_RATES
-DPCD accesses.
-
-When the MUX is powered down (seems this can happen whenever
-the display is turned off) it loses track of the snooped
-link rates so when we do the LINK_RATE_SELECT write it no
-longer knowns which link rate we're selecting, and thus it
-falls back to the lower quality redriver mode. This results
-in unstable high link rates (eg. usually 8.1Gbps link rate
-no longer works correctly).
-
-In order to avoid all that let's re-snoop SUPPORTED_LINK_RATES
-from the sink at the start of every link training.
-
-Unfortunately we don't have a way to detect the presence of
-the MUX. It looks like the set of laptops equipped with this
-MUX is fairly large and contains devices from multiple
-manufacturers. It may also still be growing with new models.
-So a quirk doesn't seem like a very easily maintainable
-option, thus we shall attempt to do this unconditionally on
-all machines that use LINK_RATE_SELECT. Hopefully this extra
-DPCD read doesn't cause issues for any unaffected machine.
-If that turns out to be the case we'll need to convert this
-into a quirk in the future.
-
-Cc: stable@vger.kernel.org
-Cc: Jason A. Donenfeld <Jason@zx2c4.com>
-Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
-Cc: Jani Nikula <jani.nikula@intel.com>
-Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/6205
-Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
-Link: https://patchwork.freedesktop.org/patch/msgid/20220902070319.15395-1-ville.syrjala@linux.intel.com
-Tested-by: Aaron Ma <aaron.ma@canonical.com>
-Tested-by: Jason A. Donenfeld <Jason@zx2c4.com>
-Reviewed-by: Jani Nikula <jani.nikula@intel.com>
-(cherry picked from commit 25899c590cb5ba9b9f284c6ca8e7e9086793d641)
-Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
-Signed-off-by: Sasha Levin <sashal@kernel.org>
----
- .../drm/i915/display/intel_dp_link_training.c | 22 +++++++++++++++++++
- 1 file changed, 22 insertions(+)
-
-diff --git a/drivers/gpu/drm/i915/display/intel_dp_link_training.c b/drivers/gpu/drm/i915/display/intel_dp_link_training.c
-index 2a1130dd1ad0..774f910364ab 100644
---- a/drivers/gpu/drm/i915/display/intel_dp_link_training.c
-+++ b/drivers/gpu/drm/i915/display/intel_dp_link_training.c
-@@ -142,6 +142,28 @@ intel_dp_link_training_clock_recovery(struct intel_dp *intel_dp)
-       intel_dp_compute_rate(intel_dp, intel_dp->link_rate,
-                             &link_bw, &rate_select);
-+      /*
-+       * WaEdpLinkRateDataReload
-+       *
-+       * Parade PS8461E MUX (used on varius TGL+ laptops) needs
-+       * to snoop the link rates reported by the sink when we
-+       * use LINK_RATE_SET in order to operate in jitter cleaning
-+       * mode (as opposed to redriver mode). Unfortunately it
-+       * loses track of the snooped link rates when powered down,
-+       * so we need to make it re-snoop often. Without this high
-+       * link rates are not stable.
-+       */
-+      if (!link_bw) {
-+              struct intel_connector *connector = intel_dp->attached_connector;
-+              __le16 sink_rates[DP_MAX_SUPPORTED_RATES];
-+
-+              drm_dbg_kms(&i915->drm, "[CONNECTOR:%d:%s] Reloading eDP link rates\n",
-+                          connector->base.base.id, connector->base.name);
-+
-+              drm_dp_dpcd_read(&intel_dp->aux, DP_SUPPORTED_LINK_RATES,
-+                               sink_rates, sizeof(sink_rates));
-+      }
-+
-       if (link_bw)
-               DRM_DEBUG_KMS("Using LINK_BW_SET value %02x\n", link_bw);
-       else
--- 
-2.35.1
-
index 4e01c43a3ed900712ee769d38027a47f04bcda95..fee155f86d9bf63095ffce35b2b45170628a7b77 100644 (file)
@@ -7,6 +7,5 @@ parisc-ccio-dma-add-missing-iounmap-in-error-path-in.patch
 efi-libstub-disable-shadow-call-stack.patch
 efi-libstub-disable-struct-randomization.patch
 alsa-pcm-oss-fix-race-at-sndctl_dsp_sync.patch
-drm-i915-implement-waedplinkratedatareload.patch
 task_stack-x86-cea-force-inline-stack-helpers.patch
 tracing-hold-caller_addr-to-hardirq_-enable-disable-.patch