From 2cac613be8d4d661edd359cdab3c474286c4f5f0 Mon Sep 17 00:00:00 2001 From: Imre Deak Date: Thu, 30 Jan 2014 16:50:42 +0200 Subject: drm/i915: vlv: fix DP PHY lockup due to invalid PP sequencer setup MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Atm we setup the HW panel power sequencer logic both for eDP and DP ports. On eDP we then go on and start the power on sequence and commence with link training when it's ready. On DP we don't do the power on sequencing but do the link training immediately. At this point the DP PHY block gets stuck, since - supposedly - it is waiting for the power on sequence to finish. The actual register write that seems to hold off the PHY is PIPEX_PP_ON_DELAYS[Panel Control Port Select]. Writing here a non-0 value eventually sets PIPEX_PP_STATUS[Require Asset Status] to 1 and blocks the PHY until the panel power on is ready. Fix this by not doing any PP sequencing setup for DP ports. Thanks to Ville Syrjälä, Jesse Barnes and Todd Previte for the help in tracking this down. Note that on older gmch platforms (where we have lvds instead of edp) we've hacked around this by writing the magic ABCD unlock key to PP registers, which disables the hw sanity checks. For edp all platforms thus far had the pch split, with the edp port in the north display complex and the PP registers on the pch the hw sanity checks (expressed through the "Require Asset Status" bit) was never functional, hence never a real issue. This regression has been introduce in commit bf13e81b904a37d94d83dd6c3b53a147719a3ead Author: Jani Nikula Date: Fri Sep 6 07:40:05 2013 +0300 drm/i915: add support for per-pipe power sequencing on vlv Signed-off-by: Imre Deak [danvet: Add note about the bigger story here.] Cc: stable@vger.kernel.org Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_dp.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c index 5ede4e8e290d..c00b6e352c2b 100644 --- a/drivers/gpu/drm/i915/intel_dp.c +++ b/drivers/gpu/drm/i915/intel_dp.c @@ -1869,10 +1869,12 @@ static void vlv_pre_enable_dp(struct intel_encoder *encoder) mutex_unlock(&dev_priv->dpio_lock); - /* init power sequencer on this pipe and port */ - intel_dp_init_panel_power_sequencer(dev, intel_dp, &power_seq); - intel_dp_init_panel_power_sequencer_registers(dev, intel_dp, - &power_seq); + if (is_edp(intel_dp)) { + /* init power sequencer on this pipe and port */ + intel_dp_init_panel_power_sequencer(dev, intel_dp, &power_seq); + intel_dp_init_panel_power_sequencer_registers(dev, intel_dp, + &power_seq); + } intel_enable_dp(encoder); -- cgit v1.2.3 From bdde5c6a258a702bdfa7d1f4ae804a7bc405e788 Mon Sep 17 00:00:00 2001 From: Jani Nikula Date: Fri, 31 Jan 2014 15:49:08 +0200 Subject: drm/i915: demote opregion excessive timeout WARN_ONCE to DRM_INFO_ONCE The WARN_ONCE is a bit too verbose, make it a DRM_INFO_ONCE. While at it, add a #define for MAX_DSLP and make the message a bit more informative. v2: use DRM_INFO_ONCE, add MAX_DSLP, pimp the message. Suggested-by: Chris Wilson Signed-off-by: Jani Nikula Reviewed-by: Chris Wilson Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_opregion.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/i915/intel_opregion.c b/drivers/gpu/drm/i915/intel_opregion.c index 37e9a96777ef..68459605bd12 100644 --- a/drivers/gpu/drm/i915/intel_opregion.c +++ b/drivers/gpu/drm/i915/intel_opregion.c @@ -227,6 +227,8 @@ struct opregion_asle { #define ACPI_DIGITAL_OUTPUT (3<<8) #define ACPI_LVDS_OUTPUT (4<<8) +#define MAX_DSLP 1500 + #ifdef CONFIG_ACPI static int swsci(struct drm_device *dev, u32 function, u32 parm, u32 *parm_out) { @@ -261,10 +263,11 @@ static int swsci(struct drm_device *dev, u32 function, u32 parm, u32 *parm_out) /* The spec says 2ms should be the default, but it's too small * for some machines. */ dslp = 50; - } else if (dslp > 500) { + } else if (dslp > MAX_DSLP) { /* Hey bios, trust must be earned. */ - WARN_ONCE(1, "excessive driver sleep timeout (DSPL) %u\n", dslp); - dslp = 500; + DRM_INFO_ONCE("ACPI BIOS requests an excessive sleep of %u ms, " + "using %u ms instead\n", dslp, MAX_DSLP); + dslp = MAX_DSLP; } /* The spec tells us to do this, but we are the only user... */ -- cgit v1.2.3