drm/i915/psr: Display WA 0884 applied broadly for more HW tracking.
authorRodrigo Vivi <rodrigo.vivi@intel.com>
Fri, 9 Mar 2018 00:52:18 +0000 (16:52 -0800)
committerRodrigo Vivi <rodrigo.vivi@intel.com>
Mon, 12 Mar 2018 18:43:49 +0000 (11:43 -0700)
WA 0884:bxt:all,cnl:*:A - "When FBC is enabled with eDP PSR,
the CPU host modify writes may not get updated on the Display
as expected.
WA: Write 0x00000000 to CUR_SURFLIVE_A with every CPU
host modify write to trigger PSR exit."

We can also find on spec other cases where they describe
bogus writes to cursor registers to force PSR exit with
HW tracking. And it was confirmed by HW engineers that
this Wa can be safely applied for any frontbuffer activity.

So let's use this more and more here instead of forcibly
disable and re-enable PSR everytime that we have a simple
reliable flush case.

Other commits improve the fbcon/fbdev use a lot, but this
approach is the only when where we can get a fully reliable
console with no slowness or missed frames and PSR still
enabled and active.

v2: - Rebase on drm-tip
    - (DK) Add a comment to explain that WA
    tells about writing 0 to CUR_SURFLIVE_A but we write to
    CUR_SURFLIVE(pipe).
v3: Wa doesn't work on PSR2.

Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180309005218.26772-1-rodrigo.vivi@intel.com
drivers/gpu/drm/i915/i915_reg.h
drivers/gpu/drm/i915/intel_psr.c

index 9e76546..60febfb 100644 (file)
@@ -6032,6 +6032,9 @@ enum {
 #define IVB_CURSOR_B_OFFSET 0x71080
 #define IVB_CURSOR_C_OFFSET 0x72080
 
+#define _CUR_SURLIVE           0x700AC
+#define CUR_SURLIVE(pipe)      _CURSOR2(pipe, _CUR_SURLIVE)
+
 /* Display A control */
 #define _DSPACNTR                              0x70180
 #define   DISPLAY_PLANE_ENABLE                 (1<<31)
index 23175c5..975ebb5 100644 (file)
@@ -1027,8 +1027,23 @@ void intel_psr_flush(struct drm_i915_private *dev_priv,
        dev_priv->psr.busy_frontbuffer_bits &= ~frontbuffer_bits;
 
        /* By definition flush = invalidate + flush */
-       if (frontbuffer_bits)
-               intel_psr_exit(dev_priv);
+       if (frontbuffer_bits) {
+               if (dev_priv->psr.psr2_support ||
+                   IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) {
+                       intel_psr_exit(dev_priv);
+               } else {
+                       /*
+                        * Display WA #0884: all
+                        * This documented WA for bxt can be safely applied
+                        * broadly so we can force HW tracking to exit PSR
+                        * instead of disabling and re-enabling.
+                        * Workaround tells us to write 0 to CUR_SURLIVE_A,
+                        * but it makes more sense write to the current active
+                        * pipe.
+                        */
+                       I915_WRITE(CUR_SURLIVE(pipe), 0);
+               }
+       }
 
        if (!dev_priv->psr.active && !dev_priv->psr.busy_frontbuffer_bits)
                if (!work_busy(&dev_priv->psr.work.work))