[Intel-gfx] [PATCH] drm/i915: disable VGA mem when reenabling the power well

[Paulo Zanoni]

From: Paulo Zanoni <paulo.r.zanoni at intel.com>

Fixes regression introduced by:
    commit bf51d5e2cda5d36d98e4b46ac7fca9461e512c41
    Author: Paulo Zanoni <paulo.r.zanoni at intel.com>
    Date:   Wed Jul 3 17:12:13 2013 -0300
        drm/i915: switch disable_power_well default value to 1

The bug I'm seeing can be reproduced with:
- Boot your Haswell machine
- Make sure the power well gets disabled, then enabled. You can check
  this by seeing the messages print by hsw_set_power_well
- Stop your display manager
- echo 0 > /sys/class/vtconsole/vtcon1/bind

I can easily reproduce this by blacklising snd_hda_intel and booting
with eDP+HDMI.

If you do this and then look at dmesg, you'll see we're printing
infinite "Unclaimed register" messages. This is happening because
we're stuck on an infinite loop inside console_unlock(), which is
calling many functions from vgacon.c. And the code that's triggering
the error messages is from vgacon_set_cursor_size().

For some reason, what fixes the problem is the code that clears
VGA_MSR_MEM_EN from VGA_MSR_WRITE. If you add some debug messages,
you'll see that we're reading a value of 0 and writing 0 back to it. I
really don't know why this fixes the problem, but I am sure that if
you remove the code line that writes 0 back the bug won't stop. I
suspect this is probably some problem with how the hardware gets
reinitialized when we reenable the power well.

Also notice that function i915_disable_vga_mem already existed and was
removed on commit 6e1b4fdad5157bb9e88777d525704aba24389bee, I just
brought it back. Notice that the intel_drv.h declaration was still
there too.

I know some people will probably request to deeply investigate why
exactly vgacon is doing the wrong thing, and ask me to replace the
"for (;;)" that exists inside console_unlock() with something else,
but my goal here is to just fix the regression. I don't think it's
worth spending that much time on vgacon code. And this is our
regression anyway.

Notice that this is just a partial fix to fd.o #67813. This fixes the
case where the power well is already enabled when we unbind, not when
it's disabled when we unbind.

V2: - Rebase (first version was sent in September).

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67813
Signed-off-by: Paulo Zanoni <paulo.r.zanoni at intel.com>
---
 drivers/gpu/drm/i915/intel_display.c | 12 ++++++++++++
 drivers/gpu/drm/i915/intel_pm.c      |  2 ++
 2 files changed, 14 insertions(+)

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 45a87d1..bf22bea 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -11064,6 +11064,18 @@ void i915_redisable_vga(struct drm_device *dev)
 	}
 }
 
+void i915_disable_vga_mem(struct drm_device *dev)
+{
+	if (HAS_PCH_SPLIT(dev)) {
+		vga_get_uninterruptible(dev->pdev, VGA_RSRC_LEGACY_IO);
+		outb(inb(VGA_MSR_READ) & ~VGA_MSR_MEM_EN, VGA_MSR_WRITE);
+		vga_set_legacy_decoding(dev->pdev, VGA_RSRC_LEGACY_IO |
+						   VGA_RSRC_NORMAL_IO |
+						   VGA_RSRC_NORMAL_MEM);
+		vga_put(dev->pdev, VGA_RSRC_LEGACY_IO);
+	}
+}
+
 static void intel_modeset_readout_hw_state(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index e6d98fe..7ebe1bd 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -5703,6 +5703,8 @@ static void hsw_set_power_well(struct drm_device *dev,
 			if (wait_for((I915_READ(HSW_PWR_WELL_DRIVER) &
 				      HSW_PWR_WELL_STATE_ENABLED), 20))
 				DRM_ERROR("Timeout enabling power well\n");
+
+			i915_disable_vga_mem(dev);
 		}
 
 		if (IS_BROADWELL(dev)) {
-- 
1.8.3.1