2012-06-25 11:50:44

by Stefan Bader

[permalink] [raw]
Subject: [2.6.32+drm33-longterm] Patch "drm/i915: Attempt to fix watermark setup on 85x (v2)" has been added to staging queue

This is a note to let you know that I have just added a patch titled

drm/i915: Attempt to fix watermark setup on 85x (v2)

to the drm-next branch of the 2.6.32+drm33-longterm tree which can be found at

http://git.kernel.org/?p=linux/kernel/git/smb/linux-2.6.32.y-drm33.z.git;a=shortlog;h=refs/heads/drm-next
or
http://kernel.ubuntu.com/git?p=smb/linux-2.6.32.y-drm33.z.git;a=shortlog;h=refs/heads/drm-next

If you, or anyone else, feels it should not be added to the drm33-longterm tree,
please reply to this email not later than 8 days after this email was sent.

Thanks.
-Stefan

------

>From 8f4695ed1c9e068772bcce4cd4ff03f88d57a008 Mon Sep 17 00:00:00 2001
From: Adam Jackson <[email protected]>
Date: Fri, 16 Apr 2010 18:20:57 -0400
Subject: [PATCH] drm/i915: Attempt to fix watermark setup on 85x (v2)

IS_MOBILE() catches 85x, so we'd always try to use the 9xx FIFO sizing;
since there's an explicit 85x version, this seems wrong.

v2: Handle 830m correctly too.

Signed-off-by: Adam Jackson <[email protected]>
Reviewed-by: Eric Anholt <[email protected]>
Signed-off-by: Eric Anholt <[email protected]>

BugLink: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=661696
Signed-off-by: Jonathan Nieder <[email protected]>
(cherry-picked from commit 8f4695ed1c9e068772bcce4cd4ff03f88d57a008 upstream)
Signed-off-by: Stefan Bader <[email protected]>
---
drivers/gpu/drm/i915/intel_display.c | 11 ++++++-----
1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index e7356fb..c7502b6 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4853,17 +4853,18 @@ static void intel_init_display(struct drm_device *dev)
dev_priv->display.update_wm = g4x_update_wm;
else if (IS_I965G(dev))
dev_priv->display.update_wm = i965_update_wm;
- else if (IS_I9XX(dev) || IS_MOBILE(dev)) {
+ else if (IS_I9XX(dev)) {
dev_priv->display.update_wm = i9xx_update_wm;
dev_priv->display.get_fifo_size = i9xx_get_fifo_size;
+ } else if (IS_I85X(dev)) {
+ dev_priv->display.update_wm = i9xx_update_wm;
+ dev_priv->display.get_fifo_size = i85x_get_fifo_size;
} else {
- if (IS_I85X(dev))
- dev_priv->display.get_fifo_size = i85x_get_fifo_size;
- else if (IS_845G(dev))
+ dev_priv->display.update_wm = i830_update_wm;
+ if (IS_845G(dev))
dev_priv->display.get_fifo_size = i845_get_fifo_size;
else
dev_priv->display.get_fifo_size = i830_get_fifo_size;
- dev_priv->display.update_wm = i830_update_wm;
}
}

--
1.7.9.5


2012-06-26 04:15:08

by Jonathan Nieder

[permalink] [raw]
Subject: [2.6.32.y] drm/i915: Attempt to fix watermark setup on 85x (v2)

From: Adam Jackson <[email protected]>
Date: Fri, 16 Apr 2010 18:20:57 -0400

commit 8f4695ed1c9e068772bcce4cd4ff03f88d57a008 upstream.

IS_MOBILE() catches 85x, so we'd always try to use the 9xx FIFO sizing;
since there's an explicit 85x version, this seems wrong.

v2: Handle 830m correctly too.

[jn: backport to 2.6.32.y to address
https://bugzilla.kernel.org/show_bug.cgi?id=42839]

Signed-off-by: Adam Jackson <[email protected]>
Signed-off-by: Eric Anholt <[email protected]>
Signed-off-by: Jonathan Nieder <[email protected]>
---
Hi Willy,

Lukas Anzinger wrote[1]:

> I've encountered a nasty bug in the drm/i915 part of Debian's stable kernel.
> The code is also present in 2.6.33.20, which is the latest stable upstream
> release of 2.6.33.y at the time of writing.
>
> During a refactoring of the i915 driver a regression has been introduced
> (commit e70236a8d3d0a4c100a0b9f7d394d9bda9c56aca): For some chipsets the wrong
> fifo size is determined which results in lot's of pixel errors when starting
> the xserver and choosing 800x600 as a resolution. If another resolution is used
> (eg. 1024x768 or 1280x1024), I don't encounter this problem.

Linux 2.6.32.y is presumably also affected because the problematic
commit was applied during the 2.6.32 merge window.

e70236a8d3d0 "drm/i915: split display functions by chip type",
2009-09-21

Lukas tested the fix against a kernel with 2.6.33.y-based drm
subsystem and found it to work:

8f4695ed1c9e drm/i915: Attempt to fix watermark setup on 85x (v2),
2010-04-16

and therefore Stefan is staging this fix for 2.6.32.y+drm33.z
(thanks!). Newer stable lines work fine already because the fix hit
mainline in v2.6.34-rc6.

The patch applies cleanly to v2.6.32.59 and is included below for
reference. I don't know if anyone has tested it there, though. All I
can say for now is that it is tested against 2.6.33-based kernels and
looks obviously correct. What do you think?

Hope that helps,
Jonathan

[1] https://bugzilla.kernel.org/show_bug.cgi?id=42839

drivers/gpu/drm/i915/intel_display.c | 11 ++++++-----
1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 79cc437af3b8..25b3e903c67c 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4355,17 +4355,18 @@ static void intel_init_display(struct drm_device *dev)
dev_priv->display.update_wm = g4x_update_wm;
else if (IS_I965G(dev))
dev_priv->display.update_wm = i965_update_wm;
- else if (IS_I9XX(dev) || IS_MOBILE(dev)) {
+ else if (IS_I9XX(dev)) {
dev_priv->display.update_wm = i9xx_update_wm;
dev_priv->display.get_fifo_size = i9xx_get_fifo_size;
+ } else if (IS_I85X(dev)) {
+ dev_priv->display.update_wm = i9xx_update_wm;
+ dev_priv->display.get_fifo_size = i85x_get_fifo_size;
} else {
- if (IS_I85X(dev))
- dev_priv->display.get_fifo_size = i85x_get_fifo_size;
- else if (IS_845G(dev))
+ dev_priv->display.update_wm = i830_update_wm;
+ if (IS_845G(dev))
dev_priv->display.get_fifo_size = i845_get_fifo_size;
else
dev_priv->display.get_fifo_size = i830_get_fifo_size;
- dev_priv->display.update_wm = i830_update_wm;
}
}

--
1.7.11.rc3

2012-06-26 16:21:43

by Willy Tarreau

[permalink] [raw]
Subject: Re: [2.6.32.y] drm/i915: Attempt to fix watermark setup on 85x (v2)

Hi Jonathan,

On Mon, Jun 25, 2012 at 11:14:58PM -0500, Jonathan Nieder wrote:
> From: Adam Jackson <[email protected]>
> Date: Fri, 16 Apr 2010 18:20:57 -0400
>
> commit 8f4695ed1c9e068772bcce4cd4ff03f88d57a008 upstream.
>
> IS_MOBILE() catches 85x, so we'd always try to use the 9xx FIFO sizing;
> since there's an explicit 85x version, this seems wrong.
>
> v2: Handle 830m correctly too.
>
> [jn: backport to 2.6.32.y to address
> https://bugzilla.kernel.org/show_bug.cgi?id=42839]

(...)

Thank you, I've queued it for 2.6.32.60.

Best regards,
Willy