Skip to contentSkip to conversation

Intel 915GM MCHBAR bug

DISCUSSION Pekka Enberg 7 participants 14 messages Jun 5, 2009 – Jun 29, 2009 2 messages not in the archive

Subsystems (2)
lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): 20090604112738.4c01d35e@jbarnes-x200

#2 lore raw

Hi Jesse,

(I am cc'ing linux-kernel.)

On Mon, 01 Jun 2009 22:31:13 +0300

Quoted 22 lines
Quoted 20 lines · Pekka Enberg

Hi Jesse,

I am seeing this on my Eee PC 701 which is running a Ubuntu 2.6.28-11.42-generic stock kernel:

[   76.826966] [drm:i915_gem_detect_bit_6_swizzle] *ERROR* Couldn't
read from MCHBAR.  Disabling tiling.
[   76.827215] [drm] Initialized i915 1.6.0 20080730 on minor 0
[   76.834037] [drm:i915_setparam] *ERROR* unknown parameter 4
[   76.834104] [drm:i915_getparam] *ERROR* Unknown parameter 6

Google turned up this patch

http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html

but I don't seem to find it mainline kernel.

Was the bug fixed in some other way? It seem that distributions have not yet picked up your patch and I am unsure if it's in any of the -stable kernels.

Quoted 3 lines · Jesse Barnes

I think Eric acked it but may not have pushed it to drm-intel-next yet. Should happen in the next week or two though as we prepare the merge window series.

OK, thanks for the info! FYI, I tested 2.6.30-rc8 with your patch applied and everything works smoothly on my EeePC 701. I did not test plain 2.6.30-rc8 as I expect it to show same kind of behavior as Ubuntu distro kernel. Is this correct or do you want me to test 2.6.30-rc8 too?

I wonder why the patch hasn't received more attention as it's a pretty critical bug. The _default_ Ubuntu 9.04 netbook remix installation is completely broken for EeePC 701 (and probably others as well) without the patch. And it's not as if I'm the only one that suffers from it. A quick Google search reveals that a lot of people are hitting it.

So I really do think we need to merge this patch to the upcoming 2.6.31 ASAP and backport it to -stable after it has gotten some more testing.

Pekka

#3 lore raw
Quoted 38 lines · On Fri, 05 Jun 2009 11:02:21 +0300 Pekka Enberg <[email protected]> wrote:

Hi Jesse,

(I am cc'ing linux-kernel.)

On Mon, 01 Jun 2009 22:31:13 +0300

Quoted 22 lines
Quoted 20 lines · Pekka Enberg

Hi Jesse,

I am seeing this on my Eee PC 701 which is running a Ubuntu 2.6.28-11.42-generic stock kernel:

[   76.826966] [drm:i915_gem_detect_bit_6_swizzle] *ERROR* Couldn't
read from MCHBAR.  Disabling tiling.
[   76.827215] [drm] Initialized i915 1.6.0 20080730 on minor 0
[   76.834037] [drm:i915_setparam] *ERROR* unknown parameter 4
[   76.834104] [drm:i915_getparam] *ERROR* Unknown parameter 6

Google turned up this patch

http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html

but I don't seem to find it mainline kernel.

Was the bug fixed in some other way? It seem that distributions have not yet picked up your patch and I am unsure if it's in any of the -stable kernels.

Quoted 3 lines · Jesse Barnes

I think Eric acked it but may not have pushed it to drm-intel-next yet. Should happen in the next week or two though as we prepare the merge window series.

OK, thanks for the info! FYI, I tested 2.6.30-rc8 with your patch applied and everything works smoothly on my EeePC 701. I did not test plain 2.6.30-rc8 as I expect it to show same kind of behavior as Ubuntu distro kernel. Is this correct or do you want me to test 2.6.30-rc8 too?

Quoted 9 lines

I wonder why the patch hasn't received more attention as it's a pretty critical bug. The _default_ Ubuntu 9.04 netbook remix installation is completely broken for EeePC 701 (and probably others as well) without the patch. And it's not as if I'm the only one that suffers from it. A quick Google search reveals that a lot of people are hitting it.

So I really do think we need to merge this patch to the upcoming 2.6.31 ASAP and backport it to -stable after it has gotten some more testing.

Wanna show us the patch?

Because the world could certainly do with more i915 bugfixes :(

#4 lore raw
Quoted 55 lines · Andrew Morton
Quoted 40 lines · On Fri, 05 Jun 2009 11:02:21 +0300 Pekka Enberg <[email protected]> wrote:

Hi Jesse,

(I am cc'ing linux-kernel.)

On Mon, 01 Jun 2009 22:31:13 +0300

Quoted 23 lines
Quoted 20 lines · Pekka Enberg

Hi Jesse,

I am seeing this on my Eee PC 701 which is running a Ubuntu 2.6.28-11.42-generic stock kernel:

[   76.826966] [drm:i915_gem_detect_bit_6_swizzle] *ERROR* Couldn't
read from MCHBAR.  Disabling tiling.
[   76.827215] [drm] Initialized i915 1.6.0 20080730 on minor 0
[   76.834037] [drm:i915_setparam] *ERROR* unknown parameter 4
[   76.834104] [drm:i915_getparam] *ERROR* Unknown parameter 6

Google turned up this patch

http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html

but I don't seem to find it mainline kernel.

Was the bug fixed in some other way? It seem that distributions have not yet picked up your patch and I am unsure if it's in any of the -stable kernels.

Quoted 4 lines · Jesse Barnes

I think Eric acked it but may not have pushed it to drm-intel-next yet. Should happen in the next week or two though as we prepare the merge window series.

OK, thanks for the info! FYI, I tested 2.6.30-rc8 with your patch applied and everything works smoothly on my EeePC 701. I did not test plain 2.6.30-rc8 as I expect it to show same kind of behavior as Ubuntu distro kernel. Is this correct or do you want me to test 2.6.30-rc8 too?

Quoted 10 lines

I wonder why the patch hasn't received more attention as it's a pretty critical bug. The _default_ Ubuntu 9.04 netbook remix installation is completely broken for EeePC 701 (and probably others as well) without the patch. And it's not as if I'm the only one that suffers from it. A quick Google search reveals that a lot of people are hitting it.

So I really do think we need to merge this patch to the upcoming 2.6.31 ASAP and backport it to -stable after it has gotten some more testing.

Quoted 9 lines

Wanna show us the patch?

Because the world could certainly do with more i915 bugfixes :( --
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Here is Jesse's patch from http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html adopted for 2.6.30-rc8... (needed to redo hunk #3 for i915_gem_tile.c)

Woody

Signature · 1 line
Woody Suwalski, Xandros, Ottawa, Canada, 1-613-842-3498 x414
drivers/gpu/drm/i915/i915_drv.h +2 -0
index 9851589..5919537 100644
@@ -143,6 +143,8 @@ typedef struct drm_i915_private {
 	drm_local_map_t hws_map;
 	struct drm_gem_object *hws_obj;
 
+	struct resource mch_res;
+
 	unsigned int cpp;
 	int back_offset;
 	int front_offset;
(quoted diff) +0 -0

b/drivers/gpu/drm/i915/i915_gem_tiling.c
index fe7877a..1441a9e 100644

drivers/gpu/drm/i915/i915_gem_tiling.c +176 -0

200 lines not on this page · raw message

#5 lore raw
Quoted 52 lines · Woody Suwalski
Quoted 51 lines · Andrew Morton
Quoted 47 lines · On Fri, 05 Jun 2009 11:02:21 +0300 Pekka Enberg <[email protected]> wrote:

Hi Jesse,

(I am cc'ing linux-kernel.)

On Mon, 01 Jun 2009 22:31:13 +0300

Quoted 24 lines
Quoted 21 lines · Pekka Enberg

Hi Jesse,

I am seeing this on my Eee PC 701 which is running a Ubuntu 2.6.28-11.42-generic stock kernel:

[   76.826966] [drm:i915_gem_detect_bit_6_swizzle] *ERROR* Couldn't
read from MCHBAR.  Disabling tiling.
[   76.827215] [drm] Initialized i915 1.6.0 20080730 on minor 0
[   76.834037] [drm:i915_setparam] *ERROR* unknown parameter 4
[   76.834104] [drm:i915_getparam] *ERROR* Unknown parameter 6

Google turned up this patch

http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html

but I don't seem to find it mainline kernel.

Was the bug fixed in some other way? It seem that distributions have not yet picked up your patch and I am unsure if it's in any of the -stable kernels.

Quoted 4 lines · Jesse Barnes

I think Eric acked it but may not have pushed it to drm-intel-next yet. Should happen in the next week or two though as we prepare the merge window series.

OK, thanks for the info! FYI, I tested 2.6.30-rc8 with your patch applied and everything works smoothly on my EeePC 701. I did not test plain 2.6.30-rc8 as I expect it to show same kind of behavior as Ubuntu distro kernel. Is this correct or do you want me to test 2.6.30-rc8 too?

I wonder why the patch hasn't received more attention as it's a pretty critical bug. The _default_ Ubuntu 9.04 netbook remix installation is completely broken for EeePC 701 (and probably others as well) without the patch. And it's not as if I'm the only one that suffers from it. A quick Google search reveals that a lot of people are hitting it.

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

So I really do think we need to merge this patch to the upcoming 2.6.31 ASAP and backport it to -stable after it has gotten some more testing.

Wanna show us the patch?

Because the world could certainly do with more i915 bugfixes :(

Here is Jesse's patch from http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html adopted for 2.6.30-rc8... (needed to redo hunk #3 for i915_gem_tile.c)

Yup. I took the two patches from here:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/349314

and did the same rediffing on top of 2.6.30-rc8 for the first one.

Andrew, do you want me to forward you the (tested) patches or should we wait for Jesse et al to take care of it?

Pekka

#6 lore raw
Quoted 8 lines · On Sat, 06 Jun 2009 09:14:41 +0300 Pekka Enberg <[email protected]> wrote:
Quoted 7 lines
Quoted 3 lines

Wanna show us the patch?

Because the world could certainly do with more i915 bugfixes :(

Here is Jesse's patch from
http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html
adopted for 2.6.30-rc8... (needed to redo hunk #3 for i915_gem_tile.c)

Quoted 8 lines

Yup. I took the two patches from here:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/349314

and did the same rediffing on top of 2.6.30-rc8 for the first one.

Andrew, do you want me to forward you the (tested) patches or should we wait for Jesse et al to take care of it?

Yes, please send them out.

These might end up being 2.6.30.1 material.

#7 lore raw

Hi Andrew,

Quoted 7 lines · On Sat, 06 Jun 2009 09:14:41 +0300 Pekka Enberg <[email protected]> wrote:
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

Wanna show us the patch?

Because the world could certainly do with more i915 bugfixes :(

Here is Jesse's patch from http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html adopted for 2.6.30-rc8... (needed to redo hunk #3 for i915_gem_tile.c)

Yup. I took the two patches from here:

Quoted 7 lines
Quoted 5 lines

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/349314

and did the same rediffing on top of 2.6.30-rc8 for the first one.

Andrew, do you want me to forward you the (tested) patches or should we wait for Jesse et al to take care of it?

Yes, please send them out.

OK, I did that. I am bit unhappy that I had to do it, though. The patch dates back to January (!) and Ubuntu folks have been carrying it for a while now.

Looking at the amount of reports out in the wild (on Ubuntu forums, etc.), I really don't understand why this wasn't submitted earlier. Instead, the bug was flagged with low priority because it's "just a performance problem." Considering the system is more or less unusable for me, I obviously think it's a critical bug fix.

Quoted 1 line · Andrew Morton

These might end up being 2.6.30.1 material.

Yes, agreed completely. We probably want ACK from Jesse and Bjorn, though.

Pekka

#8 lore raw
Quoted 33 lines · On Sat, 06 Jun 2009 09:37:00 +0300 Pekka Enberg <[email protected]> wrote:

Hi Andrew,

Quoted 10 lines · On Sat, 06 Jun 2009 09:14:41 +0300 Pekka Enberg <[email protected]> wrote:
Quoted 10 lines
Quoted 8 lines
Quoted 3 lines

Wanna show us the patch?

Because the world could certainly do with more i915 bugfixes :(

Here is Jesse's patch from
http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html
adopted for 2.6.30-rc8... (needed to redo hunk #3 for i915_gem_tile.c)

Yup. I took the two patches from here:

Quoted 8 lines
Quoted 6 lines

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/349314

and did the same rediffing on top of 2.6.30-rc8 for the first one.

Andrew, do you want me to forward you the (tested) patches or should we wait for Jesse et al to take care of it?

Yes, please send them out.

OK, I did that. I am bit unhappy that I had to do it, though. The patch dates back to January (!) and Ubuntu folks have been carrying it for a while now.

Looking at the amount of reports out in the wild (on Ubuntu forums, etc.), I really don't understand why this wasn't submitted earlier. Instead, the bug was flagged with low priority because it's "just a performance problem." Considering the system is more or less unusable for me, I obviously think it's a critical bug fix.

Quoted 5 lines
Quoted 1 line · Andrew Morton

These might end up being 2.6.30.1 material.

Yes, agreed completely. We probably want ACK from Jesse and Bjorn, though.

I'm ok with it being 2.6.30.1 stuff, but given the significance of the change we were worried about putting it into 2.6.30 late in the cycle. But as you say, it's a very important performance fix for many machines...

Anyway, Eric may already have it merged. Eric?

Jesse

#9 lore raw
Quoted 43 lines · On Sat, 6 Jun 2009 21:27:59 +0100 Jesse Barnes <[email protected]> wrote:
Quoted 34 lines · On Sat, 06 Jun 2009 09:37:00 +0300 Pekka Enberg <[email protected]> wrote:

Hi Andrew,

Quoted 12 lines · On Sat, 06 Jun 2009 09:14:41 +0300 Pekka Enberg <[email protected]> wrote:
Quoted 12 lines
Quoted 8 lines
Quoted 3 lines

Wanna show us the patch?

Because the world could certainly do with more i915 bugfixes :(

Here is Jesse's patch from
http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html
adopted for 2.6.30-rc8... (needed to redo hunk #3 for i915_gem_tile.c)

Yup. I took the two patches from here:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/349314

Quoted 7 lines
Quoted 5 lines

and did the same rediffing on top of 2.6.30-rc8 for the first one.

Andrew, do you want me to forward you the (tested) patches or should we wait for Jesse et al to take care of it?

Yes, please send them out.

OK, I did that. I am bit unhappy that I had to do it, though. The patch dates back to January (!) and Ubuntu folks have been carrying it for a while now.

Looking at the amount of reports out in the wild (on Ubuntu forums, etc.), I really don't understand why this wasn't submitted earlier. Instead, the bug was flagged with low priority because it's "just a performance problem." Considering the system is more or less unusable for me, I obviously think it's a critical bug fix.

Quoted 5 lines
Quoted 1 line · Andrew Morton

These might end up being 2.6.30.1 material.

Yes, agreed completely. We probably want ACK from Jesse and Bjorn, though.

Quoted 7 lines

I'm ok with it being 2.6.30.1 stuff, but given the significance of the change we were worried about putting it into 2.6.30 late in the cycle. But as you say, it's a very important performance fix for many machines...

Anyway, Eric may already have it merged. Eric?

It's merged now into the drm-intel-next tree, and therefore queued for the merge window.

Jeses

#10 lore raw
Quoted 11 lines · On Tue, 9 Jun 2009 15:07:57 -0700 Jesse Barnes <[email protected]> wrote:
Quoted 11 lines
Quoted 5 lines
Quoted 1 line · Andrew Morton

These might end up being 2.6.30.1 material.

Yes, agreed completely. We probably want ACK from Jesse and Bjorn, though.

I'm ok with it being 2.6.30.1 stuff, but given the significance of the change we were worried about putting it into 2.6.30 late in the cycle. But as you say, it's a very important performance fix for many machines...

Quoted 5 lines
Quoted 2 lines

Anyway, Eric may already have it merged. Eric?

It's merged now into the drm-intel-next tree, and therefore queued for the merge window.

<hunts around for a while>

This? http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=commitdiff;h=d765898970f35acef960581f678b9da9d5c779fa;hp=1b8e69662e1a086878bf930a6042daf7f8a076cc

It doesn't have cc:stable in the changelog, so there's a reasonable chance that it will get lost.

The -stable guys troll the commit list looking for mainline commits which mention [email protected] in the changelog.

#11 lore raw
Quoted 22 lines · Andrew Morton
Quoted 11 lines · On Tue, 9 Jun 2009 15:07:57 -0700 Jesse Barnes <[email protected]> wrote:
Quoted 11 lines
Quoted 5 lines
Quoted 1 line · Andrew Morton

These might end up being 2.6.30.1 material.

Yes, agreed completely. We probably want ACK from Jesse and Bjorn, though.

I'm ok with it being 2.6.30.1 stuff, but given the significance of the change we were worried about putting it into 2.6.30 late in the cycle. But as you say, it's a very important performance fix for many machines...

Quoted 5 lines
Quoted 2 lines

Anyway, Eric may already have it merged. Eric?

It's merged now into the drm-intel-next tree, and therefore queued for the merge window.

<hunts around for a while>

Quoted 7 lines

This? http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=commitdiff;h=d765898970f35acef960581f678b9da9d5c779fa;hp=1b8e69662e1a086878bf930a6042daf7f8a076cc

It doesn't have cc:stable in the changelog, so there's a reasonable chance that it will get lost.

The -stable guys troll the commit list looking for mainline commits which mention [email protected] in the changelog.

I got it for the next .30 release.

thanks,

greg k-h

#12 lore raw
Quoted 25 lines · Greg KH
Quoted 24 lines · Andrew Morton
Quoted 11 lines · On Tue, 9 Jun 2009 15:07:57 -0700 Jesse Barnes <[email protected]> wrote:
Quoted 11 lines
Quoted 5 lines
Quoted 1 line · Andrew Morton

These might end up being 2.6.30.1 material.

Yes, agreed completely. We probably want ACK from Jesse and Bjorn, though.

I'm ok with it being 2.6.30.1 stuff, but given the significance of the change we were worried about putting it into 2.6.30 late in the cycle. But as you say, it's a very important performance fix for many machines...

Quoted 5 lines
Quoted 2 lines

Anyway, Eric may already have it merged. Eric?

It's merged now into the drm-intel-next tree, and therefore queued for the merge window.

<hunts around for a while>

This? http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=commitdiff;h=d765898970f35acef960581f678b9da9d5c779fa;hp=1b8e69662e1a086878bf930a6042daf7f8a076cc

Quoted 7 lines
Quoted 5 lines

It doesn't have cc:stable in the changelog, so there's a reasonable chance that it will get lost.

The -stable guys troll the commit list looking for mainline commits which mention [email protected] in the changelog.

I got it for the next .30 release.

I take it back. It doesn't build properly on 2.6.30, can someone backport commit d765898970f35acef960581f678b9da9d5c779fa to the 2.6.30 tree and send it to [email protected] so that we can apply it to resolve this issue?

And is it relevant to 2.6.29 as well?

thanks,

greg k-h

#13 lore raw
Quoted 6 lines · Greg KH

I take it back. It doesn't build properly on 2.6.30, can someone backport commit d765898970f35acef960581f678b9da9d5c779fa to the 2.6.30 tree and send it to [email protected] so that we can apply it to resolve this issue?
ease read the FAQ at http://www.tux.org/lkml/

There was this commit on top of them which was fixed the build problem for me:
http://git.kernel.org/?p=linux%2Fkernel%2Fgit%2Fanholt%2Fdrm-intel.git;a=commitdiff_plain;h=34c8638bb5ee130b8151c57708284bcd4972e2e1;hp=003c9ca68810bc0a63a1f3ff6beec839b197e23b

#14 lore raw
Quoted 11 lines · Ozan Çağlayan
Quoted 6 lines · Greg KH

I take it back. It doesn't build properly on 2.6.30, can someone backport commit d765898970f35acef960581f678b9da9d5c779fa to the 2.6.30 tree and send it to [email protected] so that we can apply it to resolve this issue?
ease read the FAQ at http://www.tux.org/lkml/

There was this commit on top of them which was fixed the build problem for me:
http://git.kernel.org/?p=linux%2Fkernel%2Fgit%2Fanholt%2Fdrm-intel.git;a=commitdiff_plain;h=34c8638bb5ee130b8151c57708284bcd4972e2e1;hp=003c9ca68810bc0a63a1f3ff6beec839b197e23b

No, the issue is the pnp_resource() call, that's not present in 2.6.30.

Can someone backport whatever is needed in .30 and send it to us?

thanks,

greg k-h

#15 lore raw
In reply to Pekka Enberg
Quoted 17 lines · Pekka Enberg

Hi Andrew,

Quoted 13 lines · On Sat, 06 Jun 2009 09:14:41 +0300 Pekka Enberg <[email protected]> wrote:
Quoted 13 lines
Quoted 7 lines
Quoted 3 lines

Wanna show us the patch?

Because the world could certainly do with more i915 bugfixes :(

Here is Jesse's patch from
http://lists.freedesktop.org/archives/intel-gfx/2009-January/001186.html
adopted for 2.6.30-rc8... (needed to redo hunk #3 for i915_gem_tile.c)

Yup. I took the two patches from here:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/349314

and did the same rediffing on top of 2.6.30-rc8 for the first one.

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

Andrew, do you want me to forward you the (tested) patches or should we wait for Jesse et al to take care of it?

Yes, please send them out.

OK, I did that. I am bit unhappy that I had to do it, though. The patch dates back to January (!) and Ubuntu folks have been carrying it for a while now.

We cirtainly have been carrying for some time. My understanding was that the maintainers had it and had been expecting it to merge through them. Did we miss a trick somewhere?

-apw