Deadlock in intel_user_framebuffer_destroy()
Subsystems (2)
Hi,
a deadlock was introduced by commit 60a5ca015ffd2aacfe5674b5a401cd2a37159e07
Author: Ville Syrjälä <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300
drm/i915: Add locking around framebuffer_references--
The commit amended intel_display.c:intel_user_framebuffer_destroy() with mutex_lock(&dev->struct_mutex).
A few weeks prior Chris Wilson had amended intel_fbdev.c:intelfb_create() with a call to drm_framebuffer_unreference() while &dev->struct_mutex is locked (commit edd586fe705e819bc711b5ed7194a0b6f9f1a7e1, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode").
This leads to the following call chain while &dev->struct_mutex is locked:
intel_fbdev.c:intelfb_create()
-> drm_crtc.c:drm_framebuffer_unreference()
-> drm_crtc.c:drm_framebuffer_free()
-> intel_display.c:intel_user_framebuffer_destroy()The last function in that call chain attempts to re-lock the mutex. The functionality added by Chris Wilson is thus rendered broken.
Best regards,
Lukas
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 10 lines · Lukas Wunner
Hi,
a deadlock was introduced by commit
60a5ca015ffd2aacfe5674b5a401cd2a37159e07Author: Ville Syrjälä <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300drm/i915: Add locking around framebuffer_references--
Quoted 8 lines
The commit amended intel_display.c:intel_user_framebuffer_destroy() with
mutex_lock(&dev->struct_mutex).A few weeks prior Chris Wilson had amended intel_fbdev.c:intelfb_create() with a call to
drm_framebuffer_unreference()while&dev->struct_mutexis locked (commitedd586fe705e819bc711b5ed7194a0b6f9f1a7e1, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode").
Just move the mutex_lock down a step.
t a/drivers/gpu/drm/i915/intel_fbdev.c b/drivers/gpu/drm/i915/intel_fbdev.c index 6372cfc..a9b8c43 100644
drivers/gpu/drm/i915/intel_fbdev.c +3 -2
@@ -209,8 +209,6 @@ static int intelfb_create(struct drm_fb_helper *helper,int size, ret;bool prealloc = false; -mutex_lock (&dev->struct_mutex); -if (intel_fb && (sizes->fb_width > intel_fb->base.width || sizes->fb_height > intel_fb->base.height)) {
@@ -221,6 +219,9 @@ static int intelfb_create(struct drm_fb_helper *helper,drm_framebuffer_unreference (&intel_fb->base); intel_fb = ifbdev->fb =NULL ; } + +mutex_lock (&dev->struct_mutex); +if (!intel_fb ||WARN_ON (!intel_fb->obj)) {DRM_DEBUG_KMS ("no BIOS fb, allocating a new one\n" ); ret =intelfb_alloc (helper, sizes);
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerQuoted 14 lines · Chris Wilson
Quoted 13 lines · Lukas Wunner
Hi,
a deadlock was introduced by commit
60a5ca015ffd2aacfe5674b5a401cd2a37159e07Author: Ville Syrjälä <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300drm/i915: Add locking around framebuffer_references--
The commit amended intel_display.c:intel_user_framebuffer_destroy() with
mutex_lock(&dev->struct_mutex).
Quoted 7 lines
Quoted 5 lines
A few weeks prior Chris Wilson had amended intel_fbdev.c:intelfb_create() with a call to
drm_framebuffer_unreference()while&dev->struct_mutexis locked (commitedd586fe705e819bc711b5ed7194a0b6f9f1a7e1, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode").Just move the
mutex_lockdown a step.
Lucas, did you try this?
BR,
Jani.
Quoted 31 lines
t
a/drivers/gpu/drm/i915/intel_fbdev.cb/drivers/gpu/drm/i915/intel_fbdev.cindex 6372cfc..a9b8c43 100644drivers/gpu/drm/i915/intel_fbdev.c +3
-2@@ -209,8 +209,6 @@ static int intelfb_create(struct drm_fb_helper *helper,int size, ret;bool prealloc = false; -mutex_lock (&dev->struct_mutex); -if (intel_fb && (sizes->fb_width > intel_fb->base.width || sizes->fb_height > intel_fb->base.height)) {@@ -221,6 +219,9 @@ static int intelfb_create(struct drm_fb_helper *helper,drm_framebuffer_unreference (&intel_fb->base); intel_fb = ifbdev->fb =NULL ; } + +mutex_lock (&dev->struct_mutex); +if (!intel_fb ||WARN_ON (!intel_fb->obj)) {DRM_DEBUG_KMS ("no BIOS fb, allocating a new one\n" ); ret =intelfb_alloc (helper, sizes);Signature · 6 lines
-- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Hi Jani,
Quoted 3 lines · Jani Nikula
Quoted 1 line · Chris Wilson
Just move the
mutex_lockdown a step.Lucas, did you try this?
Yes, works for me. I was going to submit the attached patch as part of the MacBook Pro GPU switching patch set (once it's finished) but feel free to merge now.
Thanks for following up on this.
Lukas
Quoted 1 line
From
b502824816e20e65f7b0b17b18efec161f7560edMon Sep 17 00:00:00 2001
Message-Id: <b502824816e20e65f7b0b17b18efec161f7560ed.1434188640.git.lukas@wunner.de>
In-Reply-To: <77dc50305ee6fa2db076907496e373cc237e4901.1434188640.git.lukas@wunner.de>
References: <1dcf15ad548ee6602134a71e2e1af11609b4c32b.1434188640.git.lukas@wunner.de>
<82e0890fdef2fcfa467f78a397641c39d7d9cfdd.1434188640.git.lukas@wunner.de> <84b1a363ca1860d523ce52b11aa83239ebf2e163.1434188640.git.lukas@wunner.de> <450b5fe4557644dbc5245a386e94c6ccc6654b92.1434188640.git.lukas@wunner.de> <4327e6c5baf23370cfb1cd09105693978f6981d3.1434188640.git.lukas@wunner.de> <1eb55048182574dd0443e7f83a44dbdab99bbe86.1434188640.git.lukas@wunner.de> <e395b70b5680ec8551848550f90f563250323fe8.1434188640.git.lukas@wunner.de> <226f1ae95e50070a6b17999d46c850d3abcb4496.1434188640.git.lukas@wunner.de> <a151a053833b7e6561479b8e24ab6a669a0f6673.1434188640.git.lukas@wunner.de> <afab3acad3d94c25b7f7582d37dd4ec3b35afa3b.1434188640.git.lukas@wunner.de> <293413a2703dda0c40967031531b20fbab6214a9.1434188640.git.lukas@wunner.de> <05704f90341a290bae3701b3ca9efc8b84729e20.1434188640.git.lukas@wunner.de> <5875c0ca739463fae31618bd6a9f07d3953b14cc.1434188640.git.lukas@wunner.de> <a5bac6cead2e8f6e3e481abdf5f3476883be64d0.1434188640.git.lukas@wunner.de> <6871930e1d0f3d8dcb702bbb4f9b121d48a334ca.1434188640.git.lukas@wunner.de> <a6d60e1981004ba1e8dc4ce37b78657b2edf61c8.1434188640.git.lukas@wunner.de> <77dc50305ee6fa2db076907496e373cc237e4901.1434188640.git.lukas@wunner.de>
From: Lukas Wunner <[email protected]>
Date: Wed, 3 Jun 2015 14:57:41 +0100
Subject: [PATCH v2 18/20] drm/i915: Fix deadlock upon discarding BIOS fb
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
To: [email protected]
From: Chris Wilson <[email protected]>
A deadlock was introduced by commit 60a5ca015ffd:
Author: Ville SyrjÀlÀ <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300
drm/i915: Add locking around framebuffer_references--
The commit amended intel_user_framebuffer_destroy() with locking of dev->struct_mutex.
A few weeks prior Chris Wilson had amended intelfb_create() with a call to drm_framebuffer_unreference() while dev->struct_mutex is locked (edd586fe705e, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode"). That function calls drm_framebuffer_free(), which calls intel_user_framebuffer_destroy(), which now tries to acquire the lock once more, rendering the functionality added by Chris Wilson broken.
Leave the offending commit as is but adapt Chris Wilson's code by moving mutex_lock() below the call to drm_framebuffer_unreference().
--- drivers/gpu/drm/i915/intel_fbdev.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_fbdev.c +3 -2
@@ -183,8 +183,6 @@ static int intelfb_create(struct drm_fb_helper *helper,int size, ret;bool prealloc = false; -mutex_lock (&dev->struct_mutex); -if (intel_fb && (sizes->fb_width > intel_fb->base.width || sizes->fb_height > intel_fb->base.height)) {
@@ -195,6 +193,9 @@ static int intelfb_create(struct drm_fb_helper *helper,drm_framebuffer_unreference (&intel_fb->base); intel_fb = ifbdev->fb =NULL ; } + +mutex_lock (&dev->struct_mutex); +if (!intel_fb ||WARN_ON (!intel_fb->obj)) {DRM_DEBUG_KMS ("no BIOS fb, allocating a new one\n" ); ret =intelfb_alloc (helper, sizes);
Signature · 1 line
2.1.0
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 21 lines · Jani Nikula
Quoted 20 lines · Chris Wilson
Quoted 18 lines · Lukas Wunner
Hi,
a deadlock was introduced by commit
60a5ca015ffd2aacfe5674b5a401cd2a37159e07Author: Ville Syrjälä <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300drm/i915: Add locking around framebuffer_references--
The commit amended intel_display.c:intel_user_framebuffer_destroy() with
mutex_lock(&dev->struct_mutex).A few weeks prior Chris Wilson had amended intel_fbdev.c:intelfb_create() with a call to
drm_framebuffer_unreference()while&dev->struct_mutexis locked (commitedd586fe705e819bc711b5ed7194a0b6f9f1a7e1, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode").
Quoted 3 lines
Quoted 1 line
Just move the
mutex_lockdown a step.Lucas, did you try this?
There's a goto unlock that also needed to be disabled, such as
drivers/gpu/drm/i915/intel_fbdev.c +2 -4
@@ -213,8 +213,6 @@ static int intelfb_create(struct drm_fb_helper *helper,bool prealloc = false;int ret; -mutex_lock (&dev->struct_mutex); -if (intel_fb && (sizes->fb_width > intel_fb->base.width || sizes->fb_height > intel_fb->base.height)) {
@@ -229,7 +227,7 @@ static int intelfb_create(struct drm_fb_helper *helper,DRM_DEBUG_KMS ("no BIOS fb, allocating a new one\n" ); ret =intelfb_alloc (helper, sizes);if (ret) - goto out_unlock; +return ret; intel_fb = ifbdev->fb; }else {DRM_DEBUG_KMS ("re-using BIOS fb\n" );
@@ -241,6 +239,7 @@ static int intelfb_create(struct drm_fb_helper *helper, obj = intel_fb->obj; vma =i915_gem_obj_to_ggtt (obj,NULL ); +mutex_lock (&dev->struct_mutex); info =framebuffer_alloc (0 , &dev->pdev->dev);if (!info) { ret = -ENOMEM ;
@@ -311,7 +310,6 @@ static int intelfb_create(struct drm_fb_helper *helper, out_unpin:drm_gem_object_unreference (&obj->base); -out_unlock:mutex_unlock (&dev->struct_mutex);return ret; }
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Hi,
Quoted 59 lines · Chris Wilson
Quoted 21 lines · Jani Nikula
Quoted 20 lines · Chris Wilson
Quoted 18 lines · Lukas Wunner
Hi,
a deadlock was introduced by commit
60a5ca015ffd2aacfe5674b5a401cd2a37159e07Author: Ville Syrjälä <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300drm/i915: Add locking around framebuffer_references--
The commit amended intel_display.c:intel_user_framebuffer_destroy() with
mutex_lock(&dev->struct_mutex).A few weeks prior Chris Wilson had amended intel_fbdev.c:intelfb_create() with a call to
drm_framebuffer_unreference()while&dev->struct_mutexis locked (commitedd586fe705e819bc711b5ed7194a0b6f9f1a7e1, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode").Quoted 3 lines
Quoted 1 line
Just move the
mutex_lockdown a step.Lucas, did you try this?
There's a goto unlock that also needed to be disabled, such as
drivers/gpu/drm/i915/intel_fbdev.c +2
-3@@ -213,8 +213,6 @@ static int intelfb_create(struct drm_fb_helper *helper,bool prealloc = false;int ret; -mutex_lock (&dev->struct_mutex); -if (intel_fb && (sizes->fb_width > intel_fb->base.width || sizes->fb_height > intel_fb->base.height)) {@@ -229,7 +227,7 @@ static int intelfb_create(struct drm_fb_helper *helper,DRM_DEBUG_KMS ("no BIOS fb, allocating a new one\n" ); ret =intelfb_alloc (helper, sizes);if (ret) - goto out_unlock; +return ret; intel_fb = ifbdev->fb; }else {DRM_DEBUG_KMS ("re-using BIOS fb\n" );@@ -241,6 +239,7 @@ static int intelfb_create(struct drm_fb_helper *helper, obj = intel_fb->obj; vma =i915_gem_obj_to_ggtt (obj,NULL ); +mutex_lock (&dev->struct_mutex); info =framebuffer_alloc (0 , &dev->pdev->dev);if (!info) { ret = -ENOMEM ;
drivers/gpu/drm/i915/intel_fbdev.c static int intelfb_create(struct drm_fb_helper *helper, · line 315
Quoted 7 lines
drivers/gpu/drm/i915/intel_fbdev.c +0
-1@@ -311,7 +310,6 @@ static int intelfb_create(struct drm_fb_helper *helper, out_unpin:drm_gem_object_unreference (&obj->base); -out_unlock:mutex_unlock (&dev->struct_mutex);return ret; }
intelfb_alloc wants struct_mutex, both for __intel_framebuffer_create and pin_and_fence. And also there is that double obj unreference in the failure path from the former Jani spotted.
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 22 lines · Chris Wilson
Quoted 21 lines · Jani Nikula
Quoted 20 lines · Chris Wilson
Quoted 18 lines · Lukas Wunner
Hi,
a deadlock was introduced by commit
60a5ca015ffd2aacfe5674b5a401cd2a37159e07Author: Ville Syrjälä <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300drm/i915: Add locking around framebuffer_references--
The commit amended intel_display.c:intel_user_framebuffer_destroy() with
mutex_lock(&dev->struct_mutex).A few weeks prior Chris Wilson had amended intel_fbdev.c:intelfb_create() with a call to
drm_framebuffer_unreference()while&dev->struct_mutexis locked (commitedd586fe705e819bc711b5ed7194a0b6f9f1a7e1, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode").
Quoted 5 lines
Quoted 3 lines
Quoted 1 line
Just move the
mutex_lockdown a step.Lucas, did you try this?
There's a goto unlock that also needed to be disabled, such as
Your previous patch placed the mutex_lock before the goto out_unlock - I fail to see what has been broken with that version? Can you resubmit that as a proper patch with sob and Lucas' t-b?
Thanks, Daniel
Quoted 45 lines
drivers/gpu/drm/i915/intel_fbdev.c +2
-4@@ -213,8 +213,6 @@ static int intelfb_create(struct drm_fb_helper *helper,bool prealloc = false;int ret; -mutex_lock (&dev->struct_mutex); -if (intel_fb && (sizes->fb_width > intel_fb->base.width || sizes->fb_height > intel_fb->base.height)) {@@ -229,7 +227,7 @@ static int intelfb_create(struct drm_fb_helper *helper,DRM_DEBUG_KMS ("no BIOS fb, allocating a new one\n" ); ret =intelfb_alloc (helper, sizes);if (ret) - goto out_unlock; +return ret; intel_fb = ifbdev->fb; }else {DRM_DEBUG_KMS ("re-using BIOS fb\n" );@@ -241,6 +239,7 @@ static int intelfb_create(struct drm_fb_helper *helper, obj = intel_fb->obj; vma =i915_gem_obj_to_ggtt (obj,NULL ); +mutex_lock (&dev->struct_mutex); info =framebuffer_alloc (0 , &dev->pdev->dev);if (!info) { ret = -ENOMEM ;@@ -311,7 +310,6 @@ static int intelfb_create(struct drm_fb_helper *helper, out_unpin:drm_gem_object_unreference (&obj->base); -out_unlock:mutex_unlock (&dev->struct_mutex);return ret; }Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 25 lines · Daniel Vetter
Quoted 24 lines · Chris Wilson
Quoted 23 lines · Jani Nikula
Quoted 21 lines · Chris Wilson
Quoted 10 lines · Lukas Wunner
Hi,
a deadlock was introduced by commit
60a5ca015ffd2aacfe5674b5a401cd2a37159e07Author: Ville Syrjälä <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300drm/i915: Add locking around framebuffer_references--
Quoted 8 lines
The commit amended intel_display.c:intel_user_framebuffer_destroy() with
mutex_lock(&dev->struct_mutex).A few weeks prior Chris Wilson had amended intel_fbdev.c:intelfb_create() with a call to
drm_framebuffer_unreference()while&dev->struct_mutexis locked (commitedd586fe705e819bc711b5ed7194a0b6f9f1a7e1, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode").Just move the
mutex_lockdown a step.
Quoted 7 lines
Quoted 3 lines
Quoted 1 line
Lucas, did you try this?
There's a goto unlock that also needed to be disabled, such as
Your previous patch placed the
mutex_lockbefore the gotoout_unlock- I fail to see what has been broken with that version? Can you resubmit that as a proper patch with sob and Lucas' t-b?
There wasn't, I just rewrote it incorrectly. There's also a drm_framebuffer_remove() called by intelfb_alloc which needs to be moved out of the mutex. A much larger disentangling of the functions involved here is required. Tvrstko volunteered himself. Brave, very brave :) -Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerQuoted 30 lines · Chris Wilson
Quoted 29 lines · Daniel Vetter
Quoted 27 lines · Chris Wilson
Quoted 21 lines · Jani Nikula
Quoted 20 lines · Chris Wilson
Quoted 18 lines · Lukas Wunner
Hi,
a deadlock was introduced by commit
60a5ca015ffd2aacfe5674b5a401cd2a37159e07Author: Ville Syrjälä <[email protected]> Date: Fri Jun 13 11:10:53 2014 +0300drm/i915: Add locking around framebuffer_references--
The commit amended intel_display.c:intel_user_framebuffer_destroy() with
mutex_lock(&dev->struct_mutex).A few weeks prior Chris Wilson had amended intel_fbdev.c:intelfb_create() with a call to
drm_framebuffer_unreference()while&dev->struct_mutexis locked (commitedd586fe705e819bc711b5ed7194a0b6f9f1a7e1, "drm/i915: Discard BIOS framebuffers too small to accommodate chosen mode").Quoted 3 lines
Quoted 1 line
Just move the
mutex_lockdown a step.Lucas, did you try this?
There's a goto unlock that also needed to be disabled, such as
Quoted 8 lines
Quoted 3 lines
Your previous patch placed the
mutex_lockbefore the gotoout_unlock- I fail to see what has been broken with that version? Can you resubmit that as a proper patch with sob and Lucas' t-b?There wasn't, I just rewrote it incorrectly. There's also a
drm_framebuffer_remove()called byintelfb_allocwhich needs to be moved out of the mutex. A much larger disentangling of the functions involved here is required. Tvrstko volunteered himself. Brave, very brave :)
-ETIMEDOUT
IIUC we still need the fix, but there's no update from Lucas addressing review. Is that right?
BR,
Jani.
Quoted 4 lines
-Chris
Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Hi Jani,
Quoted 8 lines · Jani Nikula
Quoted 4 lines · Chris Wilson
There wasn't, I just rewrote it incorrectly. There's also a
drm_framebuffer_remove()called byintelfb_allocwhich needs to be moved out of the mutex. A much larger disentangling of the functions involved here is required. Tvrstko volunteered himself. Brave, very brave :)
-ETIMEDOUT
IIUC we still need the fix, but there's no update from Lucas addressing review. Is that right?
Sorry, I was not copied on Tvrtko's e-mail of June 15 ("drm/i915:
Fix failure paths around initial fbdev allocation") and though I'm subscribed to intel-gfx I didn't realize that particular e-mail required a reaction from me.
I've just sent out my review and I'll also test the patch and report back.
Thanks,
Lukas
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerQuoted 8 lines · Lukas Wunner
Hi Jani,
Quoted 5 lines · Jani Nikula
Quoted 4 lines · Chris Wilson
There wasn't, I just rewrote it incorrectly. There's also a
drm_framebuffer_remove()called byintelfb_allocwhich needs to be moved out of the mutex. A much larger disentangling of the functions involved here is required. Tvrstko volunteered himself. Brave, very brave :)
Quoted 8 lines
Quoted 3 lines
-ETIMEDOUT
IIUC we still need the fix, but there's no update from Lucas addressing review. Is that right?Sorry, I was not copied on Tvrtko's e-mail of June 15 ("drm/i915:
Fix failure paths around initial fbdev allocation") and though I'm subscribed to intel-gfx I didn't realize that particular e-mail required a reaction from me.
No worries, I didn't realize these two threads were connected!
BR,
Jani.
Quoted 7 lines
I've just sent out my review and I'll also test the patch and report back.
Thanks,
Lukas
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx