One regression in UXA

One regression in UXA

[Zhigang Gong]

[-- Attachment #1.1: Type: text/plain, Size: 2032 bytes --]

Hi Chris, 

 

I tested a browser case with latest UXA code and experienced X crashed.

I bisected it and found that it was introduced by 

commit 5d5b2b8ee203ae2274fc7d13ed38d2945facca9e.

 

The back trace of the crash is as below:

 

Backtrace:

0: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg
(xorg_backtrace+0x3b) [0x81b5fdb]

1: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg (mieqEnqueue+0x15c)
[0x8195a0c]

2: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg (0x8048000+0x40782)
[0x8088782]

3: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg
(xf86PostMotionEventM+0xf1) [0x80c5671]

4: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/input/evdev_drv.so
(0xb7623000+0x2fdf) [0xb7625fdf]

5: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/input/evdev_drv.so
(0xb7623000+0x4229) [0xb7627229]

6: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg (0x8048000+0x67fcf)
[0x80affcf]

7: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg (0x8048000+0x8e984)
[0x80d6984]

8: (vdso) (__kernel_sigreturn+0x0) [0xb7fff400]

9: /home/gongzg/gfx-dev-test/build/lib/libdrm_intel.so.1
(drm_intel_gem_bo_map_gtt+0x83) [0xb7e9cee3]

10: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/drivers/intel_drv.so
(0xb7ea1000+0xd6cb) [0xb7eae6cb]   -- intel_uxa_pixmap_put_image

11: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/drivers/intel_drv.so
(0xb7ea1000+0xf42c) [0xb7eb042c]   -- intel_uxa_put_image 

12: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/drivers/intel_drv.so
(0xb7ea1000+0x37f8e) [0xb7ed8f8e]  -- uxa_trapezoids

 

 

It's a one line patch. And there is not too much thing to do with UXA code,
I guess It may be a libdrm related problem.

 

It's can be simply reproduced in my machine, Use Mozilla 3.6.10 to open the
following URL:

http://pnp.sh.intel.com/WRTBench/Workload_3.3_rev0.13/iterations_rev11_new.h
tml

 

It will iterate many pages and finally give a report. The crash may occur
after a few seconds of running.

Uncomment that one line patch can avoid this crash. 


[-- Attachment #1.2: Type: text/html, Size: 5855 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: One regression in UXA

[Zhigang Gong]

[-- Attachment #1.1: Type: text/plain, Size: 2441 bytes --]

Forgot to mention that to access that link need to be in Intel intranet
environment.

 

From: intel-gfx-bounces+zhigang.gong=linux.intel.com@lists.freedesktop.org
[mailto:intel-gfx-bounces+zhigang.gong=linux.intel.com@lists.freedesktop.org
] On Behalf Of Zhigang Gong
Sent: Friday, December 16, 2011 4:09 PM
To: Chris Wilson
Cc: intel-gfx@lists.freedesktop.org
Subject: [Intel-gfx] One regression in UXA

 

Hi Chris, 

 

I tested a browser case with latest UXA code and experienced X crashed.

I bisected it and found that it was introduced by 

commit 5d5b2b8ee203ae2274fc7d13ed38d2945facca9e.

 

The back trace of the crash is as below:

 

Backtrace:

0: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg
(xorg_backtrace+0x3b) [0x81b5fdb]

1: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg (mieqEnqueue+0x15c)
[0x8195a0c]

2: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg (0x8048000+0x40782)
[0x8088782]

3: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg
(xf86PostMotionEventM+0xf1) [0x80c5671]

4: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/input/evdev_drv.so
(0xb7623000+0x2fdf) [0xb7625fdf]

5: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/input/evdev_drv.so
(0xb7623000+0x4229) [0xb7627229]

6: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg (0x8048000+0x67fcf)
[0x80affcf]

7: /home/gongzg/gfx-dev-test/src/xserver/hw/xfree86/Xorg (0x8048000+0x8e984)
[0x80d6984]

8: (vdso) (__kernel_sigreturn+0x0) [0xb7fff400]

9: /home/gongzg/gfx-dev-test/build/lib/libdrm_intel.so.1
(drm_intel_gem_bo_map_gtt+0x83) [0xb7e9cee3]

10: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/drivers/intel_drv.so
(0xb7ea1000+0xd6cb) [0xb7eae6cb]   -- intel_uxa_pixmap_put_image

11: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/drivers/intel_drv.so
(0xb7ea1000+0xf42c) [0xb7eb042c]   -- intel_uxa_put_image 

12: /home/gongzg/gfx-dev-test/build/lib/xorg/modules/drivers/intel_drv.so
(0xb7ea1000+0x37f8e) [0xb7ed8f8e]  -- uxa_trapezoids

 

 

It's a one line patch. And there is not too much thing to do with UXA code,
I guess It may be a libdrm related problem.

 

It's can be simply reproduced in my machine, Use Mozilla 3.6.10 to open the
following URL:

http://pnp.sh.intel.com/WRTBench/Workload_3.3_rev0.13/iterations_rev11_new.h
tml

 

It will iterate many pages and finally give a report. The crash may occur
after a few seconds of running.

Uncomment that one line patch can avoid this crash. 


[-- Attachment #1.2: Type: text/html, Size: 7120 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: One regression in UXA

[Chris Wilson]

On Fri, 16 Dec 2011 16:08:57 +0800, "Zhigang Gong" <zhigang.gong@linux.intel.com> wrote:
> Hi Chris, 
> 
>  
> 
> I tested a browser case with latest UXA code and experienced X crashed.
> 
> I bisected it and found that it was introduced by 
> 
> commit 5d5b2b8ee203ae2274fc7d13ed38d2945facca9e.

[snip]

> It's a one line patch. And there is not too much thing to do with UXA code,
> I guess It may be a libdrm related problem.

Indeed, fixed in libdrm master, 0ab2251b7.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre