[OOPS] 2.4.20-rc4-ac1 (also occurs 2.4.20-rc2-ac3) in radeon DRI for XFree86

[OOPS] 2.4.20-rc4-ac1 (also occurs 2.4.20-rc2-ac3) in radeon DRI for XFree86

[Joshua Kwan]

Ksymoops output follows.

I compiled Radeon DRM stuff into the kernel -- i845 agp support from agapgart.
I am using gcc-3.2 to compile. 100% reproducible (okay, i've been spending too much time on bugzillas...) Feel the power of the oops.

Hope this helps kernel development.
-Josh

Unable to handle kernel NULL pointer dereference at virtual address 00000002
c01bad2d
*pde = 0eb07067
Oops: 0002
CPU:    0
EIP:    0010:[<c01bad2d>]    Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00013202
eax: d49391d0   ebx: e0000000   ecx: 00000011   edx: 00000002
esi: cf725080   edi: ce427efc   ebp: c1aa2800   esp: ce427ec4
ds: 0018   es: 0018   ss: 0018
Process XFree86 (pid: 374, stackpage=ce427000)
Stack: 00020000 00200000 000000f8 00000000 cdb49340 cdb49100 ce427efc c1aa2800 
       cea5fa80 ce348280 c01bb1ad c1aa2800 ce427efc 00000054 00000001 00000898 
       00000000 40000000 00800000 00100000 00002710 00000020 00000000 00001040 
Call Trace:    [<c01bb1ad>] [<c01b55ed>] [<c0149ce9>] [<c01072f7>]
Code: 89 02 89 34 24 e8 d9 f2 ff ff 89 74 24 04 89 2c 24 e8 0d f8


>>EIP; c01bad2d <radeon_do_init_cp+4bd/7c0>   <=====

>>esi; cf725080 <_end+f4174a8/12588488>
>>edi; ce427efc <_end+e11a324/12588488>
>>ebp; c1aa2800 <_end+1794c28/12588488>
>>esp; ce427ec4 <_end+e11a2ec/12588488>

Trace; c01bb1ad <radeon_cp_init+6d/80>
Trace; c01b55ed <radeon_ioctl+cd/150>
Trace; c0149ce9 <sys_ioctl+c9/250>
Trace; c01072f7 <system_call+33/38>

Code;  c01bad2d <radeon_do_init_cp+4bd/7c0>
00000000 <_EIP>:
Code;  c01bad2d <radeon_do_init_cp+4bd/7c0>   <=====
   0:   89 02                     mov    %eax,(%edx)   <=====
Code;  c01bad2f <radeon_do_init_cp+4bf/7c0>
   2:   89 34 24                  mov    %esi,(%esp,1)
Code;  c01bad32 <radeon_do_init_cp+4c2/7c0>
   5:   e8 d9 f2 ff ff            call   fffff2e3 <_EIP+0xfffff2e3>
Code;  c01bad37 <radeon_do_init_cp+4c7/7c0>
   a:   89 74 24 04               mov    %esi,0x4(%esp,1)
Code;  c01bad3b <radeon_do_init_cp+4cb/7c0>
   e:   89 2c 24                  mov    %ebp,(%esp,1)
Code;  c01bad3e <radeon_do_init_cp+4ce/7c0>
  11:   e8 0d f8 00 00            call   f823 <_EIP+0xf823>

Re: [OOPS] 2.4.20-rc4-ac1 (also occurs 2.4.20-rc2-ac3) in radeon DRI for XFree86

[Marc-Christian Petersen]

Hi Joshua,

> Ksymoops output follows.
> I compiled Radeon DRM stuff into the kernel -- i845 agp support from 
> agapgart. I am using gcc-3.2 to compile. 100% reproducible (okay, i've been
> spending too much time on bugzillas...) Feel the power of the oops.

I've posted a similar oops with latest rc2 -AC kernel. I have an ATI Rage128 
card and also got those oops if using DRI.

I hope Arjan may find the bug :)

ciao, Marc

Re: [OOPS] 2.4.20-rc4-ac1 (also occurs 2.4.20-rc2-ac3) in radeon DRI for XFree86

[Joshua Kwan]

Some additional information - DRI would refuse to work in aforementioned versions of the -ac branch.
I had to reboot to -rc2-ac2 to play Quake III Arena in hardware mode, so I think i'll stick with it until Arjan manages to iron out the bugs. :D

Regards
-Josh

Rabid cheeseburgers forced Marc-Christian Petersen <m.c.p@wolk-project.de> to write this on Wed, 27 Nov 2002 09:39:38 +0100:	

> Hi Joshua,
> 
> > Ksymoops output follows.
> > I compiled Radeon DRM stuff into the kernel -- i845 agp support from 
> > agapgart. I am using gcc-3.2 to compile. 100% reproducible (okay, i've been
> > spending too much time on bugzillas...) Feel the power of the oops.
> 
> I've posted a similar oops with latest rc2 -AC kernel. I have an ATI Rage128 
> card and also got those oops if using DRI.
> 
> I hope Arjan may find the bug :)
> 
> ciao, Marc

Re: [OOPS] 2.4.20-rc4-ac1 (also occurs 2.4.20-rc2-ac3) in radeon DRI for XFree86

[Marc-Christian Petersen]

On Wednesday 27 November 2002 10:39, Joshua Kwan wrote:

Hi Joshua,

> Some additional information - DRI would refuse to work in aforementioned
> versions of the -ac branch. I had to reboot to -rc2-ac2 to play Quake III
> Arena in hardware mode, so I think i'll stick with it until Arjan manages
> to iron out the bugs. :D
he did. I got a private email from him some minutes ago. It fixes the problem 
:-)

Forwarded to lkml and to you.

ciao, Marc

Re: [OOPS] 2.4.20-rc4-ac1 (also occurs 2.4.20-rc2-ac3) in radeon DRI for XFree86

[Joshua Kwan]

Not good! Not good!

Booted fine with no oops or DRI init errors while starting X.

and the quake3 test, alas:

- went into 640x480 correctly but would not display the opening movie;
- keyboard stopped responding and I had to hard-shutdown.
- i was wearing headphones too and I got some big static from my sound card :X

Looked okay, but it seems to not work all the way.
Compiling the driver gave me lots of warnings which I am not entirely sure appear in the kernels before those affected:

In file included from radeon_drv.c:35:
radeon_drv.h:846:2: warning: #warning PCI posting bug
In file included from drm_dma.h:33,
                 from radeon_drv.c:42:
drm_os_linux.h:16:2: warning: #warning the author of this code needs to read up on list_entry
In file included from radeon_cp.c:35:
radeon_drv.h:846:2: warning: #warning PCI posting bug
In file included from radeon_cp.c:36:
drm_os_linux.h:16:2: warning: #warning the author of this code needs to read up on list_entry
In file included from radeon_state.c:35:
radeon_drv.h:846:2: warning: #warning PCI posting bug
In file included from radeon_state.c:36:
drm_os_linux.h:16:2: warning: #warning the author of this code needs to read up on list_entry
In file included from radeon_mem.c:36:
radeon_drv.h:846:2: warning: #warning PCI posting bug
In file included from radeon_mem.c:37:
drm_os_linux.h:16:2: warning: #warning the author of this code needs to read up on list_entry
radeon_mem.c:135: warning: `print_heap' defined but not used
In file included from radeon_irq.c:37:
radeon_drv.h:846:2: warning: #warning PCI posting bug
In file included from radeon_irq.c:38:
drm_os_linux.h:16:2: warning: #warning the author of this code needs to read up on list_entry

this may or may not have caused the problem. bummer. I'm back on 2.4.20-rc2-ac2 right now, old reliable... and boy i need bed...3am where I am. i still have the diff lying around for reference.

night all -- josh

Rabid cheeseburgers forced Marc-Christian Petersen <m.c.p@wolk-project.de> to write this on Wed, 27 Nov 2002 09:39:38 +0100:	

> Hi Joshua,
> 
> > Ksymoops output follows.
> > I compiled Radeon DRM stuff into the kernel -- i845 agp support from 
> > agapgart. I am using gcc-3.2 to compile. 100% reproducible (okay, i've been
> > spending too much time on bugzillas...) Feel the power of the oops.
> 
> I've posted a similar oops with latest rc2 -AC kernel. I have an ATI Rage128 
> card and also got those oops if using DRI.
> 
> I hope Arjan may find the bug :)
> 
> ciao, Marc

Re: [OOPS] 2.4.20-rc4-ac1 (also occurs 2.4.20-rc2-ac3) in radeon DRI for XFree86

[Marc-Christian Petersen]

On Wednesday 27 November 2002 11:57, Joshua Kwan wrote:

Hi Joshua,

> Not good! Not good!
> Booted fine with no oops or DRI init errors while starting X.
> and the quake3 test, alas:

> - went into 640x480 correctly but would not display the opening movie;
> - keyboard stopped responding and I had to hard-shutdown.
> - i was wearing headphones too and I got some big static from my sound card
> :X
hmm, I don't have any problems with it. Quake3 works fine, opening Movie 
works, no lockups with mouse/keyboard or similar.

maybe the radeon code has also another bug, rage128 works.

ciao, Marc