Bug 13640

Created attachment 13088 [details]
xorg.conf

Created attachment 13089 [details]
quake 3 log msg

Created attachment 13090 [details]
xorg.0.log

q3demo is running fine on our Q965 with the latest git code.

which the commit number are you using for those X components?  
Or it is just the Q965 is different from my 965GM chipset? 

Forget to mention my base system is FC8 IA32. of cause, the whole X system use upstream tree. 

Created attachment 13207 [details]
q3demo's output

I have tried the latest X system and installed the ko file in drm, The TTM error was not be output. and the q3demo can be opened but it fail quickly and output a error:

DRM_I830_EXECBUFFER: -16

The mesa version: 4cca760a9b7fe120ed92257a6a6c4a3a6e929ba8 
The DRM version: 2db6400396ea5c8a5ce54fe9e211b9d01a11d506
Xf86-video-intel version: f69b48fe24ef94dac44b8123884ca71df675be4b
Xserver version: bcad2a5a24f30cfdf9eca31915ed5a55ed094285

add Gordon to CC list 

Created attachment 13213 [details]
xorg.0.log

This bug also be found at G33 Video on x86_64 platform. 
the openarena or quake3 game also failed on this bug. 

Eric:
I am quite a newbie on X system.
The function stack of this bug is : 
drmCommandWriteRead
intel_exec_ioctl
do_flush_locked
intel_batchbuffer_flush
and I found you do changes on drmCommandWriteRead. Does this change cause the issue? 

Alex 
commit 7c71ef3a3d0cf2620525f468960cdc76a0fb0d33
Author: Eric Anholt <eric@anholt.net>
Date:   Wed Dec 12 10:25:19 2007 -0800

    [intel] Move bufmgr back to context instead of screen, fixing glthreads.

    Putting the bufmgr in the screen is not thread-safe since the emit_reloc
    changes.  It also led to a significant performance hit from pthread usage
    for the attempted thread-safety (up to 12% of a cpu spent on refcounting
    protection in single-threaded 965).  The motivation had been to allow
    multi-context bufmgr sharing in classic mode, but it wasn't worth the cost.

diff --git a/src/mesa/drivers/dri/i915/intel_ioctl.c b/src/mesa/drivers/dri/i915/intel_ioctl.c
index 94f7e73..37704d6 100644
--- a/src/mesa/drivers/dri/i915/intel_ioctl.c
+++ b/src/mesa/drivers/dri/i915/intel_ioctl.c
@@ -47,15 +47,14 @@
 #define FILE_DEBUG_FLAG DEBUG_IOCTL

 int
-intelEmitIrqLocked(intelScreenPrivate *intelScreen)
+intelEmitIrqLocked(struct intel_context *intel)
 {
    drmI830IrqEmit ie;
    int ret, seq;

    ie.irq_seq = &seq;

-   ret = drmCommandWriteRead(intelScreen->driScrnPriv->fd,
-                            DRM_I830_IRQ_EMIT, &ie, sizeof(ie));
+   ret = drmCommandWriteRead(intel->driFd, DRM_I830_IRQ_EMIT, &ie, sizeof(ie));
    if (ret) {
       fprintf(stderr, "%s: drmI830IrqEmit: %d\n", __FUNCTION__, ret);
       exit(1);
@@ -67,7 +66,7 @@ intelEmitIrqLocked(intelScreenPrivate *intelScreen)
 }

If you think it's that commit, did you try checking out just before it to see if 
that fixes your problem?  git-bisect is a tool to help you track down what commit broke things.

Is this duplicated with bug #13732?
q3demo only has 32-bit binary. And, Alex, you also met this bug on 64-bit system, right?

I agree with Nian
Alex, is your system x86-64?

re-clarify the bug scenes. I have met the problem twice:
1, 32bits q3demo on G965 with IA32 kernel
2, 64bits openarena on G33 with IA32e kernel. 

Alex 

should be a dup of bug# 13732...

*** This bug has been marked as a duplicate of bug 13732 ***

Mass version move, cvs -> git