Whoops, Intel SNB Is Borked At The Last Minute In Linux 2.6.39

[Kano]

Did you ever try trine? That crashes snb with ease...

[Alex W. Jackson]

Does it still hang with the above kernel patches? It's probably a different problem, maybe in Mesa rather than in the kernel.

[Kano]

I did not compile the kernel, i used an ubuntu daily build, it only has got the first patch.

[Kano]

@Alex W. Jackson

I used this kernel

http://kernel.ubuntu.com/~kernel-ppa/mainline/daily/2011-06-25-oneiric/COMMIT

which should have got both patches

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=536142f950f7ea4f3d146a138ad6938f28a34f33

But as soon as i start trine

[   68.606982] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[   68.606994] [drm] capturing error event; look for more information in /debug/dri/0/i915_error_state
[   68.611576] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 14487 at 14484, next 14489)
[   75.087725] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[   75.087751] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 14512 at 14503, next 14513)
[   81.516585] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[   81.516609] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 14549 at 14540, next 14550)
[   87.917519] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[   87.917546] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 14567 at 14558, next 14568)
[   94.346386] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[   94.346412] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 14586 at 14576, next 14587)

So something is still wrong. I do not use sna, but current ddx (with a little patch) and current mesa with xserver 1.7.7. The patch i use for ddx is

diff --git a/configure.ac b/configure.ac
index 7bde768..f5b1146 100644
--- a/configure.ac
+++ b/configure.ac
@@ -135,7 +135,7 @@ XORG_DRIVER_CHECK_EXT(DPMSExtension, xextproto)
 PKG_CHECK_MODULES(XORG, [xorg-server >= $required_xorg_xserver_version xproto fontsproto $REQUIRED_MODULES])
 PKG_CHECK_MODULES(DRM, [libdrm >= 2.4.23])
 PKG_CHECK_MODULES(DRI, [xf86driproto], , DRI=no)
-PKG_CHECK_MODULES(DRI2, [dri2proto >= 2.6])
+PKG_CHECK_MODULES(DRI2, [dri2proto >= 2.3])
 PKG_CHECK_MODULES(PCIACCESS, [pciaccess >= 0.10])
 
 sdkdir=`$PKG_CONFIG --variable=sdkdir xorg-server`
diff --git a/src/intel_dri.c b/src/intel_dri.c
index be5c6b1..2af02cd 100644
--- a/src/intel_dri.c
+++ b/src/intel_dri.c
@@ -67,6 +67,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "dri2.h"
 
+#ifndef DRI2BufferHiz
+#define DRI2BufferHiz			10
+#endif
+
 typedef struct {
 	int refcnt;
 	PixmapPtr pixmap;
diff --git a/src/sna/sna_dri.c b/src/sna/sna_dri.c
index 20b323e..0cb344a 100644
--- a/src/sna/sna_dri.c
+++ b/src/sna/sna_dri.c
@@ -66,6 +66,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "dri2.h"
 
+#ifndef DRI2BufferHiz
+#define DRI2BufferHiz                   10
+#endif
+
 #if DRI2INFOREC_VERSION <= 2
 #error DRI2 version supported by the Xserver is too old
 #endif

Btw. it does not matter if i use "i915.semaphores=1" or not.

[Alex W. Jackson]

Yeah, that isn't the same issue at all. The hangs fixed by the kernel patches produce a "missed IRQ?" message in the log, or "kicking stuck semaphore" if semaphore=1 is set. You should file a new Mesa bug.

What is that patch for?

[Kano]

Its a compile fix when you only want to update libdrm on a squeeze install.