00:18 <opticron> ooh, new debian packages

08:41 <Ry_Darcy> @all Glanzmann has kindly provided a patch to remove message flooding "Blocking due to vertex/fragment" in syslog. However, I am still being inundated with "asahi: WorkQueue: Cannot submit, but queue is empty?" messages. These errors (methinks) are generated by

08:41 <Ry_Darcy> ../linux/drivers/gpu/drm/asahi/workqueue.rs. I presume this is all debug/test noise.

10:33 <cy8aer> IMHO the messages do not come from nowhere (and Glanzsmann only deleted the messages). As I understand this is the main problem lina is just checking with the firefox and other applications massive blocks.

10:33 <cy8aer> (s/Glanszsmann/Glanzmann/g sorry)

10:55 <cy8aer> And I for myself experience firefox crashes with the new kernel.

11:12 <Ry_Darcy> @cy8aer As stated already, these messages are no longer being generated "Blocking due to vertex/fragment.." with this (Glanzmann's) patch. However, these messages ("asahi: WorkQueue: Cannot submit, but queue is empty?" are still flooding syslog.

11:14 <Ry_Darcy> @cy8aer I am also experiencing a lot of FF crashes also (Debian, XFCE, M1 Mini, kernel 6.4-3, latest Mesa).

11:18 <cy8aer> Ry_Darcy: Yes the messages are not created anymore because of this: https://tg.st/u/0001-remove-debug-spam.patch but the problem is still there of course. No messages created +1, but in background the GPU buffers still overflow. You can of course supress "cannot submit"... but this will not fix the problem. The debugging messages really only occur when there is a problem in the kernel. And this problem is big (as lina stated in #asahi-gpu)

11:26 <Ry_Darcy> @cy8aer agreed. Just back again after another FF crash and OOM.

11:27 <cy8aer> I have compiled the kernels by myself. 6.4-3 is really buggy and I now look for 6.4-1 for the messages...

11:28 <Ry_Darcy> @cy8aer I compiled the kernel also.

11:49 <cy8aer> Ok, there is a new mesa upstream...

12:08 <cy8aer> ... and with this the Blocking-Messages are way less in dmesg. There are some but this is not worth mentioning.

12:19 <Ry_Darcy> @cy8aer are you running Debian?

12:19 <cy8aer> yes

12:19 <cy8aer> And I self compiled mesa right now

12:19 <cy8aer> with Glanzmann's source

12:20 <cy8aer> (just bumped the mesa.sh to 20230725)

12:20 <Ry_Darcy> @cy8aer what does your mesa build script look like?, if you do not mind me asking..

12:23 <cy8aer> Ok it is based on https://git.zerfleddert.de/cgi-bin/gitweb.cgi/m1-debian ...

12:23 <cy8aer> https://git.g3la.de/repos/m1-debian/src/branch/baer/mesa.sh

12:24 <cy8aer> Main difference is line 28. You may need to run "dependencies.sh" and "prepare_rust.sh" before running "mesa.sh"

12:34 <Ry_Darcy> @cy8aer Thanks.

12:35 <cy8aer> It is quite easy to build it on your own. And with gitea 1.2.0 I was able to build a shadow repository too.

12:42 <cy8aer> Now I have to provoke Firefox a little...

13:41 <Ry_Darcy> I have dependency problems. apt build-dep . => The following packages have unmet dependencies: libx11-dev : Depends: libx11-6 (= 2:1.8.4-2+deb12u1) but 2:1.8.6-1 is to be installed libx11-xcb-dev : Depends: libx11-xcb1 (= 2:1.8.4-2+deb12u1) but 2:1.8.6-1 is to be installed

15:22 <cy8aer> Which DIst are you running? Testing?

15:28 <cy8aer> For me it looks like libx11-xcb1 and libx11-6 are somehow hanging with their versions. I have libx11-6=2:1.8.6-1 and libx11-xcb1=2:1.8.6-1 and the corresponding dev files installed.

15:31 <cy8aer> 2:1.8.4-2 is stable (bookworm). That looks like a mixup between stable and testing in the apt packages definitions.

15:32 <cy8aer> And the Glanzmann Repos were bound to testing.

15:33 <cy8aer> (as far as I know)

18:58 <cy8aer> But FF is still crashing...