Bliss os 15 (Alpha Android 12)

[The Tutto Far Tutto]

I tryed android 12 bliss os 15 and i got like corrupted UI alot of java errors things and after few seconds reboot is like impossible to run.

[Mauro Rossi]

Hello, 

what is your HW and Graphics card?

Could you please boot in Live DEBUG mode using gbm_gralloc, try to insert an additional  EXT4 formatted USB drive to see if it appears as sdc1 or sdd1,

then edit the kernel cmdline entry DATA=/dev/block/sdc1 (or sdd1) and you will have the /log.txt and /tombstones/tombstone_00 files

If you post log.txt and tombostone_00 in this thread we may be able to understand what is happening to you.

I saw a strange problem on ASUS Prime B250M-A with Intel Celeron G3900 with both r-x86 and s-x86 experimental builds,

the boot to Android GUI completes only in sporadic cases, most of the times there is an hard restart during the bootanimation.

I am going to try the same procedure I have described to you on my HW.

Mauro

[crw]

You have to choose a launcher before it reboots itself, that's the workaround

[Mauro Rossi]

On s-x86 experimental iso, 

I see the artifacts due to Taskbar using ResolverActivity which is using gralloc flag SW_WRITE_OFTEN for buffers that need CPU write,

but Skylake GT1 (iris) is the first Intel platform affected by the problem. Infact on a Baytrail (crocus) the GUI is rock stable and there are no ResolverActivity artifacts.

I guess that this time I can open an issue to mesa, because i965 and crocus do not have such problems and they will triage if the problem is in gbm_gralloc or in mesa gbm/dri backend.

You may also notice the disrupted cursor near the center of Navigation bar, those cursor artifacts do not depend on ResolverActivty, that is an independent additional problem, which extremely weird because gbm_gralloc does not even pass the cursor flag to mesa gbm at buffer allocation.

 

On Saturday, January 22, 2022 at 5:10:10 PM UTC+1 crw wrote:

You have to choose a launcher before it reboots itself, that's the workaround

Due to continuos GUI crashes, in order to be able to get the tombstones I had to connect via Ethernet cable with static IP address.

I believe the tombstones where system_server crashes the one correlated to GUI restarts (not attaching them based on the assumption that the problem is related to ResolverActivity window (the one garbled)

If passed that step of launcher selection there is no GUI crash, then the attached patch for framework/base project should also allow to to see the ResolverActivity window to choose the launcher.

[Mauro Rossi]

You have to choose a launcher before it reboots itself, that's the workaround

Due to continuos GUI crashes, in order to be able to get the tombstones I had to connect via Ethernet cable with static IP address.

I believe the tombstones where system_server crashes the one correlated to GUI restarts (not attaching them based on the assumption that the problem is related to ResolverActivity window (the one garbled)

If passed that step of launcher selection there is no GUI crash, then the attached patch for framework/base project should also allow to to see the ResolverActivity window to choose the launcher.

Unfortunately for me the selection of the laucher does not prevent the GUI crashes from happening later,

interestingly the GUI seem stable until the mouse is moved and then GUI crashes