https://bugzilla.kernel.org/show_bug.cgi?id=209163

--- Comment #2 from Satish patel (satish.in@outlook.in) --- Created attachment 292359 --> https://bugzilla.kernel.org/attachment.cgi?id=292359&action=edit Mesa_opencl version information

https://bugzilla.kernel.org/show_bug.cgi?id=209163

Satish patel (satish.in@outlook.in) changed:

What |Removed |Added ---------------------------------------------------------------------------- Component|Video(DRI - non Intel) |Other Product|Drivers |Memory Management

https://bugzilla.kernel.org/show_bug.cgi?id=209163

--- Comment #5 from Satish patel (satish.in@outlook.in) --- (In reply to Christian König from comment #4)

This is expected behavior, your application tries to use more memory than physical available:

[71804.930003] [drm:amdgpu_cs_ioctl [amdgpu]] *ERROR* Not enough memory for command submission!

That is most likely a bug in the application, e.g. a memory leak.

Dear Mr. Konig,

Thanks for your reply , But I would like to inform and describe same application running up to 10 days until Physical memory and swap memory not utilized in CentOS 7 (gnome display ) with kernel 3.10.0-1127.el7.x86_64.

But same application has error "amdgpu: The CS has been cancelled because the context is lost" even system utilize only 75% physical memory from Total 5.83 GB Physical memory and 1% swap memory from 15 GB swap partition. This Error , I am getting in Kernel 4.9.118. Why system crash ( Display flickering and touch screen not responding) and not utilize swap memory area ? . But CPU and memory utilization showing when monitoring from other system .

https://bugzilla.kernel.org/show_bug.cgi?id=209163

--- Comment #7 from Satish patel (satish.in@outlook.in) --- Created attachment 292449 --> https://bugzilla.kernel.org/attachment.cgi?id=292449&action=edit VRAM Utilization screen shot

It's attached VRAM Utilization error screen shot as output of - cat /sys/kernel/debug/dri/0/amdgpu_vram_mm

https://bugzilla.kernel.org/show_bug.cgi?id=209163

--- Comment #9 from Christian König (christian.koenig@amd.com) --- Try amdgpu.vramlimit=512 on the kernel command line to limit the available VRAM to 512MB.

The problem is certainly some kind of memory leak.

You need to test an up to date kernel, like 5.8 or even better the latest bleeding edge amd-staging-drm-next branch.