[Bug 213145] AMDGPU resets, timesout and crashes after "*ERROR* Waiting for fences timed out!"

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



https://bugzilla.kernel.org/show_bug.cgi?id=213145

Alex Deucher (alexdeucher@xxxxxxxxx) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |alexdeucher@xxxxxxxxx

--- Comment #13 from Alex Deucher (alexdeucher@xxxxxxxxx) ---
(In reply to Eric Wheeler from comment #12)
> Considering this is a kernel crash, why wouldn't we still consider this a
> kernel bug?
> 
> Just because it can be fixed in userspace doesn't mean we shouldn't address
> the kernel crash because userspace should not be able to crash the kernel!  
> 
> For servers it is considered a security problem when non-root users can
> crash the machine.

There is no reasonable way for the kernel to handle it.  We would need to parse
every texture, vertex buffer, command buffer, and shader that the application
submits to the GPU to try and verify if that combination of data, shader code
and pipeline state would possibly cause a hang.  A more robust solution would
for session managers to monitor the GPU for resets and then restart the user's
session so they don't lose their desktop.  This is how it works on other OSes.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.



[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux