[Skiboot] [PATCH skiboot] npu2: Clear fence on all bricks

Alexey Kardashevskiy aik at ozlabs.ru
Mon Dec 2 13:05:50 AEDT 2019



On 30/11/2019 03:47, Reza Arbab wrote:
> On Fri, Nov 22, 2019 at 11:04:22AM +1100, Alexey Kardashevskiy wrote:
>> Reza/Ryan, could you please add more details about what exactly causes
>> these UR HMIs? Thanks!
> 
> Hopefully I've pieced together the bug history correctly. As I
> understand it...
> 
> Each GPU has a 640kb protected region which will result in a
> "unsupported request" (UR) response. The root bug is that the driver
> maps and accidentally accesses that area.

Oh. Is this address range described anywhere? We could disable mapping
these as a precaution measure.


> This firmware patch helps for recovery. From our perspective it may seem
> redundant to clear the fence on all bricks instead of just the one we're
> resettting, but at a hardware level the above UR sends a fence signal to
> all the hardware units so they all need to be cleared.
> 
> Acked-by: Reza Arbab <arbab at linux.ibm.com>


Thanks!


-- 
Alexey


More information about the Skiboot mailing list