Re: [PATCH for-next] Revert "IB/mlx5: Don't return errors from poll_cq"

[Haakon Bugge]

> On 3 Mar 2022, at 20:09, Leon Romanovsky <leon@xxxxxxxxxx> wrote:
> 
> On Thu, Mar 03, 2022 at 02:50:17PM +0100, Håkon Bugge wrote:
>> This reverts commit dbdf7d4e7f911f79ceb08365a756bbf6eecac81c.
>> 
>> Commit dbdf7d4e7f91 ("IB/mlx5: Don't return errors from poll_cq") is
>> needed, when driver/fw communication gets wedged.
>> 
>> With a large fleet of systems equipped with CX-5, we have observed the
>> following mlx5 error message:
>> 
>> wait_func:945:(pid xxx): ACCESS_REG(0x805) timeout. Will cause a
>> leak of a command resource
> 
> It is arguably FW issue. Please contact your Nvidia support representative.

The RC for the whacked driver/fw communication has been raised with Nvidia support. This commit is to avoid the kernel to crash when this situation arises. And inevitable, it may happen.


Thxs, Håkon