Re: [PATCH] crypto: qce - Remove driver

[Konrad Dybcio]

On 7/28/26 5:22 PM, Eric Biggers wrote:
> On Fri, Jul 24, 2026 at 08:51:44AM -0700, Eric Biggers wrote:
>> It is consuming a lot of the community's time to help maintain,
>> including dealing with LLM-found bugs, with no clear benefit to anyone.
>> Even considering *just today* we can see someone sent a bug fix:
>> https://lore.kernel.org/linux-crypto/20260724081537.191992-2-thorsten.blum@xxxxxxxxx/
> 
> Another example:
> https://lore.kernel.org/linux-crypto/d5de5a55-eb35-4ad5-8d53-12e188e156cb@xxxxxxxxxxxxxxxx/
> 
>> It definitely *was* causing problems before it was disabled via the
>> crypto priority system (which made it unused in the kernel) and dropped
>> it from AF_ALG (which removed most of the unprivileged attack surface).
>> When anyone accidentally used it, it caused at least a huge performance
>> problem, and sometimes other problems too like filesystem hangs.  It was
>> an issue for years.
> 
> This actually seems to have been an overly-optimistic characterization!
> As per
> https://lore.kernel.org/linux-crypto/d5de5a55-eb35-4ad5-8d53-12e188e156cb@xxxxxxxxxxxxxxxx/
> this driver drains the battery on every system with the module loaded,
> even when no crypto requests are actually sent to it.

Unfortunately removing it drains way more battery, because the buses
never hit sync_state and are kept at their maximum frequencies:

[   18.021823] qnoc-glymur interconnect-1: sync_state() pending due to 1dfa000.crypto
[   18.029605] qnoc-glymur 16e0000.interconnect: sync_state() pending due to 1dfa000.crypto

(it transiently also pins DRAM and friends to max-freq which absolutely
kills battery life)

I understand your concerns given the past handling of this driver,
however I believe the nuclear option comes with way too much fallout..

Konrad