> > No strong opinion. Especially not if you say it is in the spec (although
> > "must be sufficient" would be better than "should be" ;)). Also, I
> > assume this failure is reproducible and should turn up during
> > development? Compared to "happens once in a while randomly"?
>
> At least for me, the failure happens only on some units but is
> consistent for a given unit.
I hoped for that, thanks!
> The QUIRK needs to be card specific. The software should automatically
> detect the card (from the CID) and apply the quirk. Please see patch in
> my original reply.
I understand that. What I meant was the case when you discover your card
times out on 'cache on' but it doesn't have a quirk entry yet. So, you
are the first do discover the flag is needed. I would like a pointer
there ("timeout on activating caches. Maybe your card needs $quirk?").
Because for someone just enabling the HW and not deep into MMC stuff,
the path from some -ETIMEOUT to this specific quirk may take some time
otherwise.
Attachment:
signature.asc
Description: PGP signature
