[PATCH 2/3] nvme-pci: remove cached shadow doorbell offsets

John Levon levon at movementarian.org
Mon Oct 4 02:42:28 PDT 2021


On Mon, Apr 27, 2020 at 04:52:42PM -0700, Keith Busch wrote:

> And when this feature is in use, the specification requires all queue
> updates use this mechanism, so don't don't treat the admin queue
> differently.

Also came across this patch of yours. We were wondering why Linux isn't spec
compliant for the admin queues. But this would cause implementation difficulties
for a controller, right? Currently we presume that the admin queue's shadow
doorbells aren't valid, so always look at BAR0 location. It's unclear what we'd
do if this change were made in Linux, presuming we need to support older Linux
versions. Some hack that observes a non-zero shadow value and presumes the
admin queue is using shadow doorbells?

Thoughts?

thanks
john



More information about the Linux-nvme mailing list