Linux-NVME Archive on lore.kernel.org
 help / color / mirror / Atom feed
* question about spec hw reset
@ 2017-04-20 17:44 daggs
  2017-04-20 18:00 ` Keith Busch
  0 siblings, 1 reply; 7+ messages in thread
From: daggs @ 2017-04-20 17:44 UTC (permalink / raw)


Greetings,

I have a question about the hw reset flow. I'm maintaining a kernel driver (not the one in the linux kernel) for nvme which the company I work for uses for internal benchmarking, I was asked to implement an hardware reset.

I've followed the spec and did the following:
1. disable the en bit and wait until the ctrl is off.
2. set the shutdown bit and wait until the ctrl completed shutdown.
3. did a pci reset.
4. set the num of queues in the aqa register.
5. populate the acq and asq register with the physical addr ofthe buffers.
6. configure the cc register with the default values.
7. enable the device via the en bit and wait until the ready bit it set.
8. prepate a ctrl id cmd on the sq and send a doorbell.
9. poll for completion.

my issue is that I never get a completion for that request.
am I doing something wrong? is there any way to query the controller for status that might help me figure out what I did wrong or what is the status of the controller?
are there any limitations on the addresses the admin queue can use?

Thanks,

Dagg.

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2017-04-24  7:17 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2017-04-20 17:44 question about spec hw reset daggs
2017-04-20 18:00 ` Keith Busch
2017-04-20 18:21   ` daggs
2017-04-20 19:48     ` Keith Busch
2017-04-21  5:41       ` Christoph Hellwig
2017-04-24  7:17         ` daggs
2017-04-24  7:12       ` daggs

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox