On Tue, Sep 3, 2019 at 7:19 PM Warner Losh wrote:
>
>
> On Tue, Sep 3, 2019 at 2:38 PM Ravi Pokala wrote:
>
>> Hi Warner,
>>
>> + /*
>> +* Per Section 7.6.2 of NVMe spec 1.4, to properly suspend, we
>> need to
>> +* delete the hardware I/O queues, and then shutdown. This
>>
On Tue, Sep 3, 2019 at 2:38 PM Ravi Pokala wrote:
> Hi Warner,
>
> + /*
> +* Per Section 7.6.2 of NVMe spec 1.4, to properly suspend, we
> need to
> +* delete the hardware I/O queues, and then shutdown. This properly
> +* flushes any metadata the drive may have store
rs() -- which implements steps (2) and
(3) -- before you are calling nvme_ctrlr_disable_qpairs() -- which implements
step (1).
Or am I missing something?
Thanks,
Ravi (rpokala@)
-Original Message-
From: on behalf of Warner Losh
Date: 2019-09-03, Tuesday at 08:26
To: , ,
Subjec
Author: imp
Date: Tue Sep 3 15:26:11 2019
New Revision: 351747
URL: https://svnweb.freebsd.org/changeset/base/351747
Log:
Implement nvme suspend / resume for pci attachment
When we suspend, we need to properly shutdown the NVME controller. The
controller may go into D3 state (or may have
