Thread (27 messages) 27 messages, 6 authors, 2018-09-14

Re: [PATCH v4 07/10] block, scsi: Rework runtime power management

From: Bart Van Assche <hidden>
Date: 2018-08-06 17:17:12

On Mon, 2018-08-06 at 10:46 +0800, jianchao.wang wrote:
On 08/04/2018 08:03 AM, Bart Van Assche wrote:
quoted
Instead of allowing requests that are not power management requ=
ests
quoted
to enter the queue in runtime suspended status (RPM_SUSPEND=
ED), make
quoted
the blk_get_request() caller block.=20
=20
Who will resume the runtime suspended device ?
=20
In original blk-legacy, when a request is added, blk_pm_add�
F8-request will call
pm_request_resume could do that. The request will be issued a=
fter the q is resumed
successfully.
=20
After this patch, non-pm request will be blocked and pm_request�
F8-resume is removed from
blk_pm_add_request, who does resume the runtime suspended=
 q and device ?

Hello Jianchao,

I will double check that all contexts from which requests are submitted
protect submission of non-RQF_PM requests with scsi_autopm_get�
AF8-device() /
scsi_autopm_put_device(). scsi_autopm_get_device() =
namely resumes suspended
devices.

Bart.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help