From: Keith Busch <kbusch@kernel.org>
To: Sagi Grimberg <sagi@grimberg.me>
Cc: Edmund Nadolski <edmund.nadolski@intel.com>,
Christoph Hellwig <hch@lst.de>,
linux-nvme@lists.infradead.org
Subject: Re: [PATCH 5/5] nvme: Wait for reset state when required
Date: Thu, 5 Sep 2019 14:55:01 -0600 [thread overview]
Message-ID: <20190905205501.GD25467@localhost.localdomain> (raw)
In-Reply-To: <a1895103-5906-7a46-9222-478bc76a2dbd@grimberg.me>
On Thu, Sep 05, 2019 at 01:47:26PM -0700, Sagi Grimberg wrote:
>
> > Disabling a controller in a path that resets it in a different
> > context needs to fence off other resets from occuring between these
> > actions. Provide a method to block until we've successfully entered a
> > reset state prior to disabling so that we can perform the controller
> > disabling without racing with another reset. Otherwise, that other reset
> > may either undo our teardown, or our teardown may undo the in-progress
> > reset's bringup.
>
> Why not simply flushing the reset work like nvme_reset_ctrl_sync?
>
> I might be out of context here...
It's not enough to know a current reset completes. We have to know that
another won't schedule, which the RESETTING state guarantees.
More context:
This is for when we can't re-enable the controller in the same context
as when we have to disable it.
For example, PCIe FLR occurs in two stages. Suspend/resume is another
example. An IO timeout occuring concurrently with either gets the
wrong end result because the timeout tries to incorrectly re-enable the
device. This patch appropriatly synchronizes everyone through the state
machine to prevent these things from running concurrently.
_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
next prev parent reply other threads:[~2019-09-05 20:56 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-05 14:26 [PATCH 1/5] nvme: Restart request timers in resetting state Keith Busch
2019-09-05 14:26 ` [PATCH 2/5] nvme: Prevent resets during paused states Keith Busch
2019-09-05 20:23 ` Sagi Grimberg
2019-09-05 20:35 ` Keith Busch
2019-09-05 20:42 ` Sagi Grimberg
2019-09-05 14:26 ` [PATCH 3/5] nvme-pci: Free tagset if no IO queues Keith Busch
2019-09-05 20:24 ` Sagi Grimberg
2019-09-05 20:40 ` Keith Busch
2019-09-05 20:43 ` Sagi Grimberg
2019-09-05 14:26 ` [PATCH 4/5] nvme: Remove ADMIN_ONLY state Keith Busch
2019-09-05 14:26 ` [PATCH 5/5] nvme: Wait for reset state when required Keith Busch
2019-09-05 15:57 ` James Smart
2019-09-05 20:47 ` Sagi Grimberg
2019-09-05 20:55 ` Keith Busch [this message]
2019-09-05 20:13 ` [PATCH 1/5] nvme: Restart request timers in resetting state Sagi Grimberg
2019-09-05 20:25 ` Keith Busch
2019-09-05 20:39 ` Sagi Grimberg
2019-09-05 21:36 ` James Smart
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190905205501.GD25467@localhost.localdomain \
--to=kbusch@kernel.org \
--cc=edmund.nadolski@intel.com \
--cc=hch@lst.de \
--cc=linux-nvme@lists.infradead.org \
--cc=sagi@grimberg.me \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox