From: David Sterba <dsterba@suse.cz>
To: Anand Jain <anand.jain@oracle.com>
Cc: dsterba@suse.cz, linux-btrfs@vger.kernel.org
Subject: Re: [PATCH 4/4] btrfs: sysfs, add devid/dev_state kobject and attribute
Date: Mon, 6 Jan 2020 17:00:04 +0100 [thread overview]
Message-ID: <20200106160004.GL3929@twin.jikos.cz> (raw)
In-Reply-To: <1faa5860-130a-9f3c-3e44-724ce9a26adb@oracle.com>
On Sat, Dec 14, 2019 at 08:26:24AM +0800, Anand Jain wrote:
>
>
> On 14/12/19 1:02 AM, David Sterba wrote:
> > On Fri, Dec 13, 2019 at 05:43:32PM +0100, David Sterba wrote:
> >>> Looked into this further, actually we don't need any lock here
> >>> the device delete thread which calls kobject_put() makes sure
> >>> sysfs read is closed. So an existing sysfs read thread will have
> >>> to complete before device free.
> >>>
> >>>
> >>> CPU1 CPU2
> >>>
> >>> btrfs_rm_device
> >>> open file
> >>> btrfs_sysfs_rm_device_link
> >>> call read, access freed device
> >>> sysfs waits for the open file
> >>> to close.
> >>
> >> How exactly does sysfs wait for the device? Is it eg wait_event checking
> >> number of references? If the file stays open by an evil process is it
> >> going to block the device removal indefinitelly?
> >
> > Yeah, sysfs waits until the file is closed. Eg. umount can be stalled
> > that way too.
>
> And similar to umount, I don't think we should return EBUSY
> for btrfs_rm_device if the device sysfs attribute is opened,
> as sysfs show attributes are non blocking and would be completed
> in the timely manner.
While I don't think the blocking behaviour is totally OK, returning
EBUSY could be confusing without any other explanation. Also leaving
sysfs files open but not read is kind of strange on itself.
next prev parent reply other threads:[~2020-01-06 16:00 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-12-05 11:27 [PATCH 0/4] btrfs, sysfs cleanup and add dev_state Anand Jain
2019-12-05 11:27 ` [PATCH 1/4] btrfs: sysfs, use btrfs_sysfs_remove_fsid in fail return in add_fsid Anand Jain
2019-12-05 11:27 ` [PATCH 2/4] btrfs: sysfs, add UUID/devinfo kobject Anand Jain
2020-01-14 18:32 ` David Sterba
2020-02-03 10:40 ` Anand Jain
2019-12-05 11:27 ` [PATCH v2 3/4] btrfs: sysfs, rename device_link add,remove functions Anand Jain
2019-12-05 11:27 ` [PATCH 4/4] btrfs: sysfs, add devid/dev_state kobject and attribute Anand Jain
2019-12-05 14:10 ` David Sterba
2019-12-05 14:38 ` Anand Jain
2019-12-05 14:21 ` David Sterba
2019-12-05 14:38 ` Anand Jain
2019-12-05 15:14 ` David Sterba
2019-12-06 13:49 ` Anand Jain
2019-12-09 14:05 ` Anand Jain
2019-12-09 18:31 ` David Sterba
2019-12-13 16:43 ` David Sterba
2019-12-13 17:02 ` David Sterba
2019-12-14 0:26 ` Anand Jain
2020-01-06 16:00 ` David Sterba [this message]
2019-12-09 14:06 ` [PATCH v2 " Anand Jain
2019-12-19 10:41 ` [PATCH v3 " Anand Jain
2020-01-06 11:38 ` [PATCH v4] " Anand Jain
2020-01-09 15:20 ` [PATCH v4 4/4] " David Sterba
2020-01-10 1:03 ` Anand Jain
2019-12-05 15:00 ` [PATCH 0/4] btrfs, sysfs cleanup and add dev_state David Sterba
2019-12-06 13:46 ` Anand Jain
2019-12-09 14:09 ` Anand Jain
2019-12-09 22:48 ` Anand Jain
2019-12-10 16:41 ` David Sterba
2020-01-14 18:39 ` David Sterba
2020-01-15 8:22 ` [PATCH] btrfs: update devid after replace Anand Jain
2020-01-15 16:17 ` David Sterba
2020-01-20 19:10 ` David Sterba
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=20200106160004.GL3929@twin.jikos.cz \
--to=dsterba@suse.cz \
--cc=anand.jain@oracle.com \
--cc=linux-btrfs@vger.kernel.org \
/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