From: Vivek Goyal <vgoyal@redhat.com>
To: linux-fsdevel@vger.kernel.org, virtio-fs@redhat.com
Cc: vgoyal@redhat.com, miklos@szeredi.hu, stefanha@redhat.com,
dgilbert@redhat.com, chirantan@chromium.org,
virtualization@lists.linux-foundation.org
Subject: [PATCH 0/5] virtiofs: Fix couple of deadlocks
Date: Tue, 15 Oct 2019 13:46:21 -0400 [thread overview]
Message-ID: <20191015174626.11593-1-vgoyal@redhat.com> (raw)
Hi,
We have couple of places which can result in deadlock. This patch series
fixes these.
We can be called with fc->bg_lock (for background requests) while
submitting a request. This leads to two constraints.
- We can't end requests in submitter's context and call fuse_end_request()
as it tries to take fc->bg_lock as well. So queue these requests on a
list and use a worker to end these requests.
- If virtqueue is full, we can wait with fc->bg_lock held for queue to
have space. Worker which is completing the request gets blocked on
fc->bg_lock as well. And that means requests are not completing, that
means descriptors are not being freed and that means submitter can't
make progress. Deadlock.
Fix this by punting the requests to a list and retry submission later
with the help of a worker.
Thanks
Vivek
Vivek Goyal (5):
virtiofs: Do not end request in submission context
virtiofs: No need to check fpq->connected state
virtiofs: Set FR_SENT flag only after request has been sent
virtiofs: Count pending forgets as in_flight forgets
virtiofs: Retry request submission from worker context
fs/fuse/virtio_fs.c | 165 +++++++++++++++++++++++++++++---------------
1 file changed, 111 insertions(+), 54 deletions(-)
--
2.20.1
WARNING: multiple messages have this Message-ID (diff)
From: Vivek Goyal <vgoyal@redhat.com>
To: linux-fsdevel@vger.kernel.org, virtio-fs@redhat.com
Cc: miklos@szeredi.hu, chirantan@chromium.org, dgilbert@redhat.com,
virtualization@lists.linux-foundation.org, stefanha@redhat.com,
vgoyal@redhat.com
Subject: [PATCH 0/5] virtiofs: Fix couple of deadlocks
Date: Tue, 15 Oct 2019 13:46:21 -0400 [thread overview]
Message-ID: <20191015174626.11593-1-vgoyal@redhat.com> (raw)
Hi,
We have couple of places which can result in deadlock. This patch series
fixes these.
We can be called with fc->bg_lock (for background requests) while
submitting a request. This leads to two constraints.
- We can't end requests in submitter's context and call fuse_end_request()
as it tries to take fc->bg_lock as well. So queue these requests on a
list and use a worker to end these requests.
- If virtqueue is full, we can wait with fc->bg_lock held for queue to
have space. Worker which is completing the request gets blocked on
fc->bg_lock as well. And that means requests are not completing, that
means descriptors are not being freed and that means submitter can't
make progress. Deadlock.
Fix this by punting the requests to a list and retry submission later
with the help of a worker.
Thanks
Vivek
Vivek Goyal (5):
virtiofs: Do not end request in submission context
virtiofs: No need to check fpq->connected state
virtiofs: Set FR_SENT flag only after request has been sent
virtiofs: Count pending forgets as in_flight forgets
virtiofs: Retry request submission from worker context
fs/fuse/virtio_fs.c | 165 +++++++++++++++++++++++++++++---------------
1 file changed, 111 insertions(+), 54 deletions(-)
--
2.20.1
WARNING: multiple messages have this Message-ID (diff)
From: Vivek Goyal <vgoyal@redhat.com>
To: linux-fsdevel@vger.kernel.org, virtio-fs@redhat.com
Cc: miklos@szeredi.hu, virtualization@lists.linux-foundation.org,
vgoyal@redhat.com
Subject: [Virtio-fs] [PATCH 0/5] virtiofs: Fix couple of deadlocks
Date: Tue, 15 Oct 2019 13:46:21 -0400 [thread overview]
Message-ID: <20191015174626.11593-1-vgoyal@redhat.com> (raw)
Hi,
We have couple of places which can result in deadlock. This patch series
fixes these.
We can be called with fc->bg_lock (for background requests) while
submitting a request. This leads to two constraints.
- We can't end requests in submitter's context and call fuse_end_request()
as it tries to take fc->bg_lock as well. So queue these requests on a
list and use a worker to end these requests.
- If virtqueue is full, we can wait with fc->bg_lock held for queue to
have space. Worker which is completing the request gets blocked on
fc->bg_lock as well. And that means requests are not completing, that
means descriptors are not being freed and that means submitter can't
make progress. Deadlock.
Fix this by punting the requests to a list and retry submission later
with the help of a worker.
Thanks
Vivek
Vivek Goyal (5):
virtiofs: Do not end request in submission context
virtiofs: No need to check fpq->connected state
virtiofs: Set FR_SENT flag only after request has been sent
virtiofs: Count pending forgets as in_flight forgets
virtiofs: Retry request submission from worker context
fs/fuse/virtio_fs.c | 165 +++++++++++++++++++++++++++++---------------
1 file changed, 111 insertions(+), 54 deletions(-)
--
2.20.1
next reply other threads:[~2019-10-15 17:46 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-15 17:46 Vivek Goyal [this message]
2019-10-15 17:46 ` [Virtio-fs] [PATCH 0/5] virtiofs: Fix couple of deadlocks Vivek Goyal
2019-10-15 17:46 ` Vivek Goyal
2019-10-15 17:46 ` [PATCH 1/5] virtiofs: Do not end request in submission context Vivek Goyal
2019-10-15 17:46 ` [Virtio-fs] " Vivek Goyal
2019-10-15 17:46 ` Vivek Goyal
2019-10-21 8:03 ` Miklos Szeredi
2019-10-21 8:03 ` [Virtio-fs] " Miklos Szeredi
2019-10-21 11:52 ` Vivek Goyal
2019-10-21 11:52 ` [Virtio-fs] " Vivek Goyal
2019-10-21 11:52 ` Vivek Goyal
2019-10-21 13:58 ` Miklos Szeredi
2019-10-21 13:58 ` [Virtio-fs] " Miklos Szeredi
2019-10-15 17:46 ` [PATCH 2/5] virtiofs: No need to check fpq->connected state Vivek Goyal
2019-10-15 17:46 ` [Virtio-fs] " Vivek Goyal
2019-10-15 17:46 ` Vivek Goyal
2019-10-15 17:46 ` [PATCH 3/5] virtiofs: Set FR_SENT flag only after request has been sent Vivek Goyal
2019-10-15 17:46 ` [Virtio-fs] " Vivek Goyal
2019-10-15 17:46 ` Vivek Goyal
2019-10-15 17:46 ` [PATCH 4/5] virtiofs: Count pending forgets as in_flight forgets Vivek Goyal
2019-10-15 17:46 ` [Virtio-fs] " Vivek Goyal
2019-10-15 17:46 ` Vivek Goyal
2019-10-15 17:46 ` [PATCH 5/5] virtiofs: Retry request submission from worker context Vivek Goyal
2019-10-15 17:46 ` [Virtio-fs] " Vivek Goyal
2019-10-15 17:46 ` Vivek Goyal
2019-10-21 8:15 ` Miklos Szeredi
2019-10-21 8:15 ` [Virtio-fs] " Miklos Szeredi
2019-10-21 13:01 ` Vivek Goyal
2019-10-21 13:01 ` [Virtio-fs] " Vivek Goyal
2019-10-21 13:01 ` Vivek Goyal
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=20191015174626.11593-1-vgoyal@redhat.com \
--to=vgoyal@redhat.com \
--cc=chirantan@chromium.org \
--cc=dgilbert@redhat.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=miklos@szeredi.hu \
--cc=stefanha@redhat.com \
--cc=virtio-fs@redhat.com \
--cc=virtualization@lists.linux-foundation.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.