From: Miklos Szeredi <miklos@szeredi.hu>
To: Bart Van Assche <bvanassche@acm.org>
Cc: Christoph Hellwig <hch@lst.de>,
Alexander Viro <viro@zeniv.linux.org.uk>,
linux-fsdevel@vger.kernel.org, linux-aio <linux-aio@kvack.org>,
syzbot <syzkaller@googlegroups.com>,
Avi Kivity <avi@scylladb.com>, Eric Dumazet <edumazet@google.com>,
stable <stable@vger.kernel.org>
Subject: Re: [PATCH] aio: Fix locking in aio_poll()
Date: Wed, 6 Feb 2019 09:36:26 +0100 [thread overview]
Message-ID: <20190206083626.GA19821@veci.piliscsaba.redhat.com> (raw)
In-Reply-To: <1549414384.34241.77.camel@acm.org>
On Tue, Feb 05, 2019 at 04:53:04PM -0800, Bart Van Assche wrote:
> On Tue, 2019-02-05 at 09:12 +0100, Miklos Szeredi wrote:
> > Not all variants of the waitqueue interface require irqs to be
> > disabled, and fuse has nothing whatsoever to do with irqs, so there's
> > no sane reason to disable them.
> >
> > Also, AFAICS, the fuse device does not support asynchronous IO. I
> > just don't get what this is about...
>
> Hi Miklos,
>
> Could this be what happens?
>
> aio_poll() calls vfs_poll()
> vfs_poll() calls fuse_dev_poll()
> fuse_dev_poll() calls poll_wait(file, &fiq->waitq, wait)
> poll_wait() calls aio_poll_queue_proc(file, &fiq->waitq, wait)
> aio_poll_queue_proc() stores &fiq->waitq in pt->iocb->poll.head
> aio_poll() calls spin_lock_irq(&ctx->ctx_lock)
> aio_poll() calls spin_lock(&req->head->lock) (req == &pt->iocb->poll).
>
> I think the lockdep complaint is about the FUSE fiq->waitq lock not being
> IRQ-safe and about aio_poll() creating a dependency between an IRQ-safe lock
> (ctx->ctx_lock) and a lock that is not IRQ-safe (fiq->waitq).
Yep. And is AIO doing anything in irq context? If so, why?
Would it make sense to just disable AIO on file descriptors where it makes zero
sense?
E.g. this patch:
Thanks,
Miklos
---
diff --git a/fs/aio.c b/fs/aio.c
index b906ff70c90f..6f5edd28a83c 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -1435,6 +1435,9 @@ static int aio_prep_rw(struct kiocb *req, const struct iocb *iocb)
req->ki_filp = fget(iocb->aio_fildes);
if (unlikely(!req->ki_filp))
return -EBADF;
+ ret = -EINVAL;
+ if (req->ki_filp->f_mode & FMODE_NOAIO)
+ goto out_fput;
req->ki_complete = aio_complete_rw;
req->ki_pos = iocb->aio_offset;
req->ki_flags = iocb_flags(req->ki_filp);
@@ -1607,7 +1610,8 @@ static int aio_fsync(struct fsync_iocb *req, const struct iocb *iocb,
req->file = fget(iocb->aio_fildes);
if (unlikely(!req->file))
return -EBADF;
- if (unlikely(!req->file->f_op->fsync)) {
+ if (unlikely(req->file->f_mode & FMODE_NOAIO) ||
+ unlikely(!req->file->f_op->fsync)) {
fput(req->file);
return -EINVAL;
}
@@ -1755,6 +1759,9 @@ static ssize_t aio_poll(struct aio_kiocb *aiocb, const struct iocb *iocb)
apt.iocb = aiocb;
apt.error = -EINVAL; /* same as no support for IOCB_CMD_POLL */
+ if (req->file->f_mode & FMODE_NOAIO)
+ goto out;
+
/* initialized the list so that we can do list_empty checks */
INIT_LIST_HEAD(&req->wait.entry);
init_waitqueue_func_entry(&req->wait, aio_poll_wake);
diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
index a5e516a40e7a..c00aa39ec7f0 100644
--- a/fs/fuse/dev.c
+++ b/fs/fuse/dev.c
@@ -1397,6 +1397,7 @@ static int fuse_dev_open(struct inode *inode, struct file *file)
* keep track of whether the file has been mounted already.
*/
file->private_data = NULL;
+ file->f_mode |= FMODE_NOAIO;
return 0;
}
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 29d8e2cfed0e..7cda153baad3 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -162,6 +162,9 @@ typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
/* File does not contribute to nr_files count */
#define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
+/* File does not support AIO ops */
+#define FMODE_NOAIO ((__force fmode_t)0x40000000)
+
/*
* Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
* that indicates that they should check the contents of the iovec are
next prev parent reply other threads:[~2019-02-06 8:36 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-02-04 17:45 [PATCH] aio: Fix locking in aio_poll() Bart Van Assche
2019-02-04 17:49 ` Christoph Hellwig
2019-02-05 8:12 ` Miklos Szeredi
2019-02-06 0:53 ` Bart Van Assche
2019-02-06 8:36 ` Miklos Szeredi [this message]
2019-02-06 13:47 ` Christoph Hellwig
2019-02-06 14:31 ` Miklos Szeredi
2019-02-06 13:39 ` Christoph Hellwig
2019-02-09 0:59 Bart Van Assche
2019-02-12 7:56 ` Christoph Hellwig
2019-02-21 22:28 ` Bart Van Assche
2019-02-22 3:17 ` Al Viro
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=20190206083626.GA19821@veci.piliscsaba.redhat.com \
--to=miklos@szeredi.hu \
--cc=avi@scylladb.com \
--cc=bvanassche@acm.org \
--cc=edumazet@google.com \
--cc=hch@lst.de \
--cc=linux-aio@kvack.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=stable@vger.kernel.org \
--cc=syzkaller@googlegroups.com \
--cc=viro@zeniv.linux.org.uk \
/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;
as well as URLs for NNTP newsgroup(s).