From: Al Viro <viro@zeniv.linux.org.uk>
To: Nathan Chancellor <natechancellor@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
Christoph Hellwig <hch@lst.de>,
Greg KH <gregkh@linuxfoundation.org>,
Alexey Dobriyan <adobriyan@gmail.com>,
linux-fsdevel <linux-fsdevel@vger.kernel.org>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
kys@microsoft.com, haiyangz@microsoft.com,
sthemmin@microsoft.com, wei.liu@kernel.org,
linux-hyperv@vger.kernel.org
Subject: Re: [PATCH 1/6] seq_file: add seq_read_iter
Date: Sun, 15 Nov 2020 23:38:14 +0000 [thread overview]
Message-ID: <20201115233814.GT3576660@ZenIV.linux.org.uk> (raw)
In-Reply-To: <20201115214125.GA317@Ryzen-9-3900X.localdomain>
On Sun, Nov 15, 2020 at 02:41:25PM -0700, Nathan Chancellor wrote:
> Hi Al,
>
> Apologies for the delay.
>
> On Sun, Nov 15, 2020 at 03:53:55PM +0000, Al Viro wrote:
> > On Sat, Nov 14, 2020 at 08:50:00PM +0000, Al Viro wrote:
> >
> > OK, I think I understand what's going on. Could you check if
> > reverting the variant in -next and applying the following instead
> > fixes what you are seeing?
>
> The below diff on top of d4d50710a8b46082224376ef119a4dbb75b25c56 does
> not fix my issue unfortunately.
OK... Now that I have a reproducer[1], I think I've sorted it out.
And yes, it had been too subtle for its own good ;-/
[1] I still wonder what the hell in the userland has come up with the
idea of reading through a file with readv(), each time with 2-element
iovec array, the first element covering 0 bytes and the second one - 1024.
AFAICS, nothing is systemd git appears to be _that_ weird... Makes for
a useful testcase, though...
Anyway, could you test this replacement?
diff --git a/fs/seq_file.c b/fs/seq_file.c
index 3b20e21604e7..c0dfe2861b35 100644
--- a/fs/seq_file.c
+++ b/fs/seq_file.c
@@ -168,12 +168,14 @@ EXPORT_SYMBOL(seq_read);
ssize_t seq_read_iter(struct kiocb *iocb, struct iov_iter *iter)
{
struct seq_file *m = iocb->ki_filp->private_data;
- size_t size = iov_iter_count(iter);
size_t copied = 0;
size_t n;
void *p;
int err = 0;
+ if (!iov_iter_count(iter))
+ return 0;
+
mutex_lock(&m->lock);
/*
@@ -208,34 +210,32 @@ ssize_t seq_read_iter(struct kiocb *iocb, struct iov_iter *iter)
}
/* if not empty - flush it first */
if (m->count) {
- n = min(m->count, size);
- if (copy_to_iter(m->buf + m->from, n, iter) != n)
- goto Efault;
+ n = copy_to_iter(m->buf + m->from, m->count, iter);
m->count -= n;
m->from += n;
- size -= n;
copied += n;
- if (!size)
+ if (m->count) // hadn't managed to copy everything
goto Done;
}
- /* we need at least one record in buffer */
+ /* we need at least one non-empty record in the buffer */
m->from = 0;
p = m->op->start(m, &m->index);
while (1) {
err = PTR_ERR(p);
- if (!p || IS_ERR(p))
+ if (!p || IS_ERR(p)) // EOF or an error
break;
err = m->op->show(m, p);
- if (err < 0)
+ if (err < 0) // hard error
break;
- if (unlikely(err))
+ if (unlikely(err)) // ->show() says "skip it"
m->count = 0;
- if (unlikely(!m->count)) {
+ if (unlikely(!m->count)) { // empty record
p = m->op->next(m, p, &m->index);
continue;
}
- if (m->count < m->size)
+ if (!seq_has_overflowed(m)) // got it
goto Fill;
+ // need a bigger buffer
m->op->stop(m, p);
kvfree(m->buf);
m->count = 0;
@@ -244,11 +244,14 @@ ssize_t seq_read_iter(struct kiocb *iocb, struct iov_iter *iter)
goto Enomem;
p = m->op->start(m, &m->index);
}
+ // EOF or an error
m->op->stop(m, p);
m->count = 0;
goto Done;
Fill:
- /* they want more? let's try to get some more */
+ // one non-empty record is in the buffer; if they want more,
+ // try to fit more in, but in any case we need to advance
+ // the iterator at least once.
while (1) {
size_t offs = m->count;
loff_t pos = m->index;
@@ -259,11 +262,9 @@ ssize_t seq_read_iter(struct kiocb *iocb, struct iov_iter *iter)
m->op->next);
m->index++;
}
- if (!p || IS_ERR(p)) {
- err = PTR_ERR(p);
+ if (!p || IS_ERR(p)) // no next record for us
break;
- }
- if (m->count >= size)
+ if (m->count >= iov_iter_count(iter))
break;
err = m->op->show(m, p);
if (seq_has_overflowed(m) || err) {
@@ -273,16 +274,14 @@ ssize_t seq_read_iter(struct kiocb *iocb, struct iov_iter *iter)
}
}
m->op->stop(m, p);
- n = min(m->count, size);
- if (copy_to_iter(m->buf, n, iter) != n)
- goto Efault;
+ n = copy_to_iter(m->buf, m->count, iter);
copied += n;
m->count -= n;
m->from = n;
Done:
- if (!copied)
- copied = err;
- else {
+ if (unlikely(!copied)) {
+ copied = m->count ? -EFAULT : err;
+ } else {
iocb->ki_pos += copied;
m->read_pos += copied;
}
@@ -291,9 +290,6 @@ ssize_t seq_read_iter(struct kiocb *iocb, struct iov_iter *iter)
Enomem:
err = -ENOMEM;
goto Done;
-Efault:
- err = -EFAULT;
- goto Done;
}
EXPORT_SYMBOL(seq_read_iter);
next prev parent reply other threads:[~2020-11-15 23:38 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-11-04 8:27 support splice reads on seq_file based procfs files v2 Christoph Hellwig
2020-11-04 8:27 ` [PATCH 1/6] seq_file: add seq_read_iter Christoph Hellwig
2020-11-10 21:32 ` Al Viro
2020-11-10 21:35 ` Al Viro
2020-11-10 23:20 ` Al Viro
2020-11-11 7:55 ` Christoph Hellwig
2020-11-11 17:54 ` Linus Torvalds
2020-11-11 21:52 ` Al Viro
2020-11-11 22:21 ` Al Viro
2020-11-11 22:27 ` Linus Torvalds
2020-11-11 23:00 ` Al Viro
2020-11-13 23:54 ` Nathan Chancellor
2020-11-14 1:17 ` Al Viro
2020-11-14 3:01 ` Nathan Chancellor
2020-11-14 3:54 ` Al Viro
2020-11-14 4:14 ` Nathan Chancellor
2020-11-14 5:50 ` Al Viro
2020-11-14 6:19 ` Nathan Chancellor
2020-11-14 7:00 ` Al Viro
2020-11-14 20:50 ` Al Viro
2020-11-15 15:53 ` Al Viro
2020-11-15 16:56 ` Linus Torvalds
2020-11-15 21:41 ` Nathan Chancellor
2020-11-15 23:38 ` Al Viro [this message]
2020-11-15 23:51 ` Nathan Chancellor
2020-11-16 0:25 ` Al Viro
2020-11-16 0:34 ` Nathan Chancellor
2020-11-16 3:29 ` Al Viro
2020-11-27 16:29 ` Christoph Hellwig
2020-12-08 16:35 ` Christoph Hellwig
2020-12-08 18:34 ` Linus Torvalds
2020-12-08 19:49 ` Al Viro
2020-12-08 20:25 ` Linus Torvalds
2020-12-08 20:53 ` Al Viro
2020-12-08 21:01 ` Linus Torvalds
2020-12-08 19:49 ` Greg KH
2020-11-14 21:44 ` Al Viro
2020-11-04 8:27 ` [PATCH 2/6] proc: wire up generic_file_splice_read for iter ops Christoph Hellwig
2020-11-04 8:27 ` [PATCH 3/6] proc/cpuinfo: switch to ->read_iter Christoph Hellwig
2020-11-04 8:27 ` [PATCH 4/6] proc/stat: " Christoph Hellwig
2020-11-04 8:27 ` [PATCH 5/6] proc "single files": " Christoph Hellwig
2020-11-04 8:27 ` [PATCH 6/6] proc "seq " Christoph Hellwig
2020-11-04 17:53 ` support splice reads on seq_file based procfs files v2 Linus Torvalds
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=20201115233814.GT3576660@ZenIV.linux.org.uk \
--to=viro@zeniv.linux.org.uk \
--cc=adobriyan@gmail.com \
--cc=gregkh@linuxfoundation.org \
--cc=haiyangz@microsoft.com \
--cc=hch@lst.de \
--cc=kys@microsoft.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-hyperv@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=natechancellor@gmail.com \
--cc=sthemmin@microsoft.com \
--cc=torvalds@linux-foundation.org \
--cc=wei.liu@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;
as well as URLs for NNTP newsgroup(s).