The Linux Kernel Mailing List
 help / color / mirror / Atom feed
From: "Jörn Engel" <joern@wohnheim.fh-wedel.de>
To: Gunnar Ritter <Gunnar.Ritter@pluto.uni-freiburg.de>
Cc: linux-kernel@vger.kernel.org, Andrew Morton <akpm@osdl.org>,
	Steve French <smfltc@us.ibm.com>
Subject: Re: [PATCH 1/3] copyfile: generic_sendpage
Date: Tue, 7 Sep 2004 13:45:36 +0200	[thread overview]
Message-ID: <20040907114536.GA26630@wohnheim.fh-wedel.de> (raw)
In-Reply-To: <20040907110913.GA25802@wohnheim.fh-wedel.de>

On Tue, 7 September 2004 13:09:13 +0200, Jörn Engel wrote:
> 
> I did the loop inside the kernel, so the syscall overhead is less of
> an issue.  4k is a safe bet for _really_ slow devices and if people
> want to increase it, hey, it's just a single constant to touch.

And a stupid bug I missed.  Call sendfile with count>filesize and the
loop doesn't terminate.  Fixed patch below.

Jörn

-- 
Premature optimization is the root of all evil.
-- Donald Knuth


Linus and Andrew are rightfully concerned about local DoS via a large
file->file sendfile().  This patch turns large sendfile() calls into a
loop of 4k chunks.  After each chunk, it adds a cond_resched for
interactivity and a signal check to allow aborts etc. after the user
found out what a bad idea this may be.

Signed-off-by: Jörn Engel <joern@wohnheim.fh-wedel.de>
---

 read_write.c |   31 ++++++++++++++++++++++++++++++-
 1 files changed, 30 insertions(+), 1 deletion(-)


--- linux-2.6.8cow/fs/read_write.c~sendfile_loop	2004-09-05 12:06:39.000000000 +0200
+++ linux-2.6.8cow/fs/read_write.c	2004-09-07 13:27:30.000000000 +0200
@@ -561,6 +561,35 @@
 	return ret;
 }
 
+/**
+ * sendfile() of a 2GB file over usb1-attached hard drives can take a moment.
+ * This little loop is supposed to stop now and then to check for signals,
+ * reschedule and generally play nice with others.
+ */
+ssize_t inline __vfs_sendfile(struct file *in_file, loff_t *ppos, size_t count,
+		read_actor_t actor, struct file *out_file)
+{
+	ssize_t done = 0, ret;
+	while (count) {
+		size_t n = min(count, (size_t)4096);
+		ret = in_file->f_op->sendfile(in_file, ppos, n, actor,out_file);
+		if (ret < 0) {
+			if (done)
+				return done;
+			else
+				return ret;
+		}
+
+		done += ret;
+		count -= ret;
+
+		if ((ret == 0) || signal_pending(current))
+			break;
+		cond_resched();
+	}
+	return done;
+}
+
 ssize_t vfs_sendfile(struct file *out_file, struct file *in_file, loff_t *ppos,
 		     size_t count, loff_t max)
 {
@@ -608,7 +637,7 @@
 		count = max - pos;
 	}
 
-	ret = in_file->f_op->sendfile(in_file, ppos, count, file_send_actor, out_file);
+	ret = __vfs_sendfile(in_file, ppos, count, file_send_actor, out_file);
 
 	if (*ppos > max)
 		return -EOVERFLOW;

  reply	other threads:[~2004-09-07 11:46 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-09-04 16:57 [PATCH 1/3] copyfile: generic_sendpage Jörn Engel
2004-09-04 16:59 ` [PATCH 2/3] copyfile: sendfile Jörn Engel
2004-09-04 17:04   ` [PATCH 3/3] copyfile: copyfile Jörn Engel
2004-09-04 17:08     ` Jörn Engel
2004-09-04 17:11   ` [PATCH 2/3] copyfile: sendfile Christoph Hellwig
2004-09-04 17:16     ` Jörn Engel
2004-09-04 17:12 ` [PATCH 1/3] copyfile: generic_sendpage Christoph Hellwig
2004-09-04 17:22   ` Jörn Engel
2004-09-04 17:25     ` Christoph Hellwig
2004-09-07  8:32       ` Jan Blunck
2004-09-04 22:39 ` Andrew Morton
2004-09-05  0:16   ` William Lee Irwin III
2004-09-05  0:23     ` Andrew Morton
2004-09-06 11:54   ` Jörn Engel
2004-09-06 12:45   ` Gunnar Ritter
2004-09-06 13:35     ` Jörn Engel
2004-09-06 14:32       ` Gunnar Ritter
2004-09-06 14:46         ` Oliver Neukum
2004-09-06 15:51           ` Gunnar Ritter
2004-09-07 11:09         ` Jörn Engel
2004-09-07 11:45           ` Jörn Engel [this message]
2004-09-07 12:03             ` Gunnar Ritter
2004-09-07 12:08               ` Jörn Engel
2004-09-07 12:32                 ` Gunnar Ritter
2004-09-07 11:48           ` Gunnar Ritter
2004-09-07 12:04             ` Jörn Engel

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=20040907114536.GA26630@wohnheim.fh-wedel.de \
    --to=joern@wohnheim.fh-wedel.de \
    --cc=Gunnar.Ritter@pluto.uni-freiburg.de \
    --cc=akpm@osdl.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=smfltc@us.ibm.com \
    /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