From: Christoph Hellwig <hch@lst.de>
To: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>,
Linus Torvalds <torvalds@linux-foundation.org>,
Stephen Rothwell <sfr@canb.auug.org.au>,
Luis Chamberlain <mcgrof@kernel.org>,
Matthew Wilcox <willy@infradead.org>,
Kees Cook <keescook@chromium.org>,
Iurii Zaikin <yzaikin@google.com>,
linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH 22/23] fs: default to generic_file_splice_read for files having ->read_iter
Date: Thu, 30 Jul 2020 17:20:46 +0200 [thread overview]
Message-ID: <20200730152046.GA21192@lst.de> (raw)
In-Reply-To: <20200730150826.GA1236603@ZenIV.linux.org.uk>
On Thu, Jul 30, 2020 at 04:08:26PM +0100, Al Viro wrote:
> > I think we need to fix that in the instances, as we really expect
> > ->splice_read to just work instead of the caller knowing what could
> > work and what might not.
>
> Er... generic_file_splice_read() is a library helper; the decision to use
> is up to the filesystem/driver/protocol in question, and so's making sure
> it's not used with ->read_iter() that isn't fit for it.
Yes, but.. The problem is that while right now generic_file_splice_read
is the only user of ITER_PIPE there is absolutely not guarantee that
it remains the only user. Having ->read_iter instances lingering that
can't deal with it is at best a mine field waiting for victims.
Fortunately I think the fix is pretty easy - remove the special pipe
zero copy optimization from copy_page_to_iter, and just have the
callers actually want it because they have pagecache or similar
refcountable pages use it explicitly for the ITER_PIPE case. That gives
us a safe default with an opt-in into the optimized variant. I'm
currently auditing all the users of for how it is used and that looks
pretty promising.
> Note that we *do* have instances where we have different ->splice_read()
> (sometimes using generic_file_splice_read(), sometimes not) even though
> ->read_iter() is there.
>
> Your patch ignores those (thankfully), but commit message is rather
> misleading - it strongly implies that generic_file_splice_read() is
> *always* the right thing when ->read_iter() is there, not just that
> in such cases it makes a better fallback than default_file_splice_read().
I don't think it always is right. Not without a major audit and more
work at least.
next prev parent reply other threads:[~2020-07-30 15:20 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-07 17:47 stop using ->read and ->write for kernel access v3 Christoph Hellwig
2020-07-07 17:47 ` [PATCH 01/23] cachefiles: switch to kernel_write Christoph Hellwig
2020-07-07 17:47 ` [PATCH 02/23] autofs: " Christoph Hellwig
2020-07-07 17:47 ` [PATCH 03/23] bpfilter: " Christoph Hellwig
2020-07-07 17:47 ` [PATCH 04/23] fs: unexport __kernel_write Christoph Hellwig
2020-07-07 17:47 ` [PATCH 05/23] fs: check FMODE_WRITE in __kernel_write Christoph Hellwig
2020-07-07 17:47 ` [PATCH 06/23] fs: implement kernel_write using __kernel_write Christoph Hellwig
2020-07-07 17:47 ` [PATCH 07/23] fs: remove __vfs_write Christoph Hellwig
2020-07-07 17:47 ` [PATCH 08/23] fs: don't change the address limit for ->write_iter in __kernel_write Christoph Hellwig
2020-07-29 20:50 ` Al Viro
2020-07-30 7:02 ` Christoph Hellwig
2020-07-07 17:47 ` [PATCH 09/23] fs: add a __kernel_read helper Christoph Hellwig
2020-07-07 17:47 ` [PATCH 10/23] integrity/ima: switch to using __kernel_read Christoph Hellwig
2020-07-07 17:47 ` [PATCH 11/23] fs: implement kernel_read " Christoph Hellwig
2020-07-07 17:47 ` [PATCH 12/23] fs: remove __vfs_read Christoph Hellwig
2020-07-07 17:47 ` [PATCH 13/23] fs: don't change the address limit for ->read_iter in __kernel_read Christoph Hellwig
2020-07-07 17:47 ` [PATCH 14/23] seq_file: add seq_read_iter Christoph Hellwig
2020-07-07 17:47 ` [PATCH 16/23] proc: remove a level of indentation in proc_get_inode Christoph Hellwig
2020-07-07 17:47 ` [PATCH 17/23] proc: cleanup the compat vs no compat file ops Christoph Hellwig
2020-07-07 17:47 ` [PATCH 18/23] proc: add a read_iter method to proc proc_ops Christoph Hellwig
2020-07-07 17:47 ` [PATCH 19/23] proc: switch over direct seq_read method calls to seq_read_iter Christoph Hellwig
2020-07-07 17:47 ` [PATCH 20/23] sysctl: Convert to iter interfaces Christoph Hellwig
2020-07-07 17:47 ` [PATCH 21/23] fs: don't allow kernel reads and writes without iter ops Christoph Hellwig
2020-07-07 17:48 ` [PATCH 22/23] fs: default to generic_file_splice_read for files having ->read_iter Christoph Hellwig
2020-07-30 0:05 ` Al Viro
2020-07-30 7:03 ` Christoph Hellwig
2020-07-30 15:08 ` Al Viro
2020-07-30 15:20 ` Christoph Hellwig [this message]
2020-07-30 16:17 ` Al Viro
2020-07-30 16:22 ` Al Viro
2020-07-30 16:31 ` Christoph Hellwig
2020-07-07 17:48 ` [PATCH 23/23] fs: don't allow splice read/write without explicit ops Christoph Hellwig
2020-07-07 20:24 ` stop using ->read and ->write for kernel access v3 Linus Torvalds
2020-07-08 6:07 ` Christoph Hellwig
2020-07-07 23:03 ` Stephen Rothwell
[not found] ` <20200707174801.4162712-16-hch@lst.de>
2020-07-10 12:55 ` [PATCH 15/23] seq_file: switch over direct seq_read method calls to seq_read_iter Jon Hunter
2020-07-10 12:58 ` Christoph Hellwig
2020-07-11 6:48 ` Christoph Hellwig
2020-07-11 11:47 ` Jon Hunter
2020-07-17 21:09 ` Thomas Gleixner
2020-07-20 9:33 ` Christoph Hellwig
2020-07-29 20:59 ` Al Viro
2020-07-30 7:10 ` Thomas Gleixner
-- strict thread matches above, loose matches on Subject: below --
2020-07-01 20:09 [RFC] stop using ->read and ->write for kernel access v3 Christoph Hellwig
2020-07-01 20:09 ` [PATCH 22/23] fs: default to generic_file_splice_read for files having ->read_iter Christoph Hellwig
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=20200730152046.GA21192@lst.de \
--to=hch@lst.de \
--cc=keescook@chromium.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mcgrof@kernel.org \
--cc=sfr@canb.auug.org.au \
--cc=torvalds@linux-foundation.org \
--cc=viro@zeniv.linux.org.uk \
--cc=willy@infradead.org \
--cc=yzaikin@google.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;
as well as URLs for NNTP newsgroup(s).