linux-kernel - Re: Pending splice(file -> FIFO) excludes all other FIFO operations forever (was: ... always blocks read(FIFO), regardless of O

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 10 Jul 2023 00:33:06 +0200
From:   Ahelenia Ziemiańska 
        <nabijaczleweli@...ijaczleweli.xyz>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Christian Brauner <brauner@...nel.org>,
        Jens Axboe <axboe@...nel.dk>,
        David Howells <dhowells@...hat.com>,
        Alexander Viro <viro@...iv.linux.org.uk>,
        linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: Pending splice(file -> FIFO) excludes all other FIFO operations
 forever (was: ... always blocks read(FIFO), regardless of O_NONBLOCK on read
 side?)

On Sun, Jul 09, 2023 at 03:03:22AM +0200, Ahelenia Ziemiańska wrote:
> On Sat, Jul 08, 2023 at 01:06:56PM -0700, Linus Torvalds wrote:
> > I guess combined with something like
> > 
> >         if (!(in->f_mode & FMODE_NOWAIT))
> >                 return -EINVAL;
> > 
> > it might all work.
> Yes, that makes the splice instantly -EINVAL for ttys, but doesn't
> affect the socketpair() case above, which still blocks forever.

This also triggers for regular file -> pipe splices,
which is probably a no-go.


Attaching a summary diff that does all I said in the previous mail.

filemap_get_pages() does use and inspect IOCB_NOWAIT if set in
filemap_splice_read(), but it appears to not really make much sense,
inasmuch it returns EAGAIN for the first splice() from a
blockdev and then instantly return with data on the next go-around.

This doesn't really make much sense ‒ and open(2) says blockdevs
don't have O_NONBLOCK semantics, so I'm assuming this is not supposed
to be exposed to userspace ‒ so I'm not setting it in the diff.


I've tested this for:
  * tty: -EINVAL
  * socketpair(AF_UNIX, SOCK_STREAM): works as expected
	$ wc -c fifo &
	[1] 261
	$ ./af_unix ./s > fifo
	5       Success
	6454    Resource temporarily unavailable
	5       Success
	6445    Resource temporarily unavailable
	0       Success
	10 fifo
  * socket(AF_INET, SOCK_STREAM, 0): works as expected
	$ wc fifo &
	[1] 249
	$ ./tcp ./s > fifo
	23099   Resource temporarily unavailable
	7       Success
	2099    Resource temporarily unavailable
	4       Success
	1751    Resource temporarily unavailable
	3       Success
	21655   Resource temporarily unavailable
	95      Success
	19589   Resource temporarily unavailable
	0       Success
		  4      15     109 fifo
    corresponding to
	host$ nc 127.0.0.1 14640
	żupan
	asd
	ad
	asdda dasj aiojd askdl; jasiopdij as[pkdo[p askd9p ias90dk aso[pjd 890pasid90[ jaskl;dj il;asd
	^C
  * blockdev (/dev/vda): as expected (with filemap_splice_read() unchanged), copies it all
  * regular file: -EINVAL(!)

Splicers still sleep if the pipe's full, of course,
unless SPLICE_F_NONBLOCK.

Test drivers attached as well.

View attachment "nowait.diff" of type "text/x-diff" (7196 bytes)

View attachment "s.c" of type "text/x-csrc" (485 bytes)

View attachment "tcp.c" of type "text/x-csrc" (507 bytes)

View attachment "af_unix.c" of type "text/x-csrc" (394 bytes)

Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)