linux-kernel - why does page_cache_pipe_buf_confirm() need to check page->mapping?

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>] [day] [month] [year] [list]

Date:	Thu, 19 May 2016 10:05:43 +0200
From:	Miklos Szeredi <miklos@...redi.hu>
To:	Jens Axboe <axboe@...nel.dk>
Cc:	Abhijith Das <adas@...hat.com>,
	Linux-Fsdevel <linux-fsdevel@...r.kernel.org>,
	linux-mm@...ck.org,
	Kernel Mailing List <linux-kernel@...r.kernel.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>
Subject: why does page_cache_pipe_buf_confirm() need to check page->mapping?

Hi Jens,

I haven't done a great deal of research into this, but checking
page->mapping in page_cache_pipe_buf_confirm() might be bogus.

If the page is truncated *after* being spliced into the pipe, why on
earth does the buffer become invalid?

This looks to be a problem for filesystems that invalidate pages
(because the the data is possibly stale) and the pipe read returns
-ENODATA even though the data is there, it's just possibly different
from what was spliced into the pipe.  But I don't think that's a
reason for throwing away that buffer and definitely not a reason to
return an error.

Thanks,
Miklos