linux-kernel - Re: [PATCH] Syscalls: reboot: Add options to the reboot syscall to remount filesystems ro

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]

Message-ID: <AANLkTi=UuVxtvLqWq=8gTGQyoNKmJwvs3Kr=xSWHbxtx@mail.gmail.com>
Date:	Thu, 3 Mar 2011 18:33:36 -0800
From:	Linus Torvalds <torvalds@...ux-foundation.org>
To:	Ken Sumrall <ksumrall@...roid.com>
Cc:	Mark Lord <kernel@...savvy.com>, linux-kernel@...r.kernel.org,
	Alexander Viro <viro@...iv.linux.org.uk>,
	Christoph Hellwig <hch@....de>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Jan Kara <jack@...e.cz>, Jens Axboe <axboe@...nel.dk>,
	Matthew Wilcox <matthew@....cx>,
	Eric Paris <eparis@...hat.com>,
	Dave Young <hidave.darkstar@...il.com>,
	Jiri Slaby <jslaby@...e.cz>, James Morris <jmorris@...ei.org>,
	linux-fsdevel@...r.kernel.org
Subject: Re: [PATCH] Syscalls: reboot: Add options to the reboot syscall to
 remount filesystems ro

On Thu, Mar 3, 2011 at 6:00 PM, Ken Sumrall <ksumrall@...roid.com> wrote:
>
> Writing a single byte to /proc/sysrq-trigger is an asynchronous
> operation, with no obvious way to be informed that it has completed
> the remount.

Right you are. That's something of a misfeature, but it comes from the
way sysrq works: obviously the "real" sysrq thing is about keyboard
input, so all the sysrq stuff has to be async.

The fact that that async nature then ends up also affecting the
/proc/sysrq-trigger case (which _could_ be synchronous) is a bit sad
in this case.

That said, I obviously think that just doing the read-only mount
yourself is the right thing to do regardless, and the sysrq thing
would have been just a cute/ugly hack if it had worked.

> I'm heading down the path of reading /proc/mounts and remounting all
> read-wirte filesystems backed by a block device as read-only.

So just as a practical matter: while it's quite possible that the nice
seq_printf() model of /proc/mounts means that it should work correctly
even if you read each line individually and then re-mount while
holding the file open, I would suggest that you read the whole file
into a buffer before you then start changing the mounts.

Otherwise, _if_ we ever were to actually move the filesystem on our
internal list of mounts when we re-mount it, you might otherwise end
up seeign the same filesystem multiple times (or the reverse - missing
some filesystem).

Basically, I'm saying that you should try to avoid changing mount
information in between read() calls to /proc/mounts. It might cause
confusion.

                                        Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/