linux-kernel - Re: [PATCH] reduce large do_mount stack usage with noinlines

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <20080208172337.GE27894@ZenIV.linux.org.uk>
Date:	Fri, 8 Feb 2008 17:23:37 +0000
From:	Al Viro <viro@...IV.linux.org.uk>
To:	Eric Sandeen <sandeen@...hat.com>
Cc:	Andi Kleen <andi@...stfloor.org>,
	Andrew Morton <akpm@...ux-foundation.org>,
	linux-kernel@...r.kernel.org, rth@...hat.com
Subject: Re: [PATCH] reduce large do_mount stack usage with noinlines

On Fri, Feb 08, 2008 at 10:54:09AM -0600, Eric Sandeen wrote:

> On gcc 4.1.2 and 4.3 (fedora flavors) I don't see it re-used in
> do_mount, though... *shrug*

Frankly, a wrapper for path_lookup() that would take struct path *,
refuse to do LOOKUP_PARENT (i.e. guaranteed to have nothing stored
in nameidata other than ->mnt and ->dentry) and copied result of
lookup into passed struct path * would help a lot more.

Taking it to fs/namei.c would prevent inlining just fine and we'd
be left with pair of pointers in caller's stack frame instead of
full struct nameidata.  All users in fs/namespace.c can be trivially
converted to that and AFAICS it would save a lot more than we would
get from making everything in there not inlined.

I'll do that in fs/namei.c sanitizing series (opens as soon as ro-bind
stuff goes in).
--
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/