linux-kernel - Re: sparse -Wptr-subtraction-blows: still needed?

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: <20070502000240.GI4095@ftp.linux.org.uk>
Date:	Wed, 2 May 2007 01:02:40 +0100
From:	Al Viro <viro@....linux.org.uk>
To:	Linus Torvalds <torvalds@...ux-foundation.org>
Cc:	Josh Triplett <josh@...edesktop.org>, linux-sparse@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: sparse -Wptr-subtraction-blows: still needed?

On Tue, May 01, 2007 at 02:43:30PM -0700, Linus Torvalds wrote:
> 
> 
> On Tue, 1 May 2007, Josh Triplett wrote:
> > 
> > Does this still apply?  Do current versions of GCC still have this problem?
> > If not, can the option and warning go away?
> 
> Even if current versions of gcc don't triple the build time (and for the 
> kernel, I suspect it doesn't, because we've tried to clean up our header 
> files), the generated _code_ will invariably suck.
> 
> So I'd not want to remove the warning.

Note that code *will* blow: in effect, when we have a*2^b as object size,
we have to choose between
	* generic division by a*2^b (dog-slow pretty much on anything)
	* multiplication by such c that a*c == 1 (mod 2^word_size-b), then
shift down by b.
	* shift down by b, then multiplication by such c that a*c == 1
(mod 2^word_size).
And c in the above won't be small or pretty, so doing multiplication as
series of additions won't be short.

FWIW, I've seen very nasty cases (in userland code) where hot path got blown
by factor of 5 or so in size due to that; basically, it started with
#define INDEX(p) ((p)-array)
and that sucker got used a lot in other macros, so it wasn't immediately
obvious what had been going on.  So yes, we do want to be able to see
such cases.
-
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/