linux-kernel - Re: Subject: [PATCH] net/ipv6: Use GFP

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

Message-ID: <Pine.LNX.4.64.1005302303180.19253@ask.diku.dk>
Date:	Sun, 30 May 2010 23:09:48 +0200 (CEST)
From:	Julia Lawall <julia@...u.dk>
To:	Eric Dumazet <eric.dumazet@...il.com>
Cc:	"David S. Miller" <davem@...emloft.net>,
	Alexey Kuznetsov <kuznet@....inr.ac.ru>,
	"Pekka Savola (ipv6)" <pekkas@...core.fi>,
	James Morris <jmorris@...ei.org>,
	Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
	Patrick McHardy <kaber@...sh.net>, netdev@...r.kernel.org,
	linux-kernel@...r.kernel.org, kernel-janitors@...r.kernel.org
Subject: Re: Subject: [PATCH] net/ipv6: Use GFP_ATOMIC when a lock is held

On Sun, 30 May 2010, Eric Dumazet wrote:

> Le dimanche 30 mai 2010 à 22:50 +0200, Julia Lawall a écrit :
> 
> > I think the proposed patch does not work, because the for loop overwrites 
> > p.  That use of p looks like it is completely local to the for loop, so 
> > perhaps a new variable p1 could be added to be used there?
> 
> Please do so.
> 
> I just wanted to tell you changing GFP_KERNEL to GFP_ATOMIC is not an
> appropriate way to solve this kind of problems. My patch was to get an
> idea, not a full and tested patch :)

Looking at it again, there is still a problem, because in the original 
code, the loop:

        for (p = t->prl; p; p = p->next) {
                if (p->addr == a->addr) {
                        if (chg) {
                                p->flags = a->flags;
                                goto out;
                        }
                        err = -EEXIST;
                        goto out;
                }
        }

could exit with success without the kzalloc ever being called.  If the 
kzalloc is moved up, it could fail and then it returns immediately without 
executing the loop.  A solution could be to leave the NULL test on p where 
it is, and only move up the kzalloc.  Or perhaps the change in behavior 
doesn't matter?

julia