netdev - Re: [syzbot] WARNING in u32

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 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]

Date:   Sun, 25 Sep 2022 19:39:23 -0700
From:   Kees Cook <keescook@...omium.org>
To:     Eric Dumazet <edumazet@...gle.com>
Cc:     Jamal Hadi Salim <jhs@...atatu.com>,
        syzbot <syzbot+a2c4601efc75848ba321@...kaller.appspotmail.com>,
        David Miller <davem@...emloft.net>,
        Jiri Pirko <jiri@...nulli.us>,
        Jakub Kicinski <kuba@...nel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        netdev <netdev@...r.kernel.org>, Paolo Abeni <pabeni@...hat.com>,
        syzkaller-bugs <syzkaller-bugs@...glegroups.com>,
        Cong Wang <xiyou.wangcong@...il.com>
Subject: Re: [syzbot] WARNING in u32_change

On Sun, Sep 25, 2022 at 10:34:37AM -0700, Eric Dumazet wrote:
> Sure, please look at:
> 
> commit 54d9469bc515dc5fcbc20eecbe19cea868b70d68
> Author: Kees Cook <keescook@...omium.org>
> Date:   Thu Jun 24 15:39:26 2021 -0700
> 
>     fortify: Add run-time WARN for cross-field memcpy()
> [...]
> Here, we might switch to unsafe_memcpy() instead of memcpy()

I would tend to agree. Something like:

diff --git a/net/sched/cls_u32.c b/net/sched/cls_u32.c
index 4d27300c287c..21e0e6206ecc 100644
--- a/net/sched/cls_u32.c
+++ b/net/sched/cls_u32.c
@@ -1040,7 +1040,9 @@ static int u32_change(struct net *net, struct sk_buff *in_skb,
 	}
 #endif
 
-	memcpy(&n->sel, s, sel_size);
+	unsafe_memcpy(&n->sel, s, sel_size,
+		      /* A composite flex-array structure destination,
+		       * which was correctly sized and allocated above. */);
 	RCU_INIT_POINTER(n->ht_up, ht);
 	n->handle = handle;
 	n->fshift = s->hmask ? ffs(ntohl(s->hmask)) - 1 : 0;

This alloc/partial-copy pattern is relatively common in the kernel, so
I've been considering adding a helper for it. It'd be like kmemdup(),
but more like kmemdup_offset(), which only the object from a certainly
point is copied.

-- 
Kees Cook