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: <88ba593e-97ee-44a6-9428-37b8abb1e720@nutanix.com>
Date: Thu, 26 Sep 2024 19:00:02 +0100
From: Jonathan Davies <jonathan.davies@...anix.com>
To: Eric Dumazet <edumazet@...gle.com>,
        "David S . Miller"
 <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>
Cc: David Ahern <dsahern@...nel.org>, netdev@...r.kernel.org,
        Willem de Bruijn <willemb@...gle.com>, eric.dumazet@...il.com,
        syzbot <syzkaller@...glegroups.com>
Subject: Re: [PATCH net 1/2] net: avoid potential underflow in
 qdisc_pkt_len_init() with UFO

On 24/09/2024 16:02, Eric Dumazet wrote:
> After commit 7c6d2ecbda83 ("net: be more gentle about silly gso
> requests coming from user") virtio_net_hdr_to_skb() had sanity check
> to detect malicious attempts from user space to cook a bad GSO packet.
> 
> Then commit cf9acc90c80ec ("net: virtio_net_hdr_to_skb: count
> transport header in UFO") while fixing one issue

I've tested that the issue fixed by commit cf9acc90c80ec isn't adversely 
affected by your patch, so:

Reviewed-by: Jonathan Davies <jonathan.davies@...anix.com>

>, allowed user space
> to cook a GSO packet with the following characteristic :
> 
> IPv4 SKB_GSO_UDP, gso_size=3, skb->len = 28.
> 
> When this packet arrives in qdisc_pkt_len_init(), we end up
> with hdr_len = 28 (IPv4 header + UDP header), matching skb->len
> 
> Then the following sets gso_segs to 0 :
> 
> gso_segs = DIV_ROUND_UP(skb->len - hdr_len,
>                          shinfo->gso_size);
> 
> Then later we set qdisc_skb_cb(skb)->pkt_len to back to zero :/
> 
> qdisc_skb_cb(skb)->pkt_len += (gso_segs - 1) * hdr_len;
> 
> This leads to the following crash in fq_codel [1]
> 
> qdisc_pkt_len_init() is best effort, we only want an estimation
> of the bytes sent on the wire, not crashing the kernel.
> 
> This patch is fixing this particular issue, a following one
> adds more sanity checks for another potential bug.
> 
> [1]
> [   70.724101] BUG: kernel NULL pointer dereference, address: 0000000000000000
> [   70.724561] #PF: supervisor read access in kernel mode
> [   70.724561] #PF: error_code(0x0000) - not-present page
> [   70.724561] PGD 10ac61067 P4D 10ac61067 PUD 107ee2067 PMD 0
> [   70.724561] Oops: Oops: 0000 [#1] SMP NOPTI
> [   70.724561] CPU: 11 UID: 0 PID: 2163 Comm: b358537762 Not tainted 6.11.0-virtme #991
> [   70.724561] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
> [   70.724561] RIP: 0010:fq_codel_enqueue (net/sched/sch_fq_codel.c:120 net/sched/sch_fq_codel.c:168 net/sched/sch_fq_codel.c:230) sch_fq_codel
> [ 70.724561] Code: 24 08 49 c1 e1 06 44 89 7c 24 18 45 31 ed 45 31 c0 31 ff 89 44 24 14 4c 03 8b 90 01 00 00 eb 04 39 ca 73 37 4d 8b 39 83 c7 01 <49> 8b 17 49 89 11 41 8b 57 28 45 8b 5f 34 49 c7 07 00 00 00 00 49
> All code
> ========
>     0:	24 08                	and    $0x8,%al
>     2:	49 c1 e1 06          	shl    $0x6,%r9
>     6:	44 89 7c 24 18       	mov    %r15d,0x18(%rsp)
>     b:	45 31 ed             	xor    %r13d,%r13d
>     e:	45 31 c0             	xor    %r8d,%r8d
>    11:	31 ff                	xor    %edi,%edi
>    13:	89 44 24 14          	mov    %eax,0x14(%rsp)
>    17:	4c 03 8b 90 01 00 00 	add    0x190(%rbx),%r9
>    1e:	eb 04                	jmp    0x24
>    20:	39 ca                	cmp    %ecx,%edx
>    22:	73 37                	jae    0x5b
>    24:	4d 8b 39             	mov    (%r9),%r15
>    27:	83 c7 01             	add    $0x1,%edi
>    2a:*	49 8b 17             	mov    (%r15),%rdx		<-- trapping instruction
>    2d:	49 89 11             	mov    %rdx,(%r9)
>    30:	41 8b 57 28          	mov    0x28(%r15),%edx
>    34:	45 8b 5f 34          	mov    0x34(%r15),%r11d
>    38:	49 c7 07 00 00 00 00 	movq   $0x0,(%r15)
>    3f:	49                   	rex.WB
> 
> Code starting with the faulting instruction
> ===========================================
>     0:	49 8b 17             	mov    (%r15),%rdx
>     3:	49 89 11             	mov    %rdx,(%r9)
>     6:	41 8b 57 28          	mov    0x28(%r15),%edx
>     a:	45 8b 5f 34          	mov    0x34(%r15),%r11d
>     e:	49 c7 07 00 00 00 00 	movq   $0x0,(%r15)
>    15:	49                   	rex.WB
> [   70.724561] RSP: 0018:ffff95ae85e6fb90 EFLAGS: 00000202
> [   70.724561] RAX: 0000000002000000 RBX: ffff95ae841de000 RCX: 0000000000000000
> [   70.724561] RDX: 0000000000000000 RSI: 0000000000000001 RDI: 0000000000000001
> [   70.724561] RBP: ffff95ae85e6fbf8 R08: 0000000000000000 R09: ffff95b710a30000
> [   70.724561] R10: 0000000000000000 R11: bdf289445ce31881 R12: ffff95ae85e6fc58
> [   70.724561] R13: 0000000000000000 R14: 0000000000000040 R15: 0000000000000000
> [   70.724561] FS:  000000002c5c1380(0000) GS:ffff95bd7fcc0000(0000) knlGS:0000000000000000
> [   70.724561] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [   70.724561] CR2: 0000000000000000 CR3: 000000010c568000 CR4: 00000000000006f0
> [   70.724561] Call Trace:
> [   70.724561]  <TASK>
> [   70.724561] ? __die (arch/x86/kernel/dumpstack.c:421 arch/x86/kernel/dumpstack.c:434)
> [   70.724561] ? page_fault_oops (arch/x86/mm/fault.c:715)
> [   70.724561] ? exc_page_fault (./arch/x86/include/asm/irqflags.h:26 ./arch/x86/include/asm/irqflags.h:87 ./arch/x86/include/asm/irqflags.h:147 arch/x86/mm/fault.c:1489 arch/x86/mm/fault.c:1539)
> [   70.724561] ? asm_exc_page_fault (./arch/x86/include/asm/idtentry.h:623)
> [   70.724561] ? fq_codel_enqueue (net/sched/sch_fq_codel.c:120 net/sched/sch_fq_codel.c:168 net/sched/sch_fq_codel.c:230) sch_fq_codel
> [   70.724561] dev_qdisc_enqueue (net/core/dev.c:3784)
> [   70.724561] __dev_queue_xmit (net/core/dev.c:3880 (discriminator 2) net/core/dev.c:4390 (discriminator 2))
> [   70.724561] ? irqentry_enter (kernel/entry/common.c:237)
> [   70.724561] ? sysvec_apic_timer_interrupt (./arch/x86/include/asm/hardirq.h:74 (discriminator 2) arch/x86/kernel/apic/apic.c:1043 (discriminator 2) arch/x86/kernel/apic/apic.c:1043 (discriminator 2))
> [   70.724561] ? trace_hardirqs_on (kernel/trace/trace_preemptirq.c:58 (discriminator 4))
> [   70.724561] ? asm_sysvec_apic_timer_interrupt (./arch/x86/include/asm/idtentry.h:702)
> [   70.724561] ? virtio_net_hdr_to_skb.constprop.0 (./include/linux/virtio_net.h:129 (discriminator 1))
> [   70.724561] packet_sendmsg (net/packet/af_packet.c:3145 (discriminator 1) net/packet/af_packet.c:3177 (discriminator 1))
> [   70.724561] ? _raw_spin_lock_bh (./arch/x86/include/asm/atomic.h:107 (discriminator 4) ./include/linux/atomic/atomic-arch-fallback.h:2170 (discriminator 4) ./include/linux/atomic/atomic-instrumented.h:1302 (discriminator 4) ./include/asm-generic/qspinlock.h:111 (discriminator 4) ./include/linux/spinlock.h:187 (discriminator 4) ./include/linux/spinlock_api_smp.h:127 (discriminator 4) kernel/locking/spinlock.c:178 (discriminator 4))
> [   70.724561] ? netdev_name_node_lookup_rcu (net/core/dev.c:325 (discriminator 1))
> [   70.724561] __sys_sendto (net/socket.c:730 (discriminator 1) net/socket.c:745 (discriminator 1) net/socket.c:2210 (discriminator 1))
> [   70.724561] ? __sys_setsockopt (./include/linux/file.h:34 net/socket.c:2355)
> [   70.724561] __x64_sys_sendto (net/socket.c:2222 (discriminator 1) net/socket.c:2218 (discriminator 1) net/socket.c:2218 (discriminator 1))
> [   70.724561] do_syscall_64 (arch/x86/entry/common.c:52 (discriminator 1) arch/x86/entry/common.c:83 (discriminator 1))
> [   70.724561] entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130)
> [   70.724561] RIP: 0033:0x41ae09
> 
> Fixes: cf9acc90c80ec ("net: virtio_net_hdr_to_skb: count transport header in UFO")
> Reported-by: syzbot <syzkaller@...glegroups.com>
> Signed-off-by: Eric Dumazet <edumazet@...gle.com>
> Cc: Jonathan Davies <jonathan.davies@...anix.com>
> ---
>   net/core/dev.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/net/core/dev.c b/net/core/dev.c
> index 1e740faf9e783b047b2dc7d9fd4242e4e6c7317a..f2c47da79f17d5ebe6b334b63d66c84c84c519fc 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -3758,7 +3758,7 @@ static void qdisc_pkt_len_init(struct sk_buff *skb)
>   						sizeof(_tcphdr), &_tcphdr);
>   			if (likely(th))
>   				hdr_len += __tcp_hdrlen(th);
> -		} else {
> +		} else if (shinfo->gso_type & SKB_GSO_UDP_L4) {
>   			struct udphdr _udphdr;
>   
>   			if (skb_header_pointer(skb, hdr_len,

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ