Re: KMSAN: uninit-value in batadv_interface_tx (2)

From: Sven Eckelmann <sven@narfation.org>
To: syzbot <syzbot+24458cef7d37351dd0c3@syzkaller.appspotmail.com>
Cc: a@unstable.cc, b.a.t.m.a.n@lists.open-mesh.org,
	davem@davemloft.net, glider@google.com,
	linux-kernel@vger.kernel.org, mareklindner@neomailbox.ch,
	netdev@vger.kernel.org, sw@simonwunderlich.de,
	syzkaller-bugs@googlegroups.com,
	Arvid Brodin <arvid.brodin@alten.se>
Subject: Re: KMSAN: uninit-value in batadv_interface_tx (2)
Date: Fri, 31 Jan 2020 08:17:05 +0100	[thread overview]
Message-ID: <2402337.3Z8xRsGYif@bentobox> (raw)
In-Reply-To: <00000000000038fa02059d614893@google.com>

[-- Attachment #1: Type: text/plain, Size: 3169 bytes --]

On Thursday, 30 January 2020 21:27:12 CET syzbot wrote:
[...]
> dashboard link: https://syzkaller.appspot.com/bug?extid=24458cef7d37351dd0c3
[...]
> Reported-by: syzbot+24458cef7d37351dd0c3@syzkaller.appspotmail.com
[...]
> =====================================================
> BUG: KMSAN: uninit-value in batadv_interface_tx+0x10cf/0x2450 net/batman-adv/soft-interface.c:264

This line is just comparing the eth_hdr(skb)->h_dest with the STP address 
(static const buffer). 

Is there some situation when hsr would not initialize offset for eth_hdr() 
correctly or where h_dest is not initialized from a correctly initialized 
value? At least the comparisons with eth_hdr(skb)->h_source didn't cause this 
error.

Regarding my assupmtion: it seems like the hsr_xmit is only initializing the 
h_source in master mode. And there even more conditions (see
hsr_addr_subst_dest) for initializing the h_dest in master mode.

Btw. the code in hsr which "stored this" was

	if (skb_put_padto(skb, ETH_ZLEN + HSR_HLEN))
		return;

[...]
>  batadv_interface_tx+0x10cf/0x2450 net/batman-adv/soft-interface.c:264
>  __netdev_start_xmit include/linux/netdevice.h:4447 [inline]
>  netdev_start_xmit include/linux/netdevice.h:4461 [inline]
>  xmit_one net/core/dev.c:3420 [inline]
>  dev_hard_start_xmit+0x531/0xab0 net/core/dev.c:3436
>  __dev_queue_xmit+0x37de/0x4220 net/core/dev.c:4013
>  dev_queue_xmit+0x4b/0x60 net/core/dev.c:4046
>  hsr_xmit net/hsr/hsr_forward.c:228 [inline]
>  hsr_forward_do net/hsr/hsr_forward.c:285 [inline]
>  hsr_forward_skb+0x2614/0x30d0 net/hsr/hsr_forward.c:361
>  hsr_handle_frame+0x385/0x4b0 net/hsr/hsr_slave.c:43
>  __netif_receive_skb_core+0x21de/0x5840 net/core/dev.c:5051
>  __netif_receive_skb_one_core net/core/dev.c:5148 [inline]
>  __netif_receive_skb net/core/dev.c:5264 [inline]
>  process_backlog+0x936/0x1410 net/core/dev.c:6095
>  napi_poll net/core/dev.c:6532 [inline]
>  net_rx_action+0x786/0x1ab0 net/core/dev.c:6600
>  __do_softirq+0x311/0x83d kernel/softirq.c:293
>  run_ksoftirqd+0x25/0x40 kernel/softirq.c:607
>  smpboot_thread_fn+0x493/0x980 kernel/smpboot.c:165
>  kthread+0x4b5/0x4f0 kernel/kthread.c:256
>  ret_from_fork+0x35/0x40 arch/x86/entry/entry_64.S:353
> 
> Uninit was stored to memory at:
>  kmsan_save_stack_with_flags mm/kmsan/kmsan.c:144 [inline]
>  kmsan_internal_chain_origin+0xad/0x130 mm/kmsan/kmsan.c:310
>  kmsan_memcpy_memmove_metadata+0x272/0x2e0 mm/kmsan/kmsan.c:247
>  kmsan_memcpy_metadata+0xb/0x10 mm/kmsan/kmsan.c:267
>  __msan_memcpy+0x43/0x50 mm/kmsan/kmsan_instr.c:116
>  pskb_expand_head+0x38b/0x1b00 net/core/skbuff.c:1637
>  __skb_pad+0x47f/0x900 net/core/skbuff.c:1805
>  __skb_put_padto include/linux/skbuff.h:3193 [inline]
>  skb_put_padto include/linux/skbuff.h:3212 [inline]
>  send_hsr_supervision_frame+0x122d/0x1500 net/hsr/hsr_device.c:310
>  hsr_announce+0x1e2/0x370 net/hsr/hsr_device.c:341
>  call_timer_fn+0x218/0x510 kernel/time/timer.c:1404
>  expire_timers kernel/time/timer.c:1449 [inline]
>  __run_timers+0xcff/0x1210 kernel/time/timer.c:1773
>  run_timer_softirq+0x2d/0x50 kernel/time/timer.c:1786
>  __do_softirq+0x311/0x83d kernel/softirq.c:293

Kind regards,
	Sven

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]