Re: [syzbot] [atm?] KMSAN: uninit-value in atmtcp_c_send

[Kuniyuki Iwashima]

From: syzbot <syzbot+1d3c235276f62963e93a@xxxxxxxxxxxxxxxxxxxxxxxxx>
Date: Thu, 12 Jun 2025 20:48:02 -0700
> Hello,
> 
> syzbot has tested the proposed patch and the reproducer did not trigger any issue:
> 
> Reported-by: syzbot+1d3c235276f62963e93a@xxxxxxxxxxxxxxxxxxxxxxxxx
> Tested-by: syzbot+1d3c235276f62963e93a@xxxxxxxxxxxxxxxxxxxxxxxxx
> 
> Tested on:
> 
> commit:         27605c8c Merge tag 'net-6.16-rc2' of git://git.kernel...
> git tree:       upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=16d17682580000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=42d51b7b9f9e61d
> dashboard link: https://syzkaller.appspot.com/bug?extid=1d3c235276f62963e93a
> compiler:       Debian clang version 20.1.6 (++20250514063057+1e4d39e07757-1~exp1~20250514183223.118), Debian LLD 20.1.6
> patch:          https://syzkaller.appspot.com/x/patch.diff?x=129e7682580000
> 
> Note: testing is done by a robot and is best-effort only.

I noticed the original code didn't call dev_kfree_skb() and leaked
sk_buff_head.

Also, most of the ATM drivers don't call atm_return() to revert
the memory accounting by atm_account_tx() in vcc_sendmsg(), but
that's another bug...

#syz test

diff --git a/drivers/atm/atmtcp.c b/drivers/atm/atmtcp.c
index d4aa0f353b6c..5ce3c6c066e1 100644
--- a/drivers/atm/atmtcp.c
+++ b/drivers/atm/atmtcp.c
@@ -288,7 +288,11 @@ static int atmtcp_c_send(struct atm_vcc *vcc,struct sk_buff *skb)
 	struct sk_buff *new_skb;
 	int result = 0;
 
-	if (!skb->len) return 0;
+	if (skb->len < sizeof(struct atmtcp_hdr)) {
+		dev_kfree_skb(skb);
+		return -EINVAL;
+	}
+
 	dev = vcc->dev_data;
 	hdr = (struct atmtcp_hdr *) skb->data;
 	if (hdr->length == ATMTCP_HDR_MAGIC) {