Thread (6 messages) 6 messages, 2 authors, 22d ago

Re: [PATCH v11 3/3] media: dvb-core: extirpate defective and unsafe ULE debug block

From: sashiko-bot@kernel.org
Date: 2026-09-05 14:12:33
Also in: linux-clk, linux-pci

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:

New issues:
- [Low] The patch introduces a severe compilation failure by modifying the call to `dvb_net_ule_check_crc()` to pass 3 arguments instead of 4, without updating the function's signature to match.

Pre-existing issues:
- [Critical] The network device is unregistered after flushing the workqueues, allowing callbacks to queue works during unregistration, which leads to a Use-After-Free.
- [Critical] Missing bounds checks when parsing ULE extension headers allows malformed packets to cause out-of-bounds memory reads.
--

commit e9caeb2ec9c58e63757d73d1d8a17bcadf35a678
Author: Breno Rodrigues Alves [off-list ref]

media: dvb-core: extirpate defective and unsafe ULE debug block

This commit removes the unmaintained DVB_ULE_DEBUG code block from
dvb_net.c and its Kconfig definition. It aims to eliminate race
conditions and clean up unused function signature parameters.
quoted hunk ↗ jump to hunk
diff --git a/drivers/media/dvb-core/dvb_net.c b/drivers/media/dvb-core/dvb_net.c
index a2159b2bc1766..5060b8866de8a 100644
--- a/drivers/media/dvb-core/dvb_net.c
+++ b/drivers/media/dvb-core/dvb_net.c
[ ... ]
quoted hunk ↗ jump to hunk
@@ -843,7 +803,7 @@ static void dvb_net_ule(struct net_device *dev, const u8 *buf, size_t buf_len)
 				       *(tail - 2) << 8 |
 				       *(tail - 1);
 
-			dvb_net_ule_check_crc(&h, iov, ule_crc, expected_crc);
+			dvb_net_ule_check_crc(&h, ule_crc, expected_crc);
[Severity: Low]
Does this change miss updating the function signature of
dvb_net_ule_check_crc()? 

It looks like the invocation now passes 3 arguments, but the function
definition for dvb_net_ule_check_crc() still expects the iov array as its
second argument, which causes a compilation failure due to too few arguments.
 
 			/* Prepare for next SNDU. */
 			reset_ule(h.priv);

While reviewing this code, I also noticed a couple of issues that are not
introduced by this patch, but might be worth addressing:

[Severity: Critical]
This is a pre-existing issue, but I noticed a potential use-after-free
when unregistering the network device.

In dvb_net_remove_if(), the workqueues are flushed before the device is
unregistered:

	flush_work(&priv->set_multicast_list_wq);
	flush_work(&priv->restart_net_feed_wq);
	pr_info("removed network interface %s\n", net->name);
	unregister_netdev(net);

When unregister_netdev() is called, callbacks can be triggered (such as
ndo_set_rx_mode mapping to dvb_net_set_multicast_list) which schedule work
back onto set_multicast_list_wq. Could this result in a use-after-free
when the pending work executes later, since the priv memory gets freed
during teardown?

[Severity: Critical]
This is a pre-existing issue, but it looks like there might be an
out-of-bounds read when parsing ULE extension headers.

The handle_ule_extensions() loop iterates over extension headers using
handle_one_ule_extension(), which reads memory and advances the pointer:

	p->ule_next_hdr += ext_len;
	p->ule_sndu_type = ntohs( *(__be16 *)(p->ule_next_hdr-2) );

Is it possible for a malformed payload to specify extension lengths that
push p->ule_next_hdr past the end of the allocated skb buffer? The loop
appears to continue without bounds checking against skb->len, which could
cause a page fault on malicious inputs.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260905140113.21638-1-breno3011alves@gmail.com?part=3
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help