UBUNTU-CVE-2026-43194
In the Linux kernel, the following vulnerability has been resolved: net: consume xmit errors of GSO frames udpgro_frglist.sh and udpgro_bench.sh are the flakiest tests currently in NIPA. They fail in the same exact way, TCP GRO test stalls occasionally and the test gets killed after 10min. These tests use veth to simulate GRO. They attach a trivial ("return XDP_PASS;") XDP program to the veth to force TSO off and NAPI on. Digging into the failure mode we can see that the connection is completely stuck after a burst of drops. The sender's snd_nxt is at sequence number N [1], but the receiver claims to have received (rcv_nxt) up to N + 3 * MSS [2]. Last piece of the puzzle is that senders rtx queue is not empty (let's say the block in the rtx queue is at sequence number N - 4 * MSS [3]). In this state, sender sends a retransmission from the rtx queue with a single segment, and sequence numbers N-4*MSS:N-3*MSS [3]. Receiver sees it and responds with an ACK all the way up to N + 3 * MSS [2]. But sender will reject this ack as TCP_ACK_UNSENT_DATA because it has no recollection of ever sending data that far out [1]. And we are stuck. The root cause is the mess of the xmit return codes. veth returns an error when it can't xmit a frame. We end up with a loss event like this: ------------------------------------------------- | GSO super frame 1 | GSO super frame 2 | |-----------------------------------------------| | seg | seg | seg | seg | seg | seg | seg | seg | | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | ------------------------------------------------- x ok ok <ok>| ok ok ok <x> \\ snd_nxt "x" means packet lost by veth, and "ok" means it went thru. Since veth has TSO disabled in this test it sees individual segments. Segment 1 is on the retransmit queue and will be resent. So why did the sender not advance snd_nxt even tho it clearly did send up to seg 8? tcp_write_xmit() interprets the return code from the core to mean that data has not been sent at all. Since TCP deals with GSO super frames, not individual segment the crux of the problem is that loss of a single segment can be interpreted as loss of all. TCP only sees the last return code for the last segment of the GSO frame (in <> brackets in the diagram above). Of course for the problem to occur we need a setup or a device without a Qdisc. Otherwise Qdisc layer disconnects the protocol layer from the device errors completely. We have multiple ways to fix this. 1) make veth not return an error when it lost a packet. While this is what I think we did in the past, the issue keeps reappearing and it's annoying to debug. The game of whack a mole is not great. 2) fix the damn return codes We only talk about NETDEV_TX_OK and NETDEV_TX_BUSY in the documentation, so maybe we should make the return code from ndo_start_xmit() a boolean. I like that the most, but perhaps some ancient, not-really-networking protocol would suffer. 3) make TCP ignore the errors It is not entirely clear to me what benefit TCP gets from interpreting the result of ip_queue_xmit()? Specifically once the connection is established and we're pushing data - packet loss is just packet loss? 4) this fix Ignore the rc in the Qdisc-less+GSO case, since it's unreliable. We already always return OK in the TCQ_F_CAN_BYPASS case. In the Qdisc-less case let's be a bit more conservative and only mask the GSO errors. This path is taken by non-IP-"networks" like CAN, MCTP etc, so we could regress some ancient thing. This is the simplest, but also maybe the hackiest fix? Similar fix has been proposed by Eric in the past but never committed because original reporter was working with an OOT driver and wasn't providing feedback (see Link).
02 / AFFECTED SOFTWARE
Affected packages
26 explicit affected versions
154 explicit affected versions
11 explicit affected versions
6 explicit affected versions
59 explicit affected versions
10 explicit affected versions
53 explicit affected versions
116 explicit affected versions
87 explicit affected versions
50 explicit affected versions
6 explicit affected versions
12 explicit affected versions
19 explicit affected versions
10 explicit affected versions
6 explicit affected versions
97 explicit affected versions
84 explicit affected versions
24 explicit affected versions
82 explicit affected versions
83 explicit affected versions
91 explicit affected versions
13 explicit affected versions
85 explicit affected versions
76 explicit affected versions
10 explicit affected versions
36 explicit affected versions
26 explicit affected versions
10 explicit affected versions
89 explicit affected versions
7 explicit affected versions
84 explicit affected versions
39 explicit affected versions
43 explicit affected versions
35 explicit affected versions
49 explicit affected versions
47 explicit affected versions
18 explicit affected versions
40 explicit affected versions
118 explicit affected versions
62 explicit affected versions
44 explicit affected versions
120 explicit affected versions
13 explicit affected versions
4 explicit affected versions
121 explicit affected versions
23 explicit affected versions
51 explicit affected versions
124 explicit affected versions
1 explicit affected versions
1 explicit affected versions
21 explicit affected versions
7 explicit affected versions
16 explicit affected versions
10 explicit affected versions
20 explicit affected versions
76 explicit affected versions
33 explicit affected versions
47 explicit affected versions
8 explicit affected versions
7 explicit affected versions
12 explicit affected versions
11 explicit affected versions
111 explicit affected versions
7 explicit affected versions
12 explicit affected versions
14 explicit affected versions
10 explicit affected versions
95 explicit affected versions
1 explicit affected versions
7 explicit affected versions
16 explicit affected versions
104 explicit affected versions
12 explicit affected versions
105 explicit affected versions
12 explicit affected versions
40 explicit affected versions
120 explicit affected versions
1 explicit affected versions
43 explicit affected versions
16 explicit affected versions
17 explicit affected versions
27 explicit affected versions
10 explicit affected versions
10 explicit affected versions
37 explicit affected versions
10 explicit affected versions
11 explicit affected versions
13 explicit affected versions
7 explicit affected versions
38 explicit affected versions
3 explicit affected versions
10 explicit affected versions
11 explicit affected versions
65 explicit affected versions
1 explicit affected versions
26 explicit affected versions
9 explicit affected versions
13 explicit affected versions
144 explicit affected versions
34 explicit affected versions
1 explicit affected versions
29 explicit affected versions
111 explicit affected versions
13 explicit affected versions
12 explicit affected versions
40 explicit affected versions
39 explicit affected versions
8 explicit affected versions
1 explicit affected versions
7 explicit affected versions
100 explicit affected versions
89 explicit affected versions
52 explicit affected versions
33 explicit affected versions
23 explicit affected versions
58 explicit affected versions
10 explicit affected versions
19 explicit affected versions
188 explicit affected versions
18 explicit affected versions
11 explicit affected versions
5 explicit affected versions
12 explicit affected versions
11 explicit affected versions
12 explicit affected versions
10 explicit affected versions
16 explicit affected versions
138 explicit affected versions
2 explicit affected versions
5 explicit affected versions
74 explicit affected versions
132 explicit affected versions
90 explicit affected versions
165 explicit affected versions
26 explicit affected versions
39 explicit affected versions
1 explicit affected versions
66 explicit affected versions
45 explicit affected versions
35 explicit affected versions
46 explicit affected versions
11 explicit affected versions
10 explicit affected versions
13 explicit affected versions
139 explicit affected versions
78 explicit affected versions
16 explicit affected versions
88 explicit affected versions
49 explicit affected versions
116 explicit affected versions
3 explicit affected versions
132 explicit affected versions
93 explicit affected versions
1 explicit affected versions
7 explicit affected versions
77 explicit affected versions
2 explicit affected versions
15 explicit affected versions
51 explicit affected versions
16 explicit affected versions
92 explicit affected versions
1 explicit affected versions
1 explicit affected versions
18 explicit affected versions
41 explicit affected versions
12 explicit affected versions
70 explicit affected versions
98 explicit affected versions
7 explicit affected versions
27 explicit affected versions
63 explicit affected versions
121 explicit affected versions
42 explicit affected versions
8 explicit affected versions
54 explicit affected versions
5 explicit affected versions
8 explicit affected versions
45 explicit affected versions
13 explicit affected versions
110 explicit affected versions
13 explicit affected versions
72 explicit affected versions
7 explicit affected versions
140 explicit affected versions
8 explicit affected versions
94 explicit affected versions
10 explicit affected versions
12 explicit affected versions
9 explicit affected versions
36 explicit affected versions
108 explicit affected versions
13 explicit affected versions
80 explicit affected versions
56 explicit affected versions
44 explicit affected versions
16 explicit affected versions
3 explicit affected versions
45 explicit affected versions
37 explicit affected versions
49 explicit affected versions
14 explicit affected versions
38 explicit affected versions
38 explicit affected versions
167 explicit affected versions
9 explicit affected versions
17 explicit affected versions
11 explicit affected versions
17 explicit affected versions
14 explicit affected versions
90 explicit affected versions
6 explicit affected versions
7 explicit affected versions
97 explicit affected versions
121 explicit affected versions
4 explicit affected versions
79 explicit affected versions
10 explicit affected versions
1 explicit affected versions
1 explicit affected versions
1 explicit affected versions
88 explicit affected versions
78 explicit affected versions
78 explicit affected versions
13 explicit affected versions
43 explicit affected versions
77 explicit affected versions
16 explicit affected versions
119 explicit affected versions
10 explicit affected versions
8 explicit affected versions
44 explicit affected versions
18 explicit affected versions
88 explicit affected versions
14 explicit affected versions
4 explicit affected versions
78 explicit affected versions
14 explicit affected versions
11 explicit affected versions
1 explicit affected versions
69 explicit affected versions
123 explicit affected versions
38 explicit affected versions
77 explicit affected versions
03 / CONNECTIONS
Connected vulnerabilities
04 / EVIDENCE
Source records
In the Linux kernel, the following vulnerability has been resolved: net: consume xmit errors of GSO frames udpgro_frglist.sh and udpgro_bench.sh are the flakiest tests currently in NIPA. They fail in the same exact way, TCP GRO test stalls occasionally and the test gets killed after 10min. These tests use veth to simulate GRO. They attach a trivial ("return XDP_PASS;") XDP program to the veth to force TSO off and NAPI on. Digging into the failure mode we can see that the connection is completely stuck after a burst of drops. The sender's snd_nxt is at sequence number N [1], but the receiver claims to have received (rcv_nxt) up to N + 3 * MSS [2]. Last piece of the puzzle is that senders rtx queue is not empty (let's say the block in the rtx queue is at sequence number N - 4 * MSS [3]). In this state, sender sends a retransmission from the rtx queue with a single segment, and sequence numbers N-4*MSS:N-3*MSS [3]. Receiver sees it and responds with an ACK all the way up to N + 3 * MSS [2]. But sender will reject this ack as TCP_ACK_UNSENT_DATA because it has no recollection of ever sending data that far out [1]. And we are stuck. The root cause is the mess of the xmit return codes. veth returns an error when it can't xmit a frame. We end up with a loss event like this: ------------------------------------------------- | GSO super frame 1 | GSO super frame 2 | |-----------------------------------------------| | seg | seg | seg | seg | seg | seg | seg | seg | | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | ------------------------------------------------- x ok ok <ok>| ok ok ok <x> \\ snd_nxt "x" means packet lost by veth, and "ok" means it went thru. Since veth has TSO disabled in this test it sees individual segments. Segment 1 is on the retransmit queue and will be resent. So why did the sender not advance snd_nxt even tho it clearly did send up to seg 8? tcp_write_xmit() interprets the return code from the core to mean that data has not been sent at all. Since TCP deals with GSO super frames, not individual segment the crux of the problem is that loss of a single segment can be interpreted as loss of all. TCP only sees the last return code for the last segment of the GSO frame (in <> brackets in the diagram above). Of course for the problem to occur we need a setup or a device without a Qdisc. Otherwise Qdisc layer disconnects the protocol layer from the device errors completely. We have multiple ways to fix this. 1) make veth not return an error when it lost a packet. While this is what I think we did in the past, the issue keeps reappearing and it's annoying to debug. The game of whack a mole is not great. 2) fix the damn return codes We only talk about NETDEV_TX_OK and NETDEV_TX_BUSY in the documentation, so maybe we should make the return code from ndo_start_xmit() a boolean. I like that the most, but perhaps some ancient, not-really-networking protocol would suffer. 3) make TCP ignore the errors It is not entirely clear to me what benefit TCP gets from interpreting the result of ip_queue_xmit()? Specifically once the connection is established and we're pushing data - packet loss is just packet loss? 4) this fix Ignore the rc in the Qdisc-less+GSO case, since it's unreliable. We already always return OK in the TCQ_F_CAN_BYPASS case. In the Qdisc-less case let's be a bit more conservative and only mask the GSO errors. This path is taken by non-IP-"networks" like CAN, MCTP etc, so we could regress some ancient thing. This is the simplest, but also maybe the hackiest fix? Similar fix has been proposed by Eric in the past but never committed because original reporter was working with an OOT driver and wasn't providing feedback (see Link).
05 / REFERENCES
Further evidence
- https://git.kernel.org/linus/7aa767d0d3d04e50ae94e770db7db8197f666970
- https://ubuntu.com/security/CVE-2026-43194
- https://ubuntu.com/security/notices/USN-8492-1
- https://ubuntu.com/security/notices/USN-8492-2
- https://ubuntu.com/security/notices/USN-8492-3
- https://ubuntu.com/security/notices/USN-8492-4
- https://ubuntu.com/security/notices/USN-8492-5
- https://ubuntu.com/security/notices/USN-8497-1
- https://ubuntu.com/security/notices/USN-8498-1
- https://ubuntu.com/security/notices/USN-8499-1
- https://ubuntu.com/security/notices/USN-8575-1
- https://ubuntu.com/security/notices/USN-8575-2
- https://ubuntu.com/security/notices/USN-8575-3
- https://ubuntu.com/security/notices/USN-8576-1
- https://ubuntu.com/security/notices/USN-8576-2
- https://ubuntu.com/security/notices/USN-8597-1
- https://ubuntu.com/security/notices/USN-8606-1
- https://ubuntu.com/security/notices/USN-8607-1
- https://ubuntu.com/security/notices/USN-8609-1
- https://ubuntu.com/security/notices/USN-8610-1
- https://ubuntu.com/security/notices/USN-8619-1
- https://ubuntu.com/security/notices/USN-8620-1
- https://ubuntu.com/security/notices/USN-8620-2
- https://ubuntu.com/security/notices/USN-8620-3
- https://ubuntu.com/security/notices/USN-8620-4
- https://www.cve.org/CVERecord?id=CVE-2026-43194