[PATCH bpf-next v3 1/7] bpf, sockmap: reject overflowing copy + len in bpf_msg_push_data()

[Jiayuan Chen]

From: Weiming Shi <bestswngs@xxxxxxxxx>

When the scatterlist ring is full or nearly full, bpf_msg_push_data()
enters a copy fallback path and computes copy + len for the page
allocation size. Since len comes from BPF with arg3_type = ARG_ANYTHING
and both are u32, a crafted len can wrap the sum to a small value,
causing an undersized allocation followed by an out-of-bounds memcpy.

 BUG: unable to handle page fault for address: ffffed104089a402
 Oops: Oops: 0000 [#1] SMP KASAN NOPTI
 Call Trace:
  __asan_memcpy (mm/kasan/shadow.c:105)
  bpf_msg_push_data (net/core/filter.c:2852 net/core/filter.c:2788)
  bpf_prog_9ed8b5711920a7d7+0x2e/0x36
  sk_psock_msg_verdict (net/core/skmsg.c:934)
  tcp_bpf_sendmsg (net/ipv4/tcp_bpf.c:421 net/ipv4/tcp_bpf.c:584)
  __sys_sendto (net/socket.c:2206)
  do_syscall_64 (arch/x86/entry/syscall_64.c:94)
  entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130)

Add an overflow check before the allocation.

Link: https://lore.kernel.org/all/20260424155913.A19FDC19425@xxxxxxxxxxxxxxx
Fixes: 6fff607e2f14 ("bpf: sk_msg program helper bpf_msg_push_data")
Tested-by: Xiang Mei <xmei5@xxxxxxx>
Tested-by: Xinyu Ma <mmmxny@xxxxxxxxx>
Reviewed-by: Jiayuan Chen <jiayuan.chen@xxxxxxxxx>
Reviewed-by: Emil Tsalapatis <emil@xxxxxxxxxxxxxxx>
Signed-off-by: Weiming Shi <bestswngs@xxxxxxxxx>
Signed-off-by: Jiayuan Chen <jiayuan.chen@xxxxxxxxx>
---
To sashiko:

Regarding bpf_msg_push_data() reading "copy = msg->sg.data[i].length" with
i == msg->sg.end (appending at the very end of a full/near-full ring):

This is pre-existing code, not touched by this series, and reproducing it needs
a narrow combination -- a pure append at the end so the loop exits with
i == msg->sg.end, a full/near-full ring, plus a prior push/pop history that
leaves a stale length in the otherwise-unused end slot. A freshly built ring
zeroes that slot, so copy stays 0. We don't consider it practically reproducible.

Even then it's already covered: the overflow check in patch 1 ("copy + len <
copy") rejects the dangerous case, and __GFP_ZERO in patch 3 prevents any data
exposure. Not worth fixing here.
---
 net/core/filter.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/core/filter.c b/net/core/filter.c
index 9590877b0714f..3c8f1cedb217f 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -2829,6 +2829,9 @@ BPF_CALL_4(bpf_msg_push_data, struct sk_msg *, msg, u32, start,
 	if (!space || (space == 1 && start != offset))
 		copy = msg->sg.data[i].length;
 
+	if (unlikely(copy + len < copy))
+		return -EINVAL;
+
 	page = alloc_pages(__GFP_NOWARN | GFP_ATOMIC | __GFP_COMP,
 			   get_order(copy + len));
 	if (unlikely(!page))
-- 
2.43.0