* [PATCH] ksmbd: take credits lock for SMB1 negotiate accounting
@ 2026-11-08 22:34 Noah Weinberger
2026-11-08 23:58 ` Ian Bridges
0 siblings, 1 reply; 2+ messages in thread
From: Noah Weinberger @ 2026-11-08 22:34 UTC (permalink / raw)
To: linkinjeon; +Cc: linux-cifs, linux-kernel, security
Hello,
The SMB1 multi-protocol negotiate path increments
conn->outstanding_credits outside conn->credits_lock, while every other
writer of that counter holds the lock. The race is unreachable in
practice (connections in need_neg state are serialized by
queue_ksmbd_work(), and SMB1 messages are rejected before work
allocation once the connection is upgraded), so this is a consistency
fix: move the increment under the existing critical section.
Affected version/commit: mainline fa4ad798f6f681031775266044ba1b0cf3c670bd (v7.4-rc1, 2026-11-07), fs/smb/server/smb_common.c:165-183 (ksmbd_verify_smb_message).
Pre-existing in v7.3.
This was found during an assisted (AI-supported) static source review of
mainline fa4ad798f6f681031775266044ba1b0cf3c670bd (v7.4-rc1, 2026-11-07); every code path cited below was re-verified by the
reporter against that tree. The patch below has NOT been compiled and has
NOT been runtime tested, and no reproducer was produced - the triggering
conditions are described above instead. Please treat this as a report for
verification rather than a battle-tested fix.
Reported-by: Noah Weinberger <noahweinberger2003@proton.me>
---
diff --git a/fs/smb/server/smb_common.c b/fs/smb/server/smb_common.c
index f9fb374..7453b0f 100644
--- a/fs/smb/server/smb_common.c
+++ b/fs/smb/server/smb_common.c
@@ -166,7 +166,6 @@ int ksmbd_verify_smb_message(struct ksmbd_work *work)
hdr->Command == SMB_COM_NEGOTIATE) {
struct ksmbd_conn *conn = work->conn;
- conn->outstanding_credits++;
/*
* A legacy SMB1 multi-protocol negotiate occupies sequence
* number 0 but does not pass through
@@ -175,6 +174,7 @@ int ksmbd_verify_smb_message(struct ksmbd_work *work)
* window can advance instead of staying pinned at 0.
*/
spin_lock(&conn->credits_lock);
+ conn->outstanding_credits++;
if (conn->seq_low == 0) {
__clear_bit(0, conn->seq_bitmap);
conn->seq_low = 1;
^ permalink raw reply [flat|nested] 2+ messages in thread* Re: [PATCH] ksmbd: take credits lock for SMB1 negotiate accounting
2026-11-08 22:34 [PATCH] ksmbd: take credits lock for SMB1 negotiate accounting Noah Weinberger
@ 2026-11-08 23:58 ` Ian Bridges
0 siblings, 0 replies; 2+ messages in thread
From: Ian Bridges @ 2026-11-08 23:58 UTC (permalink / raw)
To: Noah Weinberger; +Cc: linkinjeon, linux-cifs, linux-kernel, Ian Bridges
On Sun, 08 Nov 2026 22:34:09 +0000, Noah Weinberger wrote:
> The SMB1 multi-protocol negotiate path increments
> conn->outstanding_credits outside conn->credits_lock, while every other
> writer of that counter holds the lock.
I can confirm this. smb2_validate_credit_charge(),
smb2_set_rsp_credits() and the release on the send path all update
outstanding_credits under credits_lock. The increment in
ksmbd_verify_smb_message() is the only one outside it. The suggested
change moves it into the critical section that already follows it. The
increment is still returned by smb2_set_rsp_credits() as before.
I also tried to reproduce the issue, with and without the change. After
an SMB1 negotiate, an SMB2 session and a few requests, the counter is
back at 0. Two SMB1 negotiates sent in one write are answered one after
the other. An SMB1 negotiate on a connection that is already on SMB2 is
ignored. This matches your reasoning that the race cannot happen.
Thanks,
Ian
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2026-11-08 23:58 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-08 22:34 [PATCH] ksmbd: take credits lock for SMB1 negotiate accounting Noah Weinberger
2026-11-08 23:58 ` Ian Bridges
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®