FAILED: patch "[PATCH] nvmet-tcp: Fix potential UAF when ddgst mismatch" failed to apply to 5.10-stable tree

The patch below does not apply to the 5.10-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.10.y
git checkout FETCH_HEAD
git cherry-pick -x dbbd07d0a7020b80f6a7028e561908f7b83b3d5a
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<[email protected]>' --in-reply-to '2026071900-poster-swear-030a@gregkh' --subject-prefix 'PATCH 5.10.y' 'HEAD^..'

Possible dependencies:

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From dbbd07d0a7020b80f6a7028e561908f7b83b3d5a Mon Sep 17 00:00:00 2001
From: Sagi Grimberg <[email protected]>
Date: Sun, 10 May 2026 23:30:29 +0300
Subject: [PATCH] nvmet-tcp: Fix potential UAF when ddgst mismatch
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Shivam Kumar found via vulnerability testing:
When data digest is enabled on an NVMe/TCP connection and a digest mismatch occurs on a non-final H2C_DATA PDU during an R2T-based data transfer, the digest error handler in nvmet_tcp_try_recv_ddgst() calls nvmet_req_uninit() — which performs percpu_ref_put() on the submission queue — but does NOT mark the command as completed. It does not set cqe->status, does not modify rbytes_done, and does not clear any flag. When the subsequent fatal error triggers queue teardown, nvmet_tcp_uninit_data_in_cmds() iterates all commands, checks nvmet_tcp_need_data_in() for each one, and finds that the already-uninited command still appears to need data (because rbytes_done < transfer_len and cqe->status == 0). It therefore calls nvmet_req_uninit() a second time on the same command — a double percpu_ref_put against a single percpu_ref_get.

index 164a564ba3b4..20f150d17a96 100644
@@ -1321,8 +1321,10 @@ static int nvmet_tcp_try_recv_ddgst(struct nvmet_tcp_queue *queue)
 			queue->idx, cmd->req.cmd->common.command_id,
 			queue->pdu.cmd.hdr.type, le32_to_cpu(cmd->recv_ddgst),
 			le32_to_cpu(cmd->exp_ddgst));
-		if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED))
+		if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED)) {
+			cmd->req.cqe->status = NVME_SC_CMD_SEQ_ERROR;
 			nvmet_req_uninit(&cmd->req);
+		}
 		nvmet_tcp_free_cmd_buffers(cmd);
 		ret = -EPROTO;
 		goto out;

commit dbbd07d0a7020b80f6a7028e561908f7b83b3d5a upstream.

Shivam Kumar found via vulnerability testing:
When data digest is enabled on an NVMe/TCP connection and a digest mismatch occurs on a non-final H2C_DATA PDU during an R2T-based data transfer, the digest error handler in nvmet_tcp_try_recv_ddgst() calls nvmet_req_uninit() - which performs percpu_ref_put() on the submission queue - but does NOT mark the command as completed. It does not set cqe->status, does not modify rbytes_done, and does not clear any flag. When the subsequent fatal error triggers queue teardown, nvmet_tcp_uninit_data_in_cmds() iterates all commands, checks nvmet_tcp_need_data_in() for each one, and finds that the already-uninited command still appears to need data (because rbytes_done < transfer_len and cqe->status == 0). It therefore calls nvmet_req_uninit() a second time on the same command - a double percpu_ref_put against a single percpu_ref_get.

---
 drivers/nvme/target/tcp.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

index ba5817006..11e917d02 100644
@@ -1200,8 +1200,10 @@ static int nvmet_tcp_try_recv_ddgst(struct nvmet_tcp_queue *queue)
 			queue->idx, cmd->req.cmd->common.command_id,
 			queue->pdu.cmd.hdr.type, le32_to_cpu(cmd->recv_ddgst),
 			le32_to_cpu(cmd->exp_ddgst));
-		if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED))
+		if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED)) {
+			cmd->req.cqe->status = NVME_SC_CMD_SEQ_ERROR;
 			nvmet_req_uninit(&cmd->req);
+		}
 		nvmet_tcp_free_cmd_buffers(cmd);
 		nvmet_tcp_fatal_error(queue);
 		ret = -EPROTO;

2.50.1 (Apple Git-155)

This one doesn't apply to linux-5.10.y: the diff context assumes 4606467a75cf ("nvmet-tcp: check INIT_FAILED before nvmet_req_uninit in digest error path") and the nvmet_tcp_finish_cmd() removal, neither of which is in 5.10.y -- the ddgst error path there still calls nvmet_tcp_finish_cmd(cmd).

Your 5.10.y backport of 4606467a75cf was queued today, so could you send a v2 of this patch generated on top of it? The end state in your patch looks correct and can stay as-is.

Thanks,
Sasha

commit dbbd07d0a7020b80f6a7028e561908f7b83b3d5a upstream.

Shivam Kumar found via vulnerability testing:
When data digest is enabled on an NVMe/TCP connection and a digest mismatch occurs on a non-final H2C_DATA PDU during an R2T-based data transfer, the digest error handler in nvmet_tcp_try_recv_ddgst() calls nvmet_req_uninit() - which performs percpu_ref_put() on the submission queue - but does NOT mark the command as completed. It does not set cqe->status, does not modify rbytes_done, and does not clear any flag. When the subsequent fatal error triggers queue teardown, nvmet_tcp_uninit_data_in_cmds() iterates all commands, checks nvmet_tcp_need_data_in() for each one, and finds that the already-uninited command still appears to need data (because rbytes_done < transfer_len and cqe->status == 0). It therefore calls nvmet_req_uninit() a second time on the same command - a double percpu_ref_put against a single percpu_ref_get.

---
 drivers/nvme/target/tcp.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

index ba5817006..11e917d02 100644
@@ -1200,8 +1200,10 @@ static int nvmet_tcp_try_recv_ddgst(struct nvmet_tcp_queue *queue)
 			queue->idx, cmd->req.cmd->common.command_id,
 			queue->pdu.cmd.hdr.type, le32_to_cpu(cmd->recv_ddgst),
 			le32_to_cpu(cmd->exp_ddgst));
-		if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED))
+		if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED)) {
+			cmd->req.cqe->status = NVME_SC_CMD_SEQ_ERROR;
 			nvmet_req_uninit(&cmd->req);
+		}
 		nvmet_tcp_free_cmd_buffers(cmd);
 		nvmet_tcp_fatal_error(queue);
 		ret = -EPROTO;

2.50.1 (Apple Git-155)

Queued for 5.10, thanks.

Thanks,
Sasha