[PATCH 7.2 52/82] nvmet-tcp: bound SGL data length before allocating command buffers

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

7.2-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Ibrahim Hashimov <security@auditcode.ai>

commit 4a3f00262a044e8e15064b1a6860968bf0500bf4 upstream.

nvmet_tcp_map_data() reads the host-controlled 32-bit sgl->length
and, for the in-capsule offset descriptor (type 0x01), checks it
against port->inline_data_size before use. Any other SGL descriptor
type -- including the non-inline transport SGL data-block descriptor
(type (NVME_TRANSPORT_SGL_DATA_DESC << 4) | NVME_SGL_FMT_TRANSPORT_A,
the type a real host uses for out-of-capsule writes) skips that check
entirely and falls straight through to:

	cmd->req.sg = sgl_alloc(len, GFP_KERNEL, &cmd->req.sg_cnt);

with len taken directly from the wire, unbounded up to 4 GiB.

nvmet_req_init() only parses the command and never inspects
sgl->length, and nvmet_check_transfer_len() -- the only other place
transfer_len is validated -- runs later, from req->execute(), after
the allocation has already happened. For a write command the target
responds with an R2T and parks the command waiting for the host to
send the data; if the host (or an unauthenticated peer that simply
never follows up) never does, the sgl_alloc() buffer stays resident
for the life of the command. NVMe/TCP has no mandatory authentication
in the default configuration, so any peer able to reach the target
portal and complete a Fabrics connect can drive this with a single
crafted command, repeatable across queues and connections for
amplification. This is unbounded kernel memory allocation
triggered by a remote, effectively unauthenticated peer.

Validate len against the same NVMET_TCP_MAXH2CDATA ceiling this file
already uses to bound per-PDU H2C data, for every SGL descriptor type,
before doing any allocation. This closes the gap for the non-inline
descriptor while leaving the existing, tighter inline_data_size check
in place for the in-capsule case.

Runtime-verified on a v6.19 KASAN stand: with this bound in place, a
crafted write command carrying an oversized non-inline SGL length is
rejected before sgl_alloc() runs, where the same request previously
drove an unbounded ~256 MiB kernel allocation (up to 4 GiB) that
stayed resident pending an R2T the host never satisfies.

Fixes: 872d26a391da ("nvmet-tcp: add NVMe over TCP target driver")
Cc: stable@vger.kernel.org
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ibrahim Hashimov <security@auditcode.ai>
Assisted-by: AuditCode-AI:2026.07
Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 drivers/nvme/target/tcp.c |   13 +++++++++++++
 1 file changed, 13 insertions(+)

--- a/drivers/nvme/target/tcp.c
+++ b/drivers/nvme/target/tcp.c
@@ -422,6 +422,19 @@ static int nvmet_tcp_map_data(struct nvm
 	if (!len)
 		return 0;
 
+	/*
+	 * inline_data_size only bounds the in-capsule (type 0x01) SGL
+	 * descriptor below. A non-inline transport SGL data-block
+	 * descriptor skips that check entirely and would otherwise reach
+	 * sgl_alloc() with an attacker-controlled len of up to 4 GiB,
+	 * pinning that much kernel memory for a command that may never
+	 * complete. Bound every descriptor type here, before allocating
+	 * anything, using the same ceiling this file already applies to
+	 * per-PDU H2C data.
+	 */
+	if (len > NVMET_TCP_MAXH2CDATA)
+		return NVME_SC_SGL_INVALID_DATA | NVME_STATUS_DNR;
+
 	if (sgl->type == ((NVME_SGL_FMT_DATA_DESC << 4) |
 			  NVME_SGL_FMT_OFFSET)) {
 		if (!nvme_is_write(cmd->req.cmd))