nvmet-tcp: fix out-of-bounds write when receiving an over-long PDU
Subsystems (3)
PATCH 1/1 nvmet-tcp: fix out-of-bounds write when receiving an over-long PDU
In reply to a message not in the archivenvmet_tcp_try_recv_pdu() reads a PDU header into the fixed 128-byte queue->pdu union, then computes the remaining payload length as
queue->left = hdr->hlen - queue->offset + hdgst;
and reads that many more bytes into &queue->pdu + queue->offset, without ever bounding the result against sizeof(queue->pdu).
A struct nvme_tcp_icreq_pdu is itself 128 bytes, exactly the size of the union. Once a header digest has been negotiated (hdgst = 4), a second ICReq passes the hlen == nvmet_tcp_pdu_size() check but yields queue->left = 128 - 8 + 4 = 124, so bytes 8..132 are written into the 128-byte buffer -- 4 bytes past its end, over queue->hdr_digest and queue->data_digest. Those bytes are attacker-controlled (an ICReq carries no digest), and the duplicate ICReq is only rejected later, after the overflow. A remote unauthenticated host can thus corrupt kernel memory adjacent to the receive buffer.
Reject any PDU whose declared length would read past the end of queue->pdu before the second recv.
--- drivers/nvme/target/tcp.c | 2 ++ 1 file changed, 2 insertions(+)
drivers/nvme/target/tcp.c +2 -0
@@ -1229,6 +1229,8 @@ static int nvmet_tcp_try_recv_pdu(struct nvmet_tcp_queue *queue) } queue->left = hdr->hlen - queue->offset + hdgst; +if (queue->left >sizeof (queue->pdu) - queue->offset) +return -EPROTO ; goto recv; }
Signature · 1 line
2.53.0
Quoted 44 lines · Shivam Kumar
nvmet_tcp_try_recv_pdu()reads a PDU header into the fixed 128-bytequeue->pduunion, then computes the remaining payload length asqueue->left = hdr->hlen - queue->offset + hdgst;and reads that many more bytes into
&queue->pdu+queue->offset, without ever bounding the result againstsizeof(queue->pdu).A struct
nvme_tcp_icreq_pduis itself 128 bytes, exactly the size of the union. Once a header digest has been negotiated (hdgst = 4), a second ICReq passes the hlen ==nvmet_tcp_pdu_size()check but yieldsqueue->left= 128 - 8 + 4 = 124, so bytes 8..132 are written into the 128-byte buffer -- 4 bytes past its end, overqueue->hdr_digestandqueue->data_digest. Those bytes are attacker-controlled (an ICReq carries no digest), and the duplicate ICReq is only rejected later, after the overflow. A remote unauthenticated host can thus corrupt kernel memory adjacent to the receive buffer.Reject any PDU whose declared length would read past the end of
queue->pdubefore the second recv.Fixes 872d26a391da ("nvmet-tcp: add NVMe over TCP target driver")Assisted-by Claude:claude-opus-4-8Signed-off-by Shivam Kumar <[email protected]>--- drivers/nvme/target/tcp.c | 2 ++ 1 file changed, 2 insertions(+)drivers/nvme/target/tcp.c +2
-1@@ -1229,6 +1229,8 @@ static int nvmet_tcp_try_recv_pdu(struct nvmet_tcp_queue *queue) } queue->left = hdr->hlen - queue->offset + hdgst; +if (queue->left >sizeof (queue->pdu) - queue->offset) +return -EPROTO ; goto recv; } --
Quoted 2 lines
2.53.0
Gentle ping on this one.
Quoted 18 lines · Shivam Kumar
nvmet_tcp_try_recv_pdu()reads a PDU header into the fixed 128-bytequeue->pduunion, then computes the remaining payload length asqueue->left = hdr->hlen - queue->offset + hdgst;and reads that many more bytes into
&queue->pdu+queue->offset, without ever bounding the result againstsizeof(queue->pdu).A struct
nvme_tcp_icreq_pduis itself 128 bytes, exactly the size of the union. Once a header digest has been negotiated (hdgst = 4), a second ICReq passes the hlen ==nvmet_tcp_pdu_size()check but yieldsqueue->left= 128 - 8 + 4 = 124, so bytes 8..132 are written into the 128-byte buffer -- 4 bytes past its end, overqueue->hdr_digestandqueue->data_digest. Those bytes are attacker-controlled (an ICReq carries no digest), and the duplicate ICReq is only rejected later, after the overflow. A remote unauthenticated host can thus corrupt kernel memory adjacent to the receive buffer.
Quoted 2 lines
Reject any PDU whose declared length would read past the end of
queue->pdubefore the second recv.
Thanks, applied to nvme-7.3.