diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
index c13cda5..46a785c 100644
--- a/drivers/block/ublk_drv.c
+++ b/drivers/block/ublk_drv.c
@@ -3255,7 +3255,8 @@ static int ublk_ch_uring_cmd_local(struct io_uring_cmd *cmd,
 		unsigned int issue_flags)
 {
 	/* May point to userspace-mapped memory */
-	const struct ublksrv_io_cmd *ub_src = io_uring_sqe_cmd(cmd->sqe);
+	const struct ublksrv_io_cmd *ub_src = io_uring_sqe_cmd(cmd->sqe,
+							       struct ublksrv_io_cmd);
 	u16 buf_idx = UBLK_INVALID_BUF_IDX;
 	struct ublk_device *ub = cmd->file->private_data;
 	struct ublk_queue *ubq;
@@ -3833,7 +3834,8 @@ static int ublk_validate_batch_fetch_cmd(struct ublk_batch_io_data *data)
 static int ublk_handle_non_batch_cmd(struct io_uring_cmd *cmd,
 				     unsigned int issue_flags)
 {
-	const struct ublksrv_io_cmd *ub_cmd = io_uring_sqe_cmd(cmd->sqe);
+	const struct ublksrv_io_cmd *ub_cmd = io_uring_sqe_cmd(cmd->sqe,
+							       struct ublksrv_io_cmd);
 	struct ublk_device *ub = cmd->file->private_data;
 	unsigned tag = READ_ONCE(ub_cmd->tag);
 	unsigned q_id = READ_ONCE(ub_cmd->q_id);
@@ -3862,7 +3864,8 @@ static int ublk_handle_non_batch_cmd(struct io_uring_cmd *cmd,
 static int ublk_ch_batch_io_uring_cmd(struct io_uring_cmd *cmd,
 				       unsigned int issue_flags)
 {
-	const struct ublk_batch_io *uc = io_uring_sqe_cmd(cmd->sqe);
+	const struct ublk_batch_io *uc = io_uring_sqe_cmd(cmd->sqe,
+							  struct ublk_batch_io);
 	struct ublk_device *ub = cmd->file->private_data;
 	struct ublk_batch_io_data data = {
 		.ub  = ub,
@@ -5253,7 +5256,8 @@ static int ublk_ctrl_uring_cmd(struct io_uring_cmd *cmd,
 		unsigned int issue_flags)
 {
 	/* May point to userspace-mapped memory */
-	const struct ublksrv_ctrl_cmd *ub_src = io_uring_sqe_cmd(cmd->sqe);
+	const struct ublksrv_ctrl_cmd *ub_src = io_uring_sqe128_cmd(cmd->sqe,
+								    struct ublksrv_ctrl_cmd);
 	struct ublksrv_ctrl_cmd header;
 	struct ublk_device *ub = NULL;
 	u32 cmd_op = cmd->cmd_op;
diff --git a/drivers/nvme/host/ioctl.c b/drivers/nvme/host/ioctl.c
index fb62633..8844bbd 100644
--- a/drivers/nvme/host/ioctl.c
+++ b/drivers/nvme/host/ioctl.c
@@ -447,7 +447,8 @@ static int nvme_uring_cmd_io(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 		struct io_uring_cmd *ioucmd, unsigned int issue_flags, bool vec)
 {
 	struct nvme_uring_cmd_pdu *pdu = nvme_uring_cmd_pdu(ioucmd);
-	const struct nvme_uring_cmd *cmd = io_uring_sqe_cmd(ioucmd->sqe);
+	const struct nvme_uring_cmd *cmd = io_uring_sqe128_cmd(ioucmd->sqe,
+							       struct nvme_uring_cmd);
 	struct request_queue *q = ns ? ns->queue : ctrl->admin_q;
 	struct nvme_uring_data d;
 	struct nvme_command c;
diff --git a/fs/fuse/dev_uring.c b/fs/fuse/dev_uring.c
index 5ceb217..60f2058 100644
--- a/fs/fuse/dev_uring.c
+++ b/fs/fuse/dev_uring.c
@@ -879,7 +879,8 @@ static int fuse_ring_ent_set_commit(struct fuse_ring_ent *ent)
 static int fuse_uring_commit_fetch(struct io_uring_cmd *cmd, int issue_flags,
 				   struct fuse_conn *fc)
 {
-	const struct fuse_uring_cmd_req *cmd_req = io_uring_sqe_cmd(cmd->sqe);
+	const struct fuse_uring_cmd_req *cmd_req = io_uring_sqe128_cmd(cmd->sqe,
+								       struct fuse_uring_cmd_req);
 	struct fuse_ring_ent *ent;
 	int err;
 	struct fuse_ring *ring = fc->ring;
@@ -1083,7 +1084,8 @@ fuse_uring_create_ring_ent(struct io_uring_cmd *cmd,
 static int fuse_uring_register(struct io_uring_cmd *cmd,
 			       unsigned int issue_flags, struct fuse_conn *fc)
 {
-	const struct fuse_uring_cmd_req *cmd_req = io_uring_sqe_cmd(cmd->sqe);
+	const struct fuse_uring_cmd_req *cmd_req = io_uring_sqe128_cmd(cmd->sqe,
+								       struct fuse_uring_cmd_req);
 	struct fuse_ring *ring = smp_load_acquire(&fc->ring);
 	struct fuse_ring_queue *queue;
 	struct fuse_ring_ent *ent;
diff --git a/include/linux/io_uring/cmd.h b/include/linux/io_uring/cmd.h
index 375fd04..331dcbe 100644
--- a/include/linux/io_uring/cmd.h
+++ b/include/linux/io_uring/cmd.h
@@ -20,10 +20,17 @@ struct io_uring_cmd {
 	u8		unused[8];
 };
 
-static inline const void *io_uring_sqe_cmd(const struct io_uring_sqe *sqe)
-{
-	return sqe->cmd;
-}
+#define io_uring_sqe128_cmd(sqe, type)	({					\
+	BUILD_BUG_ON(sizeof(type) > ((2 * sizeof(struct io_uring_sqe)) -	\
+				     offsetof(struct io_uring_sqe, cmd)));	\
+	(const type *)(sqe)->cmd;						\
+})
+
+#define io_uring_sqe_cmd(sqe, type)	({					\
+	BUILD_BUG_ON(sizeof(type) > (sizeof(struct io_uring_sqe) -		\
+				     offsetof(struct io_uring_sqe, cmd)));	\
+	(const type *)(sqe)->cmd;						\
+})
 
 static inline void io_uring_cmd_private_sz_check(size_t cmd_sz)
 {
diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h
index 503663d..0fa844f 100644
--- a/io_uring/io_uring.h
+++ b/io_uring/io_uring.h
@@ -530,6 +530,12 @@ static inline bool io_file_can_poll(struct io_kiocb *req)
 	return false;
 }
 
+static inline bool io_is_uring_cmd(const struct io_kiocb *req)
+{
+	return req->opcode == IORING_OP_URING_CMD ||
+	       req->opcode == IORING_OP_URING_CMD128;
+}
+
 static inline ktime_t io_get_time(struct io_ring_ctx *ctx)
 {
 	if (ctx->clockid == CLOCK_MONOTONIC)
diff --git a/io_uring/kbuf.c b/io_uring/kbuf.c
index 67d4fe5..dae5b4a 100644
--- a/io_uring/kbuf.c
+++ b/io_uring/kbuf.c
@@ -171,7 +171,7 @@ static bool io_should_commit(struct io_kiocb *req, unsigned int issue_flags)
 		return true;
 
 	/* uring_cmd commits kbuf upfront, no need to auto-commit */
-	if (!io_file_can_poll(req) && req->opcode != IORING_OP_URING_CMD)
+	if (!io_file_can_poll(req) && !io_is_uring_cmd(req))
 		return true;
 	return false;
 }
diff --git a/io_uring/rw.c b/io_uring/rw.c
index b397117..1a5f262 100644
--- a/io_uring/rw.c
+++ b/io_uring/rw.c
@@ -1254,7 +1254,7 @@ static int io_uring_classic_poll(struct io_kiocb *req, struct io_comp_batch *iob
 {
 	struct file *file = req->file;
 
-	if (req->opcode == IORING_OP_URING_CMD) {
+	if (io_is_uring_cmd(req)) {
 		struct io_uring_cmd *ioucmd;
 
 		ioucmd = io_kiocb_to_cmd(req, struct io_uring_cmd);
@@ -1380,7 +1380,7 @@ int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin)
 		wq_list_add_tail(&req->comp_list, &ctx->submit_state.compl_reqs);
 		nr_events++;
 		req->cqe.flags = io_put_kbuf(req, req->cqe.res, NULL);
-		if (req->opcode != IORING_OP_URING_CMD)
+		if (!io_is_uring_cmd(req))
 			io_req_rw_cleanup(req, 0);
 	}
 	if (nr_events)
diff --git a/io_uring/zcrx.c b/io_uring/zcrx.c
index 28150c6..97984a7 100644
--- a/io_uring/zcrx.c
+++ b/io_uring/zcrx.c
@@ -349,10 +349,14 @@ static inline atomic_t *io_get_user_counter(struct net_iov *niov)
 static bool io_zcrx_put_niov_uref(struct net_iov *niov)
 {
 	atomic_t *uref = io_get_user_counter(niov);
+	int old;
+
+	old = atomic_read(uref);
+	do {
+		if (unlikely(old == 0))
+			return false;
+	} while (!atomic_try_cmpxchg(uref, &old, old - 1));
 
-	if (unlikely(!atomic_read(uref)))
-		return false;
-	atomic_dec(uref);
 	return true;
 }
 
