Sabrina Dubroca [off-list ref] writes:

2024-07-14, 16:22:36 -0400, Christian Hopps wrote:

quoted

+struct xfrm_iptfs_config {
+	u32 pkt_size;	    /* outer_packet_size or 0 */

Please convert this to kdoc.

Done.

quoted

+};
+
+struct xfrm_iptfs_data {
+	struct xfrm_iptfs_config cfg;
+
+	/* Ingress User Input */
+	struct xfrm_state *x;	    /* owning state */

And this too.

Done.

quoted

+	u32 payload_mtu;	    /* max payload size */
+};

quoted

+static int iptfs_create_state(struct xfrm_state *x)
+{
+	struct xfrm_iptfs_data *xtfs;
+	int err;
+
+	xtfs = kzalloc(sizeof(*xtfs), GFP_KERNEL);
+	if (!xtfs)
+		return -ENOMEM;
+
+	err = __iptfs_init_state(x, xtfs);
+	if (err)
+		return err;

BTW, I wrote that this was leaking xtfs in my previous review, back in
March :/

I went back through your earlier review again. The next patch set adds the queue flush/cleanup on state delete, and uses a new MIB direction independent MIB counter for failed skb alloc -- which you identified as well.

Thanks,
Chris.