From 7c2ab1d9f6ff94b01027808fde17933a54ea8c82 Mon Sep 17 00:00:00 2001
From: hselasky <hselasky@FreeBSD.org>
Date: Wed, 6 Sep 2017 13:56:18 +0000
Subject: [PATCH] Add support for generic backpressure indicator for
 ratelimited transmit queues aswell as non-ratelimited ones.

Add the required structure bits in order to support a backpressure
indication with ratelimited connections aswell as non-ratelimited
ones. The backpressure indicator is a value between zero and 65535
inclusivly, indicating if the destination transmit queue is empty or
full respectivly. Applications can use this value as a decision point
for when to stop transmitting data to avoid endless ENOBUFS error
codes upon transmitting an mbuf. This indicator is also useful to
reduce the latency for ratelimited queues.

Reviewed by:		gallatin, kib, gnn
Differential Revision:	https://reviews.freebsd.org/D11518
Sponsored by:		Mellanox Technologies
---
 sys/net/if_var.h     | 10 +++++++++-
 sys/netinet/in_pcb.c | 32 +++++++++++++++++++++++++++++++-
 sys/netinet/in_pcb.h |  1 +
 3 files changed, 41 insertions(+), 2 deletions(-)

diff --git a/sys/net/if_var.h b/sys/net/if_var.h
index d4d2ddceeff1..e1f8a21c4611 100644
--- a/sys/net/if_var.h
+++ b/sys/net/if_var.h
@@ -183,7 +183,8 @@ struct if_encap_req {
 struct m_snd_tag;
 
 #define	IF_SND_TAG_TYPE_RATE_LIMIT 0
-#define	IF_SND_TAG_TYPE_MAX 1
+#define	IF_SND_TAG_TYPE_UNLIMITED 1
+#define	IF_SND_TAG_TYPE_MAX 2
 
 struct if_snd_tag_alloc_header {
 	uint32_t type;		/* send tag type, see IF_SND_TAG_XXX */
@@ -198,19 +199,26 @@ struct if_snd_tag_alloc_rate_limit {
 
 struct if_snd_tag_rate_limit_params {
 	uint64_t max_rate;	/* in bytes/s */
+	uint32_t queue_level;	/* 0 (empty) .. 65535 (full) */
+#define	IF_SND_QUEUE_LEVEL_MIN 0
+#define	IF_SND_QUEUE_LEVEL_MAX 65535
+	uint32_t reserved;	/* padding */
 };
 
 union if_snd_tag_alloc_params {
 	struct if_snd_tag_alloc_header hdr;
 	struct if_snd_tag_alloc_rate_limit rate_limit;
+	struct if_snd_tag_alloc_rate_limit unlimited;
 };
 
 union if_snd_tag_modify_params {
 	struct if_snd_tag_rate_limit_params rate_limit;
+	struct if_snd_tag_rate_limit_params unlimited;
 };
 
 union if_snd_tag_query_params {
 	struct if_snd_tag_rate_limit_params rate_limit;
+	struct if_snd_tag_rate_limit_params unlimited;
 };
 
 typedef int (if_snd_tag_alloc_t)(struct ifnet *, union if_snd_tag_alloc_params *,
diff --git a/sys/netinet/in_pcb.c b/sys/netinet/in_pcb.c
index 3df59e85fdf1..4292ac348571 100644
--- a/sys/netinet/in_pcb.c
+++ b/sys/netinet/in_pcb.c
@@ -2796,6 +2796,35 @@ in_pcbquery_txrtlmt(struct inpcb *inp, uint32_t *p_max_pacing_rate)
 	return (error);
 }
 
+/*
+ * Query existing TX queue level based on the existing
+ * "inp->inp_snd_tag", if any.
+ */
+int
+in_pcbquery_txrlevel(struct inpcb *inp, uint32_t *p_txqueue_level)
+{
+	union if_snd_tag_query_params params = { };
+	struct m_snd_tag *mst;
+	struct ifnet *ifp;
+	int error;
+
+	mst = inp->inp_snd_tag;
+	if (mst == NULL)
+		return (EINVAL);
+
+	ifp = mst->ifp;
+	if (ifp == NULL)
+		return (EINVAL);
+
+	if (ifp->if_snd_tag_query == NULL)
+		return (EOPNOTSUPP);
+
+	error = ifp->if_snd_tag_query(mst, &params);
+	if (error == 0 &&  p_txqueue_level != NULL)
+		*p_txqueue_level = params.rate_limit.queue_level;
+	return (error);
+}
+
 /*
  * Allocate a new TX rate limit send tag from the network interface
  * given by the "ifp" argument and save it in "inp->inp_snd_tag":
@@ -2805,7 +2834,8 @@ in_pcbattach_txrtlmt(struct inpcb *inp, struct ifnet *ifp,
     uint32_t flowtype, uint32_t flowid, uint32_t max_pacing_rate)
 {
 	union if_snd_tag_alloc_params params = {
-		.rate_limit.hdr.type = IF_SND_TAG_TYPE_RATE_LIMIT,
+		.rate_limit.hdr.type = (max_pacing_rate == -1U) ?
+		    IF_SND_TAG_TYPE_UNLIMITED : IF_SND_TAG_TYPE_RATE_LIMIT,
 		.rate_limit.hdr.flowid = flowid,
 		.rate_limit.hdr.flowtype = flowtype,
 		.rate_limit.max_rate = max_pacing_rate,
diff --git a/sys/netinet/in_pcb.h b/sys/netinet/in_pcb.h
index 42fd23d0b572..059ea90c17a0 100644
--- a/sys/netinet/in_pcb.h
+++ b/sys/netinet/in_pcb.h
@@ -761,6 +761,7 @@ int	in_pcbattach_txrtlmt(struct inpcb *, struct ifnet *, uint32_t, uint32_t, uin
 void	in_pcbdetach_txrtlmt(struct inpcb *);
 int	in_pcbmodify_txrtlmt(struct inpcb *, uint32_t);
 int	in_pcbquery_txrtlmt(struct inpcb *, uint32_t *);
+int	in_pcbquery_txrlevel(struct inpcb *, uint32_t *);
 void	in_pcboutput_txrtlmt(struct inpcb *, struct ifnet *, struct mbuf *);
 void	in_pcboutput_eagain(struct inpcb *);
 #endif