[PATCH net-next] net-timestamp: support TCP GSO case for a few missing flags

[Jason Xing <kerneljasonxing@gmail.com>]

When I read through the TSO codes, I found out that we probably
miss initializing the tx_flags of last seg when TSO is turned
off, which means at the following points no more timestamp
(for this last one) will be generated. There are three flags
to be handled in this patch:
1. SKBTX_HW_TSTAMP
2. SKBTX_HW_TSTAMP_USE_CYCLES
3. SKBTX_BPF

This patch initializes the tx_flags to SKBTX_ANY_TSTAMP like what
the UDP GSO does. But flag like SKBTX_SCHED_TSTAMP is not useful
and will not be used in the remaining path since the skb has already
passed the QDisc layer.

Signed-off-by: Jason Xing <kerneljasonxing@gmail.com>
---
 net/ipv4/tcp_offload.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/net/ipv4/tcp_offload.c b/net/ipv4/tcp_offload.c
index 2308665b51c5..886582002425 100644
--- a/net/ipv4/tcp_offload.c
+++ b/net/ipv4/tcp_offload.c
@@ -13,12 +13,15 @@
 #include <net/tcp.h>
 #include <net/protocol.h>
 
-static void tcp_gso_tstamp(struct sk_buff *skb, unsigned int ts_seq,
+static void tcp_gso_tstamp(struct sk_buff *skb, struct sk_buff *gso_skb,
 			   unsigned int seq, unsigned int mss)
 {
+	u32 flags = skb_shinfo(gso_skb)->tx_flags & SKBTX_ANY_TSTAMP;
+	u32 ts_seq = skb_shinfo(gso_skb)->tskey;
+
 	while (skb) {
 		if (before(ts_seq, seq + mss)) {
-			skb_shinfo(skb)->tx_flags |= SKBTX_SW_TSTAMP;
+			skb_shinfo(skb)->tx_flags |= flags;
 			skb_shinfo(skb)->tskey = ts_seq;
 			return;
 		}
@@ -193,8 +196,8 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
 	th = tcp_hdr(skb);
 	seq = ntohl(th->seq);
 
-	if (unlikely(skb_shinfo(gso_skb)->tx_flags & SKBTX_SW_TSTAMP))
-		tcp_gso_tstamp(segs, skb_shinfo(gso_skb)->tskey, seq, mss);
+	if (unlikely(skb_shinfo(gso_skb)->tx_flags & (SKBTX_ANY_TSTAMP)))
+		tcp_gso_tstamp(segs, gso_skb, seq, mss);
 
 	newcheck = ~csum_fold(csum_add(csum_unfold(th->check), delta));
 
-- 
2.43.5