External/passt
1
0
Fork 0
mirror of https://passt.top/passt synced 2024-07-03 00:12:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

tcp: Limit TCP_INFO getsockopt() syscalls

Browse Source 
There's no need to constantly query the socket for number of
acknowledged bytes if we're far from exhausting the sending window,
just do it if we're at least down to 90% of it.

Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
This commit is contained in:
Stefano Brivio 2021-07-27 00:50:53 +02:00
parent d372c42460
commit fd5050ccba
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
tcp.c
View File

@ -1075,13 +1075,14 @@ static void tcp_rst(struct ctx *c, struct tcp_tap_conn *conn);
static int tcp_send_to_tap(struct ctx *c, struct tcp_tap_conn *conn,
int flags, char *in, int len)
{
uint32_t ack_offset = conn->seq_from_tap - conn->seq_ack_to_tap;
char buf[USHRT_MAX] = { 0 }, *data;
struct tcp_info info = { 0 };
socklen_t sl = sizeof(info);
struct tcphdr *th;
int ws = 0, err;
if (conn->seq_from_tap == conn->seq_ack_to_tap && !flags && len) {
if (ack_offset < conn->tcpi_snd_wnd / 10 && !flags) {
err = 0;
info.tcpi_bytes_acked = conn->tcpi_acked_last;
info.tcpi_snd_wnd = conn->tcpi_snd_wnd;