lupyuen/local-nuttx-update
1
0
Fork 0
forked from nuttx/nuttx-update
Code Pull requests Activity

tcp_get_recvwindow: Make this match the reality (tcp_datahandler)

Browse source 
It doesn't make much sense to advertize more than we can actually
accept, especially when we don't queue partial segments.
This commit is contained in:
YAMAMOTO Takashi 2021-06-08 10:54:04 +09:00 committed by Xiang Xiao
parent af64912833
commit ca0f2bdb95
1 changed files with 8 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

14
net/tcp/tcp_recvwindow.c
View file

@ -143,6 +143,7 @@ uint16_t tcp_get_recvwindow(FAR struct net_driver_s *dev,
recvwndo = (uint16_t)rwnd;
}
#if CONFIG_IOB_THROTTLE > 0
else if (IOB_QEMPTY(&conn->readahead))
{
/* Advertise maximum segment size for window edge if here is no
@ -150,16 +151,17 @@ uint16_t tcp_get_recvwindow(FAR struct net_driver_s *dev,
*
* Note: hopefully, a single mss-sized packet can be queued by
* the throttled=false case in tcp_datahandler().
*
* Revisit: I (yamamoto) am a bit skeptical on this logic.
* - Probably this should depend on IOB throttling.
* - IOB_BUFSIZE can be smaller than MSS. It seems wrong to
* advertize a larger window when no IOBs are available
* than when eg. a single IOB is available.
*/
int niob_avail_no_throttle = iob_navail(false);
recvwndo = tcp_rx_mss(dev);
if (recvwndo > niob_avail_no_throttle * CONFIG_IOB_BUFSIZE)
{
recvwndo = niob_avail_no_throttle * CONFIG_IOB_BUFSIZE;
}
}
#endif
else /* nqentry_avail == 0 || niob_avail == 0 */
{
/* No IOB chains or noIOBs are available.