JehanneOS/newlib
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

* net.cc (fdsock): Change default values for socket buffers on 32 bit

Browse Source 
to fix performance on 10Gb networks.
This commit is contained in:
Corinna Vinschen 2014-11-20 08:29:31 +00:00
parent 5c6a0f3117
commit f70b9c32d5
2 changed files with 14 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
winsup/cygwin/ChangeLog
View File

@ -1,3 +1,8 @@
2014-11-20 Iuliu Rus <rus.iuliu@gmail.com>
* net.cc (fdsock): Change default values for socket buffers on 32 bit
to fix performance on 10Gb networks.
2014-11-17 Corinna Vinschen <corinna@vinschen.de>
* uinfo.cc (pwdgrp::fetch_account_from_windows): Allow fetching of

11
winsup/cygwin/net.cc
View File

@ -621,13 +621,20 @@ fdsock (cygheap_fdmanip& fd, const device *dev, SOCKET soc)
this is no problem on 64 bit. So we set the default buffer size to
the default values in current 3.x Linux versions.
NOTE 3. Setting the window size to 65535 results in extremely bad
performance for apps that send data in multiples of Kb, as they
eventually end up sending 1 byte on the network and naggle + delay
ack kicks in. For example, iperf on a 10Gb network gives only 10
Mbits/sec with a 65535 send buffer. We want this to be a multiple
of 1k, but since 64k breaks WSADuplicateSocket we use 63Kb.
(*) Maximum normal TCP window size. Coincidence? */
#ifdef __x86_64__
((fhandler_socket *) fd)->rmem () = 212992;
((fhandler_socket *) fd)->wmem () = 212992;
#else
((fhandler_socket *) fd)->rmem () = 65535;
((fhandler_socket *) fd)->wmem () = 65535;
((fhandler_socket *) fd)->rmem () = 64512;
((fhandler_socket *) fd)->wmem () = 64512;
#endif
if (::setsockopt (soc, SOL_SOCKET, SO_RCVBUF,
(char *) &((fhandler_socket *) fd)->rmem (), sizeof (int)))