kern/175153: will there miss a FIN when do TSO?
Zhouyi Zhou
yizhouzhou at ict.ac.cn
Wed Jan 9 06:10:02 UTC 2013
>Number: 175153
>Category: kern
>Synopsis: will there miss a FIN when do TSO?
>Confidential: no
>Severity: non-critical
>Priority: low
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Wed Jan 09 06:10:00 UTC 2013
>Closed-Date:
>Last-Modified:
>Originator: Zhouyi Zhou
>Release: FreeBSD current
>Organization:
ICT CAS
>Environment:
FreeBSD host48 8.2-RELEASE FreeBSD 8.2-RELEASE root at host48:/usr/src/sys.orig/amd64/compile/GENERIC
>Description:
Hi,
I guess there will miss a FIN when do TSO in function tcp_output, following
is the code piece from CVS:
/*
* Adjust data length if insertion of options will
* bump the packet length beyond the t_maxopd length.
* Clear the FIN bit because we cut off the tail of
* the segment.
*/
if (len + optlen + ipoptlen > tp->t_maxopd) {
flags &= ~TH_FIN;
if (tso) {
....
if (tp->t_flags & TF_NEEDFIN)
sendalot = 1;
///////////////////////////////////////////////////////
My question is when the FreeBSD close the established tcp connection actively,
the TF_NEEDFIN will not be set (am I got wrong?), and when previous computation
has not set sendalot to 1, the the tsoed packet will be the last packet to send
out in this round, so a FIN is missed.
Thanks for your time
>How-To-Repeat:
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted:
More information about the freebsd-bugs
mailing list