cvs commit: src/sys/netinet tcp_var.h

Bjoern A. Zeeb bz at FreeBSD.org
Sun Apr 20 18:50:07 UTC 2008


On Sun, 20 Apr 2008, Bjoern A. Zeeb wrote:

> bz          2008-04-20 18:36:59 UTC
>
>  FreeBSD src repository
>
>  Modified files:
>    sys/netinet          tcp_var.h
>  Log:
>  Revert to rev. 1.161 - switch back to optimized TCP options ordering.
>
>  A lot of testing has shown that the problem people were seeing was due
>  to invalid padding after the end of option list option, which was corrected
>  in tcp_output.c rev. 1.146.
>
>  Thanks to:              anders@, s3raphi, Matt Reimer
>  Thanks to:              Doug Hardie and Randy Rose, John Mayer, Susan Guzzardi
>  Special thanks to:      dwhite@ and BitGravity
>  Discussed with:         silby
>  MFC after:              1 day
>
>  Revision  Changes    Path
>  1.162     +2 -5      src/sys/netinet/tcp_var.h

A few more words on this:

1) the short MFC time is because the past has shown that HEAD does not
    uncover any problems with changes like this and we'd like to move
    towards an errata for this so RELENG_7 exposure time should be as
    long as possible.

2) With all the testing I have only seen 1 client having problems with
    the optimized ordering but that client also had other (connectivity)
    problems so I was told to ignore that one.

3) I have seen clients failing with the RELENG_6 (now old again)
    option ordering as well.

4) The cable modems/routers blamed before in any of the threads
    actually worked correctly not accepting a TCP packet with bad
    padding violating the TCP RFC.


Lats but not least, I hope to be able to find the time to further summarize
the problems seen, the things discovered, etc. at a later time.

I have yet to look at another few strange TCP connections found during
all of the testing (some of which seem to have hit net@ or stable@ the
last days as well.)


/bz

-- 
Bjoern A. Zeeb              Stop bit received. Insert coin for new game.


More information about the cvs-src mailing list