>No Andrea, I think Philip is completely right.
Probably yes but look at this part of the trace:
55.573090 gwyn > riesz.: FP 258421:259687(1266) ack 1 win 32120 (DF) [tos 0x8]
Linux sent the last packet full of data with FIN set for the first time.
But if I remeber the TCP right (note: I could be wrong, but I really don't
have the time to check the specs now :( ), the sequence number of the data
should be 259686 and the sequence number 259687 is the one for taking care
of the FIN.
55.870201 riesz. > gwyn: . ack 259687 win 24820 (DF)
And here Solaris is acking also the FIN. Maybe I am wrong.
>I concede, nobody wants to fix Solaris, everyone wishes to code
>workarounds for other systems until they are blue in the face.
Agreed.
> Andrea's patch was "INSTALL THIS ON ALL YOUR FTP HTTP SERVERS"
> without one mention of "people with solaris systems should install
I agree. I tried to expose the thing as more clear as possible in my first
email just to avoid all possible mistakes. I quote myself again:
-------------------------------------------------------------------
[..]
Solaris has a stupid bug in its TCP stack that cause it to hang the
connection between linux-2.2.x and Solaris (with linux as server) once the
connection closes.
[..]
Sun just released a fix for Solaris but since I think there are still many
buggy Solaris 2.6 and 2.5.1 TCP stacks on the Internet, I taken the time
to implement a workaround for Linux.
[..]
Just to avoid confusion I repeat: this is a silly Solaris bug and not a
Linux bug.
[..]
-------------------------------------------------------------------
>So I'll put the hack in some form into 2.2.3, even though I still
>think this is all a crock.
I think my form is the cleaner and the shorter (zero crap) and
exactly/only for this reason I think it could be applyed to the official
kernel. I really don't think it worth to waste more time for a workaround
of a silly Solaris bug adding more code to _smart_workaround_ it (note
also that my sysctl has zero performance inpact), and I don't like that
linux TCP stack will retransmit different packets when we guess that the
other end could have been confused by us. So I take the way to avoid
confusion in first place and I still like it.
Andrea Arcangeli
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/