tcp/ip stack design/implementation in solaris 9

[Sam]

Dear all,

I would like to know more about the tcp/ip stack design and
implementation of solaris 9. Can anyone direct me to some online
reference/material?

thanks
sam

[Robert Pouliot]

From what I remember the TCP/IP stack of Solaris was developper by
Mentat and with probably quite a bit of tunning by Sun.
At least they didn't break their DLPI implementation as badly as
SCO did. See http://www.mentat.com/ that may answer many of
your questions.

--
If rabbit's feet are so lucky, what happened to the rabbit?
http://www.krynos.org/ - ICQ UIN: 2295979

[Andrew Gabriel]

In article <slrnblqh6t...@huxley.krynos.dyn.saturnus.com>,

Robert Pouliot <kry...@huxley.krynos.dyn.saturnus.com> writes:
> In article <3F5D4122...@thales-is.com.hk>, Sam wrote:
>> Dear all,
>>
>> I would like to know more about the tcp/ip stack design and
>> implementation of solaris 9. Can anyone direct me to some online
>> reference/material?

For what purpose?
This would give some clue what type of information you need.

> From what I remember the TCP/IP stack of Solaris was developper by
> Mentat and with probably quite a bit of tunning by Sun.

It was originally based on the Mentat stack, but Mentat wouldn't
recognise it now -- it's one of the things that gets constantly
worked on, and has changed very considerably over the years.

> At least they didn't break their DLPI implementation as badly as
> SCO did. See http://www.mentat.com/ that may answer many of
> your questions.

SCO's TCP/IP stack came from Lackman (both SVR3.2 Openserver and
SVR4 Unixware). However, I believe the completely bust DLPI in
Unixware was a creation of Novell. The original aim was for it to
'transliterate' Novell network driver binaries, so they didn't have
to write any for Unixware, but it didn't work -- they sort of got
it limping along for the TCP/IP stack, but pretty much no other
applications ever worked with it. I believe the scheme was ditched
in later Unixware, but most people had ditched Unixware by then,
so it hardly mattered anymore.

I don't remember who did the LLI (the DLPI predecessor in SVR3.2)
-- it could have been Lackman too (Interactive UNIX used the same
Lackman TCP/IP as SCO SVR3.2, and they both had the same LLI which
is why I guess this to be the case).

--
Andrew Gabriel

[ultra...@hotmail.com]

In <bjpe6t$ah2$1...@new-usenet.uk.sun.com> and...@cucumber.demon.co.uk (Andrew Gabriel) writes:
>SCO's TCP/IP stack came from Lackman (both SVR3.2 Openserver and

...

>Lackman TCP/IP as SCO SVR3.2, and they both had the same LLI which

Lachman. http://www.lachmangoldman.com/people.html