[9fans] Is the "cpu" command using version 1 of p9any?

[9fans] Is the "cpu" command using version 1 of p9any?

[Alberto Cortés]

Authsrv(6) sais cpu(1) uses "p9any" and also sais that "p9any
version 1" is no longer used.

I had snooped on a cpu connection, here is the payload of the
first 4 (data) tcp packets:


(1)  C->S  p9 rc4_256 sha1
(2)  S->C  (null char here)
(3)  S->C  p9sk1@my.auth.domain
(4)  C->S  p9sk1 my.auth.domain


(1) and (2) are part of an specific negotiation of cpu clients
and servers.

(3) and (4) look like a "p9any version 1" negotiation. I presume
these packets are due to /sys/src/cmd/cpu.c:524, a call to
auth_proxy(2).

I believe the above samples comes from calling auth_proxy with
"proto=p9any role=client".

Maybe cpu is using "p9any version 1" or there is an errata in
authsrv(6).


PS: The samples are from a tcp connection to 17010 port; on my server:

    cpu% cat /rc/bin/serice/tcp17010
    #!/bin/cpu -R
    cpu%

-- 
Alberto Cortés

Re: [9fans] Is the "cpu" command using version 1 of p9any?

[Alberto Cortés]

Alberto Cortés said:

> (3) and (4) look like a "p9any version 1" negotiation. I presume
> these packets are due to /sys/src/cmd/cpu.c:524, a call to
> auth_proxy(2).

Sorry I meant:

    /sys/src/cmd/cpu.c:588

> I believe the above samples comes from calling auth_proxy with
> "proto=p9any role=client".

and
    "proto=p9any role=server"


-- 
Alberto Cortés

Re: [9fans] Is the "cpu" command using version 1 of p9any?

[Russ Cox]

It appears that authsrv(6) is lying.

One could easily make it tell the truth:
in /sys/src/cmd/auth/factotum/p9any.c,
in p9anyread, change

		negstr = s_new();

to

		s->version = 2;
		negstr = s_copy("v.2 ");

Russ