[rsbac] smp boot problem

[Florian Tischler]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am Mittwoch, 24. September 2003 15:51 schrieb Amon Ott:
> On Wednesday 24 September 2003 11:44, Florian Tischler wrote:
> > I have troubles booting rsbac kernel on a dual xeon machine.
> > With boot option nosmp & noapic everything seems to work fine.
> > With rsbac_softmode as boot option it runs even with SMP enabled.
> >
> > With 2 CPUs the last lines i can see are:
> > rsbac_init(): Ready
> > Freeing unused mernel memory
> > INIT: version 2.82 booting
> > rsbac_get_attr(): auto-mounting device 00:05
> > rsbac_acl_get_single_rigth(): Could not lookup device!
> > rsbac_acl_check_right(): rsbac_acl_get_single_rigth() returned error
> RSBAC_EINVALIDDEV!
> > rsbac_adf_request(): request CLOSE ... prog_name boot, uid 0, target_type
> FIFO ...result NOT_GRANTED by ACL
> > filp_close() [sys_close]: ADF-call returned NOT GRANTED
> > the same a second time and the machine hangs.
>
> What is the device given in the rsbac_adf_request line?

I probably cutted to much from the request_line
Here is the full line: (the rest of the messages is the same like above)

rsbac_adf_request(): request CLOSE, pid 20, ppid 17, prog_name boot, uid0,
target_type FIFO, tid Device 00:05 Inode 18 Path pipe:/[18], attr , value 0,
result NOT GRANTED by ACL

then i get the filp_close() message and all messages starting from
rsbac_acl_get_single_rigth(): Could not lookup device! appear a second
time.

I gave you some wrong information - the system does not hang - it just
does not continue booting. Pressing ctrl-c continuing booting - but the
filesystems get not mounted.

With softmode i also get this errors - but it boots.
With nosmp i do not get any error. (So this must be smp-related)

> Did I get you right: You have two lines for auto-mounting the same device?
> This should never happen and can be the reason for the lockup.

Was a wrong Information with the lockup.
I have 2 times the EINVALIDDEV lines - but not 2 times the automounting
lines. I filled the boot-scripts with echo "whatever". it seems to stop after
mounting proc & dev/pts. (which are the first lines in the boot-script)

Any Ideas how i can track (or better work around) the error?
Would be fine to get this up and running.

Florian

> > I´m new to rsbac. i have used lids for years on single cpu machines.
> > The reason for switching is that rsbac is said to be smp-save and lids
>
> isn´t.
>
> It has been running on SMP systems for quite a while, but those are much
> more likely to trigger some lockup.
>
> > Rsbac also looks more powerfull to me.
>
> I'd say it is. :)
>
> Amon.

- -- 
Florian Tischler
System Administrator
*Johann Radon Institute for Computational and Applied Mathematics (RICAM)
http://www.ricam.oeaw.ac.at/
florian.tischler at oeaw.ac.at
*Industrial Mathematics Institute
http://www.indmath.uni-linz.ac.at/
tischler at indmath.uni-linz.ac.at
http://www.ricam.oeaw.ac.at/people/page.cgi?firstn=Florian&lastn=Tischler
GPG-Key: http://www.ricam.oeaw.ac.at/gpg/florian_tischler.asc
tel: +43 732 2468 9227
fax: +43 732 2468 8855
(Spezialforschungsbereich SFB F013)
(florian.tischler at sfb013.uni-linz.ac.at)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2-rc1-SuSE (GNU/Linux)

iD8DBQE/cbLKFnWoolw85SsRAkqBAKCcBH1wurDyOBbdCuQVt5tVeTuzGACfbJ9h
UG+5ekbyEoGZqcIq91NLkfE=
=KrNa
-----END PGP SIGNATURE-----