Hello!
(I resend this because I get no feedback.)
Testing UML on my project, I hit the limit for command line with something like
this:
default \
mem=64M \
ubda=COW,/data/UML4/roots/ROOT1 \
ubdb=SWAP \
ubdc=CONF.tar \
${HSSTR} \
eth0=daemon,fe:fd:39:bd:cc:d7,,/data/UML4/conf/example1/socks/SW1
eth1=daemon,fe:fd:01:01:01:12,,/data/UML4/conf/example1/socks/SW2
eth2=daemon,fe:fd:01:01:01:99,,/data/UML4/conf/example1/socks/SW4
eth3=daemon,fe:fd:01:01:01:03,,/data/UML4/conf/example1/socks/SW5
eth4=daemon,fe:fd:01:01:01:04,,/data/UML4/conf/example1/socks/SW6
eth5=daemon,fe:fd:01:01:01:81,,/data/UML4/conf/example1/socks/SW8p1
eth6=daemon,fe:fd:3f:a9:35:e2,,/data/UML4/conf/EXTERN/E1 \
con=null \
ssl0=port:9101 \
umid=example1-pc1 \
@pc1@
Patch to extend the limits (buffer and number of args/envs) is attached.
Please consider including it because UML is intended to be run with such
long lines.
I'm open to other alternatives as a Kconfig entry for this.
Thank you!
Signed-off-by: Catalin(ux aka Dino) BOIE <catab at umbrella dot ro>
---
Catalin(ux aka Dino) BOIE
catab at deuroconsult.ro
http://kernel.umbrella.ro/
On Thursday 04 November 2004 08:08, Catalin(ux aka Dino) BOIE wrote:
> Hello!
>
> (I resend this because I get no feedback.)
Sorry, miss of time.
> Testing UML on my project, I hit the limit for command line with something
> like this:
> default \
> mem=64M \
> ubda=COW,/data/UML4/roots/ROOT1 \
> ubdb=SWAP \
> ubdc=CONF.tar \
> ${HSSTR} \
> eth0=daemon,fe:fd:39:bd:cc:d7,,/data/UML4/conf/example1/socks/SW1
> eth1=daemon,fe:fd:01:01:01:12,,/data/UML4/conf/example1/socks/SW2
> eth2=daemon,fe:fd:01:01:01:99,,/data/UML4/conf/example1/socks/SW4
> eth3=daemon,fe:fd:01:01:01:03,,/data/UML4/conf/example1/socks/SW5
> eth4=daemon,fe:fd:01:01:01:04,,/data/UML4/conf/example1/socks/SW6
> eth5=daemon,fe:fd:01:01:01:81,,/data/UML4/conf/example1/socks/SW8p1
> eth6=daemon,fe:fd:3f:a9:35:e2,,/data/UML4/conf/EXTERN/E1 \
> con=null \
> ssl0=port:9101 \
> umid=example1-pc1 \
> @pc1@
> Patch to extend the limits (buffer and number of args/envs) is attached.
> Please consider including it because UML is intended to be run with such
> long lines.
> I'm open to other alternatives as a Kconfig entry for this.
> Thank you!
> Signed-off-by: Catalin(ux aka Dino) BOIE <catab at umbrella dot ro>
The idea of the patch is correct.
The UML part should be mergeable immediately - however, I don't know if that
could be a security risk or has any bug (any place still using the old size).
Also, I'd like to understand what arch/um/kernel/user_util.c wants to do near
its beginning!
For the arch-independent code, there needs to be something like a Kconfig
entry (not necessarily a user-question, see CONFIG_GENERIC_RWSEM for
instance).
Also, I'd like to cross-check this patch with the one used by Knoppix for ages
to do the same thing.
Bye
--
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
On Thu, 4 Nov 2004, Catalin(ux aka Dino) BOIE wrote:
> Hello!
>
> (I resend this because I get no feedback.)
>
> Testing UML on my project, I hit the limit for command line with something like
> this:
>
> default \
> mem=64M \
> ubda=COW,/data/UML4/roots/ROOT1 \
> ubdb=SWAP \
> ubdc=CONF.tar \
> ${HSSTR} \
> eth0=daemon,fe:fd:39:bd:cc:d7,,/data/UML4/conf/example1/socks/SW1
> eth1=daemon,fe:fd:01:01:01:12,,/data/UML4/conf/example1/socks/SW2
> eth2=daemon,fe:fd:01:01:01:99,,/data/UML4/conf/example1/socks/SW4
> eth3=daemon,fe:fd:01:01:01:03,,/data/UML4/conf/example1/socks/SW5
> eth4=daemon,fe:fd:01:01:01:04,,/data/UML4/conf/example1/socks/SW6
> eth5=daemon,fe:fd:01:01:01:81,,/data/UML4/conf/example1/socks/SW8p1
> eth6=daemon,fe:fd:3f:a9:35:e2,,/data/UML4/conf/EXTERN/E1 \
> con=null \
> ssl0=port:9101 \
> umid=example1-pc1 \
> @pc1@
>
>
> Patch to extend the limits (buffer and number of args/envs) is attached.
> Please consider including it because UML is intended to be run with such
> long lines.
> I'm open to other alternatives as a Kconfig entry for this.
Actually, I should redo/update my patch that lets uml parse a config file.
[email protected] said:
> Patch to extend the limits (buffer and number of args/envs) is
> attached. Please consider including it because UML is intended to be
> run with such long lines.
I don't have a problem with this either. The init/main.c piece needs to be
run through LKML, and so far I haven't seen any reaction.
Jeff
On Thursday 04 November 2004 15:34, Blaisorblade wrote:
> On Thursday 04 November 2004 08:08, Catalin(ux aka Dino) BOIE wrote:
> > Hello!
> >
> > (I resend this because I get no feedback.)
> Sorry, miss of time.
I'm doing it now... I'll post it soon for merging. The crash fixed by Jeff
Dike in his patch is a different things: he fixes a buffer overflow, I
cleanup things and increase cleanly the allowed cmdline size. The original
Catalin BOIE's patch should work in the meanwhile.
> Also, I'd like to cross-check this patch with the one used by Knoppix for
> ages to do the same thing.
Still to find.
--
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
http://www.user-mode-linux.org/~blaisorblade