2008-02-04 19:52:45

by Pavel Machek

[permalink] [raw]
Subject: Re: Fwd: brk randomization breaks columns

> >
> > > I still don't seem to fully understand what is happening here --
> > > aparently this is triggerable only with old programs linked against
> > > libc.so.5, and I am not able to trigger it with my trivial program when
> > > I link it against old libc.so.5, which just basically does brk() and
> > > checks whether /proc/<pid>/maps are OK. Seems to me that (at least
> > > certain versions) of libc.so.5 (wrongly) assume that end of the bss is
> > > the start of the heap, but I will try to investigate it more.
>
> I think I am getting the some problem with brand new binaries produced
> by myself :(
>
> Look at this report I posted days ago:
>
> http://marc.info/?l=linux-kernel&m=120169141603789&w=2

Your crash is in kernel, no? Anyway, I can workaround mine by
disabling randomization...
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html


2008-02-04 21:55:09

by Abel Bernabeu

[permalink] [raw]
Subject: Re: Fwd: brk randomization breaks columns

2008/2/4, Pavel Machek <[email protected]>:
> > >
> > > > I still don't seem to fully understand what is happening here --
> > > > aparently this is triggerable only with old programs linked against
> > > > libc.so.5, and I am not able to trigger it with my trivial program when
> > > > I link it against old libc.so.5, which just basically does brk() and
> > > > checks whether /proc/<pid>/maps are OK. Seems to me that (at least
> > > > certain versions) of libc.so.5 (wrongly) assume that end of the bss is
> > > > the start of the heap, but I will try to investigate it more.
> >
> > I think I am getting the some problem with brand new binaries produced
> > by myself :(
> >
> > Look at this report I posted days ago:
> >
> > http://marc.info/?l=linux-kernel&m=120169141603789&w=2
>
> Your crash is in kernel, no? Anyway, I can workaround mine by
> disabling randomization...

Yes, my crash happens at kernel space, but is due to the randomized
params passed to set_brk (the same thing that makes your game crash
later at runtime).

I did not understand why my kernel was crashing before reading your
comments, but now I think is all happening because of the brk
randomization. Sorry, I will not be able to try disabling
randomization until tomorrow at office in order to confirm you this.

I am trying to figure out what is relation the brk randomization
feature and my crash... but makes sense to me now, trying to find it.

Yours, Abel.

2008-02-04 22:48:18

by Jiri Kosina

[permalink] [raw]
Subject: Re: Fwd: brk randomization breaks columns

On Mon, 4 Feb 2008, Abel Bernabeu wrote:

> > > I think I am getting the some problem with brand new binaries produced
> > > by myself :(
> > > Look at this report I posted days ago:
> > > http://marc.info/?l=linux-kernel&m=120169141603789&w=2
> > Your crash is in kernel, no? Anyway, I can workaround mine by
> > disabling randomization...
> Yes, my crash happens at kernel space, but is due to the randomized
> params passed to set_brk (the same thing that makes your game crash
> later at runtime).

[ please don't drop me from CC, thanks ]

Abel, I actually don't think you have chance to have any issues with
randomization, as the mentioned post talks about 2.6.22.10, which doesn't
randomize the brk start at all.

Thanks,

--
Jiri Kosina
SUSE Labs

2008-02-04 23:14:15

by Abel Bernabeu

[permalink] [raw]
Subject: Re: Fwd: brk randomization breaks columns

2008/2/4, Jiri Kosina <[email protected]>:

> Abel, I actually don't think you have chance to have any issues with
> randomization, as the mentioned post talks about 2.6.22.10, which doesn't
> randomize the brk start at all.

You are rigth, my kernel just randomizes stack top.

I've got a bit sick with this crash in the latest days, trying to
understand the elf loader, trying toolchain configurations, etc... O:)

Yours, Abel.

2008-02-04 23:38:56

by Pavel Machek

[permalink] [raw]
Subject: Re: Fwd: brk randomization breaks columns

Hi!

> > Abel, I actually don't think you have chance to have any issues with
> > randomization, as the mentioned post talks about 2.6.22.10, which doesn't
> > randomize the brk start at all.
>
> You are rigth, my kernel just randomizes stack top.
>
> I've got a bit sick with this crash in the latest days, trying to
> understand the elf loader, trying toolchain configurations, etc... O:)

Well, elf loader is indeed evil.
Pavel

--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html