Not simply a Linux problem - if this is the ImageSegment issue I think  
you guys are referring to (apologies if not) then the problem appeared  
under OpenBSD 3.5 too; code sharing between L2.6 and OBSD3.5 kernels is  
of course non-existent.

But yeah, the quick-'n'-dirty hack is to avoid dynamic allocation by  
specifying a fixed heap, iirc. Goran and I submitted a few messages  
reporting such problems in July, with regards to SM update problems,  
based around ImageSegment negative pointers.

Goran's post:  
http://lists.squeakfoundation.org/pipermail/squeak-dev/2004-July/ 
079498.html
Mine:  
http://lists.squeakfoundation.org/pipermail/squeak-dev/2004-July/ 
079505.html

Hope that helps.

Regards,
Marc

On Sunday, August 29, 2004, at 04:41  am, David T. Lewis wrote:

> [...]
>
> Ah, yes, that's probably it. If I recall, the problem showed up for  
> people
> running the newer Linux 2.6 kernels (maybe only on 64 bit machines? I'm
> not sure).  If I remember right, I think that it was possible to work  
> around
> the problem by forcing the VM to use a fixed memory allocation,  
> something
> like "squeak -memory 40m myimage". I don't happen to have a Linux 2.6
> system or a 64 bit machine to confirm this, but it's worth trying the
> " -memory" option to see if it helps.
>
> Damien, essayez ceci:
>
>   $ squeak -memory 40m squeak.image
>
> Dave
>