[Help-smalltalk] Re: problems with 1.8.4 (FWD)

[Paolo Bonzini]

Jack Lenz sent me more information on the SPARC bug:

> well, I tried to see some buggy code at the assembler level: the only
> thing I noticed is that the hashfunction bombed out when called the
> first time, at the first multiplication. Multiplication looks like a
> function call on Solaris - contrary to linux pc's where there is an
> assembler instruction for that. So somewhere deep into the solaris
> multiplication a function is invoked that does'nt exist. Unless you tell
> me that you already put your hand on a wrong generated code, to me this
> behaviour still looks like a memory problem, so I tried to find
> something in the code that could confirm this idea. Of course I didn't
> succeed to find a trivial bug, but I saw you're working with mapped
> memory;  although I couldn't figure out the details of your memory
> management - especially I didn't understand, whether or not malloc is
> affected by the mmap calls - it seems to me, that further investigations
> in the memory management especially on a sun could be helpfull.
>
> Jack

|_  _  _  ___
|_)(_)| )  ,'
--------- '-._.