Re: uninitialized variable in GSFormat?

[Richard Frith-Macdonald]

On 25 Aug 2006, at 18:51, Richard Frith-Macdonald wrote:

> On 25 Aug 2006, at 18:07, Wim Oudshoorn wrote:
>
> > I don't know if this is a bug, but
> > I used valgrind on our application and I got the
> > following message
>
> > ==18223==
> > ==18223== Conditional jump or move depends on uninitialised value(s)
> > ==18223==    at 0x4B1C8FF: GSFormat (GSFormat.m:291)
> > ==18223==    by 0x4B33F85:  
> > _i_GSPlaceholderString__initWithFormat_locale_arguments_  
> > (GSString.m:502)
> > ==18223==    by 0x4C064E8: _c_NSString__stringWithFormat_  
> > (NSZone.h:281)
> > ==18223==    by 0x4BE0E0A: _i_NSProcessInfo__globallyUniqueString  
> > (NSProcessInfo.m:913)
> > ==18223==    by 0x400950: main (NSPT.m:7)
> > ==18223==
>
> I haven't a clue why valgrind is complaining about that.  Can't see  
> what might be uninitialised there.
> Anyone else?
>
> It doesn't happen on my machine (valgrind-3.2.0) in the 32bit  
> environment, though I do get a lot of similar errors reported  
> about /lib/ld-2.3.6.so
> However, if I run in the 64bit environment (intel emt64) valgrind  
> crashes in globallyUniqueString
>
> This could be a valgrind bug on 64bit machines.

Nope ... I upgraded by version of valgrind in the 64bit environment,  
and it stopped crashing and started exhibiting the same behavior you  
reported.
I eventually tracked that down to passing an int argument to  
stringWithFormat where the format string specified a long.  Fixed in  
svn.
Thanks.