Re: [LTP] [PATCH] hugeshmat01 isn't adapt to mips with 64bit ELF

[Cyril H. <ch...@su...>]

Hi!
> > > --- a/testcases/kernel/mem/hugetlb/hugeshmat/hugeshmat01.c
> > > +++ b/testcases/kernel/mem/hugetlb/hugeshmat/hugeshmat01.c
> > > 
> > >  #define CASE1		20		/* memory location.		   */
> > >  #if __WORDSIZE==64
> > > +#if defined(__mips__)
> > > +#define UNALIGNED      0x1000000eee
> > > +#else
> > >  #define UNALIGNED      0x10000000eee
> > > +#endif
> > >  #else
> > >  #define UNALIGNED      0x60000eee
> > >  #endif
> > 
> > I wonder why we need such #ifdef for 64 bit systems in the first place,
> > or is the unaligned address 0x60000eee too low for 64 bit system?
> > 
> > Garrett, Mike any clue?
> 
> i can only guess that the author was attempting to locate a highly-unlikely-
> to-be-used memory location

Hmm, okay. So what we need is to create library function that returns
highly-unlikely memory location (I've seen this code in several places
and I would like to avoid the duplication).

-- 
Cyril Hrubis
ch...@su...