Tyson D Sawyer <[EMAIL PROTECTED]> writes:
> Ollie Lho wrote:
> >
> > Ronald G Minnich wrote:
> > >
> > > I have not had this problem. I will try to test again today. I did test
> > > this before doing the commit, so I can't see what could have gone wrong.
> > >
> > > Tyson, any thoughts?
> > >
> > > ron
> >
> > It seems that the zlib has some problem uncompress a gzip -9 compressed
> > image. gzip -6 works fine.
> >
> > Ollie
>
>
> Hmmm... Did _I_ submit any changes that affected anyone other than
> myself that might have had gzip -9 in it?
>
> I am using -9 right now because the default wasn't small enough. I have
> _sometimes_ had trouble with -9 so I don't assume it will work and I
> don't think I submitted any changes to that effect.
>
> ...I certainly didn't mean to.
The only difference between the compression levels is how hard the
compressor works to find compressible data. The decompression
algorithm is the same. I suspect the bug lies in how we manage memory
during decompression. I know I increased my reliability on the l440gx
if I malloc'd the 32K window used for the kernel.
I'm not racing there but I suspect we need an audit of exactly how
we are using memory.
Eric
