mpz_mul core dump - attempt 2

[Kate Minola]

Alex wrote:

> After building a 64-bit version, your test program does
> crash for me as well.  I am using Sun's compiler, but gcc3
> is probably the same.
> 
> So, yes, it needs to be debugged.
> 
> I can try to help, but I not sure that I have the time,
> until maybe this weekend.

Much much thanks for confirming that a bug exists!  With
Torbjorn reporting negative results, I was beginning to
think I had gone crazy.

Yes, I am using a 64-bit version of gmp.  I apologize that
this was not clear.  Looking back over my original report,
I see that I did not include the compile options that I used
when I compiled the C program:

   gcc -O2 -m64 -mptr64 -Wa,-xarch=v9 -mcpu=v9

These are the compile options that gmp-4.1.3 defaulted to
on my system and used to build gmp.  So I used the same
options on the C program ...

So far, I have observed that if I use the printf at
line 502 of mpn/generic/mul_fft.c/mpn_mul_fft_internal,
I get the following output on the following machines

ultrasparc2-sun-solaris2.8

pl=2097152 k=11 K=2048 np=2112 l=1024 Mp=66 rec=0 sqr=1
pl=2112 k=7 K=128 np=34 l=16 Mp=17 rec=1 sqr=1

alphaev68-dec-osf5.1b

pl=2097152 k=11 K=2048 np=2112 l=1024 Mp=66 rec=0 sqr=1
pl=2112 k=6 K=64 np=67 l=33 Mp=67 rec=1 sqr=1

The differences on the second line seem to be causing
the crash I am observing on the sparc.  I do not
know why the differences occur ... whether this is
a symptom of the bug or something having to do with
the fft algorithm on the different architectures.

Kate Minola
University of Maryland, College Park