pari bug

["William Stein" <wstein@gmail.com>]

Hello,

I found a bug in PARI's polynomial factorization over finite fields
in pari-2.3.1 and in pari-2.4.0.alpha:

?  f = Mod(Mod(1, 100003), Mod(1, 100003)*a^2 + Mod(78244, 100003)*a  
+ Mod(99957, 100003))*x^2 + Mod(Mod(92725, 100003)*a + Mod(69647, 100003),  
Mod(1, 100003)*a^2 + Mod(78244, 100003)*a + Mod(99957, 100003))*x  
+ Mod(Mod(80881, 100003)*a + Mod(93268, 100003), Mod(1, 100003)*a^2  
+ Mod(78244, 100003)*a + Mod(99957, 100003))
%1 = Mod(Mod(1, 100003), Mod(1, 100003)*a^2 + Mod(78244, 100003)*a  
+ Mod(99957, 100003))*x^2 + Mod(Mod(92725, 100003)*a + Mod(69647, 100003),  
Mod(1, 100003)*a^2 + Mod(78244, 100003)*a + Mod(99957, 100003))*x  
+ Mod(Mod(80881, 100003)*a + Mod(93268, 100003), Mod(1, 100003)*a^2  
+ Mod(78244, 100003)*a + Mod(99957, 100003))
?
? factor(f)

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_PROTECTION_FAILURE at address: 0x00000005
modii (x=0x11402b8, y=0x29) at ../src/kernel/none/mp_indep.c:417
417     ../src/kernel/none/mp_indep.c: No such file or directory.
        in ../src/kernel/none/mp_indep.c
(gdb) bt
#0  modii (x=0x11402b8, y=0x29) at ../src/kernel/none/mp_indep.c:417
#1  0x007d5480 in ?? ()
Previous frame inner to this frame (corrupt stack?)
(gdb)

---------

I see this bug on an Intel OS X Mac and an Intel core duo Linux 32-bit  
machine.
The bug does not appear on a 64-bit opteron Linux machine.

All builds of PARI mentioned above are against gmp-4.2.1.

William