#902 Segfaults with exponentiation on amd64 (GCL problem)

closed
nobody
5
2006-08-24
2006-04-14
Anonymous
No

Maxima version: 5.9.1
Maxima build date: 3:30 2/26/2005
host type: x86_64-unknown-linux-gnu
lisp-implementation-type: GNU Common Lisp (GCL)
lisp-implementation-version: GCL 2.6.6

(%i1) is((255/256)^(2^22) < (1/(256!)));
Segmentation fault

(%i1) (2*x)^(2^24);
Segmentation fault

It doesn't seem a stack overflow.. maybe it is a
problem related to exponentiation? or maybe the amd64 arch?

Discussion

  • Raymond Toy

    Raymond Toy - 2006-04-14

    Logged In: YES
    user_id=28849

    Considering that 2^(2^24) has about 5 million digits,
    perhaps the seg fault comes from trying to convert 2^(2^24)
    from the internal binary representation to a decimal
    representation?

    Or perhaps gcl can't handle a bignum with some 16 million bits?

     
  • Raymond Toy

    Raymond Toy - 2006-04-15

    Logged In: YES
    user_id=28849

    The first example works fine with cmucl on Mac OS X. It
    does the user if he really wants to compute the large
    exponentiation, and if you say yes, it finally returns true,
    as expected.

    I think this is a gcl issue.

     
  • Robert Dodier

    Robert Dodier - 2006-08-24

    Logged In: YES
    user_id=501686

    Maxima 5.9.3.99rc2 / GCL 2.6.7 / Linux:

    (255/256)^(2^22)$
    => OK

    (2*x)^(2^24)$
    => OK

    (255^(2^22))/(256^(2^22))$
    => segfault

    This last bit has been reported to the GCL bug tracker as
    bug #17515. Closing this report as "won't fix".

     
  • Robert Dodier

    Robert Dodier - 2006-08-24
    • status: open --> closed
     
  • Robert Dodier

    Robert Dodier - 2006-08-24
    • labels: 460522 --> Problem not in Maxima
    • summary: Segfaults with exponentiation on amd64 --> Segfaults with exponentiation on amd64 (GCL problem)
     

Log in to post a comment.