On Thu, Oct 03, 2002 at 10:04:32AM +0200, tonyms@... wrote:
> When building 0.7.8 on both Solaris/Sparc (using Cristophe's binary)
> and Debian/x86 (using my own build of 0.7.7), I removed the
> declaration in with-array-data in order to work around the warning
> produced in stream.lisp.
What is this warning, by the way?
> However, both builds seem to fail in the same place during warm-init.
> LDB's backtrace suggests and infinite loop before stack exhaustion
> (backtrace and last part of build output using SHOW attached).
Thanks. Ew. *boggle*.
> First of all, apologies if this isn't a bug: the low-level stuff is
> completely over my head, and I don't think I could distinguish a
> low-level bug from reasonable behaviour if my life depended on it.
Well, I'm fairly confident that this isn't reasonable behaviour :-/
But it is a bit weird...
> <Frame 0x50ffdfc0, CODE: 0x302BE8A7, top level local call SB!BIGNUM::%DIGIT-GREATER, LRA: 0x302be917, PC: 0x58>
> <Frame 0x50ffdf80, CODE: 0x302C919F, top level local call SB!BIGNUM:BIGNUM-COMPARE, LRA: 0x302c948f, PC: 0x2d0>
> <Frame 0x50ffdf40, CODE: 0x302FAA57, top level local call SB!KERNEL:TWO-ARG->, LRA: 0x302fb76f, PC: 0xcc0>
What is %DIGIT-GREATER doing on the stack? It _should_ be expanded
inline (see src/code/bignum.lisp) in a context where it knows its
arguments are 32-bit quantities and are then optimized away... somehow,
this isn't happening...
Do you have an after-xc.core? If you do, could you touch the
src/code/bignum.lisp file and run slam (which should recompile
src/code/bignum), and look for messages about %digit-greater from the
http://www-jcsu.jesus.cam.ac.uk/~csr21/ +44 1223 510 299/+44 7729 383 757
(set-pprint-dispatch 'number (lambda (s o) (declare (special b)) (format s b)))
(defvar b "~&Just another Lisp hacker~%") (pprint #36rJesusCollegeCambridge)