Re: crash in emacs-21.1

[Richard Stallman]

    Isn't the handling of #x9e (LEADING_CODE_8_BIT_CONTROL) incompatible
    in the case?

Handa would know for certain, but it could be correct.
Emacs was changed to read multibyte strings and convert
certain nonascii codes (is it 0200 to 0237?) to these
multibyte sequences, and Fstring_as_unibyte converts
them back to the "intended" single-byte sequences.

Maybe the call to Fstring_as_unibyte was added specifically to do
that.

Meanwhile, as to debugging this problem, if someone can identify a
single function whose byte code is invalid, running M-x disassemble on
that function (instead of running it!) could show what is wrong.  Any
of us, looking at the M-x disassemble output and comparing it with the
source code, could get to the root of the problem.

Daniel, if you go up to this frame with `frame 2'

#2  0x13d244 in Ffuncall (nargs=1, args=0xefffd758) at
/home/sadind/src/emacs-21.1/src/eval.c:2712

and then type

p args[0]
pr

then you can show us which function it is.