I updated the bug,=20
removing the realloc() did make the symptoms of the bug disappear.
This bug corrupts the stack so the backtrace isn't always the same.
On Sun, 17 Mar 2002, Armin Rigo wrote:
> ----- Original Message -----
> > On FreeBSD 4.5, GCC 2.95.3 I got a
> > python in free(): warning: page is already free
> > Bus error (core dumped)
> This might be related to the reporter problem of 'realloc()' moving memor=
> when shrinking a block, although I am not sure, as you mention that it wo=
> when Psyco is compiled in debug mode. As there is no FreeBSD machine I ca=
> get access to, I cannot do more than wild guesses. Perhaps you can first =
> to see if the problem really comes from 'realloc()' by removing the
> 'realloc()' call at all in codemanager.c, in psyco_shrink_buffer(). This
> will make Psyco use an unreasonably high amount of memory, but would tell=
> if the bug comes from there.
> Also, could someone else listening on the psyco-devel list please see if =
> bug is reproductible (and gives the same backtraces)?
> A bient=F4t,