Re: crashes on x86_64

[Kim F. Storm]

Reiner Steib <address@hidden> writes:

>
>> And the same crash does NOT occur with GC checking enabled?
>
> The crash happened much faster without GC checking.  But the last gdb
> output in my previous mail (the one with "<EMACS BUG: INVALID
> DATATYPE") was with GC checking.
>

I looked at the code involved here, and it could be caused by a
stack overflow due to excessive alloca usage (there is no practical
limit on the amount of memory used here).

I have installed a change to eliminate the alloca usage.

>
> [1] (gdb) r -Q ChangeLog* -l _crash_me_.el
>
> _crash_me_.el creates upto 1000 frames...
>
> --8<---------------cut here---------------start------------->8---
> (setq file "/tmp/ste/make-frame.out")
>
> (with-temp-buffer
>   (insert "\n# New test\n")
>   (append-to-file (point-min) (point-max) file))
>
> (when t ; nil
>   (while (< (length (frame-list)) 1000)
>     (with-temp-buffer
>       (insert (format "%s\n" (length (frame-list))))
>       (append-to-file (point-min) (point-max) file))
>     (next-buffer)
>     (make-frame-command)
>     (sit-for 2)
>     (sleep-for 2))
>   (kill-emacs))
> --8<---------------cut here---------------end--------------->8---


--
Kim F. Storm <address@hidden> http://www.cua.dk