Still able to reproduce history crash (3.0.4)

Still able to reproduce history crash (3.0.4)

[Bart Schaefer]

And still unable to get a backtrace.

Easiest way to reproduce is to start zsh -f, then do:

HISTSIZE=0
HISTSIZE=5

and then execute at least five commands.  Appears to dump in malloc(), but
as the stack is complete garbage there's no way to be sure.  I've enabled
all the debug and mem-debug options but there's not even a hint of an error
until the gooey kablooey.

Is my build the only one in which this happens?

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Re: Still able to reproduce history crash (3.0.4)

[Zoltan Hidvegi]

> And still unable to get a backtrace.
> 
> Easiest way to reproduce is to start zsh -f, then do:
> 
> HISTSIZE=0
> HISTSIZE=5
> 
> and then execute at least five commands.  Appears to dump in malloc(), but
> as the stack is complete garbage there's no way to be sure.  I've enabled
> all the debug and mem-debug options but there's not even a hint of an error
> until the gooey kablooey.

I'll try to look at it, but right now, I'm busy rewriting paramsubst() to
clean up the mess around rc_expand_param and the (e) flag.  This turned
out to be a much more difficult problem than I first thought, but
hopefully the final code will be a bit simpler than the current one.

Zoltan

Re: Still able to reproduce history crash (3.0.4)

[Geoff Wing]

Bart Schaefer <schaefer@brasslantern.com> typed:
:And still unable to get a backtrace.
:Easiest way to reproduce is to start zsh -f, then do:
:
:HISTSIZE=0
:HISTSIZE=5
:
:and then execute at least five commands.  Appears to dump in malloc(), but
:as the stack is complete garbage there's no way to be sure.  I've enabled
:all the debug and mem-debug options but there's not even a hint of an error
:until the gooey kablooey.

I can't get it to dump. zsh-3.0.4 (NetBSD) w/all --enable-zsh-* set.
Have you tried it after moving /etc/zshenv since that's still read with
"zsh -f"

:Is my build the only one in which this happens?

Has anyone else tried it?
-- 
Geoff Wing [mason@primenet.com.au]                   Phone    : +61-3-9818 2977 
 Technical Manager: PrimeNet Computer Consultants    Facsimile: +61-3-9819 3788 
 Web: <URL:http://www.primenet.com.au/>              Mobile   : 0412 162 441
        [ Boulderdash: <URL:http://ciips.ee.uwa.edu.au/~williams/bd/> ]

Re: Still able to reproduce history crash (3.0.4)

[Philip Kizer]

Bart Schaefer <schaefer@brasslantern.com> typed:
>:Easiest way to reproduce is to start zsh -f, then do:
>:
>:HISTSIZE=0
>:HISTSIZE=5
>:
>:and then execute at least five commands.

mason@primenet.com.au (Geoff Wing) wrote:
>Has anyone else tried it?

Just to check, tried it with both 3.0.2 and 3.0.4.
Straight ./configure, no repetition of this here.


-philip

____________________________________________________________ Philip Kizer ___
Texas A&M CIS Operating Systems Group, Unix ( 409.862.4120 ) pckizer@tamu.edu

Re: Still able to reproduce history crash (3.0.4)

[Bart Schaefer]

I finally managed to get a stack trace by having GDB dump the stack just
before entering every malloc call, and tracked the problem to a patch I'd
applied to 3.0.3-test4 that got erroneously cvs-merged forward when I
imported the 3.0.4 sources.  Sorry for the red herring.

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com