zsh 3.0.4 core dump when HISTSIZE exceeded

zsh 3.0.4 core dump when HISTSIZE exceeded

[Bart Schaefer]

With the following options set:

autocd
cdablevars
noclobber
ignoreeof
listtypes

(In addition to the default settings of

interactive
monitor
norcs
shinstdin
zle

for zsh -f)

`zsh -f` will dump core in malloc() shortly after exceeding HISTSIZE
commands.  The stack is completely trashed at this point; even running
inside gdb I was unable to get a backtrace beyond malloc().

I'm not sure exactly which of those options is pertinent; I know only
that I was unable to get `zsh -f` to crash without first setting them,
and that it crashed rapidly after setting HISTSIZE=5 with those options
in effect.  I set those options by copying the $- value of another user
here who has repeatedly encountered this crash (he typically gets it
upon executing his 101st command with HISTSIZE=100; this most recent
time, he mailed me the output of `set`, from which I grabbed $-).

SAVEHIST is not set.  I tend to suspect ignoreeof, but I'm not sure why.

-- 
Bart Schaefer                                                   Zanshin
http://www.well.com/user/barts                   http://www.zanshin.com

Re: zsh 3.0.4 core dump when HISTSIZE exceeded

[Bart Schaefer]

On Jul 29, 11:52pm, Bart Schaefer wrote:
} Subject: zsh 3.0.4 core dump when HISTSIZE exceeded
}
} I tend to suspect ignoreeof, but I'm not sure why.

Turns out none of those options are necessary; I'm able to produce a crash
in plain zsh -f, by setting HISTSIZE=0 and then back to HISTSIZE=5, and
then executing 5 or 6 commands.

Strangely, I can't get it to happen with HISTSIZE=4 or less.

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