Re: Crash on interrupted completion with prezto

[Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>]

On Oct 10,  5:17am, Bart Polot wrote:
}
} 1. Start zsh 5.1.1 with default config of prezto enabled

Sigh.  It's really hard to debug things that require a plugin package
to be loaded before anything goes wrong.

} I re-compiled zsh with debug info and attached a quick gdb log of the
} coredump. Unfortuantely, I can't reproduce it under valgrind.

This probably indicates that it's yet another interrupt-handling race
condition.  We've been playing whack-a-mole with those for a while.

However, there have also been a number of recent optimizations in the
code around the point where your stack trace ends.  At the least it
may mean that things have sped up to the point where you don't feel
like whacking ctrl+c.  Refer to ZSH_PATCHLEVEL zsh-5.1.1-83-g83a1757.

However part 2, this makes me wonder if the new patallocstr() routine
needs queue_signals() / unqueue_signals().