Zsh Mailing List Archive
Messages sorted by:
Reverse Date,
Date,
Thread,
Author
Segfault due short overflow in hist.c
- X-seq: zsh-workers 51411
- From: donoban@xxxxxxxxxx
- To: zsh-workers@xxxxxxx
- Subject: Segfault due short overflow in hist.c
- Date: Sun, 12 Feb 2023 09:46:46 +0000
- Archived-at: <https://zsh.org/workers/51411>
- List-id: <zsh-workers.zsh.org>
Hi,
I noticed that after expanding a big directory with 'rm -fr *' I could
crash zsh in a reproducible way. Here is the backtrace from gdb:
(gdb) bt
#0 0x00005555555aa8be in hend (prog=prog@entry=0x7ffff7a5fbc8) at
hist.c:1578
#1 0x00005555555ab08f in loop (justonce=<optimized out>,
toplevel=<optimized out>) at init.c:170
#2 0x00005555555afacf in zsh_main (argc=<optimized out>,
argv=<optimized out>) at init.c:1794
#3 0x00007ffff7f809ca in libc_start_main_stage2 (main=0x55555556ad90
<main>, argc=1, argv=0x7fffffffe8c8) at src/env/__libc_start_main.c:95
#4 0x000055555556adab in _start ()
It seems that *chwords overflows with long enough strings so then it
tries to access a negative index in chline[].
I'm not sure of how it should it be properly fixed without breaking any
parts. Probably chwords should be a different type like size_t, but also
it needs some check for the next overflow scenario.
Best Regards!
Messages sorted by:
Reverse Date,
Date,
Thread,
Author