On Tuesday, 12 November 2019 19:35:11 CET Richard W.M. Jones wrote:
Easily reproducible using:
$ guestfish --key dev1:key:key1 --key dev2:key:key2
causing this stack trace (or others depending on where the memory corruption was caught):
Program received signal SIGABRT, Aborted. 0x00007ffff7905625 in raise () from /lib64/libc.so.6 (gdb) bt #0 0x00007ffff7905625 in raise () from /lib64/libc.so.6 #1 0x00007ffff78ee8d9 in abort () from /lib64/libc.so.6 #2 0x00007ffff79494af in __libc_message () from /lib64/libc.so.6 #3 0x00007ffff7950a6c in malloc_printerr () from /lib64/libc.so.6 #4 0x00007ffff79528d0 in _int_free () from /lib64/libc.so.6 #5 0x00005555555bdd6e in free_key_store () #6 0x0000555555589027 in main () (gdb) quit --- options/keys.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/options/keys.c b/options/keys.c index 7f68986..f783066 100644 --- a/options/keys.c +++ b/options/keys.c @@ -216,7 +216,8 @@ key_store_import_key (struct key_store *ks, const struct key_store_key *key) } assert (ks != NULL);
- new_keys = realloc (ks->keys, sizeof (*ks->keys) + 1); + new_keys = realloc (ks->keys, + (ks->nr_keys + 1) * sizeof (struct key_store_key));
Theoretically, sizeof (*ks->keys) should still be fine, instead of explicitly spelling the struct name. Apart from that, LGTM. -- Pino Toscano