> Methink I found the sucker - we have one place where the unix_socket
>is freed without kernel_lock hold. It is in unix_destroy_timer(). I'm
I just examined it yesterday and as far I can see such place it's just
fine. The reason is that when you are in unix_destroy_timer(), the sock is
just been removed from the unix-sockets hashtable from ages (see my email
of yesterday). The important thing is that we remove the sock from the
hashtable with the kenrel lock held and that we don't sleep in unix_gc. It
seems to me that it's just fine...
>+ if (!down_trylock(&unix_gc_marking))
>+ goto delay_more;
Happy to see that my down_trylock is going to be usful though ;).
Feel free to try the patch thought but I dubit it will make differences.
Maybe I missed something, I had really a little time to read the code.
Andrea Arcangeli
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/