Re: [PATCH] kconfig: Add kernel config option for fuzz testing.

From: Dmitry Vyukov
Date: Thu Dec 19 2019 - 12:21:48 EST


On Wed, Dec 18, 2019 at 11:30 AM Tetsuo Handa
<penguin-kernel@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On 2019/12/17 17:36, Dmitry Vyukov wrote:
> > FWIW we've just disabled sysrq entirely:
> > https://github.com/google/syzkaller/blob/master/dashboard/config/bits-syzbot.config#L182
> > because random packets over usb can trigger a panic sysrq (again
> > almost impossible to reliably filter these out on fuzzer side).
>
> Excuse me, but CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE=0x0 helps only if show_state() etc. are
> called via the __handle_sysrq() handler in drivers/tty/sysrq.c .
>
> static void sysrq_handle_showstate(int key)
> {
> show_state();
> show_workqueue_state();
> }
> static struct sysrq_key_op sysrq_showstate_op = {
> .handler = sysrq_handle_showstate,
> .help_msg = "show-task-states(t)",
> .action_msg = "Show State",
> .enable_mask = SYSRQ_ENABLE_DUMP,
> };
>
> The k_spec() handler in drivers/tty/vt/keyboard.c calls show_state() etc. without
> evaluating sysrq_enabled value.
>
> #define FN_HANDLERS\
> fn_null, fn_enter, fn_show_ptregs, fn_show_mem,\
> fn_show_state, fn_send_intr, fn_lastcons, fn_caps_toggle,\
> fn_num, fn_hold, fn_scroll_forw, fn_scroll_back,\
> fn_boot_it, fn_caps_on, fn_compose, fn_SAK,\
> fn_dec_console, fn_inc_console, fn_spawn_con, fn_bare_num
>
> typedef void (fn_handler_fn)(struct vc_data *vc);
> static fn_handler_fn FN_HANDLERS;
> static fn_handler_fn *fn_handler[] = { FN_HANDLERS };
>
> static void fn_show_state(struct vc_data *vc)
> {
> show_state();
> }
>
> static void k_spec(struct vc_data *vc, unsigned char value, char up_flag)
> {
> if (up_flag)
> return;
> if (value >= ARRAY_SIZE(fn_handler))
> return;
> if ((kbd->kbdmode == VC_RAW ||
> kbd->kbdmode == VC_MEDIUMRAW ||
> kbd->kbdmode == VC_OFF) &&
> value != KVAL(K_SAK))
> return; /* SAK is allowed even in raw mode */
> fn_handler[value](vc);
> }
>
> Therefore, we need to guard at either callee side (e.g. show_state_filter())
> or caller side (e.g. k_spec()) using kernel config (or something equivalent)
> in order to avoid forever calling show_state() from timer function.

+Andrey, please take a look if CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE
covers everything we need, or we need to disable sysrq entirely.