linux-kernel - Re: [RFC] firmware load: defer request_firmware during early boot and resume

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <201207211949.13944.rjw@sisk.pl>
Date:	Sat, 21 Jul 2012 19:49:13 +0200
From:	"Rafael J. Wysocki" <rjw@...k.pl>
To:	Ming Lei <tom.leiming@...il.com>
Cc:	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	"linux-usb" <linux-usb@...r.kernel.org>,
	Alan Stern <stern@...land.harvard.edu>,
	"Greg Kroah-Hartman" <gregkh@...uxfoundation.org>,
	Oliver Neukum <oneukum@...e.de>
Subject: Re: [RFC] firmware load: defer request_firmware during early boot and resume

On Friday, July 20, 2012, Ming Lei wrote:
> The RFC patch is just for discussing if the idea of deferring
> request_firmware is doable for addressing the issue of
> request_firmware in resume path, which is caused by driver
> unbind/rebind during resume.
> 
> At least usb bus is involved in such things, one driver may be
> unbound and rebound in resume path at several situations, and
> request_firmware is often called inside probe().
> 
> Also the idea should be helpful for other hotplug buses too,
> at least there was the similar problem report on pcmcia bus.
> 
> diff --git a/drivers/base/dd.c b/drivers/base/dd.c
> index 6cd2c6c..fb02eff 100644
> --- a/drivers/base/dd.c
> +++ b/drivers/base/dd.c
> @@ -123,7 +123,7 @@ static bool driver_deferred_probe_enable = false;
>   * list and schedules the deferred probe workqueue to process them.  It
>   * should be called anytime a driver is successfully bound to a device.
>   */
> -static void driver_deferred_probe_trigger(void)
> +void driver_deferred_probe_trigger(void)
>  {
>  	if (!driver_deferred_probe_enable)
>  		return;
> @@ -144,6 +144,7 @@ static void driver_deferred_probe_trigger(void)
>  	 */
>  	queue_work(deferred_wq, &deferred_probe_work);
>  }
> +EXPORT_SYMBOL_GPL(driver_deferred_probe_trigger);
> 
>  /**
>   * deferred_probe_initcall() - Enable probing of deferred devices
> diff --git a/drivers/base/firmware_class.c b/drivers/base/firmware_class.c
> index 5401814..4fe742f 100644
> --- a/drivers/base/firmware_class.c
> +++ b/drivers/base/firmware_class.c
> @@ -593,6 +593,9 @@ request_firmware(const struct firmware
> **firmware_p, const char *name,
>  	if (IS_ERR_OR_NULL(fw_priv))
>  		return PTR_RET(fw_priv);
> 
> +	if (system_state != SYSTEM_RUNNING)
> +		return -EPROBE_DEFER;

You can't just return here, _request_firmware_cleanup() has to be done still.

> +
>  	ret = usermodehelper_read_trylock();

So why don't you do this here, actually, like:

 	if (ret) {
 	 	ret = -EPROBE_DEFER;
 	} else {

instead of the WARN_ON()?

Arguably, all cases in which usermodehelper_read_trylock() returns error
codes will require deferred probing.

>  	if (WARN_ON(ret)) {
>  		dev_err(device, "firmware: %s will not be loaded\n", name);
> diff --git a/include/linux/device.h b/include/linux/device.h
> index d0e4d99..a63d3171 100644
> --- a/include/linux/device.h
> +++ b/include/linux/device.h
> @@ -244,7 +244,7 @@ extern struct device_driver *driver_find(const char *name,
>  					 struct bus_type *bus);
>  extern int driver_probe_done(void);
>  extern void wait_for_device_probe(void);
> -
> +extern void driver_deferred_probe_trigger(void);
> 
>  /* sysfs interface for exporting driver attributes */
> 
> diff --git a/include/linux/kernel.h b/include/linux/kernel.h
> index e07f5e0..c8d74c6 100644
> --- a/include/linux/kernel.h
> +++ b/include/linux/kernel.h
> @@ -378,6 +378,7 @@ extern enum system_states {
>  	SYSTEM_POWER_OFF,
>  	SYSTEM_RESTART,
>  	SYSTEM_SUSPEND_DISK,
> +	SYSTEM_SUSPEND,

First off, SYSTEM_SUSPEND_DISK is not used and probably should be removed.
Second, both SYSTEM_SUSPEND and SYSTEM_SUSPEND_DISK would require the same
kind of handling in the respect of device probing, so it is not sufficient
to change the state in suspend_devices_and_enter().

Moreover, there are other situations in which tasks are frozen and
request_firmware() won't work just as well, so I don't think using
system_state for that is going to work in general.

>  } system_state;
> 
>  #define TAINT_PROPRIETARY_MODULE	0
> diff --git a/init/main.c b/init/main.c
> index e60679d..237eae1 100644
> --- a/init/main.c
> +++ b/init/main.c
> @@ -809,6 +809,8 @@ static noinline int init_post(void)
>  	current->signal->flags |= SIGNAL_UNKILLABLE;
>  	flush_delayed_fput();
> 
> +	driver_deferred_probe_trigger();
> +
>  	if (ramdisk_execute_command) {
>  		run_init_process(ramdisk_execute_command);
>  		printk(KERN_WARNING "Failed to execute %s\n",
> diff --git a/kernel/power/suspend.c b/kernel/power/suspend.c
> index 1da39ea..dbf6ffb 100644
> --- a/kernel/power/suspend.c
> +++ b/kernel/power/suspend.c
> @@ -224,6 +224,9 @@ int suspend_devices_and_enter(suspend_state_t state)
>  		goto Recover_platform;
>  	}
>  	suspend_test_finish("suspend devices");
> +
> +	system_state = SYSTEM_SUSPEND;
> +
>  	if (suspend_test(TEST_DEVICES))
>  		goto Recover_platform;
> 
> @@ -301,6 +304,10 @@ static int enter_state(suspend_state_t state)
>   Finish:
>  	pr_debug("PM: Finishing wakeup.\n");
>  	suspend_finish();
> +
> +	system_state = SYSTEM_RUNNING;
> +	driver_deferred_probe_trigger();
> +
>   Unlock:
>  	mutex_unlock(&pm_mutex);
>  	return error;

Thanks,
Rafael
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/