Skip to contentSkip to conversation

Fix issues with deferred_probe_timeout being non-zero

PATCH In review Saravana Kannan 3 participants 5 messages May 26, 2022 – May 30, 2022 Driver core

lore
Share & tools
lore
#0 lore raw

cover Fix issues with deferred_probe_timeout being non-zero

Setting deferred_probe_timeout to a non-zero default used to cause issues with NFS rootfs. But the fix for that is causing other problem. So revert the fix and set the default deferred_probe_timeout back to zero.

Most of the Cc list if from commit 35a672363ab3. Please let me know if you want to be removed from future patches related to deferred_probe_timeout and NFS rootfs issues.

-Saravana

Saravana Kannan (2):
driver core: Fix wait_for_device_probe() & deferred_probe_timeout
interaction
driver core: Set default deferred_probe_timeout back to 0.

 drivers/base/dd.c | 10 ----------
 1 file changed, 10 deletions(-)
Signature · 1 line
2.36.1.124.g0e6072fb45-goog
#1 lore raw

PATCH 1/2 driver core: Fix wait_for_device_probe() & deferred_probe_timeout interaction

Mounting NFS rootfs was timing out when deferred_probe_timeout was non-zero [1]. This was because ip_auto_config() initcall times out waiting for the network interfaces to show up when deferred_probe_timeout was non-zero. While ip_auto_config() calls wait_for_device_probe() to make sure any currently running deferred probe work or asynchronous probe finishes, that wasn't sufficient to account for devices being deferred until deferred_probe_timeout.

Commit 35a672363ab3 ("driver core: Ensure wait_for_device_probe() waits until the deferred_probe_timeout fires") tried to fix that by making sure wait_for_device_probe() waits for deferred_probe_timeout to expire before returning.

However, if wait_for_device_probe() is called from the kernel_init() context:

- Before deferred_probe_initcall() [2], it causes the boot process to hang due to a deadlock.

- After deferred_probe_initcall() [3], it blocks kernel_init() from continuing till deferred_probe_timeout expires and beats the point of deferred_probe_timeout that's trying to wait for userspace to load modules.

Neither of this is good. So revert the changes to
wait_for_device_probe().

[1] - https://lore.kernel.org/lkml/TYAPR01MB45443DF63B9EF29054F7C41FD8C60@TYAPR01MB4544.jpnprd01.prod.outlook.com/
[2] - https://lore.kernel.org/lkml/[email protected]/
[3] - https://lore.kernel.org/lkml/[email protected]/

Cc John Stultz <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Alexey Kuznetsov <[email protected]>
Cc Hideaki YOSHIFUJI <[email protected]>
Cc Jakub Kicinski <[email protected]>
Cc Rafael J. Wysocki <[email protected]>
Cc Rob Herring <[email protected]>
Cc Geert Uytterhoeven <[email protected]>
Cc Yoshihiro Shimoda <[email protected]>
Cc Robin Murphy <[email protected]>
Cc Andy Shevchenko <[email protected]>
Cc Sudeep Holla <[email protected]>
Cc Andy Shevchenko <[email protected]>
Cc Naresh Kamboju <[email protected]>
Cc Basil Eljuse <[email protected]>
Cc Ferry Toth <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Anders Roxell <[email protected]>
Fixes 35a672363ab3 ("driver core: Ensure wait_for_device_probe() waits until the deferred_probe_timeout fires")

---

 drivers/base/dd.c | 5 -----
 1 file changed, 5 deletions(-)
drivers/base/dd.c +0 -5
index 2fc8507f59ee..91f63cd33b12 100644
@@ -263,7 +263,6 @@ int driver_deferred_probe_timeout;
 #endif
 
 EXPORT_SYMBOL_GPL(driver_deferred_probe_timeout);
-static DECLARE_WAIT_QUEUE_HEAD(probe_timeout_waitqueue);
 
 static int __init deferred_probe_timeout_setup(char *str)
 {
@@ -318,7 +317,6 @@ static void deferred_probe_timeout_work_func(struct work_struct *work)
 	list_for_each_entry(p, &deferred_probe_pending_list, deferred_probe)
 		dev_info(p->device, "deferred probe pending\n");
 	mutex_unlock(&deferred_probe_mutex);
-	wake_up_all(&probe_timeout_waitqueue);
 }
 static DECLARE_DELAYED_WORK(deferred_probe_timeout_work, deferred_probe_timeout_work_func);
 
@@ -736,9 +734,6 @@ int driver_probe_done(void)
  */
 void wait_for_device_probe(void)
 {
-	/* wait for probe timeout */
-	wait_event(probe_timeout_waitqueue, !driver_deferred_probe_timeout);
-
 	/* wait for the deferred probe workqueue to finish */
 	flush_work(&deferred_probe_work);
 
Signature · 1 line
2.36.1.124.g0e6072fb45-goog
Acked-by on v1 1/2
#2 lore raw

Re: [PATCH v1 1/2] driver core: Fix wait_for_device_probe() & deferred_probe_timeout interaction

Quoted 47 lines · Saravana Kannan

Mounting NFS rootfs was timing out when deferred_probe_timeout was non-zero [1]. This was because ip_auto_config() initcall times out waiting for the network interfaces to show up when deferred_probe_timeout was non-zero. While ip_auto_config() calls wait_for_device_probe() to make sure any currently running deferred probe work or asynchronous probe finishes, that wasn't sufficient to account for devices being deferred until deferred_probe_timeout.

Commit 35a672363ab3 ("driver core: Ensure wait_for_device_probe() waits until the deferred_probe_timeout fires") tried to fix that by making sure wait_for_device_probe() waits for deferred_probe_timeout to expire before returning.

However, if wait_for_device_probe() is called from the kernel_init() context:

- Before deferred_probe_initcall() [2], it causes the boot process to hang due to a deadlock.

- After deferred_probe_initcall() [3], it blocks kernel_init() from continuing till deferred_probe_timeout expires and beats the point of deferred_probe_timeout that's trying to wait for userspace to load modules.

Neither of this is good. So revert the changes to
wait_for_device_probe().

[1] - https://lore.kernel.org/lkml/TYAPR01MB45443DF63B9EF29054F7C41FD8C60@TYAPR01MB4544.jpnprd01.prod.outlook.com/
[2] - https://lore.kernel.org/lkml/[email protected]/
[3] - https://lore.kernel.org/lkml/[email protected]/

Cc John Stultz <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Alexey Kuznetsov <[email protected]>
Cc Hideaki YOSHIFUJI <[email protected]>
Cc Jakub Kicinski <[email protected]>
Cc Rafael J. Wysocki <[email protected]>
Cc Rob Herring <[email protected]>
Cc Geert Uytterhoeven <[email protected]>
Cc Yoshihiro Shimoda <[email protected]>
Cc Robin Murphy <[email protected]>
Cc Andy Shevchenko <[email protected]>
Cc Sudeep Holla <[email protected]>
Cc Andy Shevchenko <[email protected]>
Cc Naresh Kamboju <[email protected]>
Cc Basil Eljuse <[email protected]>
Quoted 8 lines
Cc Ferry Toth <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Anders Roxell <[email protected]>
Fixes 35a672363ab3 ("driver core: Ensure wait_for_device_probe() waits until the deferred_probe_timeout fires")

Apologies for the trouble this caused. Thanks for chasing it down!

thanks
-john

#3 lore raw

PATCH 2/2 driver core: Set default deferred_probe_timeout back to 0.

In reply to Saravana Kannan

Since we had to effectively reverted
commit 35a672363ab3 ("driver core: Ensure wait_for_device_probe() waits until the deferred_probe_timeout fires") in an earlier patch, a non-zero deferred_probe_timeout will break NFS rootfs mounting [1] again. So, set the default back to zero until we can fix that.

[1] - https://lore.kernel.org/lkml/TYAPR01MB45443DF63B9EF29054F7C41FD8C60@TYAPR01MB4544.jpnprd01.prod.outlook.com/

Cc Mark Brown <[email protected]>
Cc Rob Herring <[email protected]>
Fixes 2b28a1a84a0e ("driver core: Extend deferred probe timeout on driver registration")
---
 drivers/base/dd.c | 5 -----
 1 file changed, 5 deletions(-)
drivers/base/dd.c +0 -5
index 91f63cd33b12..251b5ba1b84a 100644
@@ -256,12 +256,7 @@ static int deferred_devs_show(struct seq_file *s, void *data)
 }
 DEFINE_SHOW_ATTRIBUTE(deferred_devs);
 
-#ifdef CONFIG_MODULES
-int driver_deferred_probe_timeout = 10;
-#else
 int driver_deferred_probe_timeout;
-#endif
-
 EXPORT_SYMBOL_GPL(driver_deferred_probe_timeout);
 
 static int __init deferred_probe_timeout_setup(char *str)
Signature · 1 line
2.36.1.124.g0e6072fb45-goog
Tested-by on the v1 series
#4 lore raw
In reply to Saravana Kannan

Hi Saravana,

Quoted 8 lines · Saravana Kannan

Setting deferred_probe_timeout to a non-zero default used to cause issues with NFS rootfs. But the fix for that is causing other problem. So revert the fix and set the default deferred_probe_timeout back to zero.

Most of the Cc list if from commit 35a672363ab3. Please let me know if you want to be removed from future patches related to deferred_probe_timeout and NFS rootfs issues.

Thanks for your series!

On Salvator-XS, Micrel KSZ9031 Gigabit PHY probe is no longer delayed by 9s after applying this series.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds