On 17/02/2025 22:22, Ahmad Fatoum wrote:

Hello Matti,

On 22.01.25 12:28, Matti Vaittinen wrote:

quoted

On 13/01/2025 18:25, Ahmad Fatoum wrote:

quoted

hw_protection_shutdown() will kick off an orderly shutdown and if that
takes longer than a configurable amount of time, an emergency shutdown
will occur.

Recently, hw_protection_reboot() was added for those systems that don't
implement a proper shutdown and are better served by rebooting and
having the boot firmware worry about doing something about the critical
condition.

On timeout of the orderly reboot of hw_protection_reboot(), the system
would go into shutdown, instead of reboot. This is not a good idea, as
going into shutdown was explicitly not asked for.

Fix this by always doing an emergency reboot if hw_protection_reboot()
is called and the orderly reboot takes too long.

Fixes: 79fa723ba84c ("reboot: Introduce thermal_zone_device_critical_reboot()")
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
---
   kernel/reboot.c | 70 ++++++++++++++++++++++++++++++++++++++++-----------------
   1 file changed, 49 insertions(+), 21 deletions(-)

diff --git a/kernel/reboot.c b/kernel/reboot.c
index 847ac5d17a659981c6765699eac323f5e87f48c1..222b63dfd31020d0e2bc1b1402dbfa82adc71990 100644
--- a/kernel/reboot.c
+++ b/kernel/reboot.c

@@ -932,48 +932,76 @@ void orderly_reboot(void)
   }
   EXPORT_SYMBOL_GPL(orderly_reboot);
   +static const char *hw_protection_action_str(enum hw_protection_action action)
+{
+    switch (action) {
+    case HWPROT_ACT_SHUTDOWN:
+        return "shutdown";
+    case HWPROT_ACT_REBOOT:
+        return "reboot";
+    default:
+        return "undefined";
+    }
+}
+
+static enum hw_protection_action hw_failure_emergency_action;

nit: Do we have a (theoretical) possibility that two emergency restarts get scheduled with different actions? Should the action be allocated (maybe not) for each caller, or should there be a check if an operation with conflicting action is already scheduled?

If this was already considered and thought it is not an issue:

Reviewed-by: Matti Vaittinen <mazziesaccount@gmail.com>

__hw_protection_trigger (née __hw_protection_shutdown) has this at its start:

  static atomic_t allow_proceed = ATOMIC_INIT(1);

  /* Shutdown should be initiated only once. */
  if (!atomic_dec_and_test(&allow_proceed))
          return;

It's thus not possible to have a later emergency restart race against the first.

Ah, indeed. I missed this. Thanks for the clarification! :)

Yours,
	-- Matti