[v1] rtc-cmos: use spin_lock_irqsave in cmos_interrupt

[DRAFT PATCH] rtc-cmos: use spin_lock_irqsave in cmos_interrupt

Posted by Mateusz Jończyk 6 months, 2 weeks ago

cmos_interrupt() can also be called also in non-interrupt contexts, such
as in ACPI handlers via rtc_handler(). Therefore, usage of
spin_lock(&rtc_lock) is insecure. Use spin_lock_irqsave() / etc.
instead.

Remove the local_irq_disable() hacks in cmos_check_wkalrm() and add a
comment so that these _irqsave / _irqrestore will not be disabled again,
as in

commit 6950d046eb6e ("rtc: cmos: Replace spin_lock_irqsave with spin_lock in hard IRQ")

Untested yet.

Signed-off-by: Mateusz Jończyk <mat.jonczyk@o2.pl>
Fixes: 13be2efc390a ("rtc: cmos: Disable irq around direct invocation of cmos_interrupt()")
---
 drivers/rtc/rtc-cmos.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/drivers/rtc/rtc-cmos.c b/drivers/rtc/rtc-cmos.c
index 8172869bd3d7..399bb82e6153 100644
--- a/drivers/rtc/rtc-cmos.c
+++ b/drivers/rtc/rtc-cmos.c
@@ -692,8 +692,12 @@ static irqreturn_t cmos_interrupt(int irq, void *p)
 {
 	u8		irqstat;
 	u8		rtc_control;
+	unsigned long	flags;
 
-	spin_lock(&rtc_lock);
+	/* We cannot use spin_lock() here, as cmos_interrupt() is also called
+	 * in non-irq context.
+	 */
+	spin_lock_irqsave(&rtc_lock, flags);
 
 	/* When the HPET interrupt handler calls us, the interrupt
 	 * status is passed as arg1 instead of the irq number.  But
@@ -727,7 +731,7 @@ static irqreturn_t cmos_interrupt(int irq, void *p)
 			hpet_mask_rtc_irq_bit(RTC_AIE);
 		CMOS_READ(RTC_INTR_FLAGS);
 	}
-	spin_unlock(&rtc_lock);
+	spin_unlock_irqrestore(&rtc_lock, flags);
 
 	if (is_intr(irqstat)) {
 		rtc_update_irq(p, 1, irqstat);
@@ -1295,9 +1299,7 @@ static void cmos_check_wkalrm(struct device *dev)
 	 * ACK the rtc irq here
 	 */
 	if (t_now >= cmos->alarm_expires && cmos_use_acpi_alarm()) {
-		local_irq_disable();
 		cmos_interrupt(0, (void *)cmos->rtc);
-		local_irq_enable();
 		return;
 	}
 
-- 
2.25.1

Re: [DRAFT PATCH] rtc-cmos: use spin_lock_irqsave in cmos_interrupt

Posted by Sebastian Andrzej Siewior 6 months, 2 weeks ago

On 2025-06-02 22:20:19 [+0200], Mateusz Jończyk wrote:
> cmos_interrupt() can also be called also in non-interrupt contexts, such
> as in ACPI handlers via rtc_handler(). Therefore, usage of
> spin_lock(&rtc_lock) is insecure. Use spin_lock_irqsave() / etc.
> instead.
> 
> Remove the local_irq_disable() hacks in cmos_check_wkalrm() and add a
> comment so that these _irqsave / _irqrestore will not be disabled again,
> as in

This also broke PREEMPT_RT due to disable-IRQ -> spinlock_t.

> commit 6950d046eb6e ("rtc: cmos: Replace spin_lock_irqsave with spin_lock in hard IRQ")
> 
> Untested yet.
> 
> Signed-off-by: Mateusz Jończyk <mat.jonczyk@o2.pl>
> Fixes: 13be2efc390a ("rtc: cmos: Disable irq around direct invocation of cmos_interrupt()")

Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Sebastian

Re: [DRAFT PATCH] rtc-cmos: use spin_lock_irqsave in cmos_interrupt

Posted by Chris Bainbridge 6 months, 2 weeks ago

On Mon, Jun 02, 2025 at 10:20:19PM +0200, Mateusz Jończyk wrote:
> cmos_interrupt() can also be called also in non-interrupt contexts, such
> as in ACPI handlers via rtc_handler(). Therefore, usage of
> spin_lock(&rtc_lock) is insecure. Use spin_lock_irqsave() / etc.
> instead.
> 
> Remove the local_irq_disable() hacks in cmos_check_wkalrm() and add a
> comment so that these _irqsave / _irqrestore will not be disabled again,
> as in
> 
> commit 6950d046eb6e ("rtc: cmos: Replace spin_lock_irqsave with spin_lock in hard IRQ")
> 
> Untested yet.
> 
> Signed-off-by: Mateusz Jończyk <mat.jonczyk@o2.pl>
> Fixes: 13be2efc390a ("rtc: cmos: Disable irq around direct invocation of cmos_interrupt()")

I tested this, and can confirm it does fix the "inconsistent lock state"
error in my test case.