drivers/rtc/rtc-loongson.c | 2 ++ 1 file changed, 2 insertions(+)
When an application sets and enables an alarm on Loongson RTC devices,
the alarm notification fails to propagate to userspace because the
ACPI event handler omits calling rtc_update_irq().
As a result, processes waiting via select() or poll() on RTC device
files fail to receive alarm notifications.
Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family chips")
Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
---
drivers/rtc/rtc-loongson.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
index 97e5625c064c..0c573f198f63 100644
--- a/drivers/rtc/rtc-loongson.c
+++ b/drivers/rtc/rtc-loongson.c
@@ -129,6 +129,8 @@ static u32 loongson_rtc_handler(void *id)
{
struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
+ rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
+
spin_lock(&priv->lock);
/* Disable RTC alarm wakeup and interrupt */
writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
--
2.33.0
Hi Dalin:
Thanks for your patch.
On 2025/4/29 14:27, Liu Dalin wrote:
> When an application sets and enables an alarm on Loongson RTC devices,
> the alarm notification fails to propagate to userspace because the
> ACPI event handler omits calling rtc_update_irq().
>
> As a result, processes waiting via select() or poll() on RTC device
> files fail to receive alarm notifications.
>
> Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family chips")
> Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
> ---
> drivers/rtc/rtc-loongson.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
> index 97e5625c064c..0c573f198f63 100644
> --- a/drivers/rtc/rtc-loongson.c
> +++ b/drivers/rtc/rtc-loongson.c
> @@ -129,6 +129,8 @@ static u32 loongson_rtc_handler(void *id)
> {
> struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
>
> + rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
> +
While testing the patch, I noticed that interrupts are triggered
multiple times (/proc/interrupt), not sure if you have the same issue.
I think we need a similar operation to loongson_rtc_isr() to clear the
interrupt:
regmap_write(priv->regmap, TOY_MATCH0_REG, 0);
> spin_lock(&priv->lock);
> /* Disable RTC alarm wakeup and interrupt */
> writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
Thanks.
Binbin
Hi Binbin:
The described issue does not happened with or without the patch by test, as the interrupt is managed by acpi.
The device info and test result are as follows.
1. Device info
[系统未激活][root@mail test]# dmidecode -q
BIOS Information
Vendor: ZD-TECH
Version: V09
Release Date: 06/15/2022
ROM Size: 8 MB
Characteristics:
PCI is supported
BIOS is upgradeable
BIOS shadowing is allowed
Boot from CD is supported
Selectable boot is supported
BIOS ROM is socketed
Serial services are supported (int 14h)
USB legacy is supported
Function key-initiated network boot is supported
UEFI is supported
BIOS Revision: 4.0
Firmware Revision: 0.4
System Information
Manufacturer: GEIT
Product Name: UT6000-LB5
Version: 1.0
Serial Number: TBD by OEM
UUID: 00112233-4455-6677-8899-aabbccddeeff
Wake-up Type: Power Switch
SKU Number: Loongson_SKU
Family: Type1Family
Base Board Information
Manufacturer: GEIT
Product Name: GG-3A5000-02
Version: 1.0
Serial Number: Chassis Board Serial#To Be Filled By O.E.M
Asset Tag: Type2 - Board Asset Tag
Features:
Board is a hosting board
Board is replaceable
Location In Chassis: Type2 - Board Chassis Location
Type: Motherboard
...
2. The system exhibits a timeout error when waiting for alarm signal response without the patch.
[系统未激活][root@mail test]# cat /proc/interrupts |grep acpi
21: 3 0 0 0 PCH PIC 1 acpi
[系统未激活][root@mail test]#
[系统未激活][root@mail test]# ./rtc01
RTC READ TEST:
RTC READ TEST Passed
Current RTC date/time is 8-5-2025, 05:50:05.
RTC ALARM TEST :Alarm time set to 05:50:10.
Waiting 5 seconds for the alarm...
Timed out waiting for the alarm
RTC UPDATE INTERRUPTS TEST :
RTC_UIE_ON not supported
RTC Tests Done!
[系统未激活][root@mail test]#
[系统未激活][root@mail test]# cat /proc/interrupts |grep acpi
21: 4 0 0 0 PCH PIC 1 acpi
[系统未激活][root@mail test]#
3. There is no error with the patch applied and the interrupts are triggered ok
[系统未激活][root@mail test]# cat /proc/interrupts |grep acpi
21: 0 0 0 0 PCH PIC 1 acpi
[系统未激活][root@mail test]#
[系统未激活][root@mail test]# ./rtc01
RTC READ TEST:
RTC READ TEST Passed
Current RTC date/time is 8-5-2025, 06:09:03.
RTC ALARM TEST :Alarm time set to 06:09:08.
Waiting 5 seconds for the alarm...
Alarm rang.
RTC ALARM TEST Passed
RTC UPDATE INTERRUPTS TEST :
RTC_UIE_ON not supported
RTC Tests Done!
[系统未激活][root@mail test]#
[系统未激活][root@mail test]# cat /proc/interrupts |grep acpi
21: 1 0 0 0 PCH PIC 1 acpi
[系统未激活][root@mail test]#
From: Binbin Zhou
Date: 2025-05-08 11:12
To: Liu Dalin; alexandre.belloni; wangming01
CC: chenhuacai; gaojuxin; git; jiaxun.yang; keguang.zhang; lixuefeng; linux-rtc; linux-kernel; zhoubb.aaron
Subject: Re: [PATCH] rtc: loongson: Add missing alarm notifications for ACPI RTC events
Hi Dalin:
Thanks for your patch.
On 2025/4/29 14:27, Liu Dalin wrote:
> When an application sets and enables an alarm on Loongson RTC devices,
> the alarm notification fails to propagate to userspace because the
> ACPI event handler omits calling rtc_update_irq().
>
> As a result, processes waiting via select() or poll() on RTC device
> files fail to receive alarm notifications.
>
> Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family chips")
> Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
> ---
> drivers/rtc/rtc-loongson.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
> index 97e5625c064c..0c573f198f63 100644
> --- a/drivers/rtc/rtc-loongson.c
> +++ b/drivers/rtc/rtc-loongson.c
> @@ -129,6 +129,8 @@ static u32 loongson_rtc_handler(void *id)
> {
> struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
>
> + rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
> +
While testing the patch, I noticed that interrupts are triggered
multiple times (/proc/interrupt), not sure if you have the same issue.
I think we need a similar operation to loongson_rtc_isr() to clear the
interrupt:
regmap_write(priv->regmap, TOY_MATCH0_REG, 0);
> spin_lock(&priv->lock);
> /* Disable RTC alarm wakeup and interrupt */
> writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
Thanks.
Binbin
When an application sets and enables an alarm on Loongson RTC devices,
the alarm notification fails to propagate to userspace because the
ACPI event handler omits calling rtc_update_irq().
As a result, processes waiting via select() or poll() on RTC device
files fail to receive alarm notifications.
The ACPI interrupt is also triggered multiple times. In loongson_rtc_handler,
we need to clear TOY_MATCH0_REG to resolve this issue.
Fixes: 09471d8f5b39 ("rtc: loongson: clear TOY_MATCH0_REG in loongson_rtc_isr()")
Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family chips")
Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
---
drivers/rtc/rtc-loongson.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
index 97e5625c064c..2ca7ffd5d7a9 100644
--- a/drivers/rtc/rtc-loongson.c
+++ b/drivers/rtc/rtc-loongson.c
@@ -129,6 +129,14 @@ static u32 loongson_rtc_handler(void *id)
{
struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
+ rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
+
+ /*
+ * The TOY_MATCH0_REG should be cleared 0 here,
+ * otherwise the interrupt cannot be cleared.
+ */
+ regmap_write(priv->regmap, TOY_MATCH0_REG, 0);
+
spin_lock(&priv->lock);
/* Disable RTC alarm wakeup and interrupt */
writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
--
2.33.0
On Fri, 09 May 2025 16:44:16 +0800, Liu Dalin wrote:
> When an application sets and enables an alarm on Loongson RTC devices,
> the alarm notification fails to propagate to userspace because the
> ACPI event handler omits calling rtc_update_irq().
>
> As a result, processes waiting via select() or poll() on RTC device
> files fail to receive alarm notifications.
>
> [...]
Applied, thanks!
[1/1] rtc: loongson: Add missing alarm notifications for ACPI RTC events
https://git.kernel.org/abelloni/c/5af9f1fa5768
Best regards,
--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
On 2025/5/9 16:44, Liu Dalin wrote:
> When an application sets and enables an alarm on Loongson RTC devices,
> the alarm notification fails to propagate to userspace because the
> ACPI event handler omits calling rtc_update_irq().
>
> As a result, processes waiting via select() or poll() on RTC device
> files fail to receive alarm notifications.
>
> The ACPI interrupt is also triggered multiple times. In loongson_rtc_handler,
> we need to clear TOY_MATCH0_REG to resolve this issue.
>
> Fixes: 09471d8f5b39 ("rtc: loongson: clear TOY_MATCH0_REG in loongson_rtc_isr()")
> Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family chips")
> Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
Reviewed-by: Binbin Zhou <zhoubinbin@loongson.cn>
> ---
> drivers/rtc/rtc-loongson.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
> index 97e5625c064c..2ca7ffd5d7a9 100644
> --- a/drivers/rtc/rtc-loongson.c
> +++ b/drivers/rtc/rtc-loongson.c
> @@ -129,6 +129,14 @@ static u32 loongson_rtc_handler(void *id)
> {
> struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
>
> + rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
> +
> + /*
> + * The TOY_MATCH0_REG should be cleared 0 here,
> + * otherwise the interrupt cannot be cleared.
> + */
> + regmap_write(priv->regmap, TOY_MATCH0_REG, 0);
> +
> spin_lock(&priv->lock);
> /* Disable RTC alarm wakeup and interrupt */
> writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
Thanks.
Binbin
When an application sets and enables an alarm on Loongson RTC devices,
the alarm notification fails to propagate to userspace because the
ACPI event handler omits calling rtc_update_irq().
As a result, processes waiting via select() or poll() on RTC device
files fail to receive alarm notifications.
Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family chips")
Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
---
drivers/rtc/rtc-loongson.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
index 97e5625c064c..2ca7ffd5d7a9 100644
--- a/drivers/rtc/rtc-loongson.c
+++ b/drivers/rtc/rtc-loongson.c
@@ -129,6 +129,14 @@ static u32 loongson_rtc_handler(void *id)
{
struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
+ rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
+
+ /*
+ * The TOY_MATCH0_REG should be cleared 0 here,
+ * otherwise the interrupt cannot be cleared.
+ */
+ regmap_write(priv->regmap, TOY_MATCH0_REG, 0);
+
spin_lock(&priv->lock);
/* Disable RTC alarm wakeup and interrupt */
writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
--
2.33.0
Hi:
On 2025/5/9 09:40, Liu Dalin wrote:
> When an application sets and enables an alarm on Loongson RTC devices,
> the alarm notification fails to propagate to userspace because the
> ACPI event handler omits calling rtc_update_irq().
>
> As a result, processes waiting via select() or poll() on RTC device
> files fail to receive alarm notifications.
>
> Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family chips")
> Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
Technically I don't think this is a compliant patch, firstly this patch
should be V2 and also you should add "Changlog" for describing the
differences from the previous patch.
Anyway, it's good to me.
Reviewed-by: Binbin Zhou <zhoubinbin@loongson.cn>
> ---
> drivers/rtc/rtc-loongson.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
> index 97e5625c064c..2ca7ffd5d7a9 100644
> --- a/drivers/rtc/rtc-loongson.c
> +++ b/drivers/rtc/rtc-loongson.c
> @@ -129,6 +129,14 @@ static u32 loongson_rtc_handler(void *id)
> {
> struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
>
> + rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
> +
> + /*
> + * The TOY_MATCH0_REG should be cleared 0 here,
> + * otherwise the interrupt cannot be cleared.
> + */
> + regmap_write(priv->regmap, TOY_MATCH0_REG, 0);
> +
> spin_lock(&priv->lock);
> /* Disable RTC alarm wakeup and interrupt */
> writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
Thanks.
Binbin
在 2025/5/9 15:12, Binbin Zhou 写道:
> Hi:
>
> On 2025/5/9 09:40, Liu Dalin wrote:
>> When an application sets and enables an alarm on Loongson RTC devices,
>> the alarm notification fails to propagate to userspace because the
>> ACPI event handler omits calling rtc_update_irq().
>>
>> As a result, processes waiting via select() or poll() on RTC device
>> files fail to receive alarm notifications.
>>
>> Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family chips")
>> Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
> Technically I don't think this is a compliant patch, firstly this patch
> should be V2 and also you should add "Changlog" for describing the
> differences from the previous patch.
> Anyway, it's good to me.
Ok! There is another problem, if regmap_write(priv->regmap,
TOY_MATCH0_REG, 0) move to the end of loongson_rtc_handler as
09471d8f5b39("rtc: loongson: clear TOY_MATCH0_REG in
loongson_rtc_isr()"), the interrupt also can be triggered multiple times.
So the regmap_write(priv->regmap,TOY_MATCH0_REG, 0) is move behind of
rtc_update_irq by me, then the interrupt was triggered OK.
Is there any problem ?
>
> Reviewed-by: Binbin Zhou <zhoubinbin@loongson.cn>
>
>> ---
>> drivers/rtc/rtc-loongson.c | 8 ++++++++
>> 1 file changed, 8 insertions(+)
>>
>> diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
>> index 97e5625c064c..2ca7ffd5d7a9 100644
>> --- a/drivers/rtc/rtc-loongson.c
>> +++ b/drivers/rtc/rtc-loongson.c
>> @@ -129,6 +129,14 @@ static u32 loongson_rtc_handler(void *id)
>> {
>> struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
>> + rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
>> +
>> + /*
>> + * The TOY_MATCH0_REG should be cleared 0 here,
>> + * otherwise the interrupt cannot be cleared.
>> + */
>> + regmap_write(priv->regmap, TOY_MATCH0_REG, 0);
>> +
>> spin_lock(&priv->lock);
>> /* Disable RTC alarm wakeup and interrupt */
>> writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
> Thanks.
> Binbin
>
>
>
On 2025/5/9 16:06, Liu Dalin wrote:
>
>
> 在 2025/5/9 15:12, Binbin Zhou 写道:
>> Hi:
>>
>> On 2025/5/9 09:40, Liu Dalin wrote:
>>> When an application sets and enables an alarm on Loongson RTC devices,
>>> the alarm notification fails to propagate to userspace because the
>>> ACPI event handler omits calling rtc_update_irq().
>>>
>>> As a result, processes waiting via select() or poll() on RTC device
>>> files fail to receive alarm notifications.
>>>
>>> Fixes: 1b733a9ebc3d ("rtc: Add rtc driver for the Loongson family
>>> chips")
>>> Signed-off-by: Liu Dalin <liudalin@kylinsec.com.cn>
>> Technically I don't think this is a compliant patch, firstly this
>> patch should be V2 and also you should add "Changlog" for describing
>> the differences from the previous patch.
>> Anyway, it's good to me.
>
> Ok! There is another problem, if regmap_write(priv->regmap,
> TOY_MATCH0_REG, 0) move to the end of loongson_rtc_handler as
> 09471d8f5b39("rtc: loongson: clear TOY_MATCH0_REG in
> loongson_rtc_isr()"), the interrupt also can be triggered multiple times.
>
> So the regmap_write(priv->regmap,TOY_MATCH0_REG, 0) is move behind of
> rtc_update_irq by me, then the interrupt was triggered OK.
>
> Is there any problem ?
It is ok. That's where I added it when I suggested it before.
Thanks.
Binbin
>
>>
>> Reviewed-by: Binbin Zhou <zhoubinbin@loongson.cn>
>>
>>> ---
>>> drivers/rtc/rtc-loongson.c | 8 ++++++++
>>> 1 file changed, 8 insertions(+)
>>>
>>> diff --git a/drivers/rtc/rtc-loongson.c b/drivers/rtc/rtc-loongson.c
>>> index 97e5625c064c..2ca7ffd5d7a9 100644
>>> --- a/drivers/rtc/rtc-loongson.c
>>> +++ b/drivers/rtc/rtc-loongson.c
>>> @@ -129,6 +129,14 @@ static u32 loongson_rtc_handler(void *id)
>>> {
>>> struct loongson_rtc_priv *priv = (struct loongson_rtc_priv *)id;
>>> + rtc_update_irq(priv->rtcdev, 1, RTC_AF | RTC_IRQF);
>>> +
>>> + /*
>>> + * The TOY_MATCH0_REG should be cleared 0 here,
>>> + * otherwise the interrupt cannot be cleared.
>>> + */
>>> + regmap_write(priv->regmap, TOY_MATCH0_REG, 0);
>>> +
>>> spin_lock(&priv->lock);
>>> /* Disable RTC alarm wakeup and interrupt */
>>> writel(readl(priv->pm_base + PM1_EN_REG) & ~RTC_EN,
>> Thanks.
>> Binbin
>>
>>
>>
© 2016 - 2026 Red Hat, Inc.