LinuxLists.cc - [PATCH v3] watchdog: s3c2410: Fix potential deadlock on &wdt->lock

2023-07-05 09:23:17

Subject: [PATCH v3] watchdog: s3c2410: Fix potential deadlock on &wdt->lock

As &wdt->lock is acquired by hard irq s3c2410wdt_irq(),
other acquisition of the same lock under process context should
disable irq, otherwise deadlock could happen if the
irq preempt the execution while the lock is held in process context
on the same CPU.

[Deadlock Scenario]
s3c2410wdt_suspend()
-> s3c2410wdt_stop()
-> spin_lock(&wdt->lock)
<irq iterrupt>
-> s3c2410wdt_irq()
-> s3c2410wdt_keepalive()
-> spin_lock(&wdt->lock) (deadlock here)

[Deadlock Scenario]
s3c2410wdt_probe()
-> s3c2410wdt_start()
-> spin_lock(&wdt->lock)
<irq iterrupt>
-> s3c2410wdt_irq()
-> s3c2410wdt_keepalive()
-> spin_lock(&wdt->lock) (deadlock here)

[Deadlock Scenario]
s3c2410wdt_keepalive()
-> spin_lock(&wdt->lock)
<irq iterrupt>
-> s3c2410wdt_irq()
-> s3c2410wdt_keepalive()
-> spin_lock(&wdt->lock) (deadlock here)

This flaw was found by an experimental static analysis tool I am
developing for irq-related deadlock, which reported the above
warning when analyzing the linux kernel 6.4-rc7 release.

The tentative patch fix the potential deadlock by spin_lock_irqsave()
under process context.

Signed-off-by: Chengfeng Ye <[email protected]>
---
drivers/watchdog/s3c2410_wdt.c | 15 +++++++++------
1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/drivers/watchdog/s3c2410_wdt.c b/drivers/watchdog/s3c2410_wdt.c
index 95416a9bdd4b..e1dc71ece01e 100644
--- a/drivers/watchdog/s3c2410_wdt.c
+++ b/drivers/watchdog/s3c2410_wdt.c
@@ -379,10 +379,11 @@ static int s3c2410wdt_enable(struct s3c2410_wdt *wdt, bool en)
static int s3c2410wdt_keepalive(struct watchdog_device *wdd)
{
struct s3c2410_wdt *wdt = watchdog_get_drvdata(wdd);
+ unsigned long flags;

- spin_lock(&wdt->lock);
+ spin_lock_irqsave(&wdt->lock, flags);
writel(wdt->count, wdt->reg_base + S3C2410_WTCNT);
- spin_unlock(&wdt->lock);
+ spin_unlock_irqrestore(&wdt->lock, flags);

return 0;
}
@@ -399,10 +400,11 @@ static void __s3c2410wdt_stop(struct s3c2410_wdt *wdt)
static int s3c2410wdt_stop(struct watchdog_device *wdd)
{
struct s3c2410_wdt *wdt = watchdog_get_drvdata(wdd);
+ unsigned long flags;

- spin_lock(&wdt->lock);
+ spin_lock_irqsave(&wdt->lock, flags);
__s3c2410wdt_stop(wdt);
- spin_unlock(&wdt->lock);
+ spin_unlock_irqrestore(&wdt->lock, flags);

return 0;
}
@@ -411,8 +413,9 @@ static int s3c2410wdt_start(struct watchdog_device *wdd)
{
unsigned long wtcon;
struct s3c2410_wdt *wdt = watchdog_get_drvdata(wdd);
+ unsigned long flags;

- spin_lock(&wdt->lock);
+ spin_lock_irqsave(&wdt->lock, flags);

__s3c2410wdt_stop(wdt);

@@ -433,7 +436,7 @@ static int s3c2410wdt_start(struct watchdog_device *wdd)
writel(wdt->count, wdt->reg_base + S3C2410_WTDAT);
writel(wdt->count, wdt->reg_base + S3C2410_WTCNT);
writel(wtcon, wdt->reg_base + S3C2410_WTCON);
- spin_unlock(&wdt->lock);
+ spin_unlock_irqrestore(&wdt->lock, flags);

return 0;
}
--
2.17.1

2023-07-05 10:15:41

by Krzysztof Kozlowski

[permalink] [raw]

Subject: Re: [PATCH v3] watchdog: s3c2410: Fix potential deadlock on &wdt->lock

On 05/07/2023 11:09, Chengfeng Ye wrote:
> As &wdt->lock is acquired by hard irq s3c2410wdt_irq(),
> other acquisition of the same lock under process context should
> disable irq, otherwise deadlock could happen if the
> irq preempt the execution while the lock is held in process context
> on the same CPU.

Reviewed-by: Krzysztof Kozlowski <[email protected]>

Best regards,
Krzysztof

2023-07-05 12:59:45

by Guenter Roeck

[permalink] [raw]

Subject: Re: [PATCH v3] watchdog: s3c2410: Fix potential deadlock on &wdt->lock

2023-07-05 14:49:21

by Chengfeng Ye

[permalink] [raw]

Subject: Re: [PATCH v3] watchdog: s3c2410: Fix potential deadlock on &wdt->lock

> I am sure you know what you changed in each version of your patches. I don't.
> Please provide change logs when you send new versions of your patches.

No problem, this is the change log for this patch.
---
Changes in v3:
- Also use spin_lock_irqsave() in s3c2410wdt_keepalive().
---
---
Changes in v2:
- Use function name instead of line number in the commit message.
---

Best Regards,
Chengfeng

2023-07-05 15:20:42

by Guenter Roeck

[permalink] [raw]

Subject: Re: [PATCH v3] watchdog: s3c2410: Fix potential deadlock on &wdt->lock

On 7/5/23 02:09, Chengfeng Ye wrote:
> As &wdt->lock is acquired by hard irq s3c2410wdt_irq(),
> other acquisition of the same lock under process context should
> disable irq, otherwise deadlock could happen if the
> irq preempt the execution while the lock is held in process context
> on the same CPU.
>
> [Deadlock Scenario]
> s3c2410wdt_suspend()
> -> s3c2410wdt_stop()
> -> spin_lock(&wdt->lock)
> <irq iterrupt>
> -> s3c2410wdt_irq()
> -> s3c2410wdt_keepalive()
> -> spin_lock(&wdt->lock) (deadlock here)
>
> [Deadlock Scenario]
> s3c2410wdt_probe()
> -> s3c2410wdt_start()
> -> spin_lock(&wdt->lock)
> <irq iterrupt>
> -> s3c2410wdt_irq()
> -> s3c2410wdt_keepalive()
> -> spin_lock(&wdt->lock) (deadlock here)
>
> [Deadlock Scenario]
> s3c2410wdt_keepalive()
> -> spin_lock(&wdt->lock)
> <irq iterrupt>
> -> s3c2410wdt_irq()
> -> s3c2410wdt_keepalive()
> -> spin_lock(&wdt->lock) (deadlock here)
>
> This flaw was found by an experimental static analysis tool I am
> developing for irq-related deadlock, which reported the above
> warning when analyzing the linux kernel 6.4-rc7 release.
>
> The tentative patch fix the potential deadlock by spin_lock_irqsave()
> under process context.
>
> Signed-off-by: Chengfeng Ye <[email protected]>

Reviewed-by: Guenter Roeck <[email protected]>

> ---
> drivers/watchdog/s3c2410_wdt.c | 15 +++++++++------
> 1 file changed, 9 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/watchdog/s3c2410_wdt.c b/drivers/watchdog/s3c2410_wdt.c
> index 95416a9bdd4b..e1dc71ece01e 100644
> --- a/drivers/watchdog/s3c2410_wdt.c
> +++ b/drivers/watchdog/s3c2410_wdt.c
> @@ -379,10 +379,11 @@ static int s3c2410wdt_enable(struct s3c2410_wdt *wdt, bool en)
> static int s3c2410wdt_keepalive(struct watchdog_device *wdd)
> {
> struct s3c2410_wdt *wdt = watchdog_get_drvdata(wdd);
> + unsigned long flags;
>
> - spin_lock(&wdt->lock);
> + spin_lock_irqsave(&wdt->lock, flags);
> writel(wdt->count, wdt->reg_base + S3C2410_WTCNT);
> - spin_unlock(&wdt->lock);
> + spin_unlock_irqrestore(&wdt->lock, flags);
>
> return 0;
> }
> @@ -399,10 +400,11 @@ static void __s3c2410wdt_stop(struct s3c2410_wdt *wdt)
> static int s3c2410wdt_stop(struct watchdog_device *wdd)
> {
> struct s3c2410_wdt *wdt = watchdog_get_drvdata(wdd);
> + unsigned long flags;
>
> - spin_lock(&wdt->lock);
> + spin_lock_irqsave(&wdt->lock, flags);
> __s3c2410wdt_stop(wdt);
> - spin_unlock(&wdt->lock);
> + spin_unlock_irqrestore(&wdt->lock, flags);
>
> return 0;
> }
> @@ -411,8 +413,9 @@ static int s3c2410wdt_start(struct watchdog_device *wdd)
> {
> unsigned long wtcon;
> struct s3c2410_wdt *wdt = watchdog_get_drvdata(wdd);
> + unsigned long flags;
>
> - spin_lock(&wdt->lock);
> + spin_lock_irqsave(&wdt->lock, flags);
>
> __s3c2410wdt_stop(wdt);
>
> @@ -433,7 +436,7 @@ static int s3c2410wdt_start(struct watchdog_device *wdd)
> writel(wdt->count, wdt->reg_base + S3C2410_WTDAT);
> writel(wdt->count, wdt->reg_base + S3C2410_WTCNT);
> writel(wtcon, wdt->reg_base + S3C2410_WTCON);
> - spin_unlock(&wdt->lock);
> + spin_unlock_irqrestore(&wdt->lock, flags);
>
> return 0;
> }

2023-07-19 19:28:45

by Chengfeng Ye

[permalink] [raw]

Subject: Re: [PATCH v3] watchdog: s3c2410: Fix potential deadlock on &wdt->lock

> Reviewed-by: Krzysztof Kozlowski <[email protected]>
> Reviewed-by: Guenter Roeck <[email protected]>

Thanks much for your time in reviewing the patch :)

Best Regards,
Chengfeng