[PATCH] watchdog: fix GETTIMEOUT ioctl in booke_wdt

[PATCH] watchdog: fix GETTIMEOUT ioctl in booke_wdt

[Dmitry Artamonow]

Commit dcfb748422 ([WATCHDOG] fix book E watchdog to take
WDIOC_SETTIMEOUT arg in seconds) fixed SETTIMEOUT ioctl
to use seconds as a parameter instead of some hardware-specific
"period", but missed to apply similar changes to GETTIMEOUT,
so it still returns "period" value. Let's fix it!
Also, while at it, make SETTIMEOUT ioctl return real timeout
value as it should do according to the documentation.

Signed-off-by: Dmitry Artamonow <mad_soft@inbox.ru>
---
 drivers/watchdog/booke_wdt.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletions(-)

NOTE: 
I'm no longer have a hardware, so it haven't been tested actually. 
It would be nice if anyone could test it on some BookE platform.

diff --git a/drivers/watchdog/booke_wdt.c b/drivers/watchdog/booke_wdt.c
index 337265b..7c0fdfc 100644
--- a/drivers/watchdog/booke_wdt.c
+++ b/drivers/watchdog/booke_wdt.c
@@ -198,9 +198,13 @@ static long booke_wdt_ioctl(struct file *file,
 		booke_wdt_period = tmp;
 #endif
 		booke_wdt_set();
-		return 0;
+		/* Fall */
 	case WDIOC_GETTIMEOUT:
+#ifdef	CONFIG_FSL_BOOKE
+		return put_user(period_to_sec(booke_wdt_period), p);
+#else
 		return put_user(booke_wdt_period, p);
+#endif
 	default:
 		return -ENOTTY;
 	}
-- 
1.7.5.4

Re: [PATCH] watchdog: fix GETTIMEOUT ioctl in booke_wdt

[Timur Tabi]

Dmitry Artamonow wrote:
> +#ifdef	CONFIG_FSL_BOOKE
> +		return put_user(period_to_sec(booke_wdt_period), p);
> +#else

period_to_sec() returns an unsigned long long, which means that the driver
will write 8 bytes instead of 4 to user space.  You need to use a
temporary u32, just like WDIOC_SETTIMEOUT:

		if (get_user(tmp, p))
			return -EFAULT;


-- 
Timur Tabi
Linux kernel developer at Freescale

Re: [PATCH] watchdog: fix GETTIMEOUT ioctl in booke_wdt

[Timur Tabi]

Timur Tabi wrote:
> Dmitry Artamonow wrote:
>> +#ifdef	CONFIG_FSL_BOOKE
>> +		return put_user(period_to_sec(booke_wdt_period), p);
>> +#else
> 
> period_to_sec() returns an unsigned long long, which means that the driver
> will write 8 bytes instead of 4 to user space. 

Ok, I take that back.  put_user() uses the size of the pointer, which
makes more sense anyway.


-- 
Timur Tabi
Linux kernel developer at Freescale

Re: [PATCH] watchdog: fix GETTIMEOUT ioctl in booke_wdt

[Timur Tabi]

Dmitry Artamonow wrote:
> Commit dcfb748422 ([WATCHDOG] fix book E watchdog to take
> WDIOC_SETTIMEOUT arg in seconds) fixed SETTIMEOUT ioctl
> to use seconds as a parameter instead of some hardware-specific
> "period", but missed to apply similar changes to GETTIMEOUT,
> so it still returns "period" value. Let's fix it!
> Also, while at it, make SETTIMEOUT ioctl return real timeout
> value as it should do according to the documentation.
> 
> Signed-off-by: Dmitry Artamonow <mad_soft@inbox.ru>
> ---

Acked-by: Timur Tabi <timur@freescale.com>

I don't have an easy way to test the watchdog ioctls.

-- 
Timur Tabi
Linux kernel developer at Freescale