[U-Boot] zynq: timer: Fix wrong timer calculation

From: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>

Fix wrong timer calculation in get_timer_masked incase of
overflow.
This fixes the issue of getting wrong time from get_timer()
calls.

Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
---

 arch/arm/cpu/armv7/zynq/timer.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Hello Michal,

Michal Simek wrote on 2015-04-16:
> From: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>
> 
> Fix wrong timer calculation in get_timer_masked incase of overflow. This
> fixes the issue of getting wrong time from get_timer() calls.
> 
> Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
> ---
> 
>  arch/arm/cpu/armv7/zynq/timer.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> diff --git a/arch/arm/cpu/armv7/zynq/timer.c
> b/arch/arm/cpu/armv7/zynq/timer.c index 303dbcfceafb..5ed9642df9b3
> 100644 --- a/arch/arm/cpu/armv7/zynq/timer.c +++
> b/arch/arm/cpu/armv7/zynq/timer.c @@ -93,7 +93,9 @@ ulong
> get_timer_masked(void)
>  		gd->arch.tbl += gd->arch.lastinc - now;
>  	} else {
>  		/* We have an overflow ... */
> -		gd->arch.tbl += gd->arch.lastinc + TIMER_LOAD_VAL - now + 1;
> +		gd->arch.tbl += gd->arch.lastinc + (TIMER_LOAD_VAL /
> +				(gd->arch.timer_rate_hz / CONFIG_SYS_HZ))
> -
> +				now + 1;
>  	}
>  	gd->arch.lastinc = now;

I had similar problems with the MIPS timer code some time ago.
I solved it by switching to the common timer implementation:
http://git.denx.de/cgi-bin/gitweb.cgi?p=u-boot.git;a=commitdiff;h=a18a477147ce2493ef9ee93b8981b34929fc48a5

I  don't know the arm/zynq hardware in detail, but if you use some free running counter for this,
the switch should be simple.

As you can see from the diff, the previous code was also doing some complex checks to detect an overflow.
With the common code, it could be reduced to two simple  functions (where one could be removed, 
if we update all configs to define a different macro for the timer rate).

Best Regards,
Thomas

Hi,

On 04/16/2015 02:57 PM, thomas.langer@lantiq.com wrote:
> Hello Michal,
> 
> Michal Simek wrote on 2015-04-16:
>> From: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>
>>
>> Fix wrong timer calculation in get_timer_masked incase of overflow. This
>> fixes the issue of getting wrong time from get_timer() calls.
>>
>> Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
>> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
>> ---
>>
>>  arch/arm/cpu/armv7/zynq/timer.c | 4 +++-
>>  1 file changed, 3 insertions(+), 1 deletion(-)
>> diff --git a/arch/arm/cpu/armv7/zynq/timer.c
>> b/arch/arm/cpu/armv7/zynq/timer.c index 303dbcfceafb..5ed9642df9b3
>> 100644 --- a/arch/arm/cpu/armv7/zynq/timer.c +++
>> b/arch/arm/cpu/armv7/zynq/timer.c @@ -93,7 +93,9 @@ ulong
>> get_timer_masked(void)
>>  		gd->arch.tbl += gd->arch.lastinc - now;
>>  	} else {
>>  		/* We have an overflow ... */
>> -		gd->arch.tbl += gd->arch.lastinc + TIMER_LOAD_VAL - now + 1;
>> +		gd->arch.tbl += gd->arch.lastinc + (TIMER_LOAD_VAL /
>> +				(gd->arch.timer_rate_hz / CONFIG_SYS_HZ))
>> -
>> +				now + 1;
>>  	}
>>  	gd->arch.lastinc = now;
> 
> I had similar problems with the MIPS timer code some time ago.
> I solved it by switching to the common timer implementation:
> http://git.denx.de/cgi-bin/gitweb.cgi?p=u-boot.git;a=commitdiff;h=a18a477147ce2493ef9ee93b8981b34929fc48a5
> 
> I  don't know the arm/zynq hardware in detail, but if you use some free running counter for this,
> the switch should be simple.
> 
> As you can see from the diff, the previous code was also doing some complex checks to detect an overflow.
> With the common code, it could be reduced to two simple  functions (where one could be removed, 
> if we update all configs to define a different macro for the timer rate).

Thanks - I have changed the code and will do more testing and will
update. I think make sense to apply this because it is fixes real
problem and then add that new one on the top of it.

Definitely thanks for hint we should do it in past.

Thanks,
Michal