Re: [PATCH 09/11] RISC-V: Support per-hart timebase-frequency

[Atish Patra]

On 8/2/18 4:50 AM, Christoph Hellwig wrote:
> From: Palmer Dabbelt <palmer@xxxxxxxxxx>
>
> Follow the updated DT specs and read the timebase-frequency from the
> CPU 0 node.

However, the DT in the HighFive Unleashed has the entry at the wrong place.

Even the example in github also at wrong place.
https://github.com/riscv/riscv-device-tree-doc/pull/8/commits/2461d481329c55005fcbe684f0d6bdb3b7f0a432

DT should be consistent between Documentation and the one in the 
hardware. I can fix them in bbl & submit a bbl patch. But I am not sure 
if that's an acceptable way to do it.

Regards,
Atish
> Signed-off-by: Palmer Dabbelt <palmer@xxxxxxxxxxx>
> [hch: updated changelog]
> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> ---
>   arch/riscv/kernel/time.c | 17 ++++++++++++-----
>   1 file changed, 12 insertions(+), 5 deletions(-)
>
> diff --git a/arch/riscv/kernel/time.c b/arch/riscv/kernel/time.c
> index 0df9b2cbd645..1bb01dc2d0f1 100644
> --- a/arch/riscv/kernel/time.c
> +++ b/arch/riscv/kernel/time.c
> @@ -24,17 +24,24 @@ void __init init_clockevent(void)
>   	csr_set(sie, SIE_STIE);
>   }
>   
> -void __init time_init(void)
> +static long __init timebase_frequency(void)
>   {
>   	struct device_node *cpu;
>   	u32 prop;
>   
>   	cpu = of_find_node_by_path("/cpus");
> -	if (!cpu || of_property_read_u32(cpu, "timebase-frequency", &prop))
> -		panic(KERN_WARNING "RISC-V system with no 'timebase-frequency' in DTS\n");
> -	riscv_timebase = prop;
> +	if (cpu && !of_property_read_u32(cpu, "timebase-frequency", &prop))
> +		return prop;
> +	cpu = of_find_node_by_path("/cpus/cpu@0");
> +	if (cpu && !of_property_read_u32(cpu, "timebase-frequency", &prop))
> +		return prop;
>   
> -	lpj_fine = riscv_timebase / HZ;
> +	panic(KERN_WARNING "RISC-V system with no 'timebase-frequency' in DTS\n");
> +}
>   
> +void __init time_init(void)
> +{
> +	riscv_timebase = timebase_frequency();
> +	lpj_fine = riscv_timebase / HZ;
>   	init_clockevent();
>   }