Re: [PATCH 4/4] hv_util: improve time adjustment accuracy by disabling interrupts

[Stephen Hemminger]

On Mon,  2 Jan 2017 20:41:14 +0100
Vitaly Kuznetsov <vkuznets@xxxxxxxxxx> wrote:

> If we happen to receive interrupts during hv_set_host_time() execution
> our adjustments may get inaccurate. Make the whole function atomic.
> Unfortunately, we can's call do_settimeofday64() with interrupts
> disabled as some cross-CPU work is being done but this call happens
> very rarely.
> 
> Signed-off-by: Vitaly Kuznetsov <vkuznets@xxxxxxxxxx>
> ---
>  drivers/hv/hv_util.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/hv/hv_util.c b/drivers/hv/hv_util.c
> index 4c0fbb0..233d5cb 100644
> --- a/drivers/hv/hv_util.c
> +++ b/drivers/hv/hv_util.c
> @@ -186,6 +186,9 @@ static void hv_set_host_time(struct work_struct *work)
>  	u64 newtime;
>  	struct timespec64 host_ts, our_ts;
>  	struct timex txc = {0};
> +	unsigned long flags;
> +
> +	local_irq_save(flags);
>  
>  	wrk = container_of(work, struct adj_time_work, work);
>  
> @@ -214,6 +217,7 @@ static void hv_set_host_time(struct work_struct *work)
>  
>  	/* Try adjusting time by using phase adjustment if possible */
>  	if (abs(delta) > MAXPHASE) {
> +		local_irq_restore(flags);
>  		do_settimeofday64(&host_ts);
>  		return;
>  	}
> @@ -225,6 +229,8 @@ static void hv_set_host_time(struct work_struct *work)
>  	txc.status = STA_PLL;
>  	txc.offset = delta;
>  	do_adjtimex(&txc);
> +
> +	local_irq_restore(flags);
 

Yes, it should be atomic, but local irq save/restore is not sufficient protection
because it does not protect against premptible kernel. Why not a mutex? or a spinlock?