On Tue, 14 Jul 2015 16:45:52 +0200,
Nariman Poushin wrote:
>
> We treat a delay in a sequence the same way we treat a page change as
> they are logically similar in that you can coalesce all write before
> a delay (in the same way you can coalesce all writes before a page
> change is needed)
>
> Signed-off-by: Nariman Poushin <nariman at opensource.wolfsonmicro.com>
> ---
> drivers/base/regmap/regmap.c | 65
> ++++++++++++++++++++++++++++++++++++++++----
> 1 file changed, 59 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/base/regmap/regmap.c b/drivers/base/regmap/regmap.c
> index 0a849ee..a67473c2 100644
> --- a/drivers/base/regmap/regmap.c
> +++ b/drivers/base/regmap/regmap.c
> @@ -18,6 +18,7 @@
> #include <linux/of.h>
> #include <linux/rbtree.h>
> #include <linux/sched.h>
> +#include <linux/delay.h>
>
> #define CREATE_TRACE_POINTS
> #include "trace.h"
> @@ -47,6 +48,17 @@ static int _regmap_bus_reg_write(void *context, unsigned
> int reg,
> static int _regmap_bus_raw_write(void *context, unsigned int reg,
> unsigned int val);
>
> +static void regmap_sequence_delay(unsigned int delay_us)
> +{
> + /* For small delays it isn't worth setting up the hrtimers
> + * so fall back on udelay
> + */
> + if (delay_us < 10)
> + udelay(delay_us);
> + else
> + usleep_range(delay_us, delay_us * 2);
> +}
I think usleep_range() can't be used for fast_io, which is performed
inside a spinlock. And, the locking can't be known explicitly,
e.g. the caller may set its own config->lock, so even the check for
fast_io isn't enough.
Takashi
