regmap has introduced a maple tree based register cache which makes use of
this more advanced data structure which has been added to the kernel
recently. Maple trees are much flatter than rbtrees, meaning that they do
not grow to such depths when the register map is sparse which makes access
a bit more efficient. The maple tree cache type is still a bit of a work
in progress but should be effective for some devices already.

The wm831x devices have a pretty sparse register map and being always on
devices never do cache syncs so don't hit the major current disadvantage
so they should be good candiates for using the maple tree cache.  Update
to do so, there should be little if any visible difference at runtime.

Signed-off-by: Mark Brown <broonie@kernel.org>
---
 drivers/mfd/wm831x-core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/mfd/wm831x-core.c b/drivers/mfd/wm831x-core.c
index d2f444d2ae78..e86b6a4896a6 100644
--- a/drivers/mfd/wm831x-core.c
+++ b/drivers/mfd/wm831x-core.c
@@ -1430,7 +1430,7 @@ struct regmap_config wm831x_regmap_config = {
 	.reg_bits = 16,
 	.val_bits = 16,
 
-	.cache_type = REGCACHE_RBTREE,
+	.cache_type = REGCACHE_MAPLE,
 
 	.max_register = WM831X_DBE_CHECK_DATA,
 	.readable_reg = wm831x_reg_readable,

---
base-commit: 4a670ac3e75e517c96cbd01ef870dbd598c3ce71
change-id: 20230419-mfd-wm831x-maple-8da691bc3efc

Best regards,
-- 
Mark Brown <broonie@kernel.org>

On Tue, 25 Apr 2023, Mark Brown wrote:

> regmap has introduced a maple tree based register cache which makes use of
> this more advanced data structure which has been added to the kernel
> recently. Maple trees are much flatter than rbtrees, meaning that they do
> not grow to such depths when the register map is sparse which makes access
> a bit more efficient. The maple tree cache type is still a bit of a work
> in progress but should be effective for some devices already.
> 
> The wm831x devices have a pretty sparse register map and being always on
> devices never do cache syncs so don't hit the major current disadvantage
> so they should be good candiates for using the maple tree cache.  Update
> to do so, there should be little if any visible difference at runtime.
> 
> Signed-off-by: Mark Brown <broonie@kernel.org>
> ---
>  drivers/mfd/wm831x-core.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)

Applied, thanks

-- 
Lee Jones [李琼斯]

On Thu, Apr 27, 2023 at 04:11:34PM +0100, Lee Jones wrote:
> On Tue, 25 Apr 2023, Mark Brown wrote:
> 
> > regmap has introduced a maple tree based register cache which makes use of
> > this more advanced data structure which has been added to the kernel
> > recently. Maple trees are much flatter than rbtrees, meaning that they do

> Applied, thanks

Note that the maple tree support for regmap has only landed in Linus'
tree during the current merge window, I wasn't anticipating this being
applied until -rc1 due to it being sent after the merge window opened.

On Thu, 27 Apr 2023, Mark Brown wrote:

> On Thu, Apr 27, 2023 at 04:11:34PM +0100, Lee Jones wrote:
> > On Tue, 25 Apr 2023, Mark Brown wrote:
> > 
> > > regmap has introduced a maple tree based register cache which makes use of
> > > this more advanced data structure which has been added to the kernel
> > > recently. Maple trees are much flatter than rbtrees, meaning that they do
> 
> > Applied, thanks
> 
> Note that the maple tree support for regmap has only landed in Linus'
> tree during the current merge window, I wasn't anticipating this being
> applied until -rc1 due to it being sent after the merge window opened.

It's been applied to for-mfd-next-next which will become for-mfd-next
*after* the merge-window closes.  IOW, it's due for v6.5.

-- 
Lee Jones [李琼斯]

On Tue, Apr 25, 2023 at 07:03:33PM +0100, Mark Brown wrote:
> regmap has introduced a maple tree based register cache which makes use of
> this more advanced data structure which has been added to the kernel
> recently. Maple trees are much flatter than rbtrees, meaning that they do
> not grow to such depths when the register map is sparse which makes access
> a bit more efficient. The maple tree cache type is still a bit of a work
> in progress but should be effective for some devices already.
> 
> The wm831x devices have a pretty sparse register map and being always on
> devices never do cache syncs so don't hit the major current disadvantage
> so they should be good candiates for using the maple tree cache.  Update
> to do so, there should be little if any visible difference at runtime.
> 
> Signed-off-by: Mark Brown <broonie@kernel.org>
> ---

Reviewed-by: Charles Keepax <ckeepax@opensource.cirrus.com>

Thanks,
Charles