Instead of if-else-if, simply call roundup_pow_of_two(BITS_PER_BYTES()).
Note, there is no division assumed as compiler may optimize it away.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/spi/spi.c | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
index 6d74218cf38e..125dea8fae00 100644
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -3643,13 +3643,7 @@ int spi_split_transfers_maxwords(struct spi_controller *ctlr,
 		size_t maxsize;
 		int ret;
 
-		if (xfer->bits_per_word <= 8)
-			maxsize = maxwords;
-		else if (xfer->bits_per_word <= 16)
-			maxsize = 2 * maxwords;
-		else
-			maxsize = 4 * maxwords;
-
+		maxsize = maxwords * roundup_pow_of_two(BITS_TO_BYTES(xfer->bits_per_word));
 		if (xfer->len > maxsize) {
 			ret = __spi_split_transfer_maxsize(ctlr, msg, &xfer,
 							   maxsize, gfp);
-- 
2.40.0.1.gaa8946217a0b

On Mon, Jul 10, 2023 at 06:49:20PM +0300, Andy Shevchenko wrote:

> -		if (xfer->bits_per_word <= 8)
> -			maxsize = maxwords;
> -		else if (xfer->bits_per_word <= 16)
> -			maxsize = 2 * maxwords;
> -		else
> -			maxsize = 4 * maxwords;
> -
> +		maxsize = maxwords * roundup_pow_of_two(BITS_TO_BYTES(xfer->bits_per_word));

This will change the behaviour if bits_per_word is more than 32.  That
is validated out elsewhere but I shouldn't have had to go around
checking the code to confirm that this is the case.  This is the sort of
thing that should be highlighted when doing this sort of edge case
stylistic change.

On Mon, Jul 10, 2023 at 05:56:59PM +0100, Mark Brown wrote:
> On Mon, Jul 10, 2023 at 06:49:20PM +0300, Andy Shevchenko wrote:
> 
> > -		if (xfer->bits_per_word <= 8)
> > -			maxsize = maxwords;
> > -		else if (xfer->bits_per_word <= 16)
> > -			maxsize = 2 * maxwords;
> > -		else
> > -			maxsize = 4 * maxwords;
> > -
> > +		maxsize = maxwords * roundup_pow_of_two(BITS_TO_BYTES(xfer->bits_per_word));
> 
> This will change the behaviour if bits_per_word is more than 32.  That
> is validated out elsewhere but I shouldn't have had to go around
> checking the code to confirm that this is the case.  This is the sort of
> thing that should be highlighted when doing this sort of edge case
> stylistic change.

Right, I have to add this into commit message of v3.

-- 
With Best Regards,
Andy Shevchenko