Skip to contentSkip to conversation

iio: gyro: mpu3050: Use get_unaligned_leXX()

PATCH In review Andy Shevchenko 4 participants 6 messages Oct 26, 2020 – Nov 5, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 iio: gyro: mpu3050: Use get_unaligned_leXX()

This makes the driver code slightly easier to read.

---
 drivers/iio/gyro/mpu3050-core.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
drivers/iio/gyro/mpu3050-core.c +5 -5
index 00e58060968c..42b6ccb13568 100644
@@ -827,15 +827,15 @@ static int mpu3050_hw_init(struct mpu3050 *mpu3050)
 		 "die ID: %04X, wafer ID: %02X, A lot ID: %04X, "
 		 "W lot ID: %03X, WP ID: %01X, rev ID: %02X\n",
 		 /* Die ID, bits 0-12 */
-		 (otp[1] << 8 | otp[0]) & 0x1fff,
+		 get_unaligned_le16(&otp[0]) & GENMASK(12, 0),
 		 /* Wafer ID, bits 13-17 */
-		 ((otp[2] << 8 | otp[1]) & 0x03e0) >> 5,
+		 (get_unaligned_le16(&otp[1]) & GENMASK(9, 5)) >> 5,
 		 /* A lot ID, bits 18-33 */
-		 ((otp[4] << 16 | otp[3] << 8 | otp[2]) & 0x3fffc) >> 2,
+		 (get_unaligned_le24(&otp[2]) & GENMASK(17, 2)) >> 2,
 		 /* W lot ID, bits 34-45 */
-		 ((otp[5] << 8 | otp[4]) & 0x3ffc) >> 2,
+		 (get_unaligned_le16(&otp[4]) & GENMASK(13, 2)) >> 2,
 		 /* WP ID, bits 47-49 */
-		 ((otp[6] << 8 | otp[5]) & 0x0380) >> 7,
+		 (get_unaligned_le16(&otp[5]) & GENMASK(9, 7)) >> 7,
 		 /* rev ID, bits 50-55 */
 		 otp[6] >> 2);
 
Signature · 1 line
2.28.0
maintainer
#2 lore raw
In reply to Andy Shevchenko
Quoted 4 lines · On Mon, 26 Oct 2020 19:53:40 +0200 Andy Shevchenko…

This makes the driver code slightly easier to read.

What you have is good, but could we do better?

Why not use an __le64 to grab the whole thing, convert it to a local cpu endian u64 then use FIELD_GET or similar to pull out the right bits. Would be fairly self documenting as well which is always nice.

Or am I missing something?

Jonathan

Quoted 29 lines
---
 drivers/iio/gyro/mpu3050-core.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
drivers/iio/gyro/mpu3050-core.c +5 -5
index 00e58060968c..42b6ccb13568 100644
@@ -827,15 +827,15 @@ static int mpu3050_hw_init(struct mpu3050 *mpu3050)
 		 "die ID: %04X, wafer ID: %02X, A lot ID: %04X, "
 		 "W lot ID: %03X, WP ID: %01X, rev ID: %02X\n",
 		 /* Die ID, bits 0-12 */
-		 (otp[1] << 8 | otp[0]) & 0x1fff,
+		 get_unaligned_le16(&otp[0]) & GENMASK(12, 0),
 		 /* Wafer ID, bits 13-17 */
-		 ((otp[2] << 8 | otp[1]) & 0x03e0) >> 5,
+		 (get_unaligned_le16(&otp[1]) & GENMASK(9, 5)) >> 5,
 		 /* A lot ID, bits 18-33 */
-		 ((otp[4] << 16 | otp[3] << 8 | otp[2]) & 0x3fffc) >> 2,
+		 (get_unaligned_le24(&otp[2]) & GENMASK(17, 2)) >> 2,
 		 /* W lot ID, bits 34-45 */
-		 ((otp[5] << 8 | otp[4]) & 0x3ffc) >> 2,
+		 (get_unaligned_le16(&otp[4]) & GENMASK(13, 2)) >> 2,
 		 /* WP ID, bits 47-49 */
-		 ((otp[6] << 8 | otp[5]) & 0x0380) >> 7,
+		 (get_unaligned_le16(&otp[5]) & GENMASK(9, 7)) >> 7,
 		 /* rev ID, bits 50-55 */
 		 otp[6] >> 2);
 
#3 lore raw
Quoted 9 lines · Jonathan Cameron
Quoted 4 lines · On Mon, 26 Oct 2020 19:53:40 +0200 Andy Shevchenko…

This makes the driver code slightly easier to read.

Quoted 7 lines

What you have is good, but could we do better?

Why not use an __le64 to grab the whole thing, convert it to a local cpu endian u64 then use FIELD_GET or similar to pull out the right bits. Would be fairly self documenting as well which is always nice.

Or am I missing something?

My laziness. I shared this as a leftover from a big series of converting a lot of drivers to get_unligned_.eXX(). If you consider this is not good enough, just don't apply (no hard feelings :-) I will drop it from my local branches.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#4 lore raw
Quoted 16 lines · Andy Shevchenko
Quoted 15 lines · Jonathan Cameron
Quoted 4 lines · On Mon, 26 Oct 2020 19:53:40 +0200 Andy Shevchenko…

This makes the driver code slightly easier to read.

What you have is good, but could we do better?

Why not use an __le64 to grab the whole thing, convert it to a local cpu endian u64 then use FIELD_GET or similar to pull out the right bits. Would be fairly self documenting as well which is always nice.

Quoted 6 lines
Quoted 1 line

Or am I missing something?

My laziness. I shared this as a leftover from a big series of converting a lot of drivers to get_unligned_.eXX(). If you consider this is not good enough, just don't apply (no hard feelings :-) I will drop it from my local branches.

I would suggest to apply it anyways because the kernel will look better after than before this patch.

Possibly Jonathan could just slip in a TODO comment for this?

Yours,
Linus Walleij

#5 lore raw
Quoted 25 lines · Linus Walleij
Quoted 16 lines · Andy Shevchenko
Quoted 15 lines · Jonathan Cameron
Quoted 4 lines · On Mon, 26 Oct 2020 19:53:40 +0200 Andy Shevchenko…

This makes the driver code slightly easier to read.

What you have is good, but could we do better?

Why not use an __le64 to grab the whole thing, convert it to a local cpu endian u64 then use FIELD_GET or similar to pull out the right bits. Would be fairly self documenting as well which is always nice.

Quoted 6 lines
Quoted 1 line

Or am I missing something?

My laziness. I shared this as a leftover from a big series of converting a lot of drivers to get_unligned_.eXX(). If you consider this is not good enough, just don't apply (no hard feelings :-) I will drop it from my local branches.

Quoted 4 lines

I would suggest to apply it anyways because the kernel will look better after than before this patch.

Possibly Jonathan could just slip in a TODO comment for this?

Sorry, I have already dropped, but if you convince Jonathan to apply, I agree on whatever you propose (TODO comment as I see). I think patchwork or lore still keeps a track of this.

Signature · 2 lines
With Best Regards,
Andy Shevchenko