Skip to contentSkip to conversation

leds: bcm63x8: improve read and write functions

PATCH Superseded Álvaro Fernández Rojas 3 participants 11 messages Feb 23, 2021 – Feb 24, 2021 LED

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover leds: bcm63x8: improve read and write functions

This code is proven to work in BMIPS BE/LE and ARM BE/LE. See bcm2835-rng and bcmgenet.c:
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/char/hw_random/bcm2835-rng.c#L42-L60
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/net/ethernet/broadcom/genet/bcmgenet.c#L71-L88

Álvaro Fernández Rojas (2):
leds: bcm6328: improve write and read functions
leds: bcm6358: improve write and read functions

 drivers/leds/leds-bcm6328.c | 25 +++++++++++++------------
 drivers/leds/leds-bcm6358.c | 25 +++++++++++++------------
 2 files changed, 26 insertions(+), 24 deletions(-)
Signature · 1 line
2.20.1
#1 lore raw

PATCH 1/2 leds: bcm6328: improve write and read functions

This is proven to work in BMIPS BE/LE and ARM BE/LE, as used in bcm2835-rng and bcmgenet drivers.
Both should also be inline functions.

---
 drivers/leds/leds-bcm6328.c | 25 +++++++++++++------------
 1 file changed, 13 insertions(+), 12 deletions(-)
drivers/leds/leds-bcm6328.c +13 -12
index 226d17d253ed..112c9c858432 100644
@@ -75,22 +75,23 @@ struct bcm6328_led {
 	bool active_low;
 };
 
-static void bcm6328_led_write(void __iomem *reg, unsigned long data)
+static inline void bcm6328_led_write(void __iomem *reg, unsigned long data)
 {
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	iowrite32be(data, reg);
-#else
-	writel(data, reg);
-#endif
+	/* MIPS chips strapped for BE will automagically configure the
+	 * peripheral registers for CPU-native byte order.
+	 */
+	if (IS_ENABLED(CONFIG_MIPS) && IS_ENABLED(CONFIG_CPU_BIG_ENDIAN))
+		__raw_writel(data, reg);
+	else
+		writel_relaxed(data, reg);
 }
 
-static unsigned long bcm6328_led_read(void __iomem *reg)
+static inline unsigned long bcm6328_led_read(void __iomem *reg)
 {
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	return ioread32be(reg);
-#else
-	return readl(reg);
-#endif
+	if (IS_ENABLED(CONFIG_MIPS) && IS_ENABLED(CONFIG_CPU_BIG_ENDIAN))
+		return __raw_readl(reg);
+	else
+		return readl_relaxed(reg);
 }
 
 /**
Signature · 1 line
2.20.1
#10 lore raw

PATCH 2/2 leds: bcm6358: improve write and read functions

In reply to Álvaro Fernández Rojas

This is proven to work in BMIPS BE/LE and ARM BE/LE, as used in bcm2835-rng and bcmgenet drivers.
Both should also be inline functions.

---
 drivers/leds/leds-bcm6358.c | 25 +++++++++++++------------
 1 file changed, 13 insertions(+), 12 deletions(-)
drivers/leds/leds-bcm6358.c +13 -12
index 9d2e487fa08a..775b8c8b562a 100644
@@ -43,22 +43,23 @@ struct bcm6358_led {
 	bool active_low;
 };
 
-static void bcm6358_led_write(void __iomem *reg, unsigned long data)
+static inline void bcm6358_led_write(void __iomem *reg, unsigned long data)
 {
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	iowrite32be(data, reg);
-#else
-	writel(data, reg);
-#endif
+	/* MIPS chips strapped for BE will automagically configure the
+	 * peripheral registers for CPU-native byte order.
+	 */
+	if (IS_ENABLED(CONFIG_MIPS) && IS_ENABLED(CONFIG_CPU_BIG_ENDIAN))
+		__raw_writel(data, reg);
+	else
+		writel_relaxed(data, reg);
 }
 
-static unsigned long bcm6358_led_read(void __iomem *reg)
+static inline unsigned long bcm6358_led_read(void __iomem *reg)
 {
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	return ioread32be(reg);
-#else
-	return readl(reg);
-#endif
+	if (IS_ENABLED(CONFIG_MIPS) && IS_ENABLED(CONFIG_CPU_BIG_ENDIAN))
+		return __raw_readl(reg);
+	else
+		return readl_relaxed(reg);
 }
 
 static unsigned long bcm6358_led_busy(void __iomem *mem)
Signature · 1 line
2.20.1
#2 lore raw

Re: [PATCH 1/2] leds: bcm6328: improve write and read functions

In reply to Álvaro Fernández Rojas
Quoted 6 lines · Álvaro Fernández Rojas

This is proven to work in BMIPS BE/LE and ARM BE/LE, as used in bcm2835-rng and bcmgenet drivers.
Both should also be inline functions.

drivers/leds/leds-bcm6328.c struct bcm6328_led { · line 82

Quoted 8 lines
drivers/leds/leds-bcm6328.c +3 -5
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	iowrite32be(data, reg);
-#else
-	writel(data, reg);
-#endif
+	/* MIPS chips strapped for BE will automagically configure the
+	 * peripheral registers for CPU-native byte order.
+	 */

Bad comment style.

drivers/leds/leds-bcm6328.c struct bcm6328_led { · line 87

Quoted 5 lines
drivers/leds/leds-bcm6328.c +4 -0
+	if (IS_ENABLED(CONFIG_MIPS) && IS_ENABLED(CONFIG_CPU_BIG_ENDIAN))
+		__raw_writel(data, reg);
+	else
+		writel_relaxed(data, reg);
 }

Code does not match comment (still need to do conversion on non-MIPS?), and it certainly should not be here (do all mipsen behave like that?!), and it really should not be converting to _relaxed at the same time.

Best regards,
Pavel

Signature · 1 line
http://www.livejournal.com/~pavelmachek
#3 lore raw

Re: [PATCH 1/2] leds: bcm6328: improve write and read functions

Hi Pavel,

Quoted 9 lines · Pavel Machek
Quoted 3 lines · Álvaro Fernández Rojas

This is proven to work in BMIPS BE/LE and ARM BE/LE, as used in bcm2835-rng and bcmgenet drivers.
Both should also be inline functions.

Quoted 10 lines

drivers/leds/leds-bcm6328.c struct bcm6328_led { · line 82

Quoted 8 lines
drivers/leds/leds-bcm6328.c +3 -5
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	iowrite32be(data, reg);
-#else
-	writel(data, reg);
-#endif
+	/* MIPS chips strapped for BE will automagically configure the
+	 * peripheral registers for CPU-native byte order.
+	 */

Bad comment style.

I just wanted to copy the same comment as the one in bcm2835-rng and bcmgenet…
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/char/hw_random/bcm2835-rng.c#L42-L60
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/net/ethernet/broadcom/genet/bcmgenet.c#L71-L88

Quoted 11 lines
Quoted 6 lines
drivers/leds/leds-bcm6328.c +4 -0
+	if (IS_ENABLED(CONFIG_MIPS) && IS_ENABLED(CONFIG_CPU_BIG_ENDIAN))
+		__raw_writel(data, reg);
+	else
+		writel_relaxed(data, reg);

}

Code does not match comment (still need to do conversion on non-MIPS?), and it certainly should not be here (do all mipsen behave like that?!), and it really should not be converting to _relaxed at the same time.

I think it's because non-MIPS BE exposes that as little endian, but Florian can probably help us with that…

Quoted 5 lines

Best regards,
Pavel

Signature · 1 line
http://www.livejournal.com/~pavelmachek

Best regards,
Álvaro.

#4 lore raw

Re: [PATCH 1/2] leds: bcm6328: improve write and read functions

Hi!

Quoted 17 lines
Quoted 6 lines
Quoted 3 lines

This is proven to work in BMIPS BE/LE and ARM BE/LE, as used in bcm2835-rng and bcmgenet drivers.
Both should also be inline functions.

Quoted 10 lines

drivers/leds/leds-bcm6328.c struct bcm6328_led { · line 82

Quoted 8 lines
drivers/leds/leds-bcm6328.c +3 -5
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	iowrite32be(data, reg);
-#else
-	writel(data, reg);
-#endif
+	/* MIPS chips strapped for BE will automagically configure the
+	 * peripheral registers for CPU-native byte order.
+	 */

Bad comment style.

Quoted 4 lines

I just wanted to copy the same comment as the one in bcm2835-rng and bcmgenet…
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/char/hw_random/bcm2835-rng.c#L42-L60
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/net/ethernet/broadcom/genet/bcmgenet.c#L71-L88

Yeah, but ideally you should not be copying comments; there should be one central place which does it and does it right.

Pavel

Signature · 1 line
http://www.livejournal.com/~pavelmachek
#5 lore raw

Re: [PATCH 1/2] leds: bcm6328: improve write and read functions

Quoted 23 lines · Pavel Machek

Hi!

Quoted 17 lines
Quoted 16 lines
Quoted 6 lines

This is proven to work in BMIPS BE/LE and ARM BE/LE, as used in bcm2835-rng and bcmgenet drivers.
Both should also be inline functions.

drivers/leds/leds-bcm6328.c struct bcm6328_led { · line 82

Quoted 8 lines
drivers/leds/leds-bcm6328.c +3 -5
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	iowrite32be(data, reg);
-#else
-	writel(data, reg);
-#endif
+	/* MIPS chips strapped for BE will automagically configure the
+	 * peripheral registers for CPU-native byte order.
+	 */

Bad comment style.

Quoted 7 lines
Quoted 4 lines

I just wanted to copy the same comment as the one in bcm2835-rng and bcmgenet…
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/char/hw_random/bcm2835-rng.c#L42-L60
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/net/ethernet/broadcom/genet/bcmgenet.c#L71-L88

Yeah, but ideally you should not be copying comments; there should be one central place which does it and does it right.

I’m open to suggestions :).
Which central place would be a good place for you?

Quoted 5 lines

Pavel

Signature · 1 line
http://www.livejournal.com/~pavelmachek
#6 lore raw

Re: [PATCH 1/2] leds: bcm6328: improve write and read functions

Quoted 28 lines · Álvaro Fernández Rojas
Quoted 28 lines · Pavel Machek

Hi!

Quoted 21 lines
Quoted 6 lines
Quoted 3 lines

This is proven to work in BMIPS BE/LE and ARM BE/LE, as used in bcm2835-rng and bcmgenet drivers.
Both should also be inline functions.

Quoted 10 lines

drivers/leds/leds-bcm6328.c struct bcm6328_led { · line 82

Quoted 8 lines
drivers/leds/leds-bcm6328.c +3 -5
-#ifdef CONFIG_CPU_BIG_ENDIAN
-	iowrite32be(data, reg);
-#else
-	writel(data, reg);
-#endif
+	/* MIPS chips strapped for BE will automagically configure the
+	 * peripheral registers for CPU-native byte order.
+	 */

Bad comment style.

I just wanted to copy the same comment as the one in bcm2835-rng and bcmgenet…
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/char/hw_random/bcm2835-rng.c#L42-L60
https://github.com/torvalds/linux/blob/3b9cdafb5358eb9f3790de2f728f765fef100731/drivers/net/ethernet/broadcom/genet/bcmgenet.c#L71-L88

Quoted 5 lines
Quoted 2 lines

Yeah, but ideally you should not be copying comments; there should be one central place which does it and does it right.

I’m open to suggestions :).
Which central place would be a good place for you?

I did consider creating an include/linux/brcm/brcm_io.h header or something like that but I am really not sure what the benefit would be.

As far as using _relaxed() this is absolutely correct because the bus logic that connects the CPU to its on-chip registers is non re-ordering non posted. That is true on the MIPS BE/LE and ARM when configured in LE or BE.

We need the swapping for ARM because when running in ARM BE32, the data is going to be in the host CPU endian, but the register bus is hard wired to little endian.

Signature · 1 line
Florian
#7 lore raw

Re: [PATCH 1/2] leds: bcm6328: improve write and read functions

Hi!

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines

Yeah, but ideally you should not be copying comments; there should be one central place which does it and does it right.

I’m open to suggestions :).
Which central place would be a good place for you?

I did consider creating an include/linux/brcm/brcm_io.h header or something like that but I am really not sure what the benefit would be.

Less code duplication? It is immediately clear that driver including this is specific for brcm SoCs and would not try to work somewhere else?

Quoted 4 lines

As far as using _relaxed() this is absolutely correct because the bus logic that connects the CPU to its on-chip registers is non re-ordering non posted. That is true on the MIPS BE/LE and ARM when configured in LE or BE.

If that's right on particular SoC, then _relaxed and normal versions should be same; drivers still need to use normal versions, because they may be running on different SoC...?

Quoted 3 lines

We need the swapping for ARM because when running in ARM BE32, the data is going to be in the host CPU endian, but the register bus is hard wired to little endian.

Yeah I see you need to do some byteswapping. But I'm pretty sure not all MIPS BE boxes do the magic swapping, right? And drivers/leds is not a place where you encode knowledge about SoC byte swapping.

Best regards,
Pavel

Signature · 1 line
http://www.livejournal.com/~pavelmachek
#8 lore raw

Re: [PATCH 1/2] leds: bcm6328: improve write and read functions

Quoted 8 lines · Pavel Machek

Hi!

Quoted 6 lines
Quoted 5 lines
Quoted 2 lines

Yeah, but ideally you should not be copying comments; there should be one central place which does it and does it right.

I’m open to suggestions :).
Which central place would be a good place for you?

Quoted 6 lines
Quoted 3 lines

I did consider creating an include/linux/brcm/brcm_io.h header or something like that but I am really not sure what the benefit would be.

Less code duplication? It is immediately clear that driver including this is specific for brcm SoCs and would not try to work somewhere else?

Yes maybe, there still does not feel like this deserves a shared header, but as long as the generated code is the same, why not.

Quoted 9 lines
Quoted 5 lines

As far as using _relaxed() this is absolutely correct because the bus logic that connects the CPU to its on-chip registers is non re-ordering non posted. That is true on the MIPS BE/LE and ARM when configured in LE or BE.

If that's right on particular SoC, then _relaxed and normal versions should be same; drivers still need to use normal versions, because they may be running on different SoC...?

readl() includes barriers and read_relaxed() does not, hence the difference in the name. There is no need to pay the price of a barrier when a) the bus architecture guarantees non re-ordering and posting and that statement is true on all the SoCs where these peripherals are used, and b) you have worked on fine tuning your drivers to get the most performance out of them.

Quoted 8 lines
Quoted 4 lines

We need the swapping for ARM because when running in ARM BE32, the data is going to be in the host CPU endian, but the register bus is hard wired to little endian.

Yeah I see you need to do some byteswapping. But I'm pretty sure not all MIPS BE boxes do the magic swapping, right? And drivers/leds is not a place where you encode knowledge about SoC byte swapping.

The Broadcom MIPS CPUs (we have/had an architectural license) can be strapped for BE or LE, and when that happens the bridge that connects to the registers follows the CPU's endian, which is why __raw_{read,write}l is appropriate for these specific peripherals.

Given these peripherals can only be used on CPUs/SoCs made by Broadcom, any argument about portability to other SoCs is moot.

Signature · 1 line
Florian
#9 lore raw

Re: [PATCH 1/2] leds: bcm6328: improve write and read functions

Hi!

Quoted 5 lines
Quoted 2 lines

Less code duplication? It is immediately clear that driver including this is specific for brcm SoCs and would not try to work somewhere else?

Yes maybe, there still does not feel like this deserves a shared header, but as long as the generated code is the same, why not.

Ok, it seems patch is not needed at all, after all?

Quoted 10 lines
Quoted 9 lines
Quoted 5 lines

As far as using _relaxed() this is absolutely correct because the bus logic that connects the CPU to its on-chip registers is non re-ordering non posted. That is true on the MIPS BE/LE and ARM when configured in LE or BE.

If that's right on particular SoC, then _relaxed and normal versions should be same; drivers still need to use normal versions, because they may be running on different SoC...?

Quoted 6 lines

readl() includes barriers and read_relaxed() does not, hence the difference in the name. There is no need to pay the price of a barrier when a) the bus architecture guarantees non re-ordering and posting and that statement is true on all the SoCs where these peripherals are used, and b) you have worked on fine tuning your drivers to get the most performance out of them.

Exactly. When bus architecture guarantees ... readl and read_relaxed can be the same. That knowledge should be in architecture code, not in drivers.

(But it does not matter much when the drivers are
architecture-specific).

Quoted 2 lines

Given these peripherals can only be used on CPUs/SoCs made by Broadcom, any argument about portability to other SoCs is moot.

Ok.
Pavel

Signature · 1 line
http://www.livejournal.com/~pavelmachek