Skip to contentSkip to conversation

gpio: sch: Switch to memory mapped IO accessors

PATCH In review Andy Shevchenko 4 participants 11 messages Apr 10, 2024 – Apr 17, 2024

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

PATCH 1/1 gpio: sch: Switch to memory mapped IO accessors

Convert driver to use memory mapped IO accessors.

---
 drivers/gpio/gpio-sch.c | 25 +++++++++++++++----------
 1 file changed, 15 insertions(+), 10 deletions(-)
drivers/gpio/gpio-sch.c +15 -10
index e48392074e4b8..0c765558d03fc 100644
@@ -38,8 +38,8 @@
 
 struct sch_gpio {
 	struct gpio_chip chip;
+	void __iomem *regs;
 	spinlock_t lock;
-	unsigned short iobase;
 	unsigned short resume_base;
 
 	/* GPE handling */
@@ -75,7 +75,7 @@ static int sch_gpio_reg_get(struct sch_gpio *sch, unsigned int gpio, unsigned in
 	offset = sch_gpio_offset(sch, gpio, reg);
 	bit = sch_gpio_bit(sch, gpio);
 
-	reg_val = !!(inb(sch->iobase + offset) & BIT(bit));
+	reg_val = !!(ioread8(sch->regs + offset) & BIT(bit));
 
 	return reg_val;
 }
@@ -89,12 +89,14 @@ static void sch_gpio_reg_set(struct sch_gpio *sch, unsigned int gpio, unsigned i
 	offset = sch_gpio_offset(sch, gpio, reg);
 	bit = sch_gpio_bit(sch, gpio);
 
-	reg_val = inb(sch->iobase + offset);
+	reg_val = ioread8(sch->regs + offset);
 
 	if (val)
-		outb(reg_val | BIT(bit), sch->iobase + offset);
+		reg_val |= BIT(bit);
 	else
-		outb((reg_val & ~BIT(bit)), sch->iobase + offset);
+		reg_val &= ~BIT(bit);
+
+	iowrite8(reg_val, sch->regs + offset);
 }
 
 static int sch_gpio_direction_in(struct gpio_chip *gc, unsigned int gpio_num)
@@ -267,8 +269,8 @@ static u32 sch_gpio_gpe_handler(acpi_handle gpe_device, u32 gpe, void *context)
 
 	spin_lock_irqsave(&sch->lock, flags);
 
-	core_status = inl(sch->iobase + CORE_BANK_OFFSET + GTS);
-	resume_status = inl(sch->iobase + RESUME_BANK_OFFSET + GTS);
+	core_status = ioread32(sch->regs + CORE_BANK_OFFSET + GTS);
+	resume_status = ioread32(sch->regs + RESUME_BANK_OFFSET + GTS);
 
 	spin_unlock_irqrestore(&sch->lock, flags);
 
@@ -319,9 +321,11 @@ static int sch_gpio_install_gpe_handler(struct sch_gpio *sch)
 
 static int sch_gpio_probe(struct platform_device *pdev)
 {
+	struct device *dev = &pdev->dev;
 	struct gpio_irq_chip *girq;
 	struct sch_gpio *sch;
 	struct resource *res;
+	void __iomem *regs;
 	int ret;
 
 	sch = devm_kzalloc(&pdev->dev, sizeof(*sch), GFP_KERNEL);
@@ -332,12 +336,13 @@ static int sch_gpio_probe(struct platform_device *pdev)
 	if (!res)
 		return -EBUSY;
 
-	if (!devm_request_region(&pdev->dev, res->start, resource_size(res),
-				 pdev->name))
+	regs = devm_ioport_map(dev, res->start, resource_size(res));
+	if (!regs)
 		return -EBUSY;
 
+	sch->regs = regs;
+
 	spin_lock_init(&sch->lock);
-	sch->iobase = res->start;
 	sch->chip = sch_gpio_chip;
 	sch->chip.label = dev_name(&pdev->dev);
 	sch->chip.parent = &pdev->dev;
Signature · 1 line
2.44.0
#2 lore raw
Quoted 10 lines · Linus Walleij
Quoted 3 lines · Andy Shevchenko

Convert driver to use memory mapped IO accessors.

If this is the direction we want to go with port-mapped I/O then I'm in.

Thanks!

Quoted 2 lines

An ACK from William may be in order, he's usually keeping track of the port-mapped stuff.

William, are you okay to provide one?

Signature · 2 lines
With Best Regards,
Andy Shevchenko
Acked-by on v1 1/1
#3 lore raw
In reply to Andy Shevchenko
Quoted 3 lines · Andy Shevchenko

Convert driver to use memory mapped IO accessors.

A minor suggestion below, but I find this patch accepted as-is.

drivers/gpio/gpio-sch.c static int sch_gpio_install_gpe_handler(struct sch_gpio *sch) · line 324

Quoted 3 lines

static int sch_gpio_probe(struct platform_device *pdev) {

drivers/gpio/gpio-sch.c +1 -0
+	struct device *dev = &pdev->dev;

In general I think this is a good variable to define to simplify all the &pdev->dev appearing throughout this callback, but I'd rather have seen it as its own patch so we could change all the other uses of &pdev->dev at once without distracting from the memory-mapped I/O change of this particular patch. Not really necessary, but maybe at some point in the future a follow-up patch doing such a cleanup would be nice.

William Breathitt Gray

#4 lore raw
Quoted 6 lines · William Breathitt Gray
Quoted 3 lines · Andy Shevchenko

Convert driver to use memory mapped IO accessors.

Pushed to my review and testing queue, thank you!

Quoted 12 lines

A minor suggestion below, but I find this patch accepted as-is.

drivers/gpio/gpio-sch.c static int sch_gpio_install_gpe_handler(struct sch_gpio *sch) · line 324

Quoted 3 lines

static int sch_gpio_probe(struct platform_device *pdev) {

drivers/gpio/gpio-sch.c +1 -0
+	struct device *dev = &pdev->dev;

In general I think this is a good variable to define to simplify all the &pdev->dev appearing throughout this callback, but I'd rather have seen it as its own patch so we could change all the other uses of &pdev->dev at once without distracting from the memory-mapped I/O change of this particular patch. Not really necessary, but maybe at some point in the future a follow-up patch doing such a cleanup would be nice.

I avoid making unneeded churn on a line I have updated in this patch. That's why I introduced the local variable proactively. Yet, I can do another patch to clean up the driver based on the existence of this var.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#5 lore raw
In reply to William Breathitt Gray
Quoted 14 lines · William Breathitt Gray
Quoted 3 lines · Andy Shevchenko

Convert driver to use memory mapped IO accessors.

A minor suggestion below, but I find this patch accepted as-is.

Quoted 3 lines

static int sch_gpio_probe(struct platform_device *pdev) {
+ struct device *dev = &pdev->dev;

Quoted 8 lines

In general I think this is a good variable to define to simplify all the &pdev->dev appearing throughout this callback, but I'd rather have seen it as its own patch so we could change all the other uses of &pdev->dev at once without distracting from the memory-mapped I/O change of this particular patch. Not really necessary, but maybe at some point in the future a follow-up patch doing such a cleanup would be nice.

William Breathitt Gray

I applied it as is, if anyone wants it, this can be sent on top of it.

Bart

#6 lore raw
Quoted 1 line · Bartosz Golaszewski

On Sat, Apr 13, 2024 at 10:31 AM William Breathitt Gray <[email protected]> wrote:

...

Quoted 1 line

I applied it as is, if anyone wants it, this can be sent on top of it.

Thanks, but I assumed this should go via my tree and as PR to you. At least I have it already in my for-next.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#7 lore raw
Quoted 11 lines · Andy Shevchenko
Quoted 1 line · Bartosz Golaszewski

On Sat, Apr 13, 2024 at 10:31 AM William Breathitt Gray <[email protected]> wrote:

...

Quoted 1 line

I applied it as is, if anyone wants it, this can be sent on top of it.

Thanks, but I assumed this should go via my tree and as PR to you. At least I have it already in my for-next.

You didn't respond in any way about picking it up. Can you just drop it from your branch?

Bart

#8 lore raw
Quoted 4 lines · Bartosz Golaszewski
Quoted 2 lines · Andy Shevchenko
Quoted 1 line · Bartosz Golaszewski

On Sat, Apr 13, 2024 at 10:31 AM William Breathitt Gray <[email protected]> wrote:

...

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

I applied it as is, if anyone wants it, this can be sent on top of it.

Thanks, but I assumed this should go via my tree and as PR to you. At least I have it already in my for-next.

You didn't respond in any way about picking it up.

Hmm... I'm the author of it and I'm a maintainer for that driver. I'm not sure if it's mandatory to respond for that purpose. Usually I asked the opposite, i.e. when I'm not going to pick the thing up.

Quoted 2 lines

Can you just drop
it from your branch?

It's possible, but I will need to rebase, which is not a good thing to perform. What about just leaving it as is and letting git to (nicely) solve this?

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#9 lore raw
Quoted 23 lines · Andy Shevchenko
Quoted 4 lines · Bartosz Golaszewski
Quoted 2 lines · Andy Shevchenko
Quoted 1 line · Bartosz Golaszewski

On Sat, Apr 13, 2024 at 10:31 AM William Breathitt Gray <[email protected]> wrote:

...

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

I applied it as is, if anyone wants it, this can be sent on top of it.

Thanks, but I assumed this should go via my tree and as PR to you. At least I have it already in my for-next.

You didn't respond in any way about picking it up.

Hmm... I'm the author of it and I'm a maintainer for that driver. I'm not sure if it's mandatory to respond for that purpose. Usually I asked the opposite, i.e. when I'm not going to pick the thing up.

Quoted 2 lines

Can you just drop
it from your branch?

Quoted 7 lines

It's possible, but I will need to rebase, which is not a good thing to perform. What about just leaving it as is and letting git to (nicely) solve this?

--
With Best Regards,
Andy Shevchenko

It won't be solved nicely, we'll get a warning about the same commit appearing twice with different hashes.

Whatever, I dropped it from my tree, it was the HEAD anyway.

Bart

#10 lore raw
Quoted 7 lines · Bartosz Golaszewski
Quoted 5 lines · Andy Shevchenko
Quoted 4 lines · Bartosz Golaszewski
Quoted 2 lines · Andy Shevchenko
Quoted 1 line · Bartosz Golaszewski

On Sat, Apr 13, 2024 at 10:31 AM William Breathitt Gray <[email protected]> wrote:

...

Quoted 14 lines
Quoted 14 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line

I applied it as is, if anyone wants it, this can be sent on top of it.

Thanks, but I assumed this should go via my tree and as PR to you. At least I have it already in my for-next.

You didn't respond in any way about picking it up.

Hmm... I'm the author of it and I'm a maintainer for that driver. I'm not sure if it's mandatory to respond for that purpose. Usually I asked the opposite, i.e. when I'm not going to pick the thing up.

Quoted 2 lines

Can you just drop
it from your branch?

Quoted 6 lines
Quoted 3 lines

It's possible, but I will need to rebase, which is not a good thing to perform. What about just leaving it as is and letting git to (nicely) solve this?

It won't be solved nicely, we'll get a warning about the same commit appearing twice with different hashes.

Oh, this sounds like a new check in Linux Next? Or somewhere else?

Quoted 1 line

Whatever, I dropped it from my tree, it was the HEAD anyway.

Thank you!

Signature · 2 lines
With Best Regards,
Andy Shevchenko