On Wednesday, August 26, 2015 09:00:23 AM Suravee Suthikulpanit wrote:

Hi Rafael,

On 8/26/15 06:48, Rafael J. Wysocki wrote:

quoted

[...]
On Wednesday, August 26, 2015 12:33:27 AM Suravee Suthikulpanit wrote:

quoted

diff --git a/drivers/acpi/glue.c b/drivers/acpi/glue.c
index b9657af..55cf916 100644
--- a/drivers/acpi/glue.c
+++ b/drivers/acpi/glue.c

@@ -168,7 +168,7 @@ int acpi_bind_one(struct device *dev, struct acpi_device *acpi_dev)
  	struct list_head *physnode_list;
  	unsigned int node_id;
  	int retval = -EINVAL;
-	bool coherent;
+	int coherent;

enum, anyone?  With clearly defined values?

Originally I had defined

enum acpi_dma_coherency {
	ACPI_DMA_NON_COHERENT,
	ACPI_DMA_COHERENT,
	APCI_DMA_NOT_SUPPORTED = -1,
};

Although, this would need to be defined in the include/linux/acpi.h, and 
will be used also for #ifndef CONFIG_ACPI code to return errors. I was 
not sure if this would be too much. If this is preferred, I'll add this 
back in.

quoted

quoted

  	if (has_acpi_companion(dev)) {
  		if (acpi_dev) {

@@ -225,7 +225,8 @@ int acpi_bind_one(struct device *dev, struct acpi_device *acpi_dev)
  	if (!has_acpi_companion(dev))
  		ACPI_COMPANION_SET(dev, acpi_dev);

-	if (acpi_check_dma(acpi_dev, &coherent))
+	coherent = acpi_check_dma_coherency(acpi_dev);
+	if (coherent != -1)

Like here I'm not sure why -1 is special?

It's just another value to communicate that DMA is not supported. :)

OK

So maybe use 0/1 for the coherence thing and a proper error code
(-ENOTSUPP seems to be a good candidate) for the "no DMA" case?

Then, if you rename the local variable to something like "ret",
it will be slightly less confusing.

And instead of checking against a specific negative value, you can
simply use "if (stuff < 0)" or "if (stuff >= 0)" to check whether or
not it is supported at all.

Thanks,
Rafael