regulator: core: Add intermediate cast to uintptr_t before casting to pointer

drivers/regulator/core.c static int _regulator_do_set_voltage(struct regulator_dev *rdev, · line 2183

Quoted 4 lines

drivers/regulator/core.c +1 -1

 		_notifier_call_chain(rdev, REGULATOR_EVENT_VOLTAGE_CHANGE,
-				     (void *)best_val);
+				     (void *)(uintptr_t)best_val);

This also looks problematic, you should never need to cast a pointer to or from void.

於 一，2012-07-30 於 16:33 +0100，Mark Brown 提到：

Quoted 8 lines · Axel Lin

drivers/regulator/core.c static int _regulator_do_set_voltage(struct regulator_dev *rdev, · line 2183

Quoted 4 lines

drivers/regulator/core.c +1 -1

 		_notifier_call_chain(rdev, REGULATOR_EVENT_VOLTAGE_CHANGE,
-				     (void *)best_val);
+				     (void *)(uintptr_t)best_val);

This also looks problematic, you should never need to cast a pointer to or from void.

If I remove the cast to (void *), I got below build warning:

CC drivers/regulator/core.o
drivers/regulator/core.c: In function '_regulator_do_set_voltage':
drivers/regulator/core.c:2183:10: warning: passing argument 3 of '_notifier_call_chain' makes pointer from integer without a cast [enabled by default]
drivers/regulator/core.c:94:13: note: expected 'void *' but argument is of type 'long unsigned int'

Quoted 12 lines

於 一，2012-07-30 於 16:33 +0100，Mark Brown 提到：

Quoted 8 lines · Axel Lin

drivers/regulator/core.c static int _regulator_do_set_voltage(struct regulator_dev *rdev, · line 2183

Quoted 4 lines

drivers/regulator/core.c +1 -1

 		_notifier_call_chain(rdev, REGULATOR_EVENT_VOLTAGE_CHANGE,
-				     (void *)best_val);
+				     (void *)(uintptr_t)best_val);

This also looks problematic, you should never need to cast a pointer to or from void.

If I remove the cast to (void *), I got below build warning:

Quoted 7 lines

CC drivers/regulator/core.o
drivers/regulator/core.c: In function '_regulator_do_set_voltage':
drivers/regulator/core.c:2183:10: warning: passing argument 3 of '_notifier_call_chain' makes pointer from integer without a cast [enabled by default]
drivers/regulator/core.c:94:13: note: expected 'void *' but argument is of type 'long unsigned int'

So the above cast probably isn't right... you shouldn't need a double cast here.

於 二，2012-07-31 於 16:41 +0100，Mark Brown 提到：

Quoted 2 lines

So the above cast probably isn't right... you shouldn't need a double cast here.

The issue of original warning is: ( I compile the kernel on x86_64 )
warning: cast to pointer from integer of different size

What this patch does is:
cast best_val to uintptr_t, which is an unsigned integer large enough to hold a void* pointer.

And then cast it to (void *).

Thus I did (void *)(uintptr_t)best_val.

Quoted 11 lines

於 二，2012-07-31 於 16:41 +0100，Mark Brown 提到：

Quoted 2 lines

So the above cast probably isn't right... you shouldn't need a double cast here.

The issue of original warning is: ( I compile the kernel on x86_64 )
warning: cast to pointer from integer of different size

What this patch does is:
cast best_val to uintptr_t, which is an unsigned integer large enough to hold a void* pointer.

Quoted 3 lines

And then cast it to (void *).

Thus I did (void *)(uintptr_t)best_val.

Perhaps we need a temporary variable here. The double cast just looks too horrible.

Either is ok to me.
But the double case looks better to me because the intention is clear. It seems uintptr_t is less commonly used in drivers code. Maybe just use (void *)(unsigned long)best_val is better in readability.