| Message ID | 20220927172258.62418-1-andriy.shevchenko@linux.intel.com |
|---|---|
| State | Accepted |
| Headers | show |
| Series | [v1,1/1] pwm: core: Replace custom implementation of device_match_fwnode() | expand |
On Tue, Sep 27, 2022 at 08:22:58PM +0300, Andy Shevchenko wrote: > Replace custom implementation of the device_match_fwnode(). > > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > --- > drivers/pwm/core.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) I really don't see the point in having an exported symbol for this. It's a simple comparison and the result is even longer than the original. The *only* reason why this helper exists seems to be because it is getting used in *_find_device() callbacks. Honestly, I don't see a reason why this should be applied. And frankly, why bother making all these changes. It's a waste of time, in my opinion. Thierry
On Wed, Sep 28, 2022 at 04:26:47PM +0200, Thierry Reding wrote: > On Tue, Sep 27, 2022 at 08:22:58PM +0300, Andy Shevchenko wrote: > > Replace custom implementation of the device_match_fwnode(). > I really don't see the point in having an exported symbol for this. It's > a simple comparison and the result is even longer than the original. Longer doesn't always mean worse. > The > *only* reason why this helper exists seems to be because it is getting > used in *_find_device() callbacks. Yes and no. Initially for the purpose to be a callback it can be reused. The point is that it hides the dev_fwnode() machinery behind and taking into account ongoing discussion about constification of the dev_fwnode() we might need to touch this or similar places to avoid problems with compiler. > Honestly, I don't see a reason why this should be applied. And frankly, > why bother making all these changes. It's a waste of time, in my > opinion. Obviously I will not do it if I be with you on the same page. But okay, not a big deal in this case.
On Wed, Sep 28, 2022 at 05:49:36PM +0300, Andy Shevchenko wrote: > On Wed, Sep 28, 2022 at 04:26:47PM +0200, Thierry Reding wrote: > > On Tue, Sep 27, 2022 at 08:22:58PM +0300, Andy Shevchenko wrote: > > > Replace custom implementation of the device_match_fwnode(). > > > I really don't see the point in having an exported symbol for this. It's > > a simple comparison and the result is even longer than the original. > > Longer doesn't always mean worse. > > > The > > *only* reason why this helper exists seems to be because it is getting > > used in *_find_device() callbacks. > > Yes and no. Initially for the purpose to be a callback it can be reused. > The point is that it hides the dev_fwnode() machinery behind and taking > into account ongoing discussion about constification of the dev_fwnode() > we might need to touch this or similar places to avoid problems with > compiler. Maybe next time use that argument in the commit message. That's much more convincing than a useless "replace custom implementation" because that just makes it look like you're doing this to pass the time or something. Applied, with a slightly updated commit message, thanks. Thierry
On Wed, Sep 28, 2022 at 04:57:57PM +0200, Thierry Reding wrote: > On Wed, Sep 28, 2022 at 05:49:36PM +0300, Andy Shevchenko wrote: > > On Wed, Sep 28, 2022 at 04:26:47PM +0200, Thierry Reding wrote: > > > On Tue, Sep 27, 2022 at 08:22:58PM +0300, Andy Shevchenko wrote: > > > > Replace custom implementation of the device_match_fwnode(). > > > > > I really don't see the point in having an exported symbol for this. It's > > > a simple comparison and the result is even longer than the original. > > > > Longer doesn't always mean worse. > > > > > The > > > *only* reason why this helper exists seems to be because it is getting > > > used in *_find_device() callbacks. > > > > Yes and no. Initially for the purpose to be a callback it can be reused. > > The point is that it hides the dev_fwnode() machinery behind and taking > > into account ongoing discussion about constification of the dev_fwnode() > > we might need to touch this or similar places to avoid problems with > > compiler. > > Maybe next time use that argument in the commit message. That's much > more convincing than a useless "replace custom implementation" because > that just makes it look like you're doing this to pass the time or > something. Noted. > Applied, with a slightly updated commit message, thanks. Thank you!
diff --git a/drivers/pwm/core.c b/drivers/pwm/core.c index cfe3a0327471..d333e7422f4a 100644 --- a/drivers/pwm/core.c +++ b/drivers/pwm/core.c @@ -678,7 +678,7 @@ static struct pwm_chip *fwnode_to_pwmchip(struct fwnode_handle *fwnode) mutex_lock(&pwm_lock); list_for_each_entry(chip, &pwm_chips, list) - if (chip->dev && dev_fwnode(chip->dev) == fwnode) { + if (chip->dev && device_match_fwnode(chip->dev, fwnode)) { mutex_unlock(&pwm_lock); return chip; }
Replace custom implementation of the device_match_fwnode(). Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> --- drivers/pwm/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)