| Message ID | 20240913110523.3584749-1-andriy.shevchenko@linux.intel.com (mailing list archive) |
|---|---|
| State | Handled Elsewhere, archived |
| Headers | show |
| Series | [v1,1/1] software node: Simplify swnode_register() a bit | expand |
On Fri, Sep 13, 2024 at 02:05:23PM +0300, Andy Shevchenko wrote: > By introducing two temporary variables simplify swnode_register() a bit. > No functional change intended. > > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
On Fri, Sep 13, 2024 at 02:05:23PM +0300, Andy Shevchenko wrote: > By introducing two temporary variables simplify swnode_register() a bit. > No functional change intended. > > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > --- > drivers/base/swnode.c | 7 +++---- > 1 file changed, 3 insertions(+), 4 deletions(-) > > diff --git a/drivers/base/swnode.c b/drivers/base/swnode.c > index b0be765b12da..810c27a8c9c1 100644 > --- a/drivers/base/swnode.c > +++ b/drivers/base/swnode.c > @@ -908,6 +908,7 @@ static struct fwnode_handle * > swnode_register(const struct software_node *node, struct swnode *parent, > unsigned int allocated) > { > + struct kobject *kobj_parent = parent ? &parent->kobj : NULL; I despise ?: use just so much, EXCEPT for when it's used in something like this: > struct swnode *swnode; > int ret; > > @@ -934,12 +935,10 @@ swnode_register(const struct software_node *node, struct swnode *parent, > > if (node->name) > ret = kobject_init_and_add(&swnode->kobj, &software_node_type, > - parent ? &parent->kobj : NULL, > - "%s", node->name); > + kobj_parent, "%s", node->name); Which really is the only valid way I'd put up with it :) So can you rewrite the change above to be just: struct kobject *kobj_parent = NULL; ... if (parent) kobj_parent = &parent->kobj; Which is much simpler to read, right? thanks, greg k-h
On Fri, Sep 13, 2024 at 03:34:58PM +0200, Greg Kroah-Hartman wrote: > On Fri, Sep 13, 2024 at 02:05:23PM +0300, Andy Shevchenko wrote: ... > > + struct kobject *kobj_parent = parent ? &parent->kobj : NULL; > > I despise ?: use just so much, EXCEPT for when it's used in something > like this: > > ret = kobject_init_and_add(&swnode->kobj, &software_node_type, > > - parent ? &parent->kobj : NULL, > > - "%s", node->name); > > + kobj_parent, "%s", node->name); > > Which really is the only valid way I'd put up with it :) I see your point! > So can you rewrite the change above to be just: > > struct kobject *kobj_parent = NULL; > > ... > > if (parent) > kobj_parent = &parent->kobj; > > Which is much simpler to read, right? Yeah, but the point of the patch seems to be diminished. Let's just not continue with it for now. Maybe later it will make more sense. Thank you for the review!
diff --git a/drivers/base/swnode.c b/drivers/base/swnode.c index b0be765b12da..810c27a8c9c1 100644 --- a/drivers/base/swnode.c +++ b/drivers/base/swnode.c @@ -908,6 +908,7 @@ static struct fwnode_handle * swnode_register(const struct software_node *node, struct swnode *parent, unsigned int allocated) { + struct kobject *kobj_parent = parent ? &parent->kobj : NULL; struct swnode *swnode; int ret; @@ -934,12 +935,10 @@ swnode_register(const struct software_node *node, struct swnode *parent, if (node->name) ret = kobject_init_and_add(&swnode->kobj, &software_node_type, - parent ? &parent->kobj : NULL, - "%s", node->name); + kobj_parent, "%s", node->name); else ret = kobject_init_and_add(&swnode->kobj, &software_node_type, - parent ? &parent->kobj : NULL, - "node%d", swnode->id); + kobj_parent, "node%d", swnode->id); if (ret) { kobject_put(&swnode->kobj); return ERR_PTR(ret);
By introducing two temporary variables simplify swnode_register() a bit. No functional change intended. Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> --- drivers/base/swnode.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-)