> I think there is only one object that can be modified in this function, Such a view can be reasonable. > so I didn't mention it. I suggest to reconsider the conclusion. >> I guess that an imperative wording is preferred also for this change description. > > This sentence is referenced from the code comment, so I haven't change it. > https://elixir.bootlin.com/linux/v5.7-rc7/source/lib/kobject.c#L459 I find that that there are further possibilities to consider for improvements around the presented commit message (even after the mentioned copy from the function description of this programming interface). >> How do you think about to combine this update step together with >> “nilfs2: Fix reference count leak in nilfs_sysfs_create_device_group” >> into a small patch series? > > I'd like to improve the similar issues after I reporting this bunch of bugs. Did you find questionable implementation details with the help of an evolving source code analysis tool? Regards, Markus
