Re: drivers/char/random.c: more ruminations

["Theodore Ts'o" <tytso@mit.edu>]

On Tue, Jun 10, 2014 at 08:10:03PM -0400, George Spelvin wrote:
> 
> But even I get annoyed when I have a 1-line comment typo fix and wonder
> if it really deserves its own commit or if I can just include it with
> the other changes I'm making to that file.

Unless you're actually modifying that section of code, I usually
recommend that people just not bother.  The fact that someone included
a one-line comment fix caused a merge conflict with the ext4 pull in
this merge window that Linus had to fix up.  Not that it's a big deal,
but unless it's something that's really going to confuse the reader, I
treat it as white space fixes; something that's only worth fixing if
that particular function is being modified for a "real" change.

> I have half a dozen patches to random.c already waiting.  For example,
> one is a bulk conversion of __u8 and __u32 to u8 and u32.  The underscore
> versions are only intended for public header files where namespace
> pollution is a problem.

And sorry, I consider this sort of thing is to be just code wankery.
We use __u32 in a number of places in the kernel, and it doesn't
really affect code readability.  A change like this almost guarantees
that stable patches won't apply manually, and will have to be ported
manually.  It's just not worth it.

						- Ted