[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAPcyv4g0-rxw2KXrRRe4Mr-zS4ujx6pTqApZgZrPhJetLgjRPA@mail.gmail.com>
Date: Wed, 8 Apr 2020 22:14:11 -0700
From: Dan Williams <dan.j.williams@...el.com>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: linux-nvdimm <linux-nvdimm@...ts.01.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Vishal L Verma <vishal.l.verma@...el.com>
Subject: Re: [GIT PULL] libnvdimm for v5.7
On Wed, Apr 8, 2020 at 9:11 PM Linus Torvalds
<torvalds@...ux-foundation.org> wrote:
>
> On Tue, Apr 7, 2020 at 1:12 PM Dan Williams <dan.j.williams@...el.com> wrote:
> >
> > mm/memremap_pages: Introduce memremap_compat_align()
>
> Why is this an exported function that just returns a constant?
>
> Why isn't it just a #define (or inline) in a header file?
>
> Yes, yes, it would need to be conditional on not having that
> CONFIG_ARCH_HAS_MEMREMAP_COMPAT_ALIGN, but it does look strange.
>
> I've pulled it, since it doesn't matter that much, but I find it silly
> to have full-fledged functions - and exported them GPL-only - to
> return a constant. Crazy.
Yes. tl;dr I gave up after failing to unwind a header dependency chain [1].
The source of the trouble was trying to find an existing top-level
header file that included an asm local version, but also needed to
include mmzone.h for the definition of SUBSECTION_SIZE.
include/linux/io.h fit that requirement, pulling in mmzone.h there
proved more difficult than my header unwinding skills could
accomplish.
[1]: https://git.kernel.org/pub/scm/linux/kernel/git/djbw/nvdimm.git/commit/?id=c990ae9376c15f40aff2f61f42a71be5b81f9ee1
Powered by blists - more mailing lists