lkml.org 
[lkml]   [2016]   [Jan]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRE: [PATCH V2 1/3] NTB: Add AMD PCI-Express NTB driver
Date
> From: Jon Mason <jdmason@kudzu.us>
> > On Wed, Jan 6, 2016 at 9:50 PM, Yu, Xiangliang <Xiangliang.Yu@amd.com>
> > wrote:
> > >> > +#define NTB_READ_REG(base, r) (ioread32(base + AMD_ ## r ##
> > >> _OFFSET))
> > >> > +#define NTB_WRITE_REG(base, val, r) (iowrite32(val, base + \
> > >> > + AMD_ ## r ##
> > _OFFSET))
> > >> > +#define NTB_READ_OFFSET(base, r, of) (ioread32(base + of +
> > \
> > >> > + AMD_ ## r ##
> > _OFFSET))
> > >> > +#define NTB_WRITE_OFFSET(base, val, r, of) (iowrite32(val, base
> > >> > ++
> > \
> > >> > + of + AMD_ ## r ##
> > _OFFSET))
> > >>
> > >> Please do not use marcos to hide ioread/iowrite. Call
> > >> iorwad/iowrite
> > directly.
> > >
> > > I don't see any wrong to hide ioread/iowrite, and I think the macros
> > can make code readable and easy to maintain.
> >
> > I disagree. It is an unnecessary layer and can add to confusion.
> > Please make the change.
>
> I don't like AMD_##r##_OFFSET in these macros. It hides the use of a
> globally named constant like AMD_FOO_OFFSET, since one would read only
> FOO in the code. It makes cross referencing difficult, since the reader needs
> to know FOO is really AMD_FOO_OFFSET. This would defeat automatic cross
> referencing like cscope and lxr.
>
> #define AMD_FOO_OFFSET 0xc0ff33
> vs
> NTB_READ_OFFSET(dev->foo_base, FOO, offset_in_foo) // Where is FOO
> defined?
>
> This macro would have at least been better written without ##; so cross
> referencing would still work.
>
> NTB_READ_OFFSET(dev->foo_base, FOO, offset_in_foo) vs
> NTB_READ_OFFSET(dev->foo_base, AMD_FOO_OFFSET, offset_in_foo) //
> AMD_FOO_OFFSET is 0xcoff33 (obviously)
>
> But without ##, the macro is just the addition of its parameters. Change the
> commas to addition, and the macro to ioread, and you'll see there is no
> benefit for having this macro any more.
>
> NTB_READ_OFFSET(dev->foo_base, AMD_FOO_OFFSET, offset_in_foo) vs
> ioread32(dev->foo_base + AMD_FOO_OFFSET + offset_in_foo)
>
> I second Jon's opinion. Please make the change. This would be better as
> simply ioread/write in the code.

The main aim of these macros is to focuses on the name of register (FOO),
not the whole macros of register.
and I see there are lots of these style in kernel.
If your guys still can't accept the style, I can change it.
And I'll change ioread/iowrite to readl/writel because it only access mmio space.

\
 
 \ /
  Last update: 2016-01-11 09:21    [W:0.077 / U:0.068 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site