linux-mips
[Top] [All Lists]

Re: [PATCH] [MIPS] Move several variables from .bss to .init.data

To: dmitri.vorobiev@gmail.com
Subject: Re: [PATCH] [MIPS] Move several variables from .bss to .init.data
From: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Date: Tue, 24 Nov 2009 00:19:03 +0900 (JST)
Cc: dmitri.vorobiev@movial.com, linux-mips@linux-mips.org, ralf@linux-mips.org
In-reply-to: <90edad820911230637r3f2c94adh737c9f4f5adcd197@mail.gmail.com>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <90edad820911230543i64f1e33fg86770f3ab2b6510b@mail.gmail.com> <20091123.230616.70220116.anemo@mba.ocn.ne.jp> <90edad820911230637r3f2c94adh737c9f4f5adcd197@mail.gmail.com>
Sender: linux-mips-bounce@linux-mips.org
On Mon, 23 Nov 2009 16:37:27 +0200, Dmitri Vorobiev <dmitri.vorobiev@gmail.com> 
wrote:
> >> > NAK, at least for txx9 parts.  The struct mtd_partition arrays will be
> >> > referenced by mtd map drivers via platform_data.
> >>
> >> You are right, thanks. What do you think about moving the variables to
> >> file scope then?
> >
> > Well, why?  Does it make any check-scripts or something happy?
> 
> That's just looked somehow confusing to me that a variable defined in
> the function scope was referenced from far outside this function, and
> even when the function itself isn't valid anymore. Of course, there
> are no technical disadvantage in that in itself.

I see.  But that's not so irregular case I think.  And even if the
variables are outside the function, it will not be obvious that it
cannot be marked as __initdata.

I agree that the code looks ugly and confusing, but moving it outside
does not helps, I think.

---
Atsushi Nemoto

<Prev in Thread] Current Thread [Next in Thread>