linux-mips
[Top] [All Lists]

Re: [PATCH] MIPS checksum fix

To: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Subject: Re: [PATCH] MIPS checksum fix
From: Ralf Baechle <ralf@linux-mips.org>
Date: Sat, 20 Sep 2008 17:09:14 +0200
Cc: macro@linux-mips.org, u1@terran.org, linux-mips@linux-mips.org, netdev@vger.kernel.org
In-reply-to: <20080919.230952.128619158.anemo@mba.ocn.ne.jp>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <20080919112304.GB13440@linux-mips.org> <20080919114743.GA19359@linux-mips.org> <20080919120752.GA19877@linux-mips.org> <20080919.230952.128619158.anemo@mba.ocn.ne.jp>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.5.18 (2008-05-17)
On Fri, Sep 19, 2008 at 11:09:52PM +0900, Atsushi Nemoto wrote:

> I think it would be better splitting bugfix and optimization.  This
> code is too complex to do many things at a time, isn't it?
> 
> > @@ -53,12 +53,14 @@
> >  #define UNIT(unit)  ((unit)*NBYTES)
> >  
> >  #define ADDC(sum,reg)                                              \
> > -   .set    push;                                           \
> > -   .set    noat;                                           \
> >     ADD     sum, reg;                                       \
> >     sltu    v1, sum, reg;                                   \
> >     ADD     sum, v1;                                        \
> > -   .set    pop
> 
> Is this required?  Just a cleanup?

It papers over potencially important warnings so had to go.  I argue the
caller of ADDC should set noat mode, if at all.

  Ralf

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