linux-mips
[Top] [All Lists]

Re: [PATCH] hwrng: Add TX4939 RNG driver

To: Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH] hwrng: Add TX4939 RNG driver
From: Ralf Baechle <ralf@linux-mips.org>
Date: Sun, 31 May 2009 17:18:42 +0100
Cc: Atsushi Nemoto <anemo@mba.ocn.ne.jp>, linux-mips@linux-mips.org, linux-kernel@vger.kernel.org, Herbert Xu <herbert@gondor.apana.org.au>
In-reply-to: <20090529162907.9cb1bba2.akpm@linux-foundation.org>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <1243350141-883-1-git-send-email-anemo@mba.ocn.ne.jp> <20090529162907.9cb1bba2.akpm@linux-foundation.org>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.5.18 (2008-05-17)
On Fri, May 29, 2009 at 04:29:07PM -0700, Andrew Morton wrote:

> > +static u64 read_rng(void __iomem *base, unsigned int offset)
> > +{
> > +   /* Caller must disable interrupts */
> > +   return ____raw_readq(base + offset);
> > +}
> 
> What is the reasoning behind the local_irq_disable() requirement?
> 
> Because I'm wondering whether this is safe on SMP?

The problem are interrupts, not SMP.  readq is reading a 64-bit register
using a 64-bit load.  On a 32-bit kernel however interrupts or any
other processor exception would clobber the upper 32-bit of the processor
register so interrupts need to be disabled.  The "normal" readq
functions disable interrupts as necessary on a platform.  This code does
multiple read accesses so for efficiency sake it relies on the caller
handling interrupts explicitly.

Also this platform does not do SMP.

  Ralf

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