Re: [RFC PATCH] virt_mmio: fix signature checking for BE guests

From: "Michael S. Tsirkin" <mst@redhat.com>
To: Marc Zyngier <marc.zyngier@arm.com>
Cc: Pawel Moll <Pawel.Moll@arm.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"virtualization@lists.linux-foundation.org" 
	<virtualization@lists.linux-foundation.org>,
	Rusty Russell <rusty@rustcorp.com.au>
Subject: Re: [RFC PATCH] virt_mmio: fix signature checking for BE guests
Date: Wed, 13 Feb 2013 18:53:56 +0200	[thread overview]
Message-ID: <20130213165356.GA23193@redhat.com> (raw)
In-Reply-To: <511BB134.30003@arm.com>

On Wed, Feb 13, 2013 at 03:28:52PM +0000, Marc Zyngier wrote:
> On 13/02/13 15:08, Pawel Moll wrote:
> > On Wed, 2013-02-13 at 14:25 +0000, Marc Zyngier wrote:
> >> Using readl() to read the magic value and then memcmp() to check it
> >> fails on BE, as bytes will be the other way around (by virtue of
> >> the registers to follow the endianess of the guest).
> > 
> > Hm. Interesting. I missed the fact that readl() as a "PCI operation"
> > will always assume LE values...
> > 
> >> Fix it by encoding the magic as an integer instead of a string.
> >> So I'm not completely sure this is the right fix, 
> > 
> > It seems right, however...
> > 
> >> - Using __raw_readl() instead. Is that a generic enough API?
> >>
> > ... this implies that either the spec is wrong (as it should say: the
> > device registers are always LE, in the PCI spirit) or all readl()s & co.
> > should be replaced with __raw equivalents.
> 
> Well, the spec clearly says that the registers reflect the endianess of
> the guest, and it makes sense: when performing the MMIO access, KVM
> needs to convert between host and guest endianess.
> 
> > Having said that, does the change make everything else work with a BE
> > guest? (I assume we're talking about the guest being BE, right? ;-) If
> > so it means that the host is not following the current spec and it
> > treats all the registers as LE.
> 
> Yes, I only care about a BE guest. And no, not much is actually working
> (kvmtool is not happy about the guest addresses it finds in the
> virtio-ring). Need to dive into it and understand what needs to be fixed...

Does it work for qemu? I know people were using virtio on BE there.

> >> - Reading the MAGIC register byte by byte. Is that allowed? The spec
> >>   only says it is 32bit wide.
> > 
> > And the spirit of the spec was: _exactly 32bit wide_. It's just simpler
> > to implement one access width on the host side.
> 
> I guessed as much...
> 
> 	M.
> -- 
> Jazz is not dead. It just smells funny...