diff mbox

[2/2] arm64: dts: Reduce size of virtio block device for foundation model

Message ID 1387543964-15057-2-git-send-email-broonie@kernel.org (mailing list archive)
State New, archived
Headers show

Commit Message

Mark Brown Dec. 20, 2013, 12:52 p.m. UTC 
From: Mark Brown <broonie@linaro.org>

Will Deacon observed that kvmtool uses a size of 0x200 for virtio
block memory region and that the virtio block spec only uses 31 bytes in
the device specific region at 0x100 so reduce the region to a less
wasteful 0x200.

Signed-off-by: Mark Brown <broonie@linaro.org>
---
 arch/arm64/boot/dts/foundation-v8.dts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Catalin Marinas Dec. 20, 2013, 2:25 p.m. UTC | #1 
On Fri, Dec 20, 2013 at 12:52:44PM +0000, Mark Brown wrote:
> From: Mark Brown <broonie@linaro.org>
> 
> Will Deacon observed that kvmtool uses a size of 0x200 for virtio
> block memory region and that the virtio block spec only uses 31 bytes in
> the device specific region at 0x100 so reduce the region to a less
> wasteful 0x200.

What do we waste by keeping it at 4K? That's ioremap'ed, so we use a
full page anyway.
Mark Brown Dec. 20, 2013, 4:05 p.m. UTC | #2 
On Fri, Dec 20, 2013 at 02:25:24PM +0000, Catalin Marinas wrote:
> On Fri, Dec 20, 2013 at 12:52:44PM +0000, Mark Brown wrote:
> > From: Mark Brown <broonie@linaro.org>

> > Will Deacon observed that kvmtool uses a size of 0x200 for virtio
> > block memory region and that the virtio block spec only uses 31 bytes in
> > the device specific region at 0x100 so reduce the region to a less
> > wasteful 0x200.

> What do we waste by keeping it at 4K? That's ioremap'ed, so we use a
> full page anyway.

This was done for consistency with the same change made in RTSM
motherboard patch so that we register the same device in the same way.  
That change was at Will's request, he did point out that we might use
16K pages in some systems IIRC but personally I'm not sure it makes a
huge difference either way - I do think we should keep all registrations
of the device consistent though to avoid user confusion about how to use
it.
Catalin Marinas Dec. 20, 2013, 4:34 p.m. UTC | #3 
On Fri, Dec 20, 2013 at 04:05:22PM +0000, Mark Brown wrote:
> On Fri, Dec 20, 2013 at 02:25:24PM +0000, Catalin Marinas wrote:
> > On Fri, Dec 20, 2013 at 12:52:44PM +0000, Mark Brown wrote:
> > > From: Mark Brown <broonie@linaro.org>
> 
> > > Will Deacon observed that kvmtool uses a size of 0x200 for virtio
> > > block memory region and that the virtio block spec only uses 31 bytes in
> > > the device specific region at 0x100 so reduce the region to a less
> > > wasteful 0x200.
> 
> > What do we waste by keeping it at 4K? That's ioremap'ed, so we use a
> > full page anyway.
> 
> This was done for consistency with the same change made in RTSM
> motherboard patch so that we register the same device in the same way.  
> That change was at Will's request, he did point out that we might use
> 16K pages in some systems IIRC but personally I'm not sure it makes a
> huge difference either way - I do think we should keep all registrations
> of the device consistent though to avoid user confusion about how to use
> it.

OK. Applied.
diff mbox

Patch

diff --git a/arch/arm64/boot/dts/foundation-v8.dts b/arch/arm64/boot/dts/foundation-v8.dts
index 519c4b2c0687..4a060906809d 100644
--- a/arch/arm64/boot/dts/foundation-v8.dts
+++ b/arch/arm64/boot/dts/foundation-v8.dts
@@ -224,7 +224,7 @@ 
 
 			virtio_block@0130000 {
 				compatible = "virtio,mmio";
-				reg = <0x130000 0x1000>;
+				reg = <0x130000 0x200>;
 				interrupts = <42>;
 			};
 		};