From: Antoine Tenart Subject: Re: [PATCH 13/14] arm64: dts: marvell: armada-cp110: update the crypto engine compatible Date: Mon, 23 Jul 2018 09:21:52 +0200 Message-ID: <20180723072152.GA3456@kwain> References: <20180628151544.22134-1-antoine.tenart@bootlin.com> <20180628151544.22134-14-antoine.tenart@bootlin.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Cc: Antoine Tenart , Herbert Xu , David Miller , Gregory CLEMENT , Andrew Lunn , Jason Cooper , Sebastian Hesselbarth , linux-crypto@vger.kernel.org, Linux Kernel Mailing List , Thomas Petazzoni , maxime.chevallier@bootlin.com, miquel.raynal@bootlin.com, Nadav Haklai , oferh@marvell.com, igall@marvell.com To: Olof Johansson Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-crypto.vger.kernel.org Hi Olof, On Sat, Jul 21, 2018 at 02:35:01PM -0700, Olof Johansson wrote: > On Thu, Jun 28, 2018 at 8:15 AM, Antoine Tenart > wrote: > > New compatibles are now supported by the Inside Secure SafeXcel driver. > > As they are more specific than the old ones, they should be used > > whenever possible. This patch updates the Marvell cp110 device tree > > accordingly. > > > > Signed-off-by: Antoine Tenart > > --- > > arch/arm64/boot/dts/marvell/armada-cp110.dtsi | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/arch/arm64/boot/dts/marvell/armada-cp110.dtsi b/arch/arm64/boot/dts/marvell/armada-cp110.dtsi > > index 2bf083272a87..bb2914f90048 100644 > > --- a/arch/arm64/boot/dts/marvell/armada-cp110.dtsi > > +++ b/arch/arm64/boot/dts/marvell/armada-cp110.dtsi > > @@ -427,7 +427,7 @@ > > }; > > > > CP110_LABEL(crypto): crypto@800000 { > > - compatible = "inside-secure,safexcel-eip197"; > > + compatible = "inside-secure,safexcel-eip197b"; > > So the device is still compatible with the less-specific binding, > right? If so, it should probably have both compatible properties in > there, not just the more specific one. Using "safexcel-eip197" as a compatible was a mistake as there's no such thing as an eip197, they all have minor versions (such as 'b'). I've thought about using the compatible for a less specific binding, but this isn't true for the other patch, using the compatible ending in "-eip97". The engine it supports is the most specific one (i.e. with the largest number of algorithms supported). So it would simply not work, and as we only have a few device trees supporting the engine as of now, I thought fixing this by removing the wrong compatible was a better solution (of course the driver is backward compatible, and using the old compatibles will still work). Thanks, Antoine -- Antoine T?nart, Bootlin (formerly Free Electrons) Embedded Linux and Kernel engineering https://bootlin.com