From: Arnd Bergmann Date: Fri, 23 Jun 2017 14:33:43 +0000 (+0200) Subject: Revert "arm64: dts: marvell: add dma-mask in crypto nodes for 7k/8k" X-Git-Tag: v4.13-rc1~165^2~3 X-Git-Url: https://asedeno.scripts.mit.edu/gitweb/?a=commitdiff_plain;h=2b29ca22ed8a79ef1942b6546ccab5953678028d;p=linux.git Revert "arm64: dts: marvell: add dma-mask in crypto nodes for 7k/8k" As I found by chance while merging another patch, the usage of a dma-mask in this DT node is wrong for multiple reasons: - dma-masks are a Linux specific concept, not a general hardware feature - In DT, we use the "dma-ranges" property to describe how DMA addresses related between devices. - The 40-bit mask appears to be completely unnecessary here, as the SoC cannot address that much memory anyway, so simply asking for a 64-bit mask (as supported by the device) should succeed anyway. The patch to remove the parsing of the property is getting merged through the crypto tree. Signed-off-by: Arnd Bergmann --- diff --git a/arch/arm64/boot/dts/marvell/armada-cp110-master.dtsi b/arch/arm64/boot/dts/marvell/armada-cp110-master.dtsi index f611e843094c..726528ce54e9 100644 --- a/arch/arm64/boot/dts/marvell/armada-cp110-master.dtsi +++ b/arch/arm64/boot/dts/marvell/armada-cp110-master.dtsi @@ -270,7 +270,6 @@ cpm_crypto: crypto@800000 { interrupt-names = "mem", "ring0", "ring1", "ring2", "ring3", "eip"; clocks = <&cpm_clk 1 26>; - dma-mask = <0xff 0xffffffff>; }; }; diff --git a/arch/arm64/boot/dts/marvell/armada-cp110-slave.dtsi b/arch/arm64/boot/dts/marvell/armada-cp110-slave.dtsi index 84d3bd80eb51..95f8e5f607f6 100644 --- a/arch/arm64/boot/dts/marvell/armada-cp110-slave.dtsi +++ b/arch/arm64/boot/dts/marvell/armada-cp110-slave.dtsi @@ -261,7 +261,6 @@ cps_crypto: crypto@800000 { interrupt-names = "mem", "ring0", "ring1", "ring2", "ring3", "eip"; clocks = <&cps_clk 1 26>; - dma-mask = <0xff 0xffffffff>; /* * The cryptographic engine found on the cp110 * master is enabled by default at the SoC