Le Mon, May 10, 2021 at 11:36:32PM +0200, Linus Walleij a écrit : > From: Arnd Bergmann <arnd@xxxxxxxx> > > The ixp4xx_crypto driver traditionally registers a bare platform > device without attaching it to a driver, and detects the hardware > at module init time by reading an SoC specific hardware register. > > Change this to the conventional method of registering the platform > device from the platform code itself when the device is present, > turning the module_init/module_exit functions into probe/release > driver callbacks. > > This enables compile-testing as well as potentially having ixp4xx > coexist with other ARMv5 platforms in the same kernel in the future. > > Signed-off-by: Arnd Bergmann <arnd@xxxxxxxx> > Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxx> > --- > Herbert, David: I am looking for an ACK to take this > into the ARM SoC tree as we want to change more stuff in > the machine at the same time that we want to resolve > there. > --- > arch/arm/mach-ixp4xx/common.c | 26 ++++++++++++++++++++++++ > drivers/crypto/Kconfig | 3 ++- > drivers/crypto/ixp4xx_crypto.c | 37 ++++++++++++---------------------- > 3 files changed, 41 insertions(+), 25 deletions(-) > Hello With minor editing I successfully added this series on top of my fix series https://lore.kernel.org/patchwork/cover/1421865/ With the following patch, I successfully booted my epbx100 board and the crypto driver loaded. --- a/arch/arm/boot/dts/intel-ixp4xx.dtsi +++ b/arch/arm/boot/dts/intel-ixp4xx.dtsi @@ -61,9 +61,16 @@ timer@c8005000 { interrupts = <5 IRQ_TYPE_LEVEL_HIGH>; }; - npe@c8006000 { + npe: npe@c8006000 { compatible = "intel,ixp4xx-network-processing-engine"; reg = <0xc8006000 0x1000>, <0xc8007000 0x1000>, <0xc8008000 0x1000>; + + crypto { + compatible = "intel,ixp4xx-crypto"; + intel,npe-handle = <&npe 2>; + queue-rx = <&qmgr 30>; + queue-txready = <&qmgr 29>; + }; }; }; }; So you could add Tested-by: Corentin Labbe <clabbe@xxxxxxxxxxxx> Thanks
