From: Chris.Packham@alliedtelesis.co.nz (Chris Packham) Date: Mon, 15 Dec 2014 00:57:49 +0000 Subject: [RFC/PATCH 1/4 PATCH] ARM: mvebu: use dt_fixup to provide fallback for enable-method In-Reply-To: <548DFCEA.7000801@alliedtelesis.co.nz> References: <5488A2C0.5040804@alliedtelesis.co.nz> <1418360035-27975-1-git-send-email-chris.packham@alliedtelesis.co.nz> <2005632.nq5F7SJD3S@wuerfel> <548DFCEA.7000801@alliedtelesis.co.nz> Message-ID: <548E3209.20307@alliedtelesis.co.nz> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Answering my own question On 12/15/2014 10:11 AM, Chris Packham wrote: > Hi Arnd, > > On 12/13/2014 12:35 AM, Arnd Bergmann wrote: >> On Friday 12 December 2014 17:53:55 Chris Packham wrote: >>>> I briefly explored that approach here[1]. The tricky part would be >>>> handling the fact that the enable method can be attached to either the >>>> /cpus node or and individual /cpu entry (or is that something I can >>>> ignore?). >>>> >>>> In the end I thought that the unconditional setting of smp_ops was >>>> easier to implement and would achieve the same result. >>> >>> Actually as it turns out it's not that hard to implement something that >>> checks both /cpus and /cpus/cpu at n. I'll include this with the next round >>> after I've waited for anymore feedback. >> >> Ah, very nice! >> >>> ---8<--- >>> Subject: [PATCH] ARM: mvebu: use dt_fixup to provide fallback for >>> enable-method >>> >>> When the device tree doesn't define an enable-method insert a property >>> into the flattened device tree. arm_dt_init_cpu_maps() will then parse >>> this an set smp_ops appropriately. Now that we have this fallback it is >>> no longer necessary to set .smp in the DT_MACHINE definition. >>> >>> Signed-off-by: Chris Packham >>> --- >>> arch/arm/mach-mvebu/Makefile | 2 ++ >>> arch/arm/mach-mvebu/board-v7.c | 37 >>> ++++++++++++++++++++++++++++++++++++- >>> 2 files changed, 38 insertions(+), 1 deletion(-) >>> >>> diff --git a/arch/arm/mach-mvebu/Makefile b/arch/arm/mach-mvebu/Makefile >>> index e24136b..68310f8 100644 >>> --- a/arch/arm/mach-mvebu/Makefile >>> +++ b/arch/arm/mach-mvebu/Makefile >>> @@ -14,3 +14,5 @@ endif >>> obj-$(CONFIG_MACH_DOVE) += dove.o >>> obj-$(CONFIG_MACH_KIRKWOOD) += kirkwood.o kirkwood-pm.o >>> obj-$(CONFIG_MACH_NETXBIG) += netxbig.o >>> + >>> +CFLAGS_board-v7.o = -I$(src)/../../../scripts/dtc/libfdt >>> \ No newline at end of file >> >> Why is this needed? Can't you just include ? >> > > I am already including linux/libfdt.h. Without the CFLAGS change I get > the following compile error. > > In file included from include/linux/libfdt.h:6:0, > from arch/arm/mach-mvebu/board-v7.c:21: > include/linux/../../scripts/dtc/libfdt/libfdt.h:54:24: fatal error: > libfdt_env.h: No such file or directory > > There seems to be precedence in other architectures/drivers > > $ git grep -e '-I.*libfdt' > arch/mips/cavium-octeon/Makefile:CFLAGS_octeon-platform.o = > -I$(src)/../../../scripts/dtc/libfdt > arch/mips/cavium-octeon/Makefile:CFLAGS_setup.o = > -I$(src)/../../../scripts/dtc/libfdt > arch/mips/mti-sead3/Makefile:CFLAGS_sead3-setup.o = > -I$(src)/../../../scripts/dtc/libfdt > arch/powerpc/kernel/Makefile:CFLAGS_prom.o = > -I$(src)/../../../scripts/dtc/libfdt > drivers/firmware/efi/libstub/Makefile:CFLAGS_fdt.o += > -I$(srctree)/scripts/dtc/libfdt/ > drivers/of/Makefile:CFLAGS_fdt.o = -I$(src)/../../scripts/dtc/libfdt > drivers/of/Makefile:CFLAGS_fdt_address.o = > -I$(src)/../../scripts/dtc/libfdt > >>> +static void __init armada_370_xp_dt_fixup(void) >>> +{ >>> + int offset, node; >>> + int i, len; >>> + void *prop; >>> + char buffer[20]; >>> + >>> + offset = fdt_path_offset(initial_boot_params, "/cpus"); >>> + if (offset < 0) >>> + return; >>> + >>> + prop = fdt_getprop(initial_boot_params, offset, "enable-method", >>> &len); >>> + if (prop) >>> + return; >>> + >>> + for (i = 0; i < NR_CPUS; i++) { >>> + snprintf(buffer, sizeof(buffer), "cpu@%d", i); >>> + node = fdt_subnode_offset(initial_boot_params, offset, buffer); >>> + if (node < 0) >>> + break; >>> + prop = fdt_getprop(initial_boot_params, node, >>> + "enable-method", &len); >>> + if (prop) >>> + return; >>> + } >>> + >>> + pr_info("No enable-method defined. " >>> + "Falling back to \"marvell,armada-xp-smp\"\n"); >>> + >>> + fdt_setprop(initial_boot_params, offset, "enable-method", >>> + "marvell,armada-xp-smp", sizeof("marvell,armada-xp-smp")); >>> +} >> >> I think it would be good to first check whether you are running on >> Armada XP >> or Armada 370, because the latter does not require this. >> >> Arnd >> > > Any suggestion as to what to check for. Based on the dts files in the > current source seem compatible == "marvell,armada370" seems like a good > choice. But I don't know for sure that there isn't a armada-xp variant > out using a .dts with this set. Adding the following at the top will do the trick (assuming "marvell,armada370" is the right compatible string to use). unsigned long root = of_get_flat_dt_root(); if (of_flat_dt_is_compatible(root, "marvell,armada370")) return; > > Another option might be just to count the /cpu at n nodes.