arm64: Initial SoC support (ARMv8 RTSM model)

Based on Rob Herring's patches for arch/arm, this patch adds a dtbs target to arch/arm64/boot/Makefile. In the process, also clean up the unused MACHINE variable.

Signed-off-by Catalin Marinas <[email protected]>

---

I'll drop this patch once Rob pushes his version into linux-next (or mainline).

 arch/arm64/Makefile        | 17 +++++++++++------
 arch/arm64/boot/.gitignore |  1 +
 arch/arm64/boot/Makefile   |  8 +++++++-
 3 files changed, 19 insertions(+), 7 deletions(-)
 create mode 100644 arch/arm64/boot/dts/Makefile

arch/arm64/Makefile +11 -6

index 364191f..64089bc 100644
@@ -41,20 +41,24 @@ libs-y		:= arch/arm64/lib/ $(libs-y)
 libs-y		+= $(LIBGCC)
 
 # Default target when executing plain make
-KBUILD_IMAGE := Image.gz
+KBUILD_IMAGE	:= Image.gz
+KBUILD_DTBS	:= dtbs
 
-all:	$(KBUILD_IMAGE)
+all:	$(KBUILD_IMAGE) $(KBUILD_DTBS)
 
 boot := arch/arm64/boot
 
 Image Image.gz: vmlinux
-	$(Q)$(MAKE) $(build)=$(boot) MACHINE=$(MACHINE) $(boot)/$@
+	$(Q)$(MAKE) $(build)=$(boot) $(boot)/$@
 
 zinstall install: vmlinux
-	$(Q)$(MAKE) $(build)=$(boot) MACHINE=$(MACHINE) $@
+	$(Q)$(MAKE) $(build)=$(boot) $@
 
-%.dtb:
-	$(Q)$(MAKE) $(build)=$(boot) MACHINE=$(MACHINE) $(boot)/$@
+%.dtb: scripts
+	$(Q)$(MAKE) $(build)=$(boot) $(boot)/$@
+
+dtbs: scripts
+	$(Q)$(MAKE) $(build)=$(boot) $(boot)/$@
 
 # We use MRPROPER_FILES and CLEAN_FILES now
 archclean:

@@ -63,6 +67,7 @@ archclean:
 define archhelp
   echo  '* Image.gz      - Compressed kernel image (arch/$(ARCH)/boot/Image.gz)'
   echo  '  Image         - Uncompressed kernel image (arch/$(ARCH)/boot/Image)'
+  echo  '* dtbs          - Build device tree blobs for enabled boards'
   echo  '  install       - Install uncompressed kernel'
   echo  '  zinstall      - Install compressed kernel'
   echo  '                  Install using (your) ~/bin/installkernel or'

arch/arm64/boot/.gitignore +1 -0

index 8dab0bb..98af90a 100644
@@ -1,2 +1,3 @@
 Image
 Image.gz
+*.dtb

arch/arm64/boot/Makefile +7 -1

index eca209b..1ab9bf1 100644
@@ -14,6 +14,8 @@
 # Based on the ia64 boot/Makefile.
 #
 
+include $(srctree)/arch/arm64/boot/dts/Makefile
+
 targets := Image Image.gz
 
 $(obj)/Image: vmlinux FORCE

@@ -22,9 +24,13 @@ $(obj)/Image: vmlinux FORCE
 $(obj)/Image.gz: $(obj)/Image FORCE
 	$(call if_changed,gzip)
 
-$(obj)/%.dtb: $(src)/dts/%.dts
+targets += $(dtb-y)
+
+$(obj)/%.dtb: $(src)/dts/%.dts FORCE
 	$(call cmd,dtc)
 
+$(obj)/dtbs: $(addprefix $(obj)/, $(dtb-y))
+
 install: $(obj)/Image
 	$(CONFIG_SHELL) $(srctree)/$(src)/install.sh $(KERNELRELEASE) \
 	$(obj)/Image System.map "$(INSTALL_PATH)"

arch/arm64/boot/dts/Makefile new file +0 -0

new file mode 100644
index 0000000..e69de29

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Quoted 4 lines · Catalin Marinas

Based on Rob Herring's patches for arch/arm, this patch adds a dtbs target to arch/arm64/boot/Makefile. In the process, also clean up the unused MACHINE variable.

Quoted 5 lines

Signed-off-by Catalin Marinas <[email protected]>

---

I'll drop this patch once Rob pushes his version into linux-next (or mainline).

This and Stephen's dtb rule series are in linux-next now.

From addc6b45f915f6bbab238a2edd2072f3b474e673 Mon Sep 17 00:00:00 2001

Quoted 5 lines · Rob Herring

Quoted 4 lines · Catalin Marinas

Based on Rob Herring's patches for arch/arm, this patch adds a dtbs target to arch/arm64/boot/Makefile. In the process, also clean up the unused MACHINE variable.

Quoted 7 lines

Quoted 5 lines

Signed-off-by Catalin Marinas <[email protected]>

---

I'll drop this patch once Rob pushes his version into linux-next (or mainline).

This and Stephen's dtb rule series are in linux-next now.

One minor fix needed. Please merge or fold it, whatever is easier. Thanks.

-----------------------8<------------------------------

From addc6b45f915f6bbab238a2edd2072f3b474e673 Mon Sep 17 00:00:00 2001
From: Catalin Marinas <[email protected]>
Date: Wed, 21 Nov 2012 11:44:59 +0000
Subject: [PATCH] arm64: Fix the dtbs target building

The arch/arm64/Makefile was not passing the right target to the boot/dts/Makefile.

Signed-off-by Catalin Marinas <[email protected]>

---
 arch/arm64/Makefile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

arch/arm64/Makefile +1 -1

index 11c3599..c95c5cb 100644
@@ -58,7 +58,7 @@ zinstall install: vmlinux
 	$(Q)$(MAKE) $(build)=$(boot)/dts $(boot)/dts/$@
 
 dtbs: scripts
-	$(Q)$(MAKE) $(build)=$(boot)/dts $(boot)/dts/$@
+	$(Q)$(MAKE) $(build)=$(boot)/dts dtbs
 
 # We use MRPROPER_FILES and CLEAN_FILES now
 archclean:

arch/arm64/platforms/vexpress.c new file +6 -0

+static const char *vexpress_dt_match[] __initdata = {
+       "arm,vexpress",
+       NULL,
+};
+
+static int __init vexpress_init(void)

Quoted 9 lines · Catalin Marinas

arch/arm64/platforms/Kconfig +4 -0

index e69de29..3533d71 100644
@@ -0,0 +1,11 @@
+config ARCH_VEXPRESS
+       bool "ARMv8 software model (Versatile Express)"
+       select ARCH_REQUIRE_GPIOLIB
+       select ARM_AMBA

arch/arm64/platforms/Kconfig line 8

Quoted 4 lines

arch/arm64/platforms/Kconfig +4 -0

+       select ARM_GIC
+       select COMMON_CLK_VERSATILE
+       select GENERIC_GPIO
+       select VEXPRESS_CONFIG

I would expect that we can always select ARCH_REQUIRE_GPIOLIB, ARM_AMBA, ARM_GIC and GENERIC_GPIO for ARM64. Do you expect to see systems that don't have all of those?

Quoted 7 lines

arch/arm64/platforms/vexpress.c new file +6 -0

+static const char *vexpress_dt_match[] __initdata = {
+       "arm,vexpress",
+       NULL,
+};
+
+static int __init vexpress_init(void)

arch/arm64/platforms/vexpress.c line 38

Quoted 10 lines

arch/arm64/platforms/vexpress.c new file +10 -0

+{
+       if (!of_flat_dt_match(of_get_flat_dt_root(), vexpress_dt_match))
+               return 0;
+
+       vexpress_sysreg_of_early_init();
+       vexpress_clk_of_init();
+
+       return 0;
+}
+arch_initcall(vexpress_init);

I think it would be more appropriate for these two to check the presence of the individual devices, and move the caller into the actual device driver, rather than checking for the root node of the device tree.

There may be cases where we want to check the root node, but both the clock setup and the sysreg should both be detectable.

Quoted 22 lines

Quoted 7 lines

(quoted diff) +6 -0

+static const char *vexpress_dt_match[] __initdata = {
+       "arm,vexpress",
+       NULL,
+};
+
+static int __init vexpress_init(void)

Review

Quoted 10 lines

(quoted diff) +10 -0

+{
+       if (!of_flat_dt_match(of_get_flat_dt_root(), vexpress_dt_match))
+               return 0;
+
+       vexpress_sysreg_of_early_init();
+       vexpress_clk_of_init();
+
+       return 0;
+}
+arch_initcall(vexpress_init);

I think it would be more appropriate for these two to check the presence of the individual devices, and move the caller into the actual device driver, rather than checking for the root node of the device tree.

Quoted 2 lines

There may be cases where we want to check the root node, but both the clock setup and the sysreg should both be detectable.

I cc'ed Pawel as well since it's touching his code. We've been through several versions internally and didn't find a clear winner. Currently, vexpress_clk_of_init() requires vexpress_sysreg_of_early_init() to be called first. Pawel has a patch to allow sysreg self-initialisation when called from vexpress_clk_of_init().

This leaves us with a way to call vexpress_clk_of_init() directly from clk-vexpress.c. There are two ways:

1. arch_initcall() in clk-vexpress.c, checking for the DT match.
2. driver registration (arch_initcall) and later probing when of_platform_populate() is called from the arch/arm64 code.

The 2nd option above relies on the DT order since the clocks must be registered before any AMBA device is registered, so I'm not too keen. This leaves us with option 1 or any other suggestion you may have.

From a19b22ed4d5042a2249f0aa41633cbbc7310d5f3 Mon Sep 17 00:00:00 2001

Quoted 34 lines · Catalin Marinas

Quoted 22 lines

Quoted 7 lines

(quoted diff) +6 -0

+static const char *vexpress_dt_match[] __initdata = {
+       "arm,vexpress",
+       NULL,
+};
+
+static int __init vexpress_init(void)

Review

Quoted 10 lines

(quoted diff) +10 -0

+{
+       if (!of_flat_dt_match(of_get_flat_dt_root(), vexpress_dt_match))
+               return 0;
+
+       vexpress_sysreg_of_early_init();
+       vexpress_clk_of_init();
+
+       return 0;
+}
+arch_initcall(vexpress_init);

I think it would be more appropriate for these two to check the presence of the individual devices, and move the caller into the actual device driver, rather than checking for the root node of the device tree.

Quoted 2 lines

There may be cases where we want to check the root node, but both the clock setup and the sysreg should both be detectable.

I cc'ed Pawel as well since it's touching his code. We've been through several versions internally and didn't find a clear winner. Currently, vexpress_clk_of_init() requires vexpress_sysreg_of_early_init() to be called first. Pawel has a patch to allow sysreg self-initialisation when called from vexpress_clk_of_init().

This leaves us with a way to call vexpress_clk_of_init() directly from clk-vexpress.c. There are two ways:

Quoted 7 lines

1. arch_initcall() in clk-vexpress.c, checking for the DT match.
2. driver registration (arch_initcall) and later probing when of_platform_populate() is called from the arch/arm64 code.

The 2nd option above relies on the DT order since the clocks must be registered before any AMBA device is registered, so I'm not too keen. This leaves us with option 1 or any other suggestion you may have.

The idea I had was to associate the root of the device tree with the "platform" device, see below. This makes it possible to write a platform driver for the particular platform, so the code for the vexpress could look more or less like this...

static int __devinit vexpress_probe(struct platform_device *dev)

{
        vexpress_sysreg_of_early_init();
        vexpress_clk_of_init();

        of_platform_populate(NULL, of_default_bus_match_table, NULL, NULL);

        return 0;
}

Quoted 23 lines · Arnd Bergmann

Quoted 7 lines · Catalin Marinas

arch/arm64/platforms/vexpress.c new file +6 -0

+static const char *vexpress_dt_match[] __initdata = {
+       "arm,vexpress",
+       NULL,
+};
+
+static int __init vexpress_init(void)

arch/arm64/platforms/vexpress.c line 38

Quoted 10 lines

arch/arm64/platforms/vexpress.c new file +10 -0

+{
+       if (!of_flat_dt_match(of_get_flat_dt_root(), vexpress_dt_match))
+               return 0;
+
+       vexpress_sysreg_of_early_init();
+       vexpress_clk_of_init();
+
+       return 0;
+}
+arch_initcall(vexpress_init);

I think it would be more appropriate for these two to check the presence of the individual devices, and move the caller into the actual device driver, rather than checking for the root node of the device tree.

Quoted 2 lines

There may be cases where we want to check the root node, but both the clock setup and the sysreg should both be detectable.

I cc'ed Pawel as well since it's touching his code. We've been through several versions internally and didn't find a clear winner. Currently, vexpress_clk_of_init() requires vexpress_sysreg_of_early_init() to be called first. Pawel has a patch to allow sysreg self-initialisation when called from vexpress_clk_of_init().

This leaves us with a way to call vexpress_clk_of_init() directly from clk-vexpress.c. There are two ways:

1. arch_initcall() in clk-vexpress.c, checking for the DT match.
2. driver registration (arch_initcall) and later probing when of_platform_populate() is called from the arch/arm64 code.

After some more thinking, none of these would work for arch/arm since vexpress_clk_of_init() is called very early via the time_init() ... v2m_dt_timer_init(). On arm64 we rely on the architected timers to be always present but this assumption is not valid on arm where clocks would need to be initialised early.

While I'd really like to get rid of the SoC code in arch/arm64, the best place I see for the vexpress_clk_of_init() call is still the arch code (unless we add #ifdefs in the clk-vexpress.c code).

Quoted 9 lines · Catalin Marinas

After some more thinking, none of these would work for arch/arm since vexpress_clk_of_init() is called very early via the time_init() ... v2m_dt_timer_init(). On arm64 we rely on the architected timers to be always present but this assumption is not valid on arm where clocks would need to be initialised early.

While I'd really like to get rid of the SoC code in arch/arm64, the best place I see for the vexpress_clk_of_init() call is still the arch code (unless we add #ifdefs in the clk-vexpress.c code).

I think I'd prefer the #ifdef in this case. Ideally, we would get rid of most of the early init functions in ARM as well, but that is a different battle for now. Getting the arch/arm64 to look nice IMHO is more important in this case, and we can fix the 32 bit version later.

Quoted 11 lines · Arnd Bergmann

Quoted 9 lines · Catalin Marinas

After some more thinking, none of these would work for arch/arm since vexpress_clk_of_init() is called very early via the time_init() ... v2m_dt_timer_init(). On arm64 we rely on the architected timers to be always present but this assumption is not valid on arm where clocks would need to be initialised early.

While I'd really like to get rid of the SoC code in arch/arm64, the best place I see for the vexpress_clk_of_init() call is still the arch code (unless we add #ifdefs in the clk-vexpress.c code).

Quoted 5 lines

I think I'd prefer the #ifdef in this case. Ideally, we would get rid of most of the early init functions in ARM as well, but that is a different battle for now. Getting the arch/arm64 to look nice IMHO is more important in this case, and we can fix the 32 bit version later.

I'm sure we'll be able to make the vexpress_clk_* functions behave correctly, either with a help of #ifdef or in some other way - don't concern yourself with this detail.

The bigger fish to fry is the generic initialization of the clocks (as
in: when to call the of_clk_init() and how to gather all possible of_match-es)...