Introduce RD-1 AE Platform #7065

ZiadElhanafy · 2024-10-02T14:47:37Z

1- Add initial support for RD-1 AE platform, this includes:
a- GIC and console initialization functions.
b- Memory layout.
c- Make files.
d- Assembly function get_core_pos_mpidr to compute the linear core position from MPIDR.

2- Accept GIC version 4 if CFG_ARM_GICV3 is enabled

etienne-lms · 2024-10-03T16:19:59Z

core/arch/arm/plat-rd1ae/conf.mk

+
+CFG_CORE_SEL1_SPMC      = y
+CFG_WITH_ARM_TRUSTED_FW	= y
+CFG_CORE_RESERVED_SHM   = n


replace = with ?= of $(call force,CFG_xxx,<value>)
Ditto at line 32.

etienne-lms · 2024-10-03T16:22:16Z

core/arch/arm/plat-rd1ae/platform_config.h

+#define PLATFORM_CORE_COUNT		(PLAT_RD1AE_CHIP_COUNT * \
+					PLAT_ARM_CLUSTER_COUNT * \
+					RD1AE_MAX_CPUS_PER_CLUSTER * \
+					RD1AE_MAX_PE_PER_CPU)


PLATFORM_CORE_COUNT seems unused.

etienne-lms · 2024-10-03T16:24:09Z

core/arch/arm/plat-rd1ae/main.c

+
+void boot_primary_init_intc(void)
+{
+	gic_init(GICC_BASE, GICD_BASE);


GICR unused? Use gic_init_v3()?

Yes it is unused, thanks for noticing. Removed #define GICR_BASE

etienne-lms · 2024-10-03T16:27:40Z

core/arch/arm/plat-rd1ae/conf.mk

+arm64-platform-cflags 	+= -mcpu=$(arm64-platform-cpuarch)
+arm64-platform-aflags 	+= -mcpu=$(arm64-platform-cpuarch)
+arm64-platform-cxxflags	+= -mcpu=$(arm64-platform-cpuarch)
+platform-flavor-armv9 := 1


Would it be worth creating a new core/arch/arm/cpu/cortex-armv9.mk file for these directives?

It would be a neoverse-v2.mk (Actually it is a Neoverse V3AE to be more precise but there is no arm64-platform-cpuarch that I could find yet for it). Would that be okay ?

Hmm, I don't knw. Do armv9 variants need each a specific toolachin artchitecture directives? On armv8.x, for example, we build all Arm 64bit machine with -mcpu=cortex-a53 even if not targeting a Cortex-A53 core.

That would be fine here too.

My question is, is it okay to create an .mk file as follows ?

core/arch/arm/cpu/neoverse-v2.mk

$(call force,CFG_HWSUPP_MEM_PERM_WXN,y) $(call force,CFG_HWSUPP_MEM_PERM_PXN,y) $(call force,CFG_ENABLE_SCTLR_RR,n) $(call force,CFG_ENABLE_SCTLR_Z,n) arm64-platform-cpuarch := neoverse-v2 arm64-platform-cflags += -mcpu=$(arm64-platform-cpuarch) arm64-platform-aflags += -mcpu=$(arm64-platform-cpuarch) arm64-platform-cxxflags += -mcpu=$(arm64-platform-cpuarch) platform-flavor-armv9 := 1

Could it be named core/arch/arm/cpu/neoverse-v2.mk if applicable to all armv9 platforms?
If not, maybe it's better to keep these in your plat-rd1ae/conf.mk file. There will be time for factorization when other armv9 based platforms are integrated.
@jforissier, @jenswi-linaro, what do you think?

No it will not be applicable to all armv9 platforms.

Then maybe create core/arch/arm/cpu/neoverse-v2.mk which would include core/arch/arm/cpu/armv9.mk (common part)?

Okay sounds good, just to confirm what you intend

1- core/arch/arm/cpu/cortex-armv9.mk would be:

$(call force,CFG_HWSUPP_MEM_PERM_WXN,y) $(call force,CFG_HWSUPP_MEM_PERM_PXN,y) $(call force,CFG_ENABLE_SCTLR_RR,n) $(call force,CFG_ENABLE_SCTLR_Z,n) arm64-platform-cflags += -mcpu=$(arm64-platform-cpuarch) arm64-platform-aflags += -mcpu=$(arm64-platform-cpuarch) arm64-platform-cxxflags += -mcpu=$(arm64-platform-cpuarch) platform-flavor-armv9 := 1

2- core/arch/arm/cpu/neoverse-v2.mk would be:

include core/arch/arm/cpu/cortex-armv9.mk arm64-platform-cpuarch := neoverse-v2

Is that right ?

Yes. In (2) it is preferable to swap the two lines. Although not strictly necessary since arm64-platform-cflags is currently a recursive variable (with deferred assignment) (A = ...), we tend to prefer immediate assignments (A := ...) because they are easier to follow and it is how the conditional assignment behaves (A ?= ...).

jforissier

For "core: introduce Arm Cortex-v9 and Neoverse-v2 CPU support":
Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>
See comments below on "plat-rd1ae: introduce RD-1 AE platform support"
For "core: gic: accept GIC version 4 if CFG_ARM_GICV3 is enabled":
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

core/arch/arm/plat-rd1ae/conf.mk

jforissier

For "plat-rd1ae: introduce RD-1 AE platform support":
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

jenswi-linaro · 2024-10-07T09:06:23Z

core/arch/arm/cpu/neoverse-v2.mk

@@ -0,0 +1,3 @@
+arm64-platform-cpuarch 	:= neoverse-v2


No arm32 support for EL0?

Can you please explain more why would we need arm32 support here ?

OP-TEE supports loading and executing both AArch64 and AArch32 TAs at S-EL0. I don't see a reason to stop it if the hardware supports it.

Well it wouldn't hurt to add it.
Excuse me if it is obvious, but would it be achieved by simply adding the following ?

In neoverse-v2.mk

arm32-platform-cpuarch := neoverse-v2

and in cortex-armv9.mk:

arm32-platform-cflags += -mcpu=$(arm32-platform-cpuarch) arm32-platform-aflags += -mcpu=$(arm32-platform-cpuarch) arm32-platform-cxxflags += -mcpu=$(arm32-platform-cpuarch)

Or is there something extra that needs to be done ?

Looks good to me.

jenswi-linaro · 2024-10-08T07:37:25Z

core/arch/arm/plat-rd1ae/conf.mk

+platform-cflags-debug-info = -gdwarf-4
+platform-aflags-debug-info = -gdwarf-4
+
+CFG_ARM64_core ?= y


If I remember correctly, this should be forced since armv9 doesn't support AArch32 at EL1.

jenswi-linaro · 2024-10-08T07:37:58Z

With my comment addressed, please apply:
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

etienne-lms

Acked-by: Etienne Carriere <etienne.carriere@foss.st.com> with a minor comment.

etienne-lms · 2024-10-11T12:03:27Z

core/arch/arm/plat-rd1ae/platform_config.h

+#define DRAM0_SIZE			0x80000000UL
+
+#define DRAM1_BASE			0x8080000000ULL
+#define DRAM1_SIZE			0x80000000ULL


nitpicking: would prefer UL(xxx)/ULL(xxx) instead of xxxUL/xxxULL

Introduce cortex-armv9.mk file and use it to support the Armv9 Neoverse v2 CPU. Signed-off-by: Ziad Elhanafy <ziad.elhanafy@arm.com> Acked-by: Etienne Carriere <etienne.carriere@foss.st.com> Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org> Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

Add initial support for RD-1 AE platform, this includes: 1- GIC and console initialization functions. 2- Memory layout. 3- Make files. 4- Assembly function `get_core_pos_mpidr` to compute the linear core position from MPIDR. Signed-off-by: Ziad Elhanafy <ziad.elhanafy@arm.com> Acked-by: Jerome Forissier <jerome.forissier@linaro.org> Acked-by: Etienne Carriere <etienne.carriere@foss.st.com> Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

GIC v4 is backwards compatible with GIC v3, Accept GIC version 4 if CFG_ARM_GICV3 is enabled. Signed-off-by: Ziad Elhanafy <ziad.elhanafy@arm.com> Acked-by: Jerome Forissier <jerome.forissier@linaro.org> Acked-by: Etienne Carriere <etienne.carriere@foss.st.com> Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

etienne-lms reviewed Oct 3, 2024

View reviewed changes

ZiadElhanafy force-pushed the rd1ae branch 2 times, most recently from 2d1cbdb to 6ea2226 Compare October 7, 2024 07:38

jforissier reviewed Oct 7, 2024

View reviewed changes

core/arch/arm/plat-rd1ae/conf.mk Show resolved Hide resolved

core/arch/arm/plat-rd1ae/conf.mk Outdated Show resolved Hide resolved

core/arch/arm/plat-rd1ae/conf.mk Outdated Show resolved Hide resolved

core/arch/arm/plat-rd1ae/conf.mk Outdated Show resolved Hide resolved

ZiadElhanafy force-pushed the rd1ae branch from 6ea2226 to 6dffcdf Compare October 7, 2024 08:32

jforissier reviewed Oct 7, 2024

View reviewed changes

ZiadElhanafy force-pushed the rd1ae branch from 6dffcdf to cf59076 Compare October 7, 2024 08:37

jenswi-linaro reviewed Oct 7, 2024

View reviewed changes

ZiadElhanafy force-pushed the rd1ae branch from cf59076 to fa130e8 Compare October 8, 2024 07:19

jenswi-linaro reviewed Oct 8, 2024

View reviewed changes

ZiadElhanafy force-pushed the rd1ae branch from fa130e8 to 67d5242 Compare October 8, 2024 07:55

etienne-lms reviewed Oct 11, 2024

View reviewed changes

ZiadElhanafy added 3 commits October 11, 2024 13:13

ZiadElhanafy force-pushed the rd1ae branch from 67d5242 to 67b5248 Compare October 11, 2024 12:13

jforissier merged commit dd18bd8 into OP-TEE:master Oct 11, 2024
9 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Introduce RD-1 AE Platform #7065

Introduce RD-1 AE Platform #7065

ZiadElhanafy commented Oct 2, 2024

etienne-lms Oct 3, 2024

ZiadElhanafy Oct 4, 2024

etienne-lms Oct 3, 2024

ZiadElhanafy Oct 4, 2024

etienne-lms Oct 3, 2024

ZiadElhanafy Oct 4, 2024

etienne-lms Oct 3, 2024

ZiadElhanafy Oct 4, 2024

etienne-lms Oct 4, 2024

ZiadElhanafy Oct 4, 2024 •

edited

Loading

etienne-lms Oct 4, 2024

ZiadElhanafy Oct 4, 2024

jforissier Oct 4, 2024

ZiadElhanafy Oct 4, 2024

jforissier Oct 4, 2024

ZiadElhanafy Oct 7, 2024

jforissier left a comment

jforissier left a comment

jenswi-linaro Oct 7, 2024

ZiadElhanafy Oct 7, 2024

jenswi-linaro Oct 7, 2024

ZiadElhanafy Oct 7, 2024

jenswi-linaro Oct 7, 2024

ZiadElhanafy Oct 8, 2024

jenswi-linaro Oct 8, 2024

ZiadElhanafy Oct 8, 2024

jenswi-linaro commented Oct 8, 2024

etienne-lms left a comment

etienne-lms Oct 11, 2024

ZiadElhanafy Oct 11, 2024

Introduce RD-1 AE Platform #7065

Introduce RD-1 AE Platform #7065

Conversation

ZiadElhanafy commented Oct 2, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ZiadElhanafy Oct 4, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jforissier left a comment

Choose a reason for hiding this comment

jforissier left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jenswi-linaro commented Oct 8, 2024

etienne-lms left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ZiadElhanafy Oct 4, 2024 •

edited

Loading