[PATCH 0/9] sunxi: Allow FEL capable SPLs with 32bit builds

[Andre Przywara]

At the moment we can't use the USB-OTG based Allwinner FEL booting method
with mainline U-Boot on ARMv8 Allwinner SoCs. As we run the SPL in AArch64
with the MMU enabled, returning to the BootROM in AArch32 with the MMU off
and matching the BootROM's expectation is not trivial.

To workaround the problem, we can compile at least the SPL in AArch32.
This creates FEL-capable SPLs without further ado.
A second effect is the much smaller code size, due to smaller data types
on one hand, and due to the more compact Thumb2 encoding on the other.
Whereas Aarch64 builds get dangerously close to the 32KB limit, the Thumb2
build is less than 21KB.
This is the cornerstone for more whacky experiments like DRAM autodetection
and supporting multiple SoCs in one build.

So this series adds a Kconfig symbol to toggle between AArch64 and Thumb2
builds for the 64-bit Allwinner SoCs. The latter generates pure 32-bit
binaries (including U-Boot proper), though this is not too useful at the
moment (would require an ATF hack, for instance).
However this can be the base for further 32-bit experiments.

The first patch removes the exception vectors in the SPL build (which were
accidentally disabled anyway), to keep the Aarch64 H6 SPL within the limit.
Patch 02 to 05 prepare the build system and some generic sunxi code parts
for proper 32-bit compilation of ARMv8 boards.
Patch 06 then adds the actual new Kconfig symbol, which toggles the bitness
for a build.
The final three patches add defconfigs for the respective SoCs and their
DRAM variants. They must be build with an ARM(32) (cross) compiler, the
FEL capable SPL can be plucked from spl/sunxi-spl.bin. Short of the DRAM
configuration, there is nothing really board specific in there, so one
SPL defconfig can cover multiple boards.

The 32-bit SPL is flexible enough to launch 64-bit binaries as well, so
it can be combined with a 64-bit U-Boot proper and ATF, and also works
on SD cards, from eMMC or SPI flash. Consider it a drop-in replacement
for the 64-bit SPL build, although this has to be done manually.
Also FEL booting requires the firmware components to be given separately
on the sunxi-fel command line at the moment.

This is tested on the Pine64, Pine64-LTS, Orangepi-PC2 and Pine H64,
with FEL booting, normal AArch64 boots from an SD card and a 32-bit SPL/
64-bit U-Boot proper combination on an SD card.

Looking forward to any comments!

Cheers,
Andre.

Alexander Graf (1):
armv8: Disable exception vectors in SPL by default

Andre Przywara (8):
Makefile: use "arm64" architecture for U-Boot image files
SPL: read and store arch property from U-Boot image
sunxi: introduce RMR switch to enter payloads in 64-bit mode
sunxi: allow FIT image for 32-bit builds as well
sunxi: allow 32-bit builds for 64-bit SoCs
sunxi: H5: add generic H5 32-bit SPL defconfig
sunxi: A64: add generic A64 32-bit SPL defconfigs
sunxi: H6: add generic H6 32-bit SPL defconfig

Makefile | 12 +++++--
arch/arm/cpu/armv8/Kconfig | 2 +-
arch/arm/cpu/armv8/start.S | 4 +--
arch/arm/lib/spl.c | 14 ++++++++
arch/arm/mach-sunxi/Kconfig | 13 +++++--
arch/arm/mach-sunxi/Makefile | 3 ++
arch/arm/mach-sunxi/spl_switch.c | 64 +++++++++++++++++++++++++++++++++
common/spl/spl.c | 1 +
common/spl/spl_fit.c | 8 +++++
configs/sun50i-a64-ddr3-spl_defconfig | 13 +++++++
configs/sun50i-a64-lpddr3-spl_defconfig | 17 +++++++++
configs/sun50i-h5-ddr3-spl_defconfig | 16 +++++++++
configs/sun50i-h6-lpddr3-spl_defconfig | 12 +++++++
include/spl.h | 3 +-
14 files changed, 172 insertions(+), 10 deletions(-)
create mode 100644 arch/arm/mach-sunxi/spl_switch.c
create mode 100644 configs/sun50i-a64-ddr3-spl_defconfig
create mode 100644 configs/sun50i-a64-lpddr3-spl_defconfig
create mode 100644 configs/sun50i-h5-ddr3-spl_defconfig
create mode 100644 configs/sun50i-h6-lpddr3-spl_defconfig

--
2.14.5

[Andre Przywara]

From: Alexander Graf <ag...@suse.de>

Commit 1416e2d2253 ("armv8: make SPL exception vectors optional") had a
typo in it which effectively disabled exception handling in SPL code always.

Since nobody complained, I guess we may as well disable exception handling
in SPL always by default.

So fix the bug to make the config option effective, but disable exception
handling in SPL by default. This gets us to the same functionality as before
by default, but with much less code included in the binary.

Signed-off-by: Alexander Graf <ag...@suse.de>
Reviewed-by: Matthias Brugger <mbru...@suse.com>
Reviewed-by: Andre Przywara <andre.p...@arm.com>
---
arch/arm/cpu/armv8/Kconfig | 2 +-
arch/arm/cpu/armv8/start.S | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/arm/cpu/armv8/Kconfig b/arch/arm/cpu/armv8/Kconfig
index 1c12bbde75..fb061c8d73 100644
--- a/arch/arm/cpu/armv8/Kconfig
+++ b/arch/arm/cpu/armv8/Kconfig
@@ -3,7 +3,7 @@ if ARM64
config ARMV8_SPL_EXCEPTION_VECTORS
bool "Install crash dump exception vectors"
depends on SPL
- default y
+ default n
help
The default exception vector table is only used for the crash
dump, but still takes quite a lot of space in the image size.
diff --git a/arch/arm/cpu/armv8/start.S b/arch/arm/cpu/armv8/start.S
index 12a78ee38b..fe52166e28 100644
--- a/arch/arm/cpu/armv8/start.S
+++ b/arch/arm/cpu/armv8/start.S
@@ -88,7 +88,7 @@ pie_fixup_done:
bl reset_sctrl
#endif

-#if defined(CONFIG_ARMV8__SPL_EXCEPTION_VECTORS) || !defined(CONFIG_SPL_BUILD)
+#if defined(CONFIG_ARMV8_SPL_EXCEPTION_VECTORS) || !defined(CONFIG_SPL_BUILD)
.macro set_vbar, regname, reg
msr \regname, \reg
.endm
@@ -354,7 +354,7 @@ ENDPROC(smp_kick_all_cpus)
/*-----------------------------------------------------------------------*/

ENTRY(c_runtime_cpu_setup)
-#if defined(CONFIG_ARMV8__SPL_EXCEPTION_VECTORS) || !defined(CONFIG_SPL_BUILD)
+#if defined(CONFIG_ARMV8_SPL_EXCEPTION_VECTORS) || !defined(CONFIG_SPL_BUILD)
/* Relocate vBAR */
adr x0, vectors
switch_el x1, 3f, 2f, 1f
--
2.14.5

[Andre Przywara]

At the moment we use the arch/arm directory for arm64 boards as well,
so the Makefile will pick up the "arm" name for the architecture to use
for tagging binaries in U-Boot image files.
Differentiate between the two by looking at the CPU variable being defined
to "armv8", and use the arm64 architecture name on creating the image
file if that matches.

Signed-off-by: Andre Przywara <andre.p...@arm.com>
---
Makefile | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/Makefile b/Makefile
index 75a5c7d171..d6e7c69f40 100644
--- a/Makefile
+++ b/Makefile
@@ -1167,13 +1167,19 @@ $(U_BOOT_ITS): $(U_BOOT_ITS_DEPS) FORCE
endif
endif

+ifeq ($(CPU),armv8)
+IH_ARCH := arm64
+else
+IH_ARCH := $(ARCH)
+endif
+
ifdef CONFIG_SPL_LOAD_FIT
-MKIMAGEFLAGS_u-boot.img = -f auto -A $(ARCH) -T firmware -C none -O u-boot \
+MKIMAGEFLAGS_u-boot.img = -f auto -A $(IH_ARCH) -T firmware -C none -O u-boot \
-a $(CONFIG_SYS_TEXT_BASE) -e $(CONFIG_SYS_UBOOT_START) \
-n "U-Boot $(UBOOTRELEASE) for $(BOARD) board" -E \
$(patsubst %,-b arch/$(ARCH)/dts/%.dtb,$(subst ",,$(CONFIG_OF_LIST)))
else
-MKIMAGEFLAGS_u-boot.img = -A $(ARCH) -T firmware -C none -O u-boot \
+MKIMAGEFLAGS_u-boot.img = -A $(IH_ARCH) -T firmware -C none -O u-boot \
-a $(CONFIG_SYS_TEXT_BASE) -e $(CONFIG_SYS_UBOOT_START) \
-n "U-Boot $(UBOOTRELEASE) for $(BOARD) board"
MKIMAGEFLAGS_u-boot-ivt.img = -A $(ARCH) -T firmware_ivt -C none -O u-boot \
--
2.14.5

[Andre Przywara]

Read the specified "arch" value from a legacy or FIT U-Boot image and
store it in our SPL data structure.
This allows loaders to take the target architecture in account for
custom loading procedures.
Having the complete string -> arch mapping for FIT based images in the
SPL would be too big, so we leave it up to architectures (or boards) to
overwrite the weak function that does the actual translation, possibly
covering only the required subset there.

Signed-off-by: Andre Przywara <andre.p...@arm.com>
---

common/spl/spl.c | 1 +
common/spl/spl_fit.c | 8 ++++++++
include/spl.h | 3 ++-
3 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/common/spl/spl.c b/common/spl/spl.c
index 2e2af1b28e..7d25e72bbb 100644
--- a/common/spl/spl.c
+++ b/common/spl/spl.c
@@ -272,6 +272,7 @@ int spl_parse_image_header(struct spl_image_info *spl_image,
#endif

spl_image->os = image_get_os(header);
+ spl_image->arch = image_get_arch(header);
spl_image->name = image_get_name(header);
debug(SPL_TPL_PROMPT
"payload image: %32s load addr: 0x%lx size: %d\n",
diff --git a/common/spl/spl_fit.c b/common/spl/spl_fit.c
index db436268cb..d583946571 100644
--- a/common/spl/spl_fit.c
+++ b/common/spl/spl_fit.c
@@ -340,6 +340,11 @@ static int spl_fit_image_get_os(const void *fit, int noffset, uint8_t *os)
#endif
}

+__weak u8 spl_genimg_get_arch_id(const char *arch_str)
+{
+ return IH_ARCH_DEFAULT;
+}
+
int spl_load_simple_fit(struct spl_image_info *spl_image,
struct spl_load_info *info, ulong sector, void *fit)
{
@@ -351,6 +356,7 @@ int spl_load_simple_fit(struct spl_image_info *spl_image,
int images, ret;
int base_offset, hsize, align_len = ARCH_DMA_MINALIGN - 1;
int index = 0;
+ const char *arch_str;

/*
* For FIT with external data, figure out where the external images
@@ -467,6 +473,8 @@ int spl_load_simple_fit(struct spl_image_info *spl_image,
else
spl_image->os = IH_OS_U_BOOT;
#endif
+ arch_str = fdt_getprop(fit, node, "arch", NULL);
+ spl_image->arch = spl_genimg_get_arch_id(arch_str);

/*
* Booting a next-stage U-Boot may require us to append the FDT.
diff --git a/include/spl.h b/include/spl.h
index f09909e189..aac85150a7 100644
--- a/include/spl.h
+++ b/include/spl.h
@@ -64,7 +64,6 @@ static inline bool u_boot_first_phase(void)

struct spl_image_info {
const char *name;
- u8 os;
uintptr_t load_addr;
uintptr_t entry_point;
#if CONFIG_IS_ENABLED(LOAD_FIT)
@@ -79,6 +78,8 @@ struct spl_image_info {
ulong dcrc_length;
ulong dcrc;
#endif
+ u8 os;
+ u8 arch;
};

/*
--
2.14.5

[Andre Przywara]

The ARMv8 capable Allwinner A64 SoC comes out of reset in AArch32 mode.
To run AArch64 code, we have to trigger a warm reset via the RMR register,
which proceeds with code execution at the address stored in the RVBAR
register.
If the bootable payload in the FIT image is using a different
architecture than the SPL has been compiled for, enter it via this said
RMR switch mechanism, by writing the entry point address into the MMIO
mapped, writable version of the RVBAR register.
Then the warm reset is triggered via a system register write.
If the payload architecture is the same as the SPL, we use the normal
branch as usual.

Signed-off-by: Andre Przywara <andre.p...@arm.com>
---

arch/arm/lib/spl.c | 14 +++++++++
arch/arm/mach-sunxi/Makefile | 3 ++
arch/arm/mach-sunxi/spl_switch.c | 64 ++++++++++++++++++++++++++++++++++++++++
3 files changed, 81 insertions(+)
create mode 100644 arch/arm/mach-sunxi/spl_switch.c

diff --git a/arch/arm/lib/spl.c b/arch/arm/lib/spl.c
index 33cc76ba3d..4d9370d232 100644
--- a/arch/arm/lib/spl.c
+++ b/arch/arm/lib/spl.c
@@ -73,3 +73,17 @@ void __noreturn jump_to_image_linux(struct spl_image_info *spl_image)
}
#endif /* CONFIG_ARM64 */
#endif
+
+u8 spl_genimg_get_arch_id(const char *arch_str)
+{
+ if (!arch_str)
+ return IH_ARCH_DEFAULT;
+
+ if (!strcmp(arch_str, "arm"))
+ return IH_ARCH_ARM;
+
+ if (!strcmp(arch_str, "arm64"))
+ return IH_ARCH_ARM64;

+
+ return IH_ARCH_DEFAULT;
+}

diff --git a/arch/arm/mach-sunxi/Makefile b/arch/arm/mach-sunxi/Makefile
index 43a93e3085..5a5e10a024 100644
--- a/arch/arm/mach-sunxi/Makefile
+++ b/arch/arm/mach-sunxi/Makefile
@@ -39,4 +39,7 @@ obj-$(CONFIG_SPL_SPI_SUNXI) += spl_spi_sunxi.o
obj-$(CONFIG_SUNXI_DRAM_DW) += dram_sunxi_dw.o
obj-$(CONFIG_SUNXI_DRAM_DW) += dram_timings/
obj-$(CONFIG_DRAM_SUN50I_H6) += dram_sun50i_h6.o
+obj-$(CONFIG_MACH_SUN50I) += spl_switch.o
+obj-$(CONFIG_MACH_SUN50I_H5) += spl_switch.o
+obj-$(CONFIG_MACH_SUN50I_H6) += spl_switch.o
endif
diff --git a/arch/arm/mach-sunxi/spl_switch.c b/arch/arm/mach-sunxi/spl_switch.c
new file mode 100644
index 0000000000..1de43c2396
--- /dev/null
+++ b/arch/arm/mach-sunxi/spl_switch.c
@@ -0,0 +1,64 @@
+/*
+ * (C) Copyright 2016 ARM Ltd.
+ *
+ * SPDX-License-Identifier: GPL-2.0+
+ */
+
+#include <common.h>
+#include <spl.h>
+
+#include <asm/io.h>
+#include <asm/barriers.h>
+
+static void __noreturn jump_to_image_native(struct spl_image_info *spl_image)
+{
+ typedef void __noreturn (*image_entry_noargs_t)(void);
+
+ image_entry_noargs_t image_entry =
+ (image_entry_noargs_t)spl_image->entry_point;
+
+ image_entry();
+}
+
+static void __noreturn reset_rmr_switch(void)
+{
+#ifdef CONFIG_ARM64
+ __asm__ volatile ( "mrs x0, RMR_EL3\n\t"
+ "bic x0, x0, #1\n\t" /* Clear enter-in-64 bit */
+ "orr x0, x0, #2\n\t" /* set reset request bit */
+ "msr RMR_EL3, x0\n\t"
+ "isb sy\n\t"
+ "nop\n\t"
+ "wfi\n\t"
+ "b .\n"
+ ::: "x0");
+#else
+ __asm__ volatile ( "mrc 15, 0, r0, cr12, cr0, 2\n\t"
+ "orr r0, r0, #3\n\t" /* request reset in 64 bit */
+ "mcr 15, 0, r0, cr12, cr0, 2\n\t"
+ "isb\n\t"
+ "nop\n\t"
+ "wfi\n\t"
+ "b .\n"
+ ::: "r0");
+#endif
+ while (1); /* to avoid a compiler warning about __noreturn */
+}
+
+void __noreturn jump_to_image_no_args(struct spl_image_info *spl_image)
+{
+ if (spl_image->arch == IH_ARCH_DEFAULT) {
+ debug("entering by branch\n");
+ jump_to_image_native(spl_image);
+ } else {
+ debug("entering by RMR switch\n");
+#ifdef CONFIG_MACH_SUN50I_H6
+ writel(spl_image->entry_point, 0x09010040);
+#else
+ writel(spl_image->entry_point, 0x017000a0);
+#endif
+ DSB;
+ ISB;
+ reset_rmr_switch();
+ }
+}
--
2.14.5

[Andre Przywara]

Even though we introduced FIT image support for the SPL to cover the
64-bit SoCs, there is no technical limitation to those parts.

Change the Makefile stanza to always create a FIT image if the
particular config either provided a script or an .its file.

Signed-off-by: Andre Przywara <andre.p...@arm.com>
---

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

diff --git a/Makefile b/Makefile
index d6e7c69f40..3a560e29bf 100644
--- a/Makefile
+++ b/Makefile
@@ -1350,7 +1350,7 @@ u-boot-x86-16bit.bin: u-boot FORCE
endif

ifneq ($(CONFIG_ARCH_SUNXI),)
-ifeq ($(CONFIG_ARM64),)
+ifeq ($(CONFIG_SPL_FIT_SOURCE)$(CONFIG_SPL_FIT_GENERATOR),)
u-boot-sunxi-with-spl.bin: spl/sunxi-spl.bin u-boot.img u-boot.dtb FORCE
$(call if_changed,binman)
else
--
2.14.5

[Andre Przywara]

At the moment we build the SPL and U-Boot proper for the 64-bit AArch64
instruction set. But since the cores provide an AArch32 compatibility mode
and in fact the BootROM runs in 32-bit mode, it can be useful to have at
least the SPL run in AArch32 as well. This has two advantages:
- As AArch32 features the compact Thumb2 instruction encoding, we can
get a much smaller image size, which is a relief for our SPL.
- Staying in AArch32, with the MMU turned off, allows an easy return to
the BootROM and its FEL mode. This enables FEL booting on those SoCs.

Introduce a Kconfig option which toggles between CONFIG_ARM64 and
CONFIG_CPU_V7A, to allow easy switching between the two modes. This can
be manually selected in menuconfig, but follow-up patches will introduce
a separate defconfig for that purpose.

Signed-off-by: Andre Przywara <andre.p...@arm.com>
---

arch/arm/mach-sunxi/Kconfig | 13 ++++++++++---
1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/arch/arm/mach-sunxi/Kconfig b/arch/arm/mach-sunxi/Kconfig
index 74e234cded..347d737fd0 100644
--- a/arch/arm/mach-sunxi/Kconfig
+++ b/arch/arm/mach-sunxi/Kconfig
@@ -146,6 +146,10 @@ config SUNXI_DRAM_MAX_SIZE
default 0xC0000000 if MACH_SUN50I || MACH_SUN50I_H5 || MACH_SUN50I_H6
default 0x80000000

+config SUNXI_ARMV8_32BIT_BUILD
+ bool "Build 32-bit binaries for ARMv8 SoCs"
+ default n
+
choice
prompt "Sunxi SoC Variant"
optional
@@ -275,7 +279,8 @@ config MACH_SUN9I

config MACH_SUN50I
bool "sun50i (Allwinner A64)"
- select ARM64
+ select ARM64 if !SUNXI_ARMV8_32BIT_BUILD
+ select CPU_V7A if SUNXI_ARMV8_32BIT_BUILD
select DM_I2C
select PHY_SUN4I_USB
select SUN6I_PRCM
@@ -291,14 +296,16 @@ config MACH_SUN50I

config MACH_SUN50I_H5
bool "sun50i (Allwinner H5)"
- select ARM64
+ select ARM64 if !SUNXI_ARMV8_32BIT_BUILD
+ select CPU_V7A if SUNXI_ARMV8_32BIT_BUILD
select MACH_SUNXI_H3_H5
select FIT
select SPL_LOAD_FIT

config MACH_SUN50I_H6
bool "sun50i (Allwinner H6)"
- select ARM64
+ select ARM64 if !SUNXI_ARMV8_32BIT_BUILD
+ select CPU_V7A if SUNXI_ARMV8_32BIT_BUILD
select SUPPORT_SPL
select FIT
select SPL_LOAD_FIT
--
2.14.5

[Andre Przywara]

To be able to use FEL booting (Allwinner's USB-OTG protocol), we need a
32 bit SPL, to easily return into the 32 bit BootROM.
Add a generic defconfig for that purpose, that should work on all boards.
The resulting spl/sunxi-spl.bin can be used with the sunxi-fel tool to
initialise the DRAM via FEL. This allows to load the normal versions for
the rest of the firmware components (ATF, 64 bit U-Boot proper).
This SPL will determine the architecture of the loaded binaries and will
switch the bitness if necessary.

Signed-off-by: Andre Przywara <andre.p...@arm.com>
---

configs/sun50i-h5-ddr3-spl_defconfig | 16 ++++++++++++++++
1 file changed, 16 insertions(+)
create mode 100644 configs/sun50i-h5-ddr3-spl_defconfig

diff --git a/configs/sun50i-h5-ddr3-spl_defconfig b/configs/sun50i-h5-ddr3-spl_defconfig
new file mode 100644
index 0000000000..7b720e2fc5
--- /dev/null
+++ b/configs/sun50i-h5-ddr3-spl_defconfig
@@ -0,0 +1,16 @@
+CONFIG_ARM=y
+CONFIG_ARCH_SUNXI=y
+CONFIG_SPL=y
+CONFIG_MACH_SUN50I_H5=y
+CONFIG_SUNXI_ARMV8_32BIT_BUILD=y
+CONFIG_DRAM_CLK=672
+CONFIG_DRAM_ZQ=3881977
+CONFIG_SPL_SPI_SUNXI=y
+CONFIG_NR_DRAM_BANKS=1
+# CONFIG_SYS_MALLOC_CLEAR_ON_INIT is not set
+# CONFIG_CMD_FLASH is not set
+# CONFIG_SPL_DOS_PARTITION is not set
+# CONFIG_SPL_EFI_PARTITION is not set
+CONFIG_DEFAULT_DEVICE_TREE="sun50i-h5-orangepi-pc2"
+CONFIG_MMC_SUNXI_SLOT_EXTRA=2
+CONFIG_SYS_USB_EVENT_POLL_VIA_INT_QUEUE=y
--
2.14.5

[Andre Przywara]

To be able to use FEL booting (Allwinner's USB-OTG protocol), we need a
32 bit SPL, to easily return into the 32 bit BootROM.

Add two generic defconfigs for that purpose, one for boards with DDR3
DRAM, the other for those with LPDDR3 DRAM chips.

The resulting spl/sunxi-spl.bin can be used with the sunxi-fel tool to
initialise the DRAM via FEL. This allows to load the normal versions for
the rest of the firmware components (ATF, 64 bit U-Boot proper).
This SPL will determine the architecture of the loaded binaries and will
switch the bitness if necessary.

Signed-off-by: Andre Przywara <andre.p...@arm.com>
---

configs/sun50i-a64-ddr3-spl_defconfig | 13 +++++++++++++
configs/sun50i-a64-lpddr3-spl_defconfig | 17 +++++++++++++++++
2 files changed, 30 insertions(+)

create mode 100644 configs/sun50i-a64-ddr3-spl_defconfig
create mode 100644 configs/sun50i-a64-lpddr3-spl_defconfig

diff --git a/configs/sun50i-a64-ddr3-spl_defconfig b/configs/sun50i-a64-ddr3-spl_defconfig
new file mode 100644
index 0000000000..fda15e2491
--- /dev/null
+++ b/configs/sun50i-a64-ddr3-spl_defconfig
@@ -0,0 +1,13 @@
+CONFIG_ARM=y
+CONFIG_ARCH_SUNXI=y
+CONFIG_SPL=y
+CONFIG_MACH_SUN50I=y
+CONFIG_SUNXI_ARMV8_32BIT_BUILD=y

+CONFIG_NR_DRAM_BANKS=1
+# CONFIG_SYS_MALLOC_CLEAR_ON_INIT is not set
+# CONFIG_CMD_FLASH is not set
+# CONFIG_SPL_DOS_PARTITION is not set
+# CONFIG_SPL_EFI_PARTITION is not set

+CONFIG_DEFAULT_DEVICE_TREE="sun50i-a64-pine64-plus"
+CONFIG_SPL_SPI_SUNXI=y
+CONFIG_MMC_SUNXI_SLOT_EXTRA=2
diff --git a/configs/sun50i-a64-lpddr3-spl_defconfig b/configs/sun50i-a64-lpddr3-spl_defconfig
new file mode 100644
index 0000000000..20bc376cbd
--- /dev/null
+++ b/configs/sun50i-a64-lpddr3-spl_defconfig
@@ -0,0 +1,17 @@
+CONFIG_ARM=y
+CONFIG_ARCH_SUNXI=y
+CONFIG_SPL=y
+CONFIG_MACH_SUN50I=y
+CONFIG_SUNXI_ARMV8_32BIT_BUILD=y
+CONFIG_SUNXI_DRAM_LPDDR3_STOCK=y
+CONFIG_DRAM_CLK=552
+CONFIG_DRAM_ZQ=3881949

+CONFIG_NR_DRAM_BANKS=1
+# CONFIG_SYS_MALLOC_CLEAR_ON_INIT is not set
+# CONFIG_CMD_FLASH is not set
+# CONFIG_SPL_DOS_PARTITION is not set
+# CONFIG_SPL_EFI_PARTITION is not set

+CONFIG_DEFAULT_DEVICE_TREE="sun50i-a64-pine64-lts"
+CONFIG_MMC0_CD_PIN=""
+CONFIG_SPL_SPI_SUNXI=y
+CONFIG_MMC_SUNXI_SLOT_EXTRA=2
--
2.14.5

[Andre Przywara]

To be able to use FEL booting (Allwinner's USB-OTG protocol), we need a
32 bit SPL, to easily return into the 32 bit BootROM.

Add a generic defconfig for that purpose, that should work on all boards.

The resulting spl/sunxi-spl.bin can be used with the sunxi-fel tool to
initialise the DRAM via FEL. This allows to load the normal versions for
the rest of the firmware components (ATF, 64 bit U-Boot proper).
This SPL will determine the architecture of the loaded binaries and will
switch the bitness if necessary.

Signed-off-by: Andre Przywara <andre.p...@arm.com>
---

configs/sun50i-h6-lpddr3-spl_defconfig | 12 ++++++++++++
1 file changed, 12 insertions(+)
create mode 100644 configs/sun50i-h6-lpddr3-spl_defconfig

diff --git a/configs/sun50i-h6-lpddr3-spl_defconfig b/configs/sun50i-h6-lpddr3-spl_defconfig
new file mode 100644
index 0000000000..8480d027a5
--- /dev/null
+++ b/configs/sun50i-h6-lpddr3-spl_defconfig
@@ -0,0 +1,12 @@
+CONFIG_ARM=y
+CONFIG_ARCH_SUNXI=y
+CONFIG_SPL=y
+CONFIG_MACH_SUN50I_H6=y
+CONFIG_SUNXI_ARMV8_32BIT_BUILD=y
+CONFIG_MMC_SUNXI_SLOT_EXTRA=2

+CONFIG_NR_DRAM_BANKS=1
+# CONFIG_SYS_MALLOC_CLEAR_ON_INIT is not set
+# CONFIG_CMD_FLASH is not set
+# CONFIG_SPL_DOS_PARTITION is not set
+# CONFIG_SPL_EFI_PARTITION is not set

+CONFIG_DEFAULT_DEVICE_TREE="sun50i-h6-pine-h64"
--
2.14.5

[Andre Przywara]

On Thu, 21 Feb 2019 12:47:04 +0100
Alexander Graf <ag...@suse.de> wrote:

Hi Alex,

> On 02/21/2019 02:30 AM, Andre Przywara wrote:
> > Read the specified "arch" value from a legacy or FIT U-Boot image and
> > store it in our SPL data structure.
> > This allows loaders to take the target architecture in account for
> > custom loading procedures.
> > Having the complete string -> arch mapping for FIT based images in the
> > SPL would be too big, so we leave it up to architectures (or boards) to
> > overwrite the weak function that does the actual translation, possibly
> > covering only the required subset there.
> >
> > Signed-off-by: Andre Przywara <andre.p...@arm.com>
>

> Reviewed-by: Alexander Graf <ag...@suse.de>

Thanks for having a look!

> I don't fully buy the argument that the generic mapping would be too big
> though. Realistically you should be able to get away with 1 or 2
> branches per case, no? So that would be maybe 40 instructions?

You are apparently not in the Allwinner SPL mindset ;-)
I get excited when I find a way of saving 100 bytes (see the writel_relaxed
patch), so wasting 160 bytes when we will probably never need to return
IH_ARCH_X86_64 doesn't sound right to me.

And Philipp is right: the canonical way would be to use uimage_arch[] from
common/image.c, which is quite big.

We could have a big #ifdef cascade, something like:
+#ifdef CONFIG_ARM

+ if (!strcmp(arch_str, "arm"))
+ return IH_ARCH_ARM;
+
+ if (!strcmp(arch_str, "arm64"))
+ return IH_ARCH_ARM64;

+#endif
+#ifdef CONFIG_X86
+ if (!strcmp(arch_str, "x86"))
+ return IH_ARCH_I386;
+
+ if (!strcmp(arch_str, "x86_64"))
+ return IH_ARCH_X86_64;
+#endif
....

I am just not sure it's worth to introduce this out of the blue *now*,
without use cases and without proper testing.

Cheers,
Andre.

[Andre Przywara]

On Thu, 21 Feb 2019 12:52:50 +0100
Alexander Graf <ag...@suse.de> wrote:

> On 02/21/2019 02:30 AM, Andre Przywara wrote:

> How does the original entry point get preserved over this? Don't you
> have to set the reset vector register as well?

The RVBAR_ELx (Reset Vector Base Address Register) registers are
architecturally read only. Fortunately Allwinner decided to have an MMIO
mapped writable alias, which we exploit here. I *believe* this gets reset
on a hard reset (external signal, for instance via the watchdog) to the
initial 0x0 value.

Anyway we use this method already in the AArch64 SPL today [1], and it
happens to work ;-)

Cheers,
Andre.

[1] http://git.denx.de/?p=u-boot.git;a=blob;f=arch/arm/include/asm/arch-sunxi/boot0.h;hb=HEAD

[André Przywara]

On 22/02/2019 08:06, Alexander Graf wrote:

>
>
> On 21.02.19 02:30, Andre Przywara wrote:
>> Even though we introduced FIT image support for the SPL to cover the
>> 64-bit SoCs, there is no technical limitation to those parts.
>>
>> Change the Makefile stanza to always create a FIT image if the
>> particular config either provided a script or an .its file.
>>
>> Signed-off-by: Andre Przywara <andre.p...@arm.com>
>> ---
>> Makefile | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/Makefile b/Makefile
>> index d6e7c69f40..3a560e29bf 100644
>> --- a/Makefile
>> +++ b/Makefile
>> @@ -1350,7 +1350,7 @@ u-boot-x86-16bit.bin: u-boot FORCE
>> endif
>>
>> ifneq ($(CONFIG_ARCH_SUNXI),)
>> -ifeq ($(CONFIG_ARM64),)
>> +ifeq ($(CONFIG_SPL_FIT_SOURCE)$(CONFIG_SPL_FIT_GENERATOR),)
>

> The commit message sounds as if this should be an "ifneq"?

If find those ifeq(something,) constructs always confusing, but I think
it's right here: The line in the if-clause creates the *legacy* U-Boot
image, so the two strings must be both empty for it to match and to
execute those commands. If only one of them is "y", the comparison with
the empty string will fail and the else clause will be called, which
creates the FIT image.

Or did I trick myself?

Cheers,
Andre.

[Andre Przywara]

Hi Alex,

On 2/22/19 2:13 AM, Alexander Graf wrote:
>
>
> On 21.02.19 02:30, Andre Przywara wrote:

> Can't you make this "imply ARM64" instead and then just set
> ARM64=n;CPU_V7A=y in the defconfig?

Ha, that indeed sounds tempting, and actually I tried this before, but
it doesn't work. The kernel doc says that an "imply" setting can be
overwritten by a direct dependency or a visible prompt. ARM64 is not
visible (empty bool), so I figured that introducing this simple symbol
is the smallest pain to tackle this issue.

Cheers,
Andre.

>
> Alex