[PATCH u-boot v2019.04-aspeed-openbmc 10/11] crypto: Add driver for Aspeed HACE

[Klaus Heinrich Kiwi]

On 4/13/2021 5:41 PM, Klaus Heinrich Kiwi wrote:
> I've tested your patchset with Cedric's Aspeed-6.0 but looks
> like the probe function is never called. Reading through the code
> a bit more, looks like you need to explicitly probe this device
> somewhere in board_init_r (that is, after sdram was initialized),
> since functions like dm_scan_fdt() and dm_extended_scan_fdt() will
> only scan subnodes of the top level, and the clocks node.
> 
> This is what I get (with some added printfs of mine):

I've played around a bit more, and got it to work on Qemu with the following changes:

  * Added a board-specific spl_board_init() initializing the HACE driver at
    the SPL's board_init_r() timeframe. Enabled that on the defconfig file.
  * Because the driver model is using some pre-sdram malloc pool space,
    the changes above were causing the probing of the sdram itself to fail.
    Corrected by increasing the pre-sdram malloc pool to 0x1000.

 From 4ddc9fe2727a54f2d8c088d1c7046f4a3ab56314 Mon Sep 17 00:00:00 2001
From: Klaus Heinrich Kiwi <klaus at linux.vnet.ibm.com>
Date: Wed, 14 Apr 2021 16:29:15 -0300
Subject: [PATCH] fixup: Enable HACE probing on SPL, expand malloc pool

Signed-off-by: Klaus Heinrich Kiwi <klaus at linux.vnet.ibm.com>
---
  arch/arm/mach-aspeed/ast2600/spl.c         | 15 +++++++++++++++
  configs/ast2600_openbmc_spl_emmc_defconfig |  3 ++-
  2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/arch/arm/mach-aspeed/ast2600/spl.c b/arch/arm/mach-aspeed/ast2600/spl.c
index 527e14f8e3..2b1e8477eb 100644
--- a/arch/arm/mach-aspeed/ast2600/spl.c
+++ b/arch/arm/mach-aspeed/ast2600/spl.c
@@ -86,3 +86,18 @@ struct image_header *spl_get_load_buffer(ssize_t offset, size_t size)
  {
  	return (struct image_header *)(CONFIG_SYS_LOAD_ADDR);
  }
+
+#if CONFIG_IS_ENABLED(BOARD_INIT)
+void spl_board_init(void)
+{
+#ifdef CONFIG_ASPEED_HACE
+	struct udevice *dev;
+
+	if (uclass_get_device_by_driver(UCLASS_MISC, DM_GET_DRIVER(aspeed_hace),
+									&dev))
+		debug("Warning: HACE initialization failure\n");
+#endif
+
+	return;
+}
+#endif /* CONFIG_IS_ENABLED(BOARD_INIT) */
\ No newline at end of file
diff --git a/configs/ast2600_openbmc_spl_emmc_defconfig b/configs/ast2600_openbmc_spl_emmc_defconfig
index 05069084cb..e856df2398 100644
--- a/configs/ast2600_openbmc_spl_emmc_defconfig
+++ b/configs/ast2600_openbmc_spl_emmc_defconfig
@@ -24,7 +24,7 @@ CONFIG_ASPEED_KERNEL_FIT_DRAM_BASE=0x83000000
  CONFIG_TARGET_EVB_AST2600A1=y
  CONFIG_SPL_LIBCOMMON_SUPPORT=y
  CONFIG_SPL_LIBGENERIC_SUPPORT=y
-CONFIG_SYS_MALLOC_F_LEN=0x800
+CONFIG_SYS_MALLOC_F_LEN=0x1000
  CONFIG_SPL_MMC_SUPPORT=y
  CONFIG_SPL_SERIAL_SUPPORT=y
  CONFIG_SPL_DRIVERS_MISC_SUPPORT=y
@@ -48,6 +48,7 @@ CONFIG_SYS_CONSOLE_ENV_OVERWRITE=y
  CONFIG_DISPLAY_BOARDINFO_LATE=y
  CONFIG_ARCH_EARLY_INIT_R=y
  CONFIG_BOARD_EARLY_INIT_F=y
+CONFIG_SPL_BOARD_INIT=y
  # CONFIG_SPL_LEGACY_IMAGE_SUPPORT is not set
  CONFIG_SPL_SYS_MALLOC_SIMPLE=y
  CONFIG_SPL_STACK_R=y
-- 
2.25.1


However, when I tried to test it on a Rainier, it failed:
U-Boot SPL 2019.04 (Apr 14 2021 - 19:31:59 +0000)
already initialized, Trying to boot from MMC1
## Checking hash(es) for Image uboot ... sha512,rsa4096:autogenerated-uboot-4096-key- Failed to verify required signature 'key-autogenera'
  error!
Unable to verify required signature for '' hash node in 'uboot' image node
mmc_load_image_raw_sector: mmc block read error
Trying to boot from UART
CCCCC�P

(and yes, I had since disabled my debugging printf's). I wonder if the HASH_ISR
may need to be explicitly cleared, although I'd expect it to work for the first
command at least.

Another interesting thing is that the SPL tries to boot from UART, but neither
my fitImages, Legacy images or even RAW images are working.. Not sure if we need
some special handling of those images before feeding them to the spl ymodem loader?

  -Klaus

-- 
Klaus Heinrich Kiwi <klaus at linux.vnet.ibm.com>