plat: renesas: Add support for RZ/A series #6

nhutnguyenkc · 2025-06-26T06:54:01Z

Add support for Renesas RZ/A series
The original work is from https://github.com/renesas/rza-initial-program-loader

MichaelAllportWM · 2025-09-22T08:49:02Z

Hi @binhnguyen2434, @nhutnguyenkc
I see that this pull request is in draft mode and has merge conflicts.
I'm assuming that I should only be reviewing the files added by the last commit.
I will continue with that whilst I wait for confirmation.

Add support for Renesas RZ/A series Signed-off-by: Nhut Nguyen <[email protected]>

nhutnguyenkc · 2025-09-23T09:00:50Z

Hi @MichaelAllportWM ,
Thanks for your review. I have rebased on master branch to resolve conflicts. Now you can see and review changes.
This PR is in draft mode; the official contribution will follow TF-A contribution guide via gerrit.

MichaelAllportWM · 2025-09-24T10:25:06Z

Param swizzle
There are a few param swizzle DDR configs that could be removed.
As far as I can tell the only ones used are:

./plat/renesas/rza/common/drivers/ddr/param_swizzle_T1.c
./plat/renesas/rza/common/drivers/ddr/param_swizzle_T3bcud2.c

The ones below mention the V2L in the comments and are not used in your code:

./plat/renesas/rza/common/drivers/ddr/param_swizzle_T1vbc.c
./plat/renesas/rza/common/drivers/ddr/param_swizzle_T1vc.c
./plat/renesas/rza/common/drivers/ddr/param_swizzle_T2vc.c

I assume that they have remained in error.

There are also other param swizzle configurations which are not used but I understand if you want to keep them for possible use by customers with different board setups for example param_swizzle_T1b.c is not used by any of our boards but could possibly be used by other setups.

MichaelAllportWM · 2025-09-24T10:25:33Z

Checkpatch
There are quite a few checkpatch errors and warnings that could possibly be resolved.
I ran:
git diff master | ~/scripts/checkpatch/checkpatch.pl
And got:
total: 2 errors, 317 warnings, 22173 lines checked

MichaelAllportWM · 2025-09-24T10:25:54Z

Param MC
Again there are files included that are not used, however I also understand if you want to keep them in for customers with other setups. e.g. ./plat/renesas/rza/soc/a3ul/drivers/ddr/param_mc_C-011_D3-01-2.c is not used by any makefile. No change necessary if you are happy.

MichaelAllportWM · 2025-09-24T10:26:08Z

Strict aliasing rule
Some sections of the code that do not adhere to this rule, however there also appears to be plenty of places in TF-A where this rule is ignored. Perhaps then it is not a requirement for this code considering the architecture. I thought it was best to mention it though. No change necessary if you are happy.
e.g. plat/renesas/rza/common/drivers/auth/sblib/sblib_parser.c:67

MichaelAllportWM · 2025-09-24T10:27:03Z

#include <string.h>
A lot of the xspi, qspi, octa header files have this include. I'm not sure that it is necessary.

plat/renesas/rza/common/drivers/xspidevice/octaflash_mx66uw/octaflash_mx66uw_api.h
plat/renesas/rza/common/drivers/xspidevice/octaram_apsxx/octaram_apsxx_api.h
plat/renesas/rza/common/drivers/xspidevice/qspiflash_at25/qspiflash_at25_api.h
plat/renesas/rza/common/drivers/xspidevice/qspiflash_mx25l25645g/qspiflash_mx25l25645g_api.h
plat/renesas/rza/common/include/xspidevice_api.h

MichaelAllportWM · 2025-09-24T10:27:27Z

Dates in copywrite header
I'm not sure what the preferred conditions are for updating these, I've not yet checked any of these.

MichaelAllportWM · 2025-09-19T13:39:43Z

plat/renesas/rza/common/include/drivers/auth/auth_common.h

+	AUTH_METHOD_HASH, /* Authenticate by hash matching */
+	AUTH_METHOD_SIG, /* Authenticate by PK operation */
+	AUTH_METHOD_NV_CTR, /* Authenticate by Non-Volatile Counter */
+	AUTH_METHOD_SBLIB, /* Authenticate by Renesas RZ/G2L SBLib */


The G2L is not included in this repository so I think that this comment should probably change.

MichaelAllportWM · 2025-09-19T15:06:17Z

plat/renesas/rza/common/drivers/auth/sblib/sblib_api_private.h

+ */
+#define SB_RET_SAME_IMAGE_VERSION		((sb_ret_t)0x55005501UL)
+
+/** A internal failure */


Minor spelling correction: An internal failure

MichaelAllportWM · 2025-09-19T15:07:03Z

plat/renesas/rza/common/drivers/auth/sblib/sblib_api_private.h

+/** Unsupported algorithm */
+#define SB_RET_ERR_CRYPTO_UNSUPPORTED_ALGORITHM	((sb_ret_t)0xAAAA0202UL)
+
+/** Other resorece is using CryptoIP. */


Minor spelling correction: Other resource

MichaelAllportWM · 2025-09-19T15:07:58Z

plat/renesas/rza/common/drivers/auth/sblib/sblib_api_private.h

+/** CRC mismatch */
+#define SB_RET_ERR_CRC_MISMATCH			((sb_ret_t)0xAAAA0300UL)
+
+/** Unsupported polynominal */


Minor spelling correction: Unsupported polynomial

MichaelAllportWM · 2025-09-19T15:12:47Z

plat/renesas/rza/common/drivers/auth/sblib/crypto_sblib.c

+
+static sb_secure_boot_api_t secure_boot_api;
+
+int crypto_sblib_auth(void *data_ptr, size_t len, const void *key_cert,


*data_ptr and len are unused.

MichaelAllportWM · 2025-09-23T08:17:48Z

plat/renesas/rza/common/drivers/pfc.c

Again consider changing large arrays to static const.

MichaelAllportWM · 2025-09-23T09:39:59Z

plat/renesas/rza/common/drivers/scifa.S

+func console_rza_flush
+	ldr x0, [x0, #CONSOLE_T_BASE]
+1:
+	/* Check TEND flag */
+	ldrh w1, [x0, #FSR]
+	and w1, w1, #FSR_TEND
+	cmp	w1, #FSR_TEND
+	bne	1b
+
+	mov	w0, #0
+	ret
+endfunc console_rza_flush


Potential issue if built with LOG_LEVEL=0
The TEND flag will never be set, thus the device may loop infinitely.

This is catered for in this driver:
https://github.com/renesas-rz/rzg_trusted-firmware-a/blob/2.10.5/rzv2h_1.1.0/plat/renesas/rz/common/drivers/scifa.S

MichaelAllportWM · 2025-09-23T09:53:27Z

plat/renesas/rza/common/drivers/spi_multi/AT25QL128A/spi_multi_device.c

+	while (1) {
+		read_status =
+			spi_multi_cmd_read(SMCMR_CMD_READ_STATUS_REGISTER_1);
+		if ((read_status & STATUS_1_BUSY_BIT) == STATUS_1_BUSY) {
+			udelay(STATUS_BUSY_READ_DELAY_TIME);
+			continue;
+		} else {
+			break;
+		}
+	}


Minor point- It might be possible to write this section with fewer lines

do { read_status = spi_multi_cmd_read(SMCMR_CMD_READ_STATUS_REGISTER_1); if (!(read_status & STATUS_1_BUSY_BIT)) { break; } udelay(STATUS_BUSY_READ_DELAY_TIME); } while (1);

MichaelAllportWM · 2025-09-23T12:41:27Z

plat/renesas/rza/common/drivers/xspidevice/octaflash_mx66uw/octaflash_mx66uw.c

+		if (false)
+			flash_read_status_register_spi(
+				myctrl); /* Avoiding build error */


Upstream reviewers may ask for this to be changed.
Consider using a typedef'ed version of __attribute__((unused)) or removing the unused function

MichaelAllportWM · 2025-09-23T15:26:01Z

plat/renesas/rza/common/drivers/io/io_emmcdrv.c

+	//	Temporarily disable DMA.
+	//	if ((buffer + length - 1U) <= (uintptr_t)UINT32_MAX) {
+	//		emmc_dma = LOADIMAGE_FLAGS_DMA_ENABLE;
+	//	}


Commented out code. Could this possibly be removed?

nhutnguyenkc force-pushed the rza_add_ipl_support branch from 6474002 to f890e68 Compare August 11, 2025 03:46

nhutnguyenkc force-pushed the rza_add_ipl_support branch from f890e68 to 2adc306 Compare September 4, 2025 06:06

nhutnguyenkc force-pushed the rza_add_ipl_support branch from 2adc306 to a8ec691 Compare September 12, 2025 10:46

nashif force-pushed the master branch from 713ffbf to 0a29cac Compare September 19, 2025 11:54

plat: renesas: Add support for RZ/A series

3a21f74

Add support for Renesas RZ/A series Signed-off-by: Nhut Nguyen <[email protected]>

nhutnguyenkc force-pushed the rza_add_ipl_support branch from a8ec691 to 3a21f74 Compare September 23, 2025 08:55

MichaelAllportWM reviewed Sep 24, 2025

View reviewed changes


		static sb_secure_boot_api_t secure_boot_api;

		int crypto_sblib_auth(void data_ptr, size_t len, const void key_cert,

plat: renesas: Add support for RZ/A series #6

Are you sure you want to change the base?

plat: renesas: Add support for RZ/A series #6

Uh oh!

Conversation

nhutnguyenkc commented Jun 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MichaelAllportWM commented Sep 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

nhutnguyenkc commented Sep 23, 2025

Uh oh!

MichaelAllportWM commented Sep 24, 2025

Uh oh!

MichaelAllportWM commented Sep 24, 2025

Uh oh!

MichaelAllportWM commented Sep 24, 2025

Uh oh!

MichaelAllportWM commented Sep 24, 2025

Uh oh!

MichaelAllportWM commented Sep 24, 2025

Uh oh!

MichaelAllportWM commented Sep 24, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

nhutnguyenkc commented Jun 26, 2025 •

edited

Loading

MichaelAllportWM commented Sep 22, 2025 •

edited

Loading