[PATCH 1/3] configs: rzg2_beacon: Disable the ability to remove devices
In order to save some space, disable the ability to dynamically remove devices. Without this, U-Boot is too large to load without recompiling TF-A.
Signed-off-by: Adam Ford aford173@gmail.com
diff --git a/configs/rzg2_beacon_defconfig b/configs/rzg2_beacon_defconfig index 73abe966b8..7b14d225b5 100644 --- a/configs/rzg2_beacon_defconfig +++ b/configs/rzg2_beacon_defconfig @@ -48,6 +48,7 @@ CONFIG_ENV_OVERWRITE=y CONFIG_ENV_IS_IN_MMC=y CONFIG_SYS_MMC_ENV_PART=2 CONFIG_VERSION_VARIABLE=y +# CONFIG_DM_DEVICE_REMOVE is not set CONFIG_REGMAP=y CONFIG_SYSCON=y CONFIG_CLK=y
The ENV size and offset were changed to different values in Beacon's downstream release. Change them to the same values as the downstream for consistent behavior.
Signed-off-by: Adam Ford aford173@gmail.com
diff --git a/configs/rzg2_beacon_defconfig b/configs/rzg2_beacon_defconfig index 7b14d225b5..534f641e84 100644 --- a/configs/rzg2_beacon_defconfig +++ b/configs/rzg2_beacon_defconfig @@ -3,7 +3,8 @@ CONFIG_ARCH_RMOBILE=y CONFIG_TEXT_BASE=0x50000000 CONFIG_SYS_MALLOC_LEN=0x4000000 CONFIG_SYS_MALLOC_F_LEN=0x2000 -CONFIG_ENV_OFFSET=0x0 +CONFIG_ENV_SIZE=0x2000 +CONFIG_ENV_OFFSET=0xFFFFE000 CONFIG_DM_GPIO=y CONFIG_DEFAULT_DEVICE_TREE="r8a774a1-beacon-rzg2m-kit" CONFIG_RCAR_GEN3=y
On Wed, Oct 25, 2023 at 6:13 PM Adam Ford aford173@gmail.com wrote:
The ENV size and offset were changed to different values in Beacon's downstream release. Change them to the same values as the downstream for consistent behavior.
Signed-off-by: Adam Ford aford173@gmail.com
Marek,
I know you had some feedback on other patches in this series. I'll drop one, and refactor another, but is this one ok to merge as-is or should I re-post without the other patches?
adam
diff --git a/configs/rzg2_beacon_defconfig b/configs/rzg2_beacon_defconfig index 7b14d225b5..534f641e84 100644 --- a/configs/rzg2_beacon_defconfig +++ b/configs/rzg2_beacon_defconfig @@ -3,7 +3,8 @@ CONFIG_ARCH_RMOBILE=y CONFIG_TEXT_BASE=0x50000000 CONFIG_SYS_MALLOC_LEN=0x4000000 CONFIG_SYS_MALLOC_F_LEN=0x2000 -CONFIG_ENV_OFFSET=0x0 +CONFIG_ENV_SIZE=0x2000 +CONFIG_ENV_OFFSET=0xFFFFE000 CONFIG_DM_GPIO=y CONFIG_DEFAULT_DEVICE_TREE="r8a774a1-beacon-rzg2m-kit" CONFIG_RCAR_GEN3=y -- 2.40.1
On 10/31/23 16:13, Adam Ford wrote:
On Wed, Oct 25, 2023 at 6:13 PM Adam Ford aford173@gmail.com wrote:
The ENV size and offset were changed to different values in Beacon's downstream release. Change them to the same values as the downstream for consistent behavior.
Signed-off-by: Adam Ford aford173@gmail.com
Marek,
I know you had some feedback on other patches in this series. I'll drop one, and refactor another, but is this one ok to merge as-is or should I re-post without the other patches?
Please repost, and use the right To: address (+renesas@mailbox) , I'll pick it up shortly.
There is a function inside the board file to autodetect which device tree is needed by U-Boot to properly load its own device tree, but it currently defaults to always loading RZ/G2M for Linux. This is not correct for other SoC variants. Add board_late_init function to query the SoC name and use that to determine which device tree is loaded for booting Linux.
Signed-off-by: Adam Ford aford173@gmail.com
diff --git a/board/beacon/beacon-rzg2m/beacon-rzg2m.c b/board/beacon/beacon-rzg2m/beacon-rzg2m.c index 99fe1edfb3..6d37ff6ac9 100644 --- a/board/beacon/beacon-rzg2m/beacon-rzg2m.c +++ b/board/beacon/beacon-rzg2m/beacon-rzg2m.c @@ -6,6 +6,7 @@ #include <common.h> #include <asm/global_data.h> #include <asm/io.h> +#include <env.h>
DECLARE_GLOBAL_DATA_PTR;
@@ -17,6 +18,38 @@ int board_init(void) return 0; }
+#if IS_ENABLED(CONFIG_BOARD_LATE_INIT) +static u8 get_SoC_letter(void) +{ + const u8 *name = rzg_get_cpu_name(); + + if (*name) + return name[6]; + + return 0; +} + +int board_late_init(void) +{ + /* If already defined, exit */ + if (!env_get("fdt_file")) { + switch (get_SoC_letter()) { + case 'A': + env_set("fdt_file", "r8a774a1-beacon-rzg2m-kit.dtb"); + break; + case 'B': + env_set("fdt_file", "r8a774b1-beacon-rzg2n-kit.dtb"); + break; + case 'E': + env_set("fdt_file", "r8a774e1-beacon-rzg2h-kit.dtb"); + break; + } + } + + return 0; +} +#endif /* CONFIG_BOARD_LATE_INIT */ + #if IS_ENABLED(CONFIG_MULTI_DTB_FIT) int board_fit_config_name_match(const char *name) { diff --git a/configs/rzg2_beacon_defconfig b/configs/rzg2_beacon_defconfig index 534f641e84..6894448b22 100644 --- a/configs/rzg2_beacon_defconfig +++ b/configs/rzg2_beacon_defconfig @@ -21,6 +21,7 @@ CONFIG_USE_BOOTCOMMAND=y CONFIG_BOOTCOMMAND="mmc dev ${mmcdev}; if mmc rescan; then if run loadbootscript; then run bootscript; else if run loadimage; then run mmcboot; else run netboot; fi; fi; else booti ${loadaddr} - ${fdt_addr}; fi" CONFIG_DEFAULT_FDT_FILE="r8a774a1-beacon-rzg2m-kit.dtb" # CONFIG_BOARD_EARLY_INIT_F is not set +CONFIG_BOARD_LATE_INIT=y CONFIG_SYS_MALLOC_BOOTPARAMS=y CONFIG_HUSH_PARSER=y CONFIG_SYS_MAXARGS=64 diff --git a/include/configs/beacon-rzg2m.h b/include/configs/beacon-rzg2m.h index 65c01835cc..493b98fbdf 100644 --- a/include/configs/beacon-rzg2m.h +++ b/include/configs/beacon-rzg2m.h @@ -18,7 +18,6 @@ "fdt_addr=0x48000000\0" \ "loadaddr=0x48080000\0" \ "boot_fdt=try\0" \ - "fdt_file=" CONFIG_DEFAULT_FDT_FILE "\0" \ "initrd_addr=0x43800000\0" \ "mmcdev=1\0" \ "mmcpart=1\0" \
On 10/26/23 01:13, Adam Ford wrote:
There is a function inside the board file to autodetect which device tree is needed by U-Boot to properly load its own device tree, but it currently defaults to always loading RZ/G2M for Linux. This is not correct for other SoC variants. Add board_late_init function to query the SoC name and use that to determine which device tree is loaded for booting Linux.
Can't this be scripted in environment instead ?
On Thu, Oct 26, 2023 at 8:44 PM Marek Vasut marex@denx.de wrote:
On 10/26/23 01:13, Adam Ford wrote:
There is a function inside the board file to autodetect which device tree is needed by U-Boot to properly load its own device tree, but it currently defaults to always loading RZ/G2M for Linux. This is not correct for other SoC variants. Add board_late_init function to query the SoC name and use that to determine which device tree is loaded for booting Linux.
Can't this be scripted in environment instead ?
Do you have a recommendation for how to read the SoC type from a script?
adam
On 10/27/23 04:12, Adam Ford wrote:
On Thu, Oct 26, 2023 at 8:44 PM Marek Vasut marex@denx.de wrote:
On 10/26/23 01:13, Adam Ford wrote:
There is a function inside the board file to autodetect which device tree is needed by U-Boot to properly load its own device tree, but it currently defaults to always loading RZ/G2M for Linux. This is not correct for other SoC variants. Add board_late_init function to query the SoC name and use that to determine which device tree is loaded for booting Linux.
Can't this be scripted in environment instead ?
Do you have a recommendation for how to read the SoC type from a script?
If you want to read PRR from script:
$ setexpr prrcontent *0xFFF00044 $ printenv prrcontent
On 10/26/23 01:13, Adam Ford wrote:
In order to save some space, disable the ability to dynamically remove devices. Without this, U-Boot is too large to load without recompiling TF-A.
Did you enable LTO yet ?
Striping DM_REMOVE seems like a really bad idea, doesn't that break 'usb reset' and its removal of devices ?
On Thu, Oct 26, 2023 at 8:44 PM Marek Vasut marex@denx.de wrote:
On 10/26/23 01:13, Adam Ford wrote:
In order to save some space, disable the ability to dynamically remove devices. Without this, U-Boot is too large to load without recompiling TF-A.
Did you enable LTO yet ?
Yes.
Striping DM_REMOVE seems like a really bad idea, doesn't that break 'usb reset' and its removal of devices ?
I did not, but we can drop this one, and I'll just have to re-compile TF-A to load a larger U-Boot.
adam
participants (3)
-
Adam Ford -
Marek Vasut -
Marek Vasut