Hi Dinesh,
On Tue, 30 Jun 2026 14:55:22 +0800, [email protected] wrote:
diff --git a/arch/arm/mach-socfpga/rsu_spl.c b/arch/arm/mach-socfpga/rsu_spl.c
@@ -0,0 +228,16 @@ finding 232-239
+ if ((strlen(UBOOT_PREFIX) + MAX_PART_NAME_LENGTH +
strlen(UBOOT_ITB_EXT))
+ > (size_t)max_size)
+ return -ENAMETOOLONG;
+
+ ret = get_spl_slot(&rsu_spt, sizeof(rsu_spt), &crt_spt_index);
+ if (ret) {
+ if (ret == -EOPNOTSUPP) {
+ puts("RSU: Error - mbox_rsu_status failed! Check for RSU
image.\n");
+ return -EOPNOTSUPP;
+ }
+
+ panic("ERROR: could not find u-boot proper (SSBL) for MMC
load");
+ }
+
+ if (IS_ENABLED(CONFIG_SPL_LOAD_FIT))
+ ret = snprintf(filename, (size_t)max_size, "%s%s%s",
UBOOT_PREFIX,
SPL panics if the RSU SPT lookup fails during MMC payload name selection
(patman-adjacent: boot hard-stop).
When CONFIG_SOCFPGA_RSU_MULTIBOOT is on, rsu_spl_mmc_filename() calls
panic() if get_spl_slot() fails for any reason other than -EOPNOTSUPP.
That hangs SPL instead of returning an error the generic loader can
recover from. common/spl/spl_mmc.c already treats any non-zero return as
fatal and never falls back to CONFIG_SPL_FS_LOAD_PAYLOAD_NAME, so a
missing SSBL.* SPT entry (or a transient mailbox miss that is not
EOPNOTSUPP) bricks MMC boot. Kconfig for SPL_SOCFPGA_RSU_SSBL_FALLBACK
claims a bounded fallback exists for env-offset math; MMC filename
selection has no equivalent.
Please: Return the errno (or -ENOENT) instead of panic(). In
spl_mmc_load_image(), if rsu_spl_mmc_filename() fails, fall back to
CONFIG_SPL_FS_LOAD_PAYLOAD_NAME rather than aborting the boot.
diff --git a/env/sf.c b/env/sf.c
@@ -98,18 +109,9 @@ finding 110-115
if (gd->env_valid == ENV_VALID) {
env_new_offset = CONFIG_ENV_OFFSET_REDUND;
- env_offset = CONFIG_ENV_OFFSET;
+ env_offset = env_sf_get_env_offset();
} else {
- env_new_offset = CONFIG_ENV_OFFSET;
+ env_new_offset = env_sf_get_env_offset();
SPI env redundant offset stays CONFIG_ENV_OFFSET_REDUND while the
primary offset becomes runtime.
env_sf_get_env_offset() is used for the primary copy, but
CONFIG_ENV_OFFSET_REDUND is still a compile-time address. With RSU
multiboot the SSBL slot moves; the redundant environment does not move
with it, so save/load can erase or read a slot that is no longer paired
with the running image.
Please: Either document that redundant SPI env is not used with
MULTIBOOT, or add env_sf_get_env_offset_redund() (primary +
CONFIG_ENV_SIZE, or SPT-derived) and use it everywhere
CONFIG_ENV_OFFSET_REDUND is read in this file.
env_offset = CONFIG_ENV_OFFSET_REDUND;
}
- read1_fail = spi_flash_read(env_flash, CONFIG_ENV_OFFSET,
- saved_offset = CONFIG_ENV_OFFSET + CONFIG_ENV_SIZE;
- ret = spi_flash_erase(env_flash, CONFIG_ENV_OFFSET,
- sector * sect_size);
- ret = spi_flash_write(env_flash, CONFIG_ENV_OFFSET,
- CONFIG_ENV_SIZE, &env_new);
- CONFIG_ENV_OFFSET, CONFIG_ENV_SIZE, buf);
- ret = spi_flash_write(env_flash, CONFIG_ENV_OFFSET, CONFIG_ENV_SIZE,
&env);
- read1_fail = spi_flash_read(env_flash, CONFIG_ENV_OFFSET,
Fix the SPL panic and MMC fallback before v2. Mailbox length check and
SPI redundant-env offset should ship in the same respin.
Style/checkpatch warnings are not merge-blocking.
Best regards,
Tien Fong
