efi: Various tidy-ups and drop the default
Subsystems (9)
cover efi: Various tidy-ups and drop the default
It has come to light that EFI_LOADER adds an extraordinary amount of code to U-Boot. For example, with nokia_rx51 the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.
The current situation is affecting U-Boot's image as a svelt bootloader.
EFI_LOADER is required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.
EBBR should be truly optional, enabled only by boards that use it. Most don't use it but it is enabled anyway. The default boot path should be one that makes use of the existing U-Boot support.
To try to retify this situation, this series adds a new Kconfig option for EBBR so that the naming is more explicit. Then EFI_LOADER is updated to depend on it.
For now, only sandbox enables EBBR. Other boards can be added as needed, presumably by distributions that require it. Another approach would be to add 'CONFIG_EBBR=y' to the .config before building, in the build system. That might be more friendly to U-Boot users.
This series also fixes a minor issue noticed during testing.
Simon Glass (7):
configs: Resync with savedefconfig
Makefile: Drop include/asm directory as well as symlink
disk: Tidy up #ifdefs in part_efi
Use LIB_UUID with ACPIGEN and FS_BTRFS
Allow efi_loader header to be included always
lib: Create a new Kconfig option for charset conversion
efi: Make EBBR optional
Makefile | 2 +- common/Kconfig.boot | 15 ++ configs/am335x_igep003x_defconfig | 1 - configs/am335x_pdu001_defconfig | 1 - configs/am64x_evm_a53_defconfig | 31 ++- configs/am64x_evm_r5_defconfig | 6 +- configs/apalis-imx8_defconfig | 1 - configs/apalis-imx8x_defconfig | 1 - configs/aristainetos2c_defconfig | 1 - configs/aristainetos2ccslb_defconfig | 1 - ...edev_cc_v1_0_ultrazedev_som_v1_0_defconfig | 1 - configs/bcm7260_defconfig | 1 - configs/bcm7445_defconfig | 1 - configs/bcm963158_ram_defconfig | 1 - configs/bcm968580xref_ram_defconfig | 1 - configs/bitmain_antminer_s9_defconfig | 1 - configs/bk4r1_defconfig | 1 - configs/brppt1_mmc_defconfig | 1 - configs/brppt1_nand_defconfig | 1 - configs/brppt1_spi_defconfig | 1 - configs/brppt2_defconfig | 1 - configs/brsmarc1_defconfig | 1 - configs/brxre1_defconfig | 1 - configs/chromebook_coral_defconfig | 1 - configs/chromebook_link_defconfig | 1 - configs/colibri-imx8x_defconfig | 1 - configs/colibri_vf_defconfig | 1 - configs/controlcenterdc_defconfig | 1 - configs/crs305-1g-4s-bit_defconfig | 1 - configs/crs305-1g-4s_defconfig | 1 - configs/crs326-24g-2s-bit_defconfig | 1 - configs/crs326-24g-2s_defconfig | 1 - configs/crs328-4c-20s-4s-bit_defconfig | 1 - configs/crs328-4c-20s-4s_defconfig | 1 - configs/deneb_defconfig | 1 - configs/draco_defconfig | 2 +- configs/dragonboard820c_defconfig | 1 - configs/efi-x86_app_defconfig | 1 - configs/etamin_defconfig | 2 +- configs/evb-ast2600_defconfig | 1 - configs/evb-px30_defconfig | 1 - configs/evb-rk3308_defconfig | 1 - configs/firefly-px30_defconfig | 1 - configs/ge_bx50v3_defconfig | 1 - configs/giedi_defconfig | 1 - configs/grpeach_defconfig | 1 - configs/imx8mm-cl-iot-gate_defconfig | 8 - configs/imx8qm_mek_defconfig | 1 - configs/imx8qm_rom7720_a1_4G_defconfig | 1 - configs/imx8qxp_mek_defconfig | 1 - configs/j7200_evm_r5_defconfig | 15 +- configs/j721e_evm_r5_defconfig | 13 +- configs/kontron_sl28_defconfig | 2 - configs/kp_imx53_defconfig | 1 - configs/ls1028aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1028aqds_tfa_defconfig | 1 - configs/ls1028aqds_tfa_lpuart_defconfig | 1 - configs/ls1028ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1028ardb_tfa_defconfig | 1 - configs/ls1043aqds_defconfig | 1 - configs/ls1043aqds_lpuart_defconfig | 1 - configs/ls1043aqds_nand_defconfig | 1 - configs/ls1043aqds_nor_ddr3_defconfig | 1 - configs/ls1043aqds_qspi_defconfig | 1 - configs/ls1043aqds_sdcard_ifc_defconfig | 1 - configs/ls1043aqds_sdcard_qspi_defconfig | 1 - configs/ls1043aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1043aqds_tfa_defconfig | 1 - configs/ls1043ardb_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_defconfig | 1 - configs/ls1043ardb_nand_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_nand_defconfig | 1 - .../ls1043ardb_sdcard_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_sdcard_defconfig | 1 - configs/ls1043ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_tfa_defconfig | 1 - configs/ls1046aqds_SECURE_BOOT_defconfig | 1 - configs/ls1046aqds_defconfig | 1 - configs/ls1046aqds_lpuart_defconfig | 1 - configs/ls1046aqds_nand_defconfig | 1 - configs/ls1046aqds_qspi_defconfig | 1 - configs/ls1046aqds_sdcard_ifc_defconfig | 1 - configs/ls1046aqds_sdcard_qspi_defconfig | 1 - configs/ls1046aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1046aqds_tfa_defconfig | 1 - configs/ls1046ardb_emmc_defconfig | 1 - configs/ls1046ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_qspi_defconfig | 1 - configs/ls1046ardb_qspi_spl_defconfig | 1 - .../ls1046ardb_sdcard_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_sdcard_defconfig | 1 - configs/ls1046ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_tfa_defconfig | 1 - configs/ls1088aqds_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1088aqds_qspi_defconfig | 1 - configs/ls1088aqds_tfa_defconfig | 1 - configs/ls1088ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1088ardb_qspi_defconfig | 1 - configs/ls1088ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1088ardb_tfa_defconfig | 1 - configs/ls2080aqds_SECURE_BOOT_defconfig | 1 - configs/ls2080aqds_defconfig | 1 - configs/ls2080aqds_nand_defconfig | 1 - configs/ls2080aqds_qspi_defconfig | 1 - configs/ls2080aqds_sdcard_defconfig | 1 - configs/ls2080ardb_SECURE_BOOT_defconfig | 1 - configs/ls2080ardb_defconfig | 1 - configs/ls2080ardb_nand_defconfig | 1 - configs/ls2081ardb_defconfig | 1 - configs/ls2088aqds_tfa_defconfig | 1 - configs/ls2088ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls2088ardb_qspi_defconfig | 1 - configs/ls2088ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls2088ardb_tfa_defconfig | 1 - configs/lx2160aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2160aqds_tfa_defconfig | 1 - configs/lx2160ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2160ardb_tfa_defconfig | 1 - configs/lx2160ardb_tfa_stmm_defconfig | 4 - configs/lx2162aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2162aqds_tfa_defconfig | 1 - .../lx2162aqds_tfa_verified_boot_defconfig | 1 - configs/mt7623n_bpir2_defconfig | 1 - configs/mt7629_rfb_defconfig | 1 - configs/mt8183_pumpkin_defconfig | 1 - configs/mt8516_pumpkin_defconfig | 1 - configs/mvebu_puzzle-m801-88f8040_defconfig | 1 - configs/mx6memcal_defconfig | 1 - configs/octeontx2_95xx_defconfig | 1 - configs/octeontx2_96xx_defconfig | 1 - configs/octeontx_81xx_defconfig | 1 - configs/octeontx_83xx_defconfig | 1 - configs/omap4_sdp4430_defconfig | 1 - configs/opos6uldev_defconfig | 1 - configs/pcm052_defconfig | 1 - configs/phycore-am335x-r2-regor_defconfig | 1 - configs/phycore-am335x-r2-wega_defconfig | 1 - configs/pxm2_defconfig | 2 +- configs/qemu-riscv32_defconfig | 2 - configs/qemu-riscv32_smode_defconfig | 2 - configs/qemu-riscv64_defconfig | 2 - configs/qemu-riscv64_smode_defconfig | 2 - configs/qemu-x86_64_defconfig | 2 - configs/qemu-x86_defconfig | 2 - configs/qemu_arm64_defconfig | 2 - configs/qemu_arm_defconfig | 2 - configs/rastaban_defconfig | 2 +- configs/roc-cc-rk3308_defconfig | 1 - configs/rock-pi-n10-rk3399pro_defconfig | 1 - configs/rock-pi-n8-rk3288_defconfig | 1 - configs/rpi_0_w_defconfig | 1 - configs/rpi_defconfig | 1 - configs/rut_defconfig | 2 +- configs/sama5d27_wlsom1_ek_mmc_defconfig | 1 - .../sama5d27_wlsom1_ek_qspiflash_defconfig | 1 - configs/sama5d2_icp_mmc_defconfig | 1 - configs/sama7g5ek_mmc1_defconfig | 1 - configs/sama7g5ek_mmc_defconfig | 1 - configs/sipeed_maix_bitm_defconfig | 1 - configs/sipeed_maix_smode_defconfig | 1 - configs/socfpga_de1_soc_defconfig | 1 - configs/somlabs_visionsom_6ull_defconfig | 1 - configs/stemmy_defconfig | 1 - configs/stm32mp15_basic_defconfig | 3 - configs/stm32mp15_trusted_defconfig | 3 - configs/tbs2910_defconfig | 1 - configs/thuban_defconfig | 2 +- configs/vf610twr_defconfig | 1 - configs/vf610twr_nand_defconfig | 1 - configs/xenguest_arm64_defconfig | 1 - configs/xilinx_versal_mini_defconfig | 1 - configs/xilinx_versal_mini_emmc0_defconfig | 1 - configs/xilinx_versal_mini_emmc1_defconfig | 1 - configs/xilinx_versal_virt_defconfig | 2 - configs/xilinx_zynqmp_mini_defconfig | 1 - configs/xilinx_zynqmp_mini_emmc0_defconfig | 1 - configs/xilinx_zynqmp_mini_emmc1_defconfig | 1 - configs/xilinx_zynqmp_mini_nand_defconfig | 1 - .../xilinx_zynqmp_mini_nand_single_defconfig | 1 - configs/xilinx_zynqmp_mini_qspi_defconfig | 1 - configs/xilinx_zynqmp_r5_defconfig | 1 - configs/xilinx_zynqmp_virt_defconfig | 9 - configs/zynq_cse_nand_defconfig | 1 - configs/zynq_cse_nor_defconfig | 1 - configs/zynq_cse_qspi_defconfig | 1 - disk/part_efi.c | 11 +- drivers/core/Kconfig | 1 + fs/btrfs/Kconfig | 1 + include/efi_loader.h | 186 +++++++++--------- lib/Kconfig | 8 + lib/Makefile | 2 +- lib/efi_loader/Kconfig | 5 +- 192 files changed, 160 insertions(+), 353 deletions(-)
Signature · 1 line
2.32.0.93.g670b81a890-goog
PATCH 1/7 configs: Resync with savedefconfig
Resync all defconfig files using moveconfig.py
---
configs/am64x_evm_a53_defconfig | 31 +++++++++++-------------------- configs/am64x_evm_r5_defconfig | 6 ++---- configs/draco_defconfig | 2 +- configs/etamin_defconfig | 2 +- configs/j7200_evm_r5_defconfig | 15 +++++---------- configs/j721e_evm_r5_defconfig | 13 +++++-------- configs/pxm2_defconfig | 2 +- configs/rastaban_defconfig | 2 +- configs/rut_defconfig | 2 +- configs/thuban_defconfig | 2 +- 10 files changed, 29 insertions(+), 48 deletions(-)
configs/am64x_evm_a53_defconfig +11 -20
@@ -30,8 +30,8 @@ CONFIG_SPL_SYS_MALLOC_SIMPLE=y CONFIG_SPL_STACK_R=y CONFIG_SPL_SEPARATE_BSS=y CONFIG_SPL_DMA=y -CONFIG_SPL_I2C_SUPPORT=y CONFIG_SPL_ENV_SUPPORT=y +CONFIG_SPL_I2C_SUPPORT=y CONFIG_SPL_DM_MAILBOX=y CONFIG_SPL_DM_SPI_FLASH=y CONFIG_SPL_POWER_DOMAIN=y
@@ -44,8 +44,11 @@ CONFIG_SPL_USB_GADGET=y CONFIG_SPL_DFU=y CONFIG_SPL_YMODEM_SUPPORT=y CONFIG_CMD_ASKENV=y +CONFIG_CMD_DFU=y +CONFIG_CMD_DM=y CONFIG_CMD_I2C=y CONFIG_CMD_MMC=y +CONFIG_CMD_USB=y CONFIG_CMD_TIME=y CONFIG_OF_CONTROL=y CONFIG_SPL_OF_CONTROL=y
@@ -66,6 +69,11 @@ CONFIG_SPL_OF_TRANSLATE=y CONFIG_CLK=y CONFIG_SPL_CLK=y CONFIG_CLK_TI_SCI=y +CONFIG_DFU_MMC=y +CONFIG_DFU_RAM=y +CONFIG_DFU_SF=y +CONFIG_SYS_DFU_DATA_BUF_SIZE=0x40000 +CONFIG_SYS_DFU_MAX_FILE_SIZE=0x800000 CONFIG_DMA_CHANNELS=y CONFIG_TI_K3_NAVSS_UDMA=y CONFIG_TI_SCI_PROTOCOL=y
@@ -107,37 +115,20 @@ CONFIG_SYSRESET_TI_SCI=y CONFIG_TIMER=y CONFIG_SPL_TIMER=y CONFIG_OMAP_TIMER=y -CONFIG_FS_FAT_MAX_CLUSTSIZE=16384 -CONFIG_CMD_DFU=y -CONFIG_CMD_DM=y -CONFIG_CMD_USB=y -CONFIG_DFU=y -CONFIG_DFU_OVER_USB=y -# CONFIG_DFU_TFTP is not set -CONFIG_DFU_MMC=y -CONFIG_DFU_RAM=y -CONFIG_DFU_SF=y -# CONFIG_DFU_VIRT is not set -CONFIG_SYS_DFU_DATA_BUF_SIZE=0x40000 -CONFIG_SYS_DFU_MAX_FILE_SIZE=0x800000 CONFIG_USB=y CONFIG_DM_USB=y CONFIG_DM_USB_GADGET=y CONFIG_SPL_DM_USB_GADGET=y -CONFIG_USB_HOST=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_CDNS3=y CONFIG_USB_CDNS3_GADGET=y -CONFIG_SPL_USB_CDNS3_GADGET=y CONFIG_USB_CDNS3_HOST=y +CONFIG_SPL_USB_CDNS3_GADGET=y CONFIG_SPL_USB_CDNS3_HOST=y -CONFIG_USB_CDNS3_TI=y -CONFIG_USB_STORAGE=y CONFIG_USB_GADGET=y CONFIG_USB_GADGET_MANUFACTURER="Texas Instruments" CONFIG_USB_GADGET_VENDOR_NUM=0x0451 CONFIG_USB_GADGET_PRODUCT_NUM=0x6165 -CONFIG_USB_GADGET_VBUS_DRAW=2 -CONFIG_USB_GADGET_DUALSPEED=y CONFIG_USB_GADGET_DOWNLOAD=y CONFIG_USB_FUNCTION_MASS_STORAGE=y +CONFIG_FS_FAT_MAX_CLUSTSIZE=16384
configs/am64x_evm_r5_defconfig +2 -4
@@ -28,11 +28,11 @@ CONFIG_SPL_LOAD_FIT_ADDRESS=0x80080000 CONFIG_SPL_SIZE_LIMIT_SUBTRACT_GD=y CONFIG_SPL_SIZE_LIMIT_SUBTRACT_MALLOC=y CONFIG_SPL_SYS_REPORT_STACK_F_USAGE=y +CONFIG_SPL_BOARD_INIT=y CONFIG_SPL_SYS_MALLOC_SIMPLE=y CONFIG_SPL_STACK_R=y CONFIG_SPL_SEPARATE_BSS=y CONFIG_SPL_EARLY_BSS=y -CONFIG_SPL_BOARD_INIT=y CONFIG_SPL_ENV_SUPPORT=y CONFIG_SPL_I2C_SUPPORT=y CONFIG_SPL_DM_MAILBOX=y
@@ -117,16 +117,14 @@ CONFIG_SPL_TIMER=y CONFIG_OMAP_TIMER=y CONFIG_USB=y CONFIG_DM_USB=y -CONFIG_SPL_DM_USB=y CONFIG_DM_USB_GADGET=y CONFIG_SPL_DM_USB_GADGET=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_CDNS3=y CONFIG_USB_CDNS3_GADGET=y -CONFIG_SPL_USB_CDNS3_GADGET=y CONFIG_USB_CDNS3_HOST=y +CONFIG_SPL_USB_CDNS3_GADGET=y CONFIG_SPL_USB_CDNS3_HOST=y -CONFIG_USB_CDNS3_TI=y CONFIG_USB_STORAGE=y CONFIG_USB_GADGET=y CONFIG_USB_GADGET_MANUFACTURER="Texas Instruments"
configs/draco_defconfig +1 -1
@@ -75,8 +75,8 @@ CONFIG_SPL_DM=y CONFIG_BOOTCOUNT_LIMIT=y CONFIG_BOOTCOUNT_ENV=y CONFIG_DFU_NAND=y -# CONFIG_SPL_DM_MMC is not set CONFIG_SYS_DFU_DATA_BUF_SIZE=0x100000 +# CONFIG_SPL_DM_MMC is not set CONFIG_MMC_OMAP_HS=y CONFIG_MTD=y CONFIG_MTD_RAW_NAND=y
configs/etamin_defconfig +1 -1
@@ -76,8 +76,8 @@ CONFIG_SPL_DM=y CONFIG_BOOTCOUNT_LIMIT=y CONFIG_BOOTCOUNT_ENV=y CONFIG_DFU_NAND=y -# CONFIG_SPL_DM_MMC is not set CONFIG_SYS_DFU_DATA_BUF_SIZE=0x100000 +# CONFIG_SPL_DM_MMC is not set CONFIG_MMC_OMAP_HS=y CONFIG_MTD=y CONFIG_MTD_RAW_NAND=y
configs/j7200_evm_r5_defconfig +5 -10
@@ -24,6 +24,7 @@ CONFIG_DEFAULT_DEVICE_TREE="k3-j7200-r5-common-proc-board" # CONFIG_SYS_MALLOC_CLEAR_ON_INIT is not set CONFIG_SPL_LOAD_FIT=y CONFIG_SPL_LOAD_FIT_ADDRESS=0x80080000 +CONFIG_SPL_FIT_IMAGE_POST_PROCESS=y # CONFIG_USE_SPL_FIT_GENERATOR is not set CONFIG_USE_BOOTCOMMAND=y # CONFIG_DISPLAY_CPUINFO is not set
@@ -75,7 +76,9 @@ CONFIG_SPL_SYSCON=y CONFIG_SPL_OF_TRANSLATE=y CONFIG_CLK=y CONFIG_SPL_CLK=y -# CONFIG_CLK_TI_SCI is not set +CONFIG_SPL_CLK_CCF=y +CONFIG_SPL_CLK_K3_PLL=y +CONFIG_SPL_CLK_K3=y CONFIG_DMA_CHANNELS=y CONFIG_TI_K3_NAVSS_UDMA=y CONFIG_TI_SCI_PROTOCOL=y
@@ -110,7 +113,7 @@ CONFIG_SPL_PINCTRL=y # CONFIG_SPL_PINCTRL_GENERIC is not set CONFIG_PINCTRL_SINGLE=y CONFIG_POWER_DOMAIN=y -# CONFIG_TI_SCI_POWER_DOMAIN is not set +CONFIG_TI_POWER_DOMAIN=y CONFIG_K3_SYSTEM_CONTROLLER=y CONFIG_REMOTEPROC_TI_K3_ARM64=y CONFIG_DM_RESET=y
@@ -142,13 +145,5 @@ CONFIG_USB_GADGET_PRODUCT_NUM=0x6164 CONFIG_USB_GADGET_DOWNLOAD=y CONFIG_FS_EXT4=y CONFIG_FS_FAT_MAX_CLUSTSIZE=16384 -CONFIG_SOC_DEVICE=y -CONFIG_SOC_DEVICE_TI_K3=y -CONFIG_TI_POWER_DOMAIN=y -CONFIG_SPL_CLK_CCF=y CONFIG_LIB_RATIONAL=y CONFIG_SPL_LIB_RATIONAL=y -CONFIG_SPL_CLK_K3_PLL=y -CONFIG_SPL_CLK_K3=y -CONFIG_K3_DM_FW=y -CONFIG_SPL_FIT_IMAGE_POST_PROCESS=y
configs/j721e_evm_r5_defconfig +5 -8
@@ -24,6 +24,7 @@ CONFIG_DEFAULT_DEVICE_TREE="k3-j721e-r5-common-proc-board" # CONFIG_SYS_MALLOC_CLEAR_ON_INIT is not set CONFIG_SPL_LOAD_FIT=y CONFIG_SPL_LOAD_FIT_ADDRESS=0x80080000 +CONFIG_SPL_FIT_IMAGE_POST_PROCESS=y # CONFIG_USE_SPL_FIT_GENERATOR is not set CONFIG_USE_BOOTCOMMAND=y # CONFIG_DISPLAY_CPUINFO is not set
@@ -72,7 +73,9 @@ CONFIG_SPL_REGMAP=y CONFIG_SPL_OF_TRANSLATE=y CONFIG_CLK=y CONFIG_SPL_CLK=y -# CONFIG_CLK_TI_SCI is not set +CONFIG_SPL_CLK_CCF=y +CONFIG_SPL_CLK_K3_PLL=y +CONFIG_SPL_CLK_K3=y CONFIG_DMA_CHANNELS=y CONFIG_TI_K3_NAVSS_UDMA=y CONFIG_TI_SCI_PROTOCOL=y
@@ -102,7 +105,7 @@ CONFIG_SPL_PINCTRL=y # CONFIG_SPL_PINCTRL_GENERIC is not set CONFIG_PINCTRL_SINGLE=y CONFIG_POWER_DOMAIN=y -# CONFIG_TI_SCI_POWER_DOMAIN is not set +CONFIG_TI_POWER_DOMAIN=y CONFIG_DM_PMIC=y CONFIG_PMIC_TPS65941=y CONFIG_DM_REGULATOR=y
@@ -140,11 +143,5 @@ CONFIG_USB_GADGET_PRODUCT_NUM=0x6163 CONFIG_USB_GADGET_DOWNLOAD=y CONFIG_FS_EXT4=y CONFIG_FS_FAT_MAX_CLUSTSIZE=16384 -CONFIG_TI_POWER_DOMAIN=y -CONFIG_SPL_CLK_CCF=y CONFIG_LIB_RATIONAL=y CONFIG_SPL_LIB_RATIONAL=y -CONFIG_SPL_CLK_K3_PLL=y -CONFIG_SPL_CLK_K3=y -CONFIG_K3_DM_FW=y -CONFIG_SPL_FIT_IMAGE_POST_PROCESS=y
configs/pxm2_defconfig +1 -1
@@ -74,8 +74,8 @@ CONFIG_SPL_DM=y CONFIG_BOOTCOUNT_LIMIT=y CONFIG_BOOTCOUNT_ENV=y CONFIG_DFU_NAND=y -# CONFIG_SPL_DM_MMC is not set CONFIG_SYS_DFU_DATA_BUF_SIZE=0x100000 +# CONFIG_SPL_DM_MMC is not set CONFIG_MMC_OMAP_HS=y CONFIG_MTD=y CONFIG_MTD_RAW_NAND=y
configs/rastaban_defconfig +1 -1
@@ -75,8 +75,8 @@ CONFIG_SPL_DM=y CONFIG_BOOTCOUNT_LIMIT=y CONFIG_BOOTCOUNT_ENV=y CONFIG_DFU_NAND=y -# CONFIG_SPL_DM_MMC is not set CONFIG_SYS_DFU_DATA_BUF_SIZE=0x100000 +# CONFIG_SPL_DM_MMC is not set CONFIG_MMC_OMAP_HS=y CONFIG_MTD=y CONFIG_MTD_RAW_NAND=y
configs/rut_defconfig +1 -1
@@ -74,8 +74,8 @@ CONFIG_SPL_DM=y CONFIG_BOOTCOUNT_LIMIT=y CONFIG_BOOTCOUNT_ENV=y CONFIG_DFU_NAND=y -# CONFIG_SPL_DM_MMC is not set CONFIG_SYS_DFU_DATA_BUF_SIZE=0x100000 +# CONFIG_SPL_DM_MMC is not set CONFIG_MMC_OMAP_HS=y CONFIG_MTD=y CONFIG_MTD_RAW_NAND=y
configs/thuban_defconfig +1 -1
@@ -75,8 +75,8 @@ CONFIG_SPL_DM=y CONFIG_BOOTCOUNT_LIMIT=y CONFIG_BOOTCOUNT_ENV=y CONFIG_DFU_NAND=y -# CONFIG_SPL_DM_MMC is not set CONFIG_SYS_DFU_DATA_BUF_SIZE=0x100000 +# CONFIG_SPL_DM_MMC is not set CONFIG_MMC_OMAP_HS=y CONFIG_MTD=y CONFIG_MTD_RAW_NAND=y
Signature · 1 line
2.32.0.93.g670b81a890-goog
PATCH 2/7 Makefile: Drop include/asm directory as well as symlink
In reply to Simon GlassAt present when using 'make mrproper' on an out-of-tree build, a warning is shown about include/asm being a directory. With old versions of U-Boot it is a file, but more recently it has become a directory.
Remove this directory first, since that covers both cases.
---
Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Makefile +1 -1
@@ -2078,7 +2078,7 @@ CLEAN_FILES += include/bmp_logo.h include/bmp_logo_data.h tools/version.h \ # Directories & files removed with 'make mrproper' MRPROPER_DIRS += include/config include/generated spl tpl \ - .tmp_objdiff doc/output + .tmp_objdiff doc/output include/asm # Remove include/asm symlink created by U-Boot before v2014.01 MRPROPER_FILES += .config .config.old include/autoconf.mk* include/config.h \
Signature · 1 line
2.32.0.93.g670b81a890-goog
PATCH 3/7 disk: Tidy up #ifdefs in part_efi
In reply to Simon GlassThis file does not correctly handle the various cases, sometimes producing warnings about partition_basic_data_guid being defined but not used. Fix it.
---
disk/part_efi.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
disk/part_efi.c +6 -5
@@ -29,12 +29,13 @@DECLARE_GLOBAL_DATA_PTR ; -/* -* GUID for basic data partions. -*/ +#ifdef CONFIG_HAVE_BLOCK_DEVICE + +/* GUID for basic data partitons */ +#if CONFIG_IS_ENABLED (EFI_PARTITION )static const efi_guid_t partition_basic_data_guid =PARTITION_BASIC_DATA_GUID ; +#endif -#ifdef CONFIG_HAVE_BLOCK_DEVICE /** * efi_crc32() - EFI version of crc32 function * @buf: buffer to calculate crc32 of
@@ -1126,4 +1127,4 @@ U_BOOT_PART_TYPE(a_efi) = { .print = part_print_ptr(part_print_efi), .test = part_test_efi, }; -#endif +#endif /* CONFIG_HAVE_BLOCK_DEVICE */
Signature · 1 line
2.32.0.93.g670b81a890-goog
Re: [PATCH 3/7] disk: Tidy up #ifdefs in part_efi
Quoted 15 lines · Simon Glass
This file does not correctly handle the various cases, sometimes producing warnings about
partition_basic_data_guidbeing defined but not used. Fix it.Signed-off-by Simon Glass <[email protected]>---
disk/part_efi.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)disk/part_efi.c +0
-0
disk/part_efi.c line 32
Quoted 7 lines
disk/part_efi.c +1
-3@@ -29,12 +29,13 @@DECLARE_GLOBAL_DATA_PTR ; -/* -* GUID for basic data partions. -*/ +#ifdef CONFIG_HAVE_BLOCK_DEVICE
This #ifdef should be removed. Make CONFIG_HAVE_BLOCK_DEVICE a prerequisite for CONFIG_PARTITIONS instead.
disk/part_efi.c line 35
Quoted 3 lines
disk/part_efi.c +3
-0+ +/* GUID for basic data partitons */ +#if CONFIG_IS_ENABLED (EFI_PARTITION )
part_efi.c is not compiled without CONFIG_$(SPL_)EFI_PARTITION=y. Why put an #if on EFI_PARTITION here?
Best regards
Heinrich
Quoted 14 lines
disk/part_efi.c +2
-2static const efi_guid_t partition_basic_data_guid =PARTITION_BASIC_DATA_GUID ; +#endif -#ifdef CONFIG_HAVE_BLOCK_DEVICE /** * efi_crc32() - EFI version of crc32 function * @buf: buffer to calculate crc32 of @@ -1126,4 +1127,4 @@ U_BOOT_PART_TYPE(a_efi) = { .print = part_print_ptr(part_print_efi), .test = part_test_efi, }; -#endif +#endif /* CONFIG_HAVE_BLOCK_DEVICE */
Re: [PATCH 3/7] disk: Tidy up #ifdefs in part_efi
Quoted 19 lines · Heinrich Schuchardt
Quoted 18 lines · Simon Glass
This file does not correctly handle the various cases, sometimes producing warnings about
partition_basic_data_guidbeing defined but not used. Fix it.Signed-off-by Simon Glass <[email protected]>---
disk/part_efi.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)disk/part_efi.c +0
-0@@ -29,12 +29,13 @@DECLARE_GLOBAL_DATA_PTR ;
Quoted 7 lines
disk/part_efi.c line 32
Quoted 4 lines
disk/part_efi.c +1
-3-/* -* GUID for basic data partions. -*/ +#ifdef CONFIG_HAVE_BLOCK_DEVICE This #ifdef should be removed. Make
CONFIG_HAVE_BLOCK_DEVICEa prerequisite forCONFIG_PARTITIONSinstead.
Ah, this is where things get funny. No, you can't do that as you can use partitions without block devices. I think it was some xilinx setup that has this?
Signature · 1 line
Tom
Re: [PATCH 3/7] disk: Tidy up #ifdefs in part_efi
Quoted 28 lines · Tom Rini
Quoted 19 lines · Heinrich Schuchardt
Quoted 18 lines · Simon Glass
This file does not correctly handle the various cases, sometimes producing warnings about
partition_basic_data_guidbeing defined but not used. Fix it.Signed-off-by Simon Glass <[email protected]>---
disk/part_efi.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)disk/part_efi.c +0
-0@@ -29,12 +29,13 @@DECLARE_GLOBAL_DATA_PTR ;Quoted 7 lines
disk/part_efi.c line 32
Quoted 4 lines
disk/part_efi.c +1
-3-/* -* GUID for basic data partions. -*/ +#ifdef CONFIG_HAVE_BLOCK_DEVICE This #ifdef should be removed. Make
CONFIG_HAVE_BLOCK_DEVICEa prerequisite forCONFIG_PARTITIONSinstead.
Quoted 4 lines
Ah, this is where things get funny. No, you can't do that as you can use partitions without block devices. I think it was some xilinx setup that has this?
How can you have a partition without a block device? There must be some backing storage for the partition.
Anyway this #ifdef should be in Kconfig or in Makefile and not here.
Best regards
Heinrich
PATCH 4/7 Use LIB_UUID with ACPIGEN and FS_BTRFS
In reply to Simon GlassSince the ACPI-generation code makes use of UUIDs we typically need to enabled UUID support for it to build. Add a new Kconfig condition.
Use it for BTRFS also.
---
drivers/core/Kconfig | 1 + fs/btrfs/Kconfig | 1 + 2 files changed, 2 insertions(+)
drivers/core/Kconfig +1 -0
@@ -325,6 +325,7 @@ config DM_DEV_READ_INLINE config ACPIGEN bool "Support ACPI table generation in driver model" default y if SANDBOX || (GENERATE_ACPI_TABLE && !QEMU) + select LIB_UUID help This option enables generation of ACPI tables using driver-model devices. It adds a new operation struct to each driver, to support
fs/btrfs/Kconfig +1 -0
@@ -1,6 +1,7 @@ config FS_BTRFS bool "Enable BTRFS filesystem support" select CRC32C + select LIB_UUID select LZO select ZSTD select RBTREE
Signature · 1 line
2.32.0.93.g670b81a890-goog
PATCH 5/7 Allow efi_loader header to be included always
In reply to Simon GlassIt is bad practice to put function declarations behind an #ifdef since it makes it impossible to use IS_ENABLED() in the C code.
This header provides two different versions of various functions. Collect them together in one place for clarity. Allow all the rest of the header to be included, regardless of the setting of EFI_LOADER.
---
include/efi_loader.h | 186 ++++++++++++++++++++++--------------------- 1 file changed, 95 insertions(+), 91 deletions(-)
include/efi_loader.h +95 -91
@@ -28,12 +28,104 @@ static inline void *guidcpy(void *dst, const void *src)return memcpy (dst, src,sizeof (efi_guid_t)); } -/* No need for efi loader support in SPL */ -#if CONFIG_IS_ENABLED (EFI_LOADER ) -#include <linux/list.h> #include <linux/oid_registry.h> +#if CONFIG_IS_ENABLED (EFI_LOADER ) + +/** +* __efi_runtime_data - declares a non-const variable for EFI runtime section +* +* This macro indicates that a variable is non-const and should go into the +* EFI runtime section, and thus still be available when the OS is running. +* +* Only use on variables not declared const. +* +* Example: +* +* :: +* +* static __efi_runtime_data my_computed_table[256]; +*/ +#define __efi_runtime_data __section(".data.efi_runtime") + +/** +* __efi_runtime_rodata - declares a read-only variable for EFI runtime section +* +* This macro indicates that a variable is read-only (const) and should go into +* the EFI runtime section, and thus still be available when the OS is running. +* +* Only use on variables also declared const. +* +* Example: +* +* :: +* +* static const __efi_runtime_rodata my_const_table[] = { 1, 2, 3 }; +*/ +#define __efi_runtime_rodata __section(".rodata.efi_runtime") + +/** +* __efi_runtime - declares a function for EFI runtime section +* +* This macro indicates that a function should go into the EFI runtime section, +* and thus still be available when the OS is running. +* +* Example: +* +* :: +* +* static __efi_runtime compute_my_table(void); +*/ +#define __efi_runtime __section(".text.efi_runtime") + +/* +* Call this with mmio_ptr as the _pointer_ to a pointer to an MMIO region +* to make it available at runtime +*/ +efi_status_t efi_add_runtime_mmio (void *mmio_ptr,u64 len); + +/* +* Special case handler for error/abort that just tries to dtrt to get +* back to u-boot world +*/ +void efi_restore_gd (void ); +/* Call this to set the current device name */ +void efi_set_bootdev (const char *dev,const char *devnr,const char *path, +void *buffer,size_t buffer_size); +/* Called by networking code to memorize the dhcp ack package */ +void efi_net_set_dhcp_ack (void *pkt,int len); +/* Print information about all loaded images */ +void efi_print_image_infos (void *pc); + +/* Hook at initialization */ +efi_status_t efi_launch_capsules (void ); + +#else /* CONFIG_IS_ENABLED(EFI_LOADER) */ + +/* Without CONFIG_EFI_LOADER we don't have a runtime section, stub it out */ +#define __efi_runtime_data +#define __efi_runtime_rodata +#define __efi_runtime +static inline efi_status_t efi_add_runtime_mmio (void *mmio_ptr,u64 len) +{ +return EFI_SUCCESS ; +} + +/* No loader configured, stub out EFI_ENTRY */ +static inline void efi_restore_gd (void ) { } +static inline void efi_set_bootdev (const char *dev,const char *devnr, +const char *path,void *buffer, +size_t buffer_size) { } +static inline void efi_net_set_dhcp_ack (void *pkt,int len) { } +static inline void efi_print_image_infos (void *pc) { } +static inline efi_status_t efi_launch_capsules (void ) +{ +return EFI_SUCCESS ; +} + +#endif /* CONFIG_IS_ENABLED(EFI_LOADER) */ +/* Maximum number of configuration tables */ #define EFI_MAX_CONFIGURATION_TABLES 16
@@ -465,8 +557,6 @@ efi_status_t efi_smbios_register(void);struct efi_simple_file_system_protocol *efi_fs_from_path (struct efi_device_path *fp); -/* Called by networking code to memorize the dhcp ack package */ -void efi_net_set_dhcp_ack (void *pkt,int len);/* Called by efi_set_watchdog_timer to reset the timer */ efi_status_t efi_set_watchdog (unsigned long timeout);
@@ -480,14 +570,8 @@ efi_status_t efi_load_pe(struct efi_loaded_image_obj *handle, struct efi_loaded_image *loaded_image_info);/* Called once to store the pristine gd pointer */ void efi_save_gd (void ); -/* Special case handler for error/abort that just tries to dtrt to get -* back to u-boot world */ -void efi_restore_gd (void );/* Call this to relocate the runtime section to an address space */ void efi_runtime_relocate (ulong offset,struct efi_mem_desc *map); -/* Call this to set the current device name */ -void efi_set_bootdev (const char *dev,const char *devnr,const char *path, -void *buffer,size_t buffer_size);/* Add a new object to the object list. */ void efi_add_handle (efi_handle_t obj);/* Create handle */
@@ -619,8 +703,6 @@ efi_status_t efi_setup_loaded_image(struct efi_device_path *device_path,struct efi_device_path *file_path, struct efi_loaded_image_obj **handle_ptr, struct efi_loaded_image **info_ptr); -/* Print information about all loaded images */ -void efi_print_image_infos (void *pc);#ifdef CONFIG_EFI_LOADER_BOUNCE_BUFFER extern void *efi_bounce_buffer;
@@ -682,62 +764,12 @@ ssize_t efi_dp_check_length(const struct efi_device_path *dp, (((_dp)->type == DEVICE_PATH_TYPE_##_type) && \ ((_dp)->sub_type == DEVICE_PATH_SUB_TYPE_##_subtype)) -/** -* __efi_runtime_data - declares a non-const variable for EFI runtime section -* -* This macro indicates that a variable is non-const and should go into the -* EFI runtime section, and thus still be available when the OS is running. -* -* Only use on variables not declared const. -* -* Example: -* -* :: -* -* static __efi_runtime_data my_computed_table[256]; - */ -#define __efi_runtime_data __section(".data.efi_runtime") - -/** -* __efi_runtime_rodata - declares a read-only variable for EFI runtime section -* -* This macro indicates that a variable is read-only (const) and should go into -* the EFI runtime section, and thus still be available when the OS is running. -* -* Only use on variables also declared const. -* -* Example: -* -* :: -* -* static const __efi_runtime_rodata my_const_table[] = { 1, 2, 3 }; -*/ -#define __efi_runtime_rodata __section(".rodata.efi_runtime") - -/** -* __efi_runtime - declares a function for EFI runtime section -* -* This macro indicates that a function should go into the EFI runtime section, -* and thus still be available when the OS is running. -* -* Example: -* -* :: -* -* static __efi_runtime compute_my_table(void); -*/ -#define __efi_runtime __section(".text.efi_runtime") - /* Indicate supported runtime services */efi_status_t efi_init_runtime_supported (void );/* Update CRC32 in table header */ void __efi_runtime efi_update_table_header_crc32(struct efi_table_hdr *table); -/* Call this with mmio_ptr as the _pointer_ to a pointer to an MMIO region -* to make it available at runtime */ -efi_status_t efi_add_runtime_mmio (void *mmio_ptr,u64 len); -/* Boards may provide the functions below to implement RTS functionality */ void __efi_runtime EFIAPI efi_reset_system(
@@ -926,34 +958,6 @@ efi_status_t efi_capsule_authenticate(const void *capsule,#define EFI_CAPSULE_DIR L"\\EFI\\UpdateCapsule\\" -/* Hook at initialization */ -efi_status_t efi_launch_capsules (void ); - -#else /* CONFIG_IS_ENABLED(EFI_LOADER) */ - -/* Without CONFIG_EFI_LOADER we don't have a runtime section, stub it out */ -#define __efi_runtime_data -#define __efi_runtime_rodata -#define __efi_runtime -static inline efi_status_t efi_add_runtime_mmio (void *mmio_ptr,u64 len) -{ -return EFI_SUCCESS ; -} - -/* No loader configured, stub out EFI_ENTRY */ -static inline void efi_restore_gd (void ) { } -static inline void efi_set_bootdev (const char *dev,const char *devnr, -const char *path,void *buffer, -size_t buffer_size) { } -static inline void efi_net_set_dhcp_ack (void *pkt,int len) { } -static inline void efi_print_image_infos (void *pc) { } -static inline efi_status_t efi_launch_capsules (void ) -{ -return EFI_SUCCESS ; -} - -#endif /* CONFIG_IS_ENABLED(EFI_LOADER) */ -/** * Install the ESRT system table. *
Signature · 1 line
2.32.0.93.g670b81a890-goog
Re: [PATCH 5/7] Allow efi_loader header to be included always
Quoted 18 lines · Simon Glass
It is bad practice to put function declarations behind an #ifdef since it makes it impossible to use
IS_ENABLED()in the C code.This header provides two different versions of various functions. Collect them together in one place for clarity. Allow all the rest of the header to be included, regardless of the setting of
EFI_LOADER.Signed-off-by Simon Glass <[email protected]>---
include/efi_loader.h | 186 ++++++++++++++++++++++--------------------- 1 file changed, 95 insertions(+), 91 deletions(-)include/efi_loader.h +0
-0
include/efi_loader.h static inline void *guidcpy(void *dst, const void *src) · line 32
Quoted 8 lines
include/efi_loader.h +0
-3@@ -28,12 +28,104 @@ static inline void *guidcpy(void *dst, const void *src) return memcpy(dst src, sizeof(efi_guid_t)); } -/* No need for efi loader support in SPL */ -#if CONFIG_IS_ENABLED (EFI_LOADER ) -#include <linux/list.h> #include <linux/oid_registry.h>
Why are these includes not listed with the other includes? Please, move them up.
Why should we have both "#include <blk.h>" and "struct blk_desc;"? I assume "struct blk_desc;" can be eliminated. Cf. e6f6f9e64882 ("common:
Drop part.h from common header")
Hiding "efi_status_t efi_add_runtime_mmio()" through "efi_status_t efi_launch_capsules()" behind "#if CONFIG_IS_ENABLED(EFI_LOADER)" contradicts the commit message.
+CC Alex, reviewer EFI PAYLOAD
Best regards
Heinrich
Quoted 231 lines
include/efi_loader.h +95
-88+#if CONFIG_IS_ENABLED (EFI_LOADER ) + +/** +* __efi_runtime_data - declares a non-const variable for EFI runtime section +* +* This macro indicates that a variable is non-const and should go into the +* EFI runtime section, and thus still be available when the OS is running. +* +* Only use on variables not declared const. +* +* Example: +* +* :: +* +* static __efi_runtime_data my_computed_table[256]; +*/ +#define __efi_runtime_data __section(".data.efi_runtime") + +/** +* __efi_runtime_rodata - declares a read-only variable for EFI runtime section +* +* This macro indicates that a variable is read-only (const) and should go into +* the EFI runtime section, and thus still be available when the OS is running. +* +* Only use on variables also declared const. +* +* Example: +* +* :: +* +* static const __efi_runtime_rodata my_const_table[] = { 1, 2, 3 }; +*/ +#define __efi_runtime_rodata __section(".rodata.efi_runtime") + +/** +* __efi_runtime - declares a function for EFI runtime section +* +* This macro indicates that a function should go into the EFI runtime section, +* and thus still be available when the OS is running. +* +* Example: +* +* :: +* +* static __efi_runtime compute_my_table(void); +*/ +#define __efi_runtime __section(".text.efi_runtime") + +/* +* Call this with mmio_ptr as the _pointer_ to a pointer to an MMIO region +* to make it available at runtime +*/ +efi_status_t efi_add_runtime_mmio (void *mmio_ptr,u64 len); + +/* +* Special case handler for error/abort that just tries to dtrt to get +* back to u-boot world +*/ +void efi_restore_gd (void ); +/* Call this to set the current device name */ +void efi_set_bootdev (const char *dev,const char *devnr,const char *path, +void *buffer,size_t buffer_size); +/* Called by networking code to memorize the dhcp ack package */ +void efi_net_set_dhcp_ack (void *pkt,int len); +/* Print information about all loaded images */ +void efi_print_image_infos (void *pc); + +/* Hook at initialization */ +efi_status_t efi_launch_capsules (void ); + +#else /* CONFIG_IS_ENABLED(EFI_LOADER) */ + +/* Without CONFIG_EFI_LOADER we don't have a runtime section, stub it out */ +#define __efi_runtime_data +#define __efi_runtime_rodata +#define __efi_runtime +static inline efi_status_t efi_add_runtime_mmio (void *mmio_ptr,u64 len) +{ +return EFI_SUCCESS ; +} + +/* No loader configured, stub out EFI_ENTRY */ +static inline void efi_restore_gd (void ) { } +static inline void efi_set_bootdev (const char *dev,const char *devnr, +const char *path,void *buffer, +size_t buffer_size) { } +static inline void efi_net_set_dhcp_ack (void *pkt,int len) { } +static inline void efi_print_image_infos (void *pc) { } +static inline efi_status_t efi_launch_capsules (void ) +{ +return EFI_SUCCESS ; +} + +#endif /* CONFIG_IS_ENABLED(EFI_LOADER) */ +/* Maximum number of configuration tables */ #define EFI_MAX_CONFIGURATION_TABLES 16@@ -465,8 +557,6 @@ efi_status_t efi_smbios_register(void);struct efi_simple_file_system_protocol *efi_fs_from_path (struct efi_device_path *fp); -/* Called by networking code to memorize the dhcp ack package */ -void efi_net_set_dhcp_ack (void *pkt,int len);/* Called by efi_set_watchdog_timer to reset the timer */ efi_status_t efi_set_watchdog (unsigned long timeout);@@ -480,14 +570,8 @@ efi_status_t efi_load_pe(struct efi_loaded_image_obj *handle, struct efi_loaded_image *loaded_image_info);/* Called once to store the pristine gd pointer */ void efi_save_gd (void ); -/* Special case handler for error/abort that just tries to dtrt to get -* back to u-boot world */ -void efi_restore_gd (void );/* Call this to relocate the runtime section to an address space */ void efi_runtime_relocate (ulong offset,struct efi_mem_desc *map); -/* Call this to set the current device name */ -void efi_set_bootdev (const char *dev,const char *devnr,const char *path, -void *buffer,size_t buffer_size);/* Add a new object to the object list. */ void efi_add_handle (efi_handle_t obj);/* Create handle */ @@ -619,8 +703,6 @@ efi_status_t efi_setup_loaded_image(struct efi_device_path *device_path,struct efi_device_path *file_path, struct efi_loaded_image_obj **handle_ptr, struct efi_loaded_image **info_ptr); -/* Print information about all loaded images */ -void efi_print_image_infos (void *pc);#ifdef CONFIG_EFI_LOADER_BOUNCE_BUFFER extern void *efi_bounce_buffer;@@ -682,62 +764,12 @@ ssize_t efi_dp_check_length(const struct efi_device_path *dp, (((_dp)->type == DEVICE_PATH_TYPE_##_type) && \ ((_dp)->sub_type == DEVICE_PATH_SUB_TYPE_##_subtype)) -/** -* __efi_runtime_data - declares a non-const variable for EFI runtime section -* -* This macro indicates that a variable is non-const and should go into the -* EFI runtime section, and thus still be available when the OS is running. -* -* Only use on variables not declared const. -* -* Example: -* -* :: -* -* static __efi_runtime_data my_computed_table[256]; - */ -#define __efi_runtime_data __section(".data.efi_runtime") - -/** -* __efi_runtime_rodata - declares a read-only variable for EFI runtime section -* -* This macro indicates that a variable is read-only (const) and should go into -* the EFI runtime section, and thus still be available when the OS is running. -* -* Only use on variables also declared const. -* -* Example: -* -* :: -* -* static const __efi_runtime_rodata my_const_table[] = { 1, 2, 3 }; -*/ -#define __efi_runtime_rodata __section(".rodata.efi_runtime") - -/** -* __efi_runtime - declares a function for EFI runtime section -* -* This macro indicates that a function should go into the EFI runtime section, -* and thus still be available when the OS is running. -* -* Example: -* -* :: -* -* static __efi_runtime compute_my_table(void); -*/ -#define __efi_runtime __section(".text.efi_runtime") - /* Indicate supported runtime services */efi_status_t efi_init_runtime_supported (void );/* Update CRC32 in table header */ void __efi_runtime efi_update_table_header_crc32(struct efi_table_hdr *table); -/* Call this with mmio_ptr as the _pointer_ to a pointer to an MMIO region -* to make it available at runtime */ -efi_status_t efi_add_runtime_mmio (void *mmio_ptr,u64 len); -/* Boards may provide the functions below to implement RTS functionality */ void __efi_runtime EFIAPI efi_reset_system(@@ -926,34 +958,6 @@ efi_status_t efi_capsule_authenticate(const void *capsule,#define EFI_CAPSULE_DIR L"\\EFI\\UpdateCapsule\\" -/* Hook at initialization */ -efi_status_t efi_launch_capsules (void ); - -#else /* CONFIG_IS_ENABLED(EFI_LOADER) */ - -/* Without CONFIG_EFI_LOADER we don't have a runtime section, stub it out */ -#define __efi_runtime_data -#define __efi_runtime_rodata -#define __efi_runtime -static inline efi_status_t efi_add_runtime_mmio (void *mmio_ptr,u64 len) -{ -return EFI_SUCCESS ; -} - -/* No loader configured, stub out EFI_ENTRY */ -static inline void efi_restore_gd (void ) { } -static inline void efi_set_bootdev (const char *dev,const char *devnr, -const char *path,void *buffer, -size_t buffer_size) { } -static inline void efi_net_set_dhcp_ack (void *pkt,int len) { } -static inline void efi_print_image_infos (void *pc) { } -static inline efi_status_t efi_launch_capsules (void ) -{ -return EFI_SUCCESS ; -} - -#endif /* CONFIG_IS_ENABLED(EFI_LOADER) */ -/** * Install the ESRT system table. *
PATCH 6/7 lib: Create a new Kconfig option for charset conversion
In reply to Simon GlassRather than looking at two KConfig options in the Makefile, create a new Kconfig option for compiling lib/charset.c
Enable it for UFS also, which needs this support.
---
lib/Kconfig | 8 ++++++++ lib/Makefile | 2 +- 2 files changed, 9 insertions(+), 1 deletion(-)
lib/Kconfig +8 -0
@@ -40,6 +40,14 @@ config CC_OPTIMIZE_LIBS_FOR_SPEED If unsure, say N. +config CHARSET + bool + default y if UT_UNICODE || EFI_LOADER || UFS + help + Enables support for various conversions between different + character sets, such as between unicode representations and + different 'code pages'. + config DYNAMIC_CRC_TABLE bool "Enable Dynamic tables for CRC" help
lib/Makefile +1 -1
@@ -25,7 +25,7 @@ obj-$(CONFIG_AES) += aes/ obj-$(CONFIG_$(SPL_TPL_)BINMAN_FDT) += binman.o ifndef API_BUILD -ifneq ($(CONFIG_UT_UNICODE)$(CONFIG_EFI_LOADER),) +ifneq ($(CONFIG_CHARSET),) obj-y += charset.o endif endif
Signature · 1 line
2.32.0.93.g670b81a890-goog
Re: [PATCH 6/7] lib: Create a new Kconfig option for charset conversion
Quoted 4 lines · Simon Glass
Rather than looking at two KConfig options in the
Makefile, create a newKconfigoption for compilinglib/charset.cEnable it for UFS also, which needs this support.
+CC Faiz, maintainer UFS
Function utf16_to_utf8() is used in ufshcd_read_string_desc(). It assumes that UTF-16 is using CPU endianness. What does UFS require on big-endian systems?
Quoted 2 lines
Signed-off-by Simon Glass <[email protected]>
Quoted 39 lines
---
lib/Kconfig | 8 ++++++++ lib/Makefile | 2 +- 2 files changed, 9 insertions(+), 1 deletion(-)lib/Kconfig +8
-0@@ -40,6 +40,14 @@ config CC_OPTIMIZE_LIBS_FOR_SPEED If unsure, say N. +config CHARSET + bool + default y if UT_UNICODE || EFI_LOADER || UFS + help + Enables support for various conversions between different + character sets, such as between unicode representations and + different 'code pages'. + config DYNAMIC_CRC_TABLE bool "Enable Dynamic tables for CRC" helplib/Makefile +1
-1@@ -25,7 +25,7 @@ obj-$(CONFIG_AES) += aes/ obj-$(CONFIG_$(SPL_TPL_)BINMAN_FDT) += binman.o ifndef API_BUILD -ifneq ($(CONFIG_UT_UNICODE)$(CONFIG_EFI_LOADER),) +ifneq ($(CONFIG_CHARSET),) obj-y += charset.o endif endif
PATCH 7/7 efi: Make EBBR optional
In reply to Simon GlassAdd a new Kconfig option for EBBR so that the naming is more explicit. Make EFI_LOADER depend on it.
Avoid enabling the options (EBBR / EFI_LOADER) by default, to save space.
Also add dependencies on driver model and OF_CONTROL, since boards which have not migrated to these should not be using new features like EBBR. The unwillingness to use driver model has resulted in the duplication of driver model code in EFI, which is part of the reason for this bloat.
Resync the CONFIG options also.
---
common/Kconfig.boot | 15 +++++++++++++++ configs/am335x_igep003x_defconfig | 1 - configs/am335x_pdu001_defconfig | 1 - configs/apalis-imx8_defconfig | 1 - configs/apalis-imx8x_defconfig | 1 - configs/aristainetos2c_defconfig | 1 - configs/aristainetos2ccslb_defconfig | 1 - ...trazedev_cc_v1_0_ultrazedev_som_v1_0_defconfig | 1 - configs/bcm7260_defconfig | 1 - configs/bcm7445_defconfig | 1 - configs/bcm963158_ram_defconfig | 1 - configs/bcm968580xref_ram_defconfig | 1 - configs/bitmain_antminer_s9_defconfig | 1 - configs/bk4r1_defconfig | 1 - configs/brppt1_mmc_defconfig | 1 - configs/brppt1_nand_defconfig | 1 - configs/brppt1_spi_defconfig | 1 - configs/brppt2_defconfig | 1 - configs/brsmarc1_defconfig | 1 - configs/brxre1_defconfig | 1 - configs/chromebook_coral_defconfig | 1 - configs/chromebook_link_defconfig | 1 - configs/colibri-imx8x_defconfig | 1 - configs/colibri_vf_defconfig | 1 - configs/controlcenterdc_defconfig | 1 - configs/crs305-1g-4s-bit_defconfig | 1 - configs/crs305-1g-4s_defconfig | 1 - configs/crs326-24g-2s-bit_defconfig | 1 - configs/crs326-24g-2s_defconfig | 1 - configs/crs328-4c-20s-4s-bit_defconfig | 1 - configs/crs328-4c-20s-4s_defconfig | 1 - configs/deneb_defconfig | 1 - configs/dragonboard820c_defconfig | 1 - configs/efi-x86_app_defconfig | 1 - configs/evb-ast2600_defconfig | 1 - configs/evb-px30_defconfig | 1 - configs/evb-rk3308_defconfig | 1 - configs/firefly-px30_defconfig | 1 - configs/ge_bx50v3_defconfig | 1 - configs/giedi_defconfig | 1 - configs/grpeach_defconfig | 1 - configs/imx8mm-cl-iot-gate_defconfig | 8 -------- configs/imx8qm_mek_defconfig | 1 - configs/imx8qm_rom7720_a1_4G_defconfig | 1 - configs/imx8qxp_mek_defconfig | 1 - configs/kontron_sl28_defconfig | 2 -- configs/kp_imx53_defconfig | 1 - configs/ls1028aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1028aqds_tfa_defconfig | 1 - configs/ls1028aqds_tfa_lpuart_defconfig | 1 - configs/ls1028ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1028ardb_tfa_defconfig | 1 - configs/ls1043aqds_defconfig | 1 - configs/ls1043aqds_lpuart_defconfig | 1 - configs/ls1043aqds_nand_defconfig | 1 - configs/ls1043aqds_nor_ddr3_defconfig | 1 - configs/ls1043aqds_qspi_defconfig | 1 - configs/ls1043aqds_sdcard_ifc_defconfig | 1 - configs/ls1043aqds_sdcard_qspi_defconfig | 1 - configs/ls1043aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1043aqds_tfa_defconfig | 1 - configs/ls1043ardb_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_defconfig | 1 - configs/ls1043ardb_nand_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_nand_defconfig | 1 - configs/ls1043ardb_sdcard_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_sdcard_defconfig | 1 - configs/ls1043ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_tfa_defconfig | 1 - configs/ls1046aqds_SECURE_BOOT_defconfig | 1 - configs/ls1046aqds_defconfig | 1 - configs/ls1046aqds_lpuart_defconfig | 1 - configs/ls1046aqds_nand_defconfig | 1 - configs/ls1046aqds_qspi_defconfig | 1 - configs/ls1046aqds_sdcard_ifc_defconfig | 1 - configs/ls1046aqds_sdcard_qspi_defconfig | 1 - configs/ls1046aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1046aqds_tfa_defconfig | 1 - configs/ls1046ardb_emmc_defconfig | 1 - configs/ls1046ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_qspi_defconfig | 1 - configs/ls1046ardb_qspi_spl_defconfig | 1 - configs/ls1046ardb_sdcard_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_sdcard_defconfig | 1 - configs/ls1046ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_tfa_defconfig | 1 - configs/ls1088aqds_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1088aqds_qspi_defconfig | 1 - configs/ls1088aqds_tfa_defconfig | 1 - configs/ls1088ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1088ardb_qspi_defconfig | 1 - configs/ls1088ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1088ardb_tfa_defconfig | 1 - configs/ls2080aqds_SECURE_BOOT_defconfig | 1 - configs/ls2080aqds_defconfig | 1 - configs/ls2080aqds_nand_defconfig | 1 - configs/ls2080aqds_qspi_defconfig | 1 - configs/ls2080aqds_sdcard_defconfig | 1 - configs/ls2080ardb_SECURE_BOOT_defconfig | 1 - configs/ls2080ardb_defconfig | 1 - configs/ls2080ardb_nand_defconfig | 1 - configs/ls2081ardb_defconfig | 1 - configs/ls2088aqds_tfa_defconfig | 1 - configs/ls2088ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls2088ardb_qspi_defconfig | 1 - configs/ls2088ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls2088ardb_tfa_defconfig | 1 - configs/lx2160aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2160aqds_tfa_defconfig | 1 - configs/lx2160ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2160ardb_tfa_defconfig | 1 - configs/lx2160ardb_tfa_stmm_defconfig | 4 ---- configs/lx2162aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2162aqds_tfa_defconfig | 1 - configs/lx2162aqds_tfa_verified_boot_defconfig | 1 - configs/mt7623n_bpir2_defconfig | 1 - configs/mt7629_rfb_defconfig | 1 - configs/mt8183_pumpkin_defconfig | 1 - configs/mt8516_pumpkin_defconfig | 1 - configs/mvebu_puzzle-m801-88f8040_defconfig | 1 - configs/mx6memcal_defconfig | 1 - configs/octeontx2_95xx_defconfig | 1 - configs/octeontx2_96xx_defconfig | 1 - configs/octeontx_81xx_defconfig | 1 - configs/octeontx_83xx_defconfig | 1 - configs/omap4_sdp4430_defconfig | 1 - configs/opos6uldev_defconfig | 1 - configs/pcm052_defconfig | 1 - configs/phycore-am335x-r2-regor_defconfig | 1 - configs/phycore-am335x-r2-wega_defconfig | 1 - configs/qemu-riscv32_defconfig | 2 -- configs/qemu-riscv32_smode_defconfig | 2 -- configs/qemu-riscv64_defconfig | 2 -- configs/qemu-riscv64_smode_defconfig | 2 -- configs/qemu-x86_64_defconfig | 2 -- configs/qemu-x86_defconfig | 2 -- configs/qemu_arm64_defconfig | 2 -- configs/qemu_arm_defconfig | 2 -- configs/roc-cc-rk3308_defconfig | 1 - configs/rock-pi-n10-rk3399pro_defconfig | 1 - configs/rock-pi-n8-rk3288_defconfig | 1 - configs/rpi_0_w_defconfig | 1 - configs/rpi_defconfig | 1 - configs/sama5d27_wlsom1_ek_mmc_defconfig | 1 - configs/sama5d27_wlsom1_ek_qspiflash_defconfig | 1 - configs/sama5d2_icp_mmc_defconfig | 1 - configs/sama7g5ek_mmc1_defconfig | 1 - configs/sama7g5ek_mmc_defconfig | 1 - configs/sipeed_maix_bitm_defconfig | 1 - configs/sipeed_maix_smode_defconfig | 1 - configs/socfpga_de1_soc_defconfig | 1 - configs/somlabs_visionsom_6ull_defconfig | 1 - configs/stemmy_defconfig | 1 - configs/stm32mp15_basic_defconfig | 3 --- configs/stm32mp15_trusted_defconfig | 3 --- configs/tbs2910_defconfig | 1 - configs/vf610twr_defconfig | 1 - configs/vf610twr_nand_defconfig | 1 - configs/xenguest_arm64_defconfig | 1 - configs/xilinx_versal_mini_defconfig | 1 - configs/xilinx_versal_mini_emmc0_defconfig | 1 - configs/xilinx_versal_mini_emmc1_defconfig | 1 - configs/xilinx_versal_virt_defconfig | 2 -- configs/xilinx_zynqmp_mini_defconfig | 1 - configs/xilinx_zynqmp_mini_emmc0_defconfig | 1 - configs/xilinx_zynqmp_mini_emmc1_defconfig | 1 - configs/xilinx_zynqmp_mini_nand_defconfig | 1 - configs/xilinx_zynqmp_mini_nand_single_defconfig | 1 - configs/xilinx_zynqmp_mini_qspi_defconfig | 1 - configs/xilinx_zynqmp_r5_defconfig | 1 - configs/xilinx_zynqmp_virt_defconfig | 9 --------- configs/zynq_cse_nand_defconfig | 1 - configs/zynq_cse_nor_defconfig | 1 - configs/zynq_cse_qspi_defconfig | 1 - lib/efi_loader/Kconfig | 5 +++-- 175 files changed, 18 insertions(+), 207 deletions(-)
common/Kconfig.boot +15 -0
@@ -300,6 +300,21 @@ config LEGACY_IMAGE_FORMAT loaded. If a board needs the legacy image format support in this case, enable it here. +config EBBR + bool "Enable support for Embeeded Boot Base Requirements (EBBR)" + select EFI_LOADER + help + Enable this to support ARM's EBBR boot method. This bases everything + on UEFI protocols. + + This Embedded Base Boot Requirements (EBBR) specification defines an + interface between platform firmware and an operating system that is + suitable for embedded platforms. EBBR-compliant platforms present a + consistent interface that will boot an EBBR-compliant operating + system without any custom tailoring required. For example, an Arm + A-class embedded platform will benefit from a standard interface that + supports features such as secure boot and firmware update. + config SUPPORT_RAW_INITRD bool "Enable raw initrd images" help
configs/am335x_igep003x_defconfig +0 -1
@@ -79,4 +79,3 @@ CONFIG_SPI=y CONFIG_DM_SPI=y CONFIG_OMAP3_SPI=y CONFIG_FDT_FIXUP_PARTITIONS=y -# CONFIG_GENERATE_SMBIOS_TABLE is not set
configs/am335x_pdu001_defconfig +0 -1
@@ -53,4 +53,3 @@ CONFIG_DM_REGULATOR_FIXED=y CONFIG_DM_REGULATOR_TPS65910=y CONFIG_CONS_INDEX=4 # CONFIG_SPL_USE_TINY_PRINTF is not set -# CONFIG_EFI_LOADER is not set
configs/apalis-imx8_defconfig +0 -1
@@ -65,4 +65,3 @@ CONFIG_DM_SERIAL=y CONFIG_FSL_LPUART=y CONFIG_DM_THERMAL=y CONFIG_IMX_SCU_THERMAL=y -# CONFIG_EFI_LOADER is not set
configs/apalis-imx8x_defconfig +0 -1
@@ -74,4 +74,3 @@ CONFIG_DM_SERIAL=y CONFIG_FSL_LPUART=y CONFIG_DM_THERMAL=y CONFIG_IMX_SCU_THERMAL=y -# CONFIG_EFI_LOADER is not set
configs/aristainetos2c_defconfig +0 -1
@@ -122,4 +122,3 @@ CONFIG_SPLASH_SCREEN_ALIGN=y CONFIG_VIDEO_BMP_RLE8=y CONFIG_BMP_16BPP=y CONFIG_IMX_WATCHDOG=y -# CONFIG_EFI_LOADER is not set
configs/aristainetos2ccslb_defconfig +0 -1
@@ -122,4 +122,3 @@ CONFIG_SPLASH_SCREEN_ALIGN=y CONFIG_VIDEO_BMP_RLE8=y CONFIG_BMP_16BPP=y CONFIG_IMX_WATCHDOG=y -# CONFIG_EFI_LOADER is not set
configs/avnet_ultrazedev_cc_v1_0_ultrazedev_som_v1_0_defconfig +0 -1
@@ -63,4 +63,3 @@ CONFIG_SPI=y CONFIG_ZYNQMP_GQSPI=y CONFIG_PANIC_HANG=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/bcm7260_defconfig +0 -1
@@ -31,4 +31,3 @@ CONFIG_SYS_RELOC_GD_ENV_ADDR=y CONFIG_MMC_SDHCI=y CONFIG_MMC_SDHCI_BCMSTB=y CONFIG_MTD=y -# CONFIG_EFI_LOADER is not set
configs/bcm7445_defconfig +0 -1
@@ -36,4 +36,3 @@ CONFIG_DM_SPI_FLASH=y CONFIG_SPI=y CONFIG_DM_SPI=y CONFIG_BCMSTB_SPI=y -# CONFIG_EFI_LOADER is not set
configs/bcm963158_ram_defconfig +0 -1
@@ -54,4 +54,3 @@ CONFIG_BCM63XX_HSSPI=y CONFIG_SYSRESET=y CONFIG_SYSRESET_WATCHDOG=y CONFIG_WDT_BCM6345=y -# CONFIG_GENERATE_SMBIOS_TABLE is not set
configs/bcm968580xref_ram_defconfig +0 -1
@@ -51,4 +51,3 @@ CONFIG_BCM63XX_HSSPI=y CONFIG_SYSRESET=y CONFIG_SYSRESET_WATCHDOG=y CONFIG_WDT_BCM6345=y -# CONFIG_GENERATE_SMBIOS_TABLE is not set
configs/bitmain_antminer_s9_defconfig +0 -1
@@ -74,4 +74,3 @@ CONFIG_ZYNQ_SERIAL=y # CONFIG_WATCHDOG is not set CONFIG_WDT=y CONFIG_WDT_CDNS=y -# CONFIG_EFI_LOADER is not set
configs/bk4r1_defconfig +0 -1
@@ -86,4 +86,3 @@ CONFIG_FSL_LPUART=y CONFIG_SPI=y CONFIG_DM_SPI=y CONFIG_FSL_QSPI=y -# CONFIG_EFI_LOADER is not set
configs/brppt1_mmc_defconfig +0 -1
@@ -98,4 +98,3 @@ CONFIG_USB_GADGET=y CONFIG_FAT_WRITE=y CONFIG_LZO=y # CONFIG_OF_LIBFDT_OVERLAY is not set -# CONFIG_EFI_LOADER is not set
configs/brppt1_nand_defconfig +0 -1
@@ -104,4 +104,3 @@ CONFIG_USB_GADGET=y CONFIG_FAT_WRITE=y CONFIG_LZO=y # CONFIG_OF_LIBFDT_OVERLAY is not set -# CONFIG_EFI_LOADER is not set
configs/brppt1_spi_defconfig +0 -1
@@ -114,4 +114,3 @@ CONFIG_USB_GADGET=y CONFIG_FAT_WRITE=y CONFIG_LZO=y # CONFIG_OF_LIBFDT_OVERLAY is not set -# CONFIG_EFI_LOADER is not set
configs/brppt2_defconfig +0 -1
@@ -97,4 +97,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_STORAGE=y CONFIG_SPL_TINY_MEMSET=y -# CONFIG_EFI_LOADER is not set
configs/brsmarc1_defconfig +0 -1
@@ -114,4 +114,3 @@ CONFIG_USB_GADGET=y CONFIG_SPL_TINY_MEMSET=y CONFIG_SHA1=y CONFIG_SHA256=y -# CONFIG_EFI_LOADER is not set
configs/brxre1_defconfig +0 -1
@@ -94,4 +94,3 @@ CONFIG_USB_GADGET=y CONFIG_SYS_WHITE_ON_BLACK=y CONFIG_SPL_TINY_MEMSET=y # CONFIG_OF_LIBFDT_OVERLAY is not set -# CONFIG_EFI_LOADER is not set
configs/chromebook_coral_defconfig +0 -1
@@ -115,4 +115,3 @@ CONFIG_CMD_DHRYSTONE=y CONFIG_TPM=y # CONFIG_GZIP is not set CONFIG_BLOBLIST_TABLES=y -# CONFIG_EFI_LOADER is not set
configs/chromebook_link_defconfig +0 -1
@@ -71,4 +71,3 @@ CONFIG_VIDEO_IVYBRIDGE_IGD=y CONFIG_CMD_DHRYSTONE=y CONFIG_TPM=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/colibri-imx8x_defconfig +0 -1
@@ -62,4 +62,3 @@ CONFIG_DM_SERIAL=y CONFIG_FSL_LPUART=y CONFIG_DM_THERMAL=y CONFIG_IMX_SCU_THERMAL=y -# CONFIG_EFI_LOADER is not set
configs/colibri_vf_defconfig +0 -1
@@ -101,4 +101,3 @@ CONFIG_VIDEO_FSL_DCU_FB=y CONFIG_SPLASH_SCREEN_ALIGN=y CONFIG_OF_LIBFDT_OVERLAY=y CONFIG_FDT_FIXUP_PARTITIONS=y -# CONFIG_EFI_UNICODE_CAPITALIZATION is not set
configs/controlcenterdc_defconfig +0 -1
@@ -83,4 +83,3 @@ CONFIG_DM_USB=y CONFIG_USB_EHCI_HCD=y CONFIG_USB_STORAGE=y CONFIG_TPM=y -# CONFIG_EFI_LOADER is not set
configs/crs305-1g-4s-bit_defconfig +0 -1
@@ -44,4 +44,3 @@ CONFIG_PCI=y CONFIG_PCI_MVEBU=y CONFIG_SYS_NS16550=y CONFIG_KIRKWOOD_SPI=y -# CONFIG_EFI_LOADER is not set
configs/crs305-1g-4s_defconfig +0 -1
@@ -45,4 +45,3 @@ CONFIG_PCI=y CONFIG_PCI_MVEBU=y CONFIG_SYS_NS16550=y CONFIG_KIRKWOOD_SPI=y -# CONFIG_EFI_LOADER is not set
configs/crs326-24g-2s-bit_defconfig +0 -1
@@ -44,4 +44,3 @@ CONFIG_PCI=y CONFIG_PCI_MVEBU=y CONFIG_SYS_NS16550=y CONFIG_KIRKWOOD_SPI=y -# CONFIG_EFI_LOADER is not set
configs/crs326-24g-2s_defconfig +0 -1
@@ -44,4 +44,3 @@ CONFIG_PCI=y CONFIG_PCI_MVEBU=y CONFIG_SYS_NS16550=y CONFIG_KIRKWOOD_SPI=y -# CONFIG_EFI_LOADER is not set
configs/crs328-4c-20s-4s-bit_defconfig +0 -1
@@ -44,4 +44,3 @@ CONFIG_PCI=y CONFIG_PCI_MVEBU=y CONFIG_SYS_NS16550=y CONFIG_KIRKWOOD_SPI=y -# CONFIG_EFI_LOADER is not set
configs/crs328-4c-20s-4s_defconfig +0 -1
@@ -44,4 +44,3 @@ CONFIG_PCI=y CONFIG_PCI_MVEBU=y CONFIG_SYS_NS16550=y CONFIG_KIRKWOOD_SPI=y -# CONFIG_EFI_LOADER is not set
configs/deneb_defconfig +0 -1
@@ -106,4 +106,3 @@ CONFIG_DM_THERMAL=y CONFIG_IMX_SCU_THERMAL=y # CONFIG_SPL_WDT is not set CONFIG_SPL_TINY_MEMSET=y -# CONFIG_EFI_LOADER is not set
configs/dragonboard820c_defconfig +0 -1
@@ -14,7 +14,6 @@ CONFIG_BOOTARGS="console=ttyMSM0,115200n8" # CONFIG_DISPLAY_BOARDINFO is not set CONFIG_MISC_INIT_R=y CONFIG_SYS_PROMPT="dragonboard820c => " -CONFIG_CMD_BOOTEFI_HELLO=y CONFIG_CMD_MD5SUM=y CONFIG_CMD_MEMINFO=y CONFIG_CMD_GPIO=y
configs/efi-x86_app_defconfig +0 -1
@@ -36,4 +36,3 @@ CONFIG_SYSCON=y # CONFIG_REGEX is not set # CONFIG_GZIP is not set CONFIG_EFI=y -# CONFIG_EFI_LOADER is not set
configs/evb-ast2600_defconfig +0 -1
@@ -65,4 +65,3 @@ CONFIG_SPL_SYSRESET=y CONFIG_WDT=y CONFIG_HEXDUMP=y # CONFIG_SPL_HEXDUMP is not set -# CONFIG_EFI_LOADER is not set
configs/evb-px30_defconfig +0 -1
@@ -105,4 +105,3 @@ CONFIG_SPL_TINY_MEMSET=y CONFIG_TPL_TINY_MEMSET=y CONFIG_LZO=y CONFIG_ERRNO_STR=y -# CONFIG_EFI_LOADER is not set
configs/evb-rk3308_defconfig +0 -1
@@ -73,4 +73,3 @@ CONFIG_USB_GADGET_DOWNLOAD=y CONFIG_SPL_TINY_MEMSET=y CONFIG_LZO=y CONFIG_ERRNO_STR=y -# CONFIG_EFI_LOADER is not set
configs/firefly-px30_defconfig +0 -1
@@ -104,4 +104,3 @@ CONFIG_SPL_TINY_MEMSET=y CONFIG_TPL_TINY_MEMSET=y CONFIG_LZO=y CONFIG_ERRNO_STR=y -# CONFIG_EFI_LOADER is not set
configs/ge_bx50v3_defconfig +0 -1
@@ -98,4 +98,3 @@ CONFIG_VIDEO_IPUV3=y CONFIG_WATCHDOG_TIMEOUT_MSECS=8000 CONFIG_IMX_WATCHDOG=y CONFIG_BCH=y -# CONFIG_EFI_LOADER is not set
configs/giedi_defconfig +0 -1
@@ -106,4 +106,3 @@ CONFIG_DM_THERMAL=y CONFIG_IMX_SCU_THERMAL=y # CONFIG_SPL_WDT is not set CONFIG_SPL_TINY_MEMSET=y -# CONFIG_EFI_LOADER is not set
configs/grpeach_defconfig +0 -1
@@ -65,4 +65,3 @@ CONFIG_DM_USB=y CONFIG_USB_R8A66597_HCD=y CONFIG_USB_STORAGE=y CONFIG_OF_LIBFDT_OVERLAY=y -# CONFIG_EFI_LOADER is not set
configs/imx8mm-cl-iot-gate_defconfig +0 -8
@@ -35,8 +35,6 @@ CONFIG_SPL_I2C_SUPPORT=y CONFIG_SPL_POWER_SUPPORT=y CONFIG_SPL_WATCHDOG_SUPPORT=y CONFIG_SYS_PROMPT="u-boot=> " -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_EEPROM=y CONFIG_CMD_SHA1SUM=y CONFIG_CMD_BIND=y
@@ -51,7 +49,6 @@ CONFIG_CMD_USB=y CONFIG_CMD_USB_MASS_STORAGE=y CONFIG_CMD_SNTP=y CONFIG_CMD_CACHE=y -CONFIG_CMD_EFIDEBUG=y CONFIG_CMD_RTC=y CONFIG_CMD_TIME=y CONFIG_CMD_GETTIME=y
@@ -141,8 +138,3 @@ CONFIG_TPM=y CONFIG_LZO=y CONFIG_BZIP2=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_SET_TIME=y -CONFIG_EFI_RUNTIME_UPDATE_CAPSULE=y -CONFIG_EFI_CAPSULE_ON_DISK=y -CONFIG_EFI_CAPSULE_FIRMWARE_FIT=y -CONFIG_EFI_SECURE_BOOT=y
configs/imx8qm_mek_defconfig +0 -1
@@ -84,4 +84,3 @@ CONFIG_SPL_DM_REGULATOR_GPIO=y CONFIG_DM_SERIAL=y CONFIG_FSL_LPUART=y CONFIG_SPL_TINY_MEMSET=y -# CONFIG_EFI_LOADER is not set
configs/imx8qm_rom7720_a1_4G_defconfig +0 -1
@@ -81,4 +81,3 @@ CONFIG_SPL_DM_REGULATOR_GPIO=y CONFIG_DM_SERIAL=y CONFIG_FSL_LPUART=y CONFIG_SPL_TINY_MEMSET=y -# CONFIG_EFI_LOADER is not set
configs/imx8qxp_mek_defconfig +0 -1
@@ -88,4 +88,3 @@ CONFIG_FSL_LPUART=y CONFIG_DM_THERMAL=y CONFIG_IMX_SCU_THERMAL=y CONFIG_SPL_TINY_MEMSET=y -# CONFIG_EFI_LOADER is not set
configs/kontron_sl28_defconfig +0 -2
@@ -36,7 +36,6 @@ CONFIG_SPL_MPC8XXX_INIT_DDR_SUPPORT=y CONFIG_SPL_SPI_LOAD=y CONFIG_CMD_ASKENV=y CONFIG_CMD_GREPENV=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_DM=y CONFIG_CMD_GPT=y CONFIG_CMD_I2C=y
@@ -44,7 +43,6 @@ CONFIG_CMD_MMC=y CONFIG_CMD_PCI=y CONFIG_CMD_USB=y CONFIG_CMD_CACHE=y -CONFIG_CMD_EFIDEBUG=y CONFIG_CMD_RNG=y CONFIG_MP=y CONFIG_OF_CONTROL=y
configs/kp_imx53_defconfig +0 -1
@@ -57,4 +57,3 @@ CONFIG_USB=y CONFIG_USB_EHCI_MX5=y CONFIG_USB_STORAGE=y CONFIG_HEXDUMP=y -# CONFIG_EFI_LOADER is not set
configs/ls1028aqds_tfa_SECURE_BOOT_defconfig +0 -1
@@ -87,4 +87,3 @@ CONFIG_WDT=y CONFIG_WDT_SP805=y CONFIG_RSA=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1028aqds_tfa_defconfig +0 -1
@@ -92,4 +92,3 @@ CONFIG_USB_XHCI_DWC3=y CONFIG_WDT=y CONFIG_WDT_SP805=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1028aqds_tfa_lpuart_defconfig +0 -1
@@ -92,4 +92,3 @@ CONFIG_USB_XHCI_DWC3=y CONFIG_WDT=y CONFIG_WDT_SP805=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1028ardb_tfa_SECURE_BOOT_defconfig +0 -1
@@ -82,4 +82,3 @@ CONFIG_WDT=y CONFIG_WDT_SP805=y CONFIG_RSA=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1028ardb_tfa_defconfig +0 -1
@@ -91,4 +91,3 @@ CONFIG_USB_ETHER_RTL8152=y CONFIG_WDT=y CONFIG_WDT_SP805=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_defconfig +0 -1
@@ -68,4 +68,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_lpuart_defconfig +0 -1
@@ -70,4 +70,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_nand_defconfig +0 -1
@@ -84,4 +84,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_nor_ddr3_defconfig +0 -1
@@ -69,4 +69,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_qspi_defconfig +0 -1
@@ -65,4 +65,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_sdcard_ifc_defconfig +0 -1
@@ -85,4 +85,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_sdcard_qspi_defconfig +0 -1
@@ -79,4 +79,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_tfa_SECURE_BOOT_defconfig +0 -1
@@ -71,4 +71,3 @@ CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_SPL_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043aqds_tfa_defconfig +0 -1
@@ -78,4 +78,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043ardb_SECURE_BOOT_defconfig +0 -1
@@ -61,4 +61,3 @@ CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_SPL_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043ardb_defconfig +0 -1
@@ -61,4 +61,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043ardb_nand_SECURE_BOOT_defconfig +0 -1
@@ -81,4 +81,3 @@ CONFIG_USB_XHCI_DWC3=y # CONFIG_SPL_USE_TINY_PRINTF is not set CONFIG_RSA=y CONFIG_SPL_RSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043ardb_nand_defconfig +0 -1
@@ -80,4 +80,3 @@ CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y # CONFIG_SPL_USE_TINY_PRINTF is not set -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043ardb_sdcard_SECURE_BOOT_defconfig +0 -1
@@ -83,4 +83,3 @@ CONFIG_USB_XHCI_DWC3=y # CONFIG_SPL_USE_TINY_PRINTF is not set CONFIG_RSA=y CONFIG_SPL_RSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043ardb_sdcard_defconfig +0 -1
@@ -80,4 +80,3 @@ CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y # CONFIG_SPL_USE_TINY_PRINTF is not set -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043ardb_tfa_SECURE_BOOT_defconfig +0 -1
@@ -62,4 +62,3 @@ CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_SPL_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1043ardb_tfa_defconfig +0 -1
@@ -65,4 +65,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_SECURE_BOOT_defconfig +0 -1
@@ -69,4 +69,3 @@ CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_defconfig +0 -1
@@ -71,4 +71,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_lpuart_defconfig +0 -1
@@ -73,4 +73,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_nand_defconfig +0 -1
@@ -79,4 +79,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_qspi_defconfig +0 -1
@@ -69,4 +69,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_sdcard_ifc_defconfig +0 -1
@@ -89,4 +89,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_sdcard_qspi_defconfig +0 -1
@@ -84,4 +84,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_tfa_SECURE_BOOT_defconfig +0 -1
@@ -71,4 +71,3 @@ CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046aqds_tfa_defconfig +0 -1
@@ -81,4 +81,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046ardb_emmc_defconfig +0 -1
@@ -81,4 +81,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046ardb_qspi_SECURE_BOOT_defconfig +0 -1
@@ -64,4 +64,3 @@ CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046ardb_qspi_defconfig +0 -1
@@ -67,4 +67,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046ardb_qspi_spl_defconfig +0 -1
@@ -86,4 +86,3 @@ CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_SPL_GZIP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046ardb_sdcard_SECURE_BOOT_defconfig +0 -1
@@ -81,4 +81,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_SPL_RSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046ardb_sdcard_defconfig +0 -1
@@ -80,4 +80,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046ardb_tfa_SECURE_BOOT_defconfig +0 -1
@@ -63,4 +63,3 @@ CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1046ardb_tfa_defconfig +0 -1
@@ -68,4 +68,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1088aqds_qspi_SECURE_BOOT_defconfig +0 -1
@@ -74,4 +74,3 @@ CONFIG_USB_DWC3=y CONFIG_USB_GADGET=y CONFIG_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1088aqds_qspi_defconfig +0 -1
@@ -75,4 +75,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_USB_DWC3=y CONFIG_USB_GADGET=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1088aqds_tfa_defconfig +0 -1
@@ -99,4 +99,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_USB_DWC3=y CONFIG_USB_GADGET=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1088ardb_qspi_SECURE_BOOT_defconfig +0 -1
@@ -76,4 +76,3 @@ CONFIG_USB_DWC3=y CONFIG_USB_GADGET=y CONFIG_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1088ardb_qspi_defconfig +0 -1
@@ -77,4 +77,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_USB_DWC3=y CONFIG_USB_GADGET=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1088ardb_tfa_SECURE_BOOT_defconfig +0 -1
@@ -85,4 +85,3 @@ CONFIG_USB_GADGET=y CONFIG_RSA=y CONFIG_SPL_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls1088ardb_tfa_defconfig +0 -1
@@ -91,4 +91,3 @@ CONFIG_USB_HOST_ETHER=y CONFIG_USB_ETHER_ASIX=y CONFIG_USB_ETHER_ASIX88179=y CONFIG_USB_ETHER_RTL8152=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2080aqds_SECURE_BOOT_defconfig +0 -1
@@ -67,4 +67,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2080aqds_defconfig +0 -1
@@ -68,4 +68,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2080aqds_nand_defconfig +0 -1
@@ -75,4 +75,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2080aqds_qspi_defconfig +0 -1
@@ -67,4 +67,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2080aqds_sdcard_defconfig +0 -1
@@ -74,4 +74,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2080ardb_SECURE_BOOT_defconfig +0 -1
@@ -65,4 +65,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2080ardb_defconfig +0 -1
@@ -66,4 +66,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2080ardb_nand_defconfig +0 -1
@@ -71,4 +71,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2081ardb_defconfig +0 -1
@@ -64,4 +64,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2088aqds_tfa_defconfig +0 -1
@@ -89,4 +89,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2088ardb_qspi_SECURE_BOOT_defconfig +0 -1
@@ -66,4 +66,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2088ardb_qspi_defconfig +0 -1
@@ -71,4 +71,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2088ardb_tfa_SECURE_BOOT_defconfig +0 -1
@@ -82,4 +82,3 @@ CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_SPL_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/ls2088ardb_tfa_defconfig +0 -1
@@ -87,4 +87,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/lx2160aqds_tfa_SECURE_BOOT_defconfig +0 -1
@@ -85,4 +85,3 @@ CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_SPL_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/lx2160aqds_tfa_defconfig +0 -1
@@ -91,4 +91,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_WDT=y CONFIG_WDT_SBSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/lx2160ardb_tfa_SECURE_BOOT_defconfig +0 -1
@@ -76,4 +76,3 @@ CONFIG_USB_XHCI_DWC3=y CONFIG_RSA=y CONFIG_SPL_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/lx2160ardb_tfa_defconfig +0 -1
@@ -86,4 +86,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_WDT=y CONFIG_WDT_SBSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/lx2160ardb_tfa_stmm_defconfig +0 -4
@@ -25,7 +25,6 @@ CONFIG_BOOTARGS="console=ttyAMA0,115200 root=/dev/ram0 earlycon=pl011,mmio32,0x2 # CONFIG_USE_BOOTCOMMAND is not set CONFIG_MISC_INIT_R=y CONFIG_CMD_GREPENV=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_EEPROM=y CONFIG_CMD_DM=y CONFIG_CMD_GPIO=y
@@ -35,7 +34,6 @@ CONFIG_CMD_MMC=y CONFIG_CMD_PCI=y CONFIG_CMD_USB=y CONFIG_CMD_CACHE=y -CONFIG_CMD_EFIDEBUG=y CONFIG_MP=y CONFIG_OF_CONTROL=y CONFIG_ENV_OVERWRITE=y
@@ -84,5 +82,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y -CONFIG_EFI_MM_COMM_TEE=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/lx2162aqds_tfa_SECURE_BOOT_defconfig +0 -1
@@ -88,4 +88,3 @@ CONFIG_WDT_SBSA=y CONFIG_RSA=y CONFIG_SPL_RSA=y CONFIG_RSA_SOFTWARE_EXP=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/lx2162aqds_tfa_defconfig +0 -1
@@ -95,4 +95,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_WDT=y CONFIG_WDT_SBSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/lx2162aqds_tfa_verified_boot_defconfig +0 -1
@@ -96,4 +96,3 @@ CONFIG_USB_XHCI_HCD=y CONFIG_USB_XHCI_DWC3=y CONFIG_WDT=y CONFIG_WDT_SBSA=y -CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
configs/mt7623n_bpir2_defconfig +0 -1
@@ -52,4 +52,3 @@ CONFIG_TIMER=y CONFIG_MTK_TIMER=y CONFIG_WDT_MTK=y CONFIG_LZMA=y -# CONFIG_EFI_GRUB_ARM32_WORKAROUND is not set
configs/mt7629_rfb_defconfig +0 -1
@@ -93,4 +93,3 @@ CONFIG_USB_KEYBOARD=y CONFIG_WDT_MTK=y CONFIG_LZMA=y CONFIG_SPL_LZMA=y -# CONFIG_EFI_LOADER is not set
configs/mt8183_pumpkin_defconfig +0 -1
@@ -76,4 +76,3 @@ CONFIG_WDT=y CONFIG_WDT_MTK=y # CONFIG_REGEX is not set CONFIG_OF_LIBFDT_OVERLAY=y -# CONFIG_EFI_LOADER is not set
configs/mt8516_pumpkin_defconfig +0 -1
@@ -74,4 +74,3 @@ CONFIG_USB_GADGET_VENDOR_NUM=0x0e8d CONFIG_USB_GADGET_PRODUCT_NUM=0x201c CONFIG_WDT=y CONFIG_WDT_MTK=y -# CONFIG_EFI_LOADER is not set
configs/mvebu_puzzle-m801-88f8040_defconfig +0 -1
@@ -77,4 +77,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_XHCI_HCD=y CONFIG_USB_EHCI_HCD=y -# CONFIG_EFI_LOADER is not set
configs/mx6memcal_defconfig +0 -1
@@ -52,4 +52,3 @@ CONFIG_USB_GADGET_VENDOR_NUM=0x0525 CONFIG_USB_GADGET_PRODUCT_NUM=0xa4a5 CONFIG_CI_UDC=y CONFIG_OF_LIBFDT=y -# CONFIG_EFI_LOADER is not set
configs/octeontx2_95xx_defconfig +0 -1
@@ -24,7 +24,6 @@ CONFIG_BOOTARGS="console=ttyAMA0,115200n8 earlycon=pl011,0x87e028000000 maxcpus= CONFIG_BOARD_EARLY_INIT_R=y CONFIG_HUSH_PARSER=y CONFIG_SYS_PROMPT="Marvell> " -# CONFIG_CMD_BOOTEFI_HELLO_COMPILE is not set CONFIG_CMD_MD5SUM=y CONFIG_MD5SUM_VERIFY=y CONFIG_CMD_MX_CYCLIC=y
configs/octeontx2_96xx_defconfig +0 -1
@@ -24,7 +24,6 @@ CONFIG_BOOTARGS="console=ttyAMA0,115200n8 earlycon=pl011,0x87e028000000 maxcpus= CONFIG_BOARD_EARLY_INIT_R=y CONFIG_HUSH_PARSER=y CONFIG_SYS_PROMPT="Marvell> " -# CONFIG_CMD_BOOTEFI_HELLO_COMPILE is not set CONFIG_CMD_MD5SUM=y CONFIG_MD5SUM_VERIFY=y CONFIG_CMD_MX_CYCLIC=y
configs/octeontx_81xx_defconfig +0 -1
@@ -26,7 +26,6 @@ CONFIG_BOOTARGS="console=ttyAMA0,115200n8 earlycon=pl011,0x87e028000000 maxcpus= CONFIG_BOARD_EARLY_INIT_R=y CONFIG_HUSH_PARSER=y CONFIG_SYS_PROMPT="Marvell> " -# CONFIG_CMD_BOOTEFI_HELLO_COMPILE is not set CONFIG_CMD_MD5SUM=y CONFIG_MD5SUM_VERIFY=y CONFIG_CMD_MX_CYCLIC=y
configs/octeontx_83xx_defconfig +0 -1
@@ -24,7 +24,6 @@ CONFIG_BOOTARGS="console=ttyAMA0,115200n8 earlycon=pl011,0x87e028000000 maxcpus= CONFIG_BOARD_EARLY_INIT_R=y CONFIG_HUSH_PARSER=y CONFIG_SYS_PROMPT="Marvell> " -# CONFIG_CMD_BOOTEFI_HELLO_COMPILE is not set CONFIG_CMD_MD5SUM=y CONFIG_MD5SUM_VERIFY=y CONFIG_CMD_MX_CYCLIC=y
configs/omap4_sdp4430_defconfig +0 -1
@@ -44,4 +44,3 @@ CONFIG_USB_OMAP3=y CONFIG_USB_GADGET=y CONFIG_FAT_WRITE=y # CONFIG_REGEX is not set -# CONFIG_EFI_LOADER is not set
configs/opos6uldev_defconfig +0 -1
@@ -114,4 +114,3 @@ CONFIG_BMP_16BPP=y CONFIG_BMP_24BPP=y CONFIG_BMP_32BPP=y CONFIG_OF_LIBFDT_OVERLAY=y -# CONFIG_EFI_LOADER is not set
configs/pcm052_defconfig +0 -1
@@ -70,4 +70,3 @@ CONFIG_FSL_LPUART=y CONFIG_SPI=y CONFIG_DM_SPI=y CONFIG_FSL_QSPI=y -# CONFIG_EFI_LOADER is not set
configs/phycore-am335x-r2-regor_defconfig +0 -1
@@ -84,4 +84,3 @@ CONFIG_USB_MUSB_TI=y CONFIG_USB_GADGET=y CONFIG_USB_ETHER=y CONFIG_FDT_FIXUP_PARTITIONS=y -# CONFIG_EFI_LOADER is not set
configs/phycore-am335x-r2-wega_defconfig +0 -1
@@ -85,4 +85,3 @@ CONFIG_USB_MUSB_TI=y CONFIG_USB_GADGET=y CONFIG_USB_ETHER=y CONFIG_FDT_FIXUP_PARTITIONS=y -# CONFIG_EFI_LOADER is not set
configs/qemu-riscv32_defconfig +0 -2
@@ -6,8 +6,6 @@ CONFIG_DISTRO_DEFAULTS=y CONFIG_FIT=y CONFIG_DISPLAY_CPUINFO=y CONFIG_DISPLAY_BOARDINFO=y -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y # CONFIG_CMD_MII is not set CONFIG_OF_PRIOR_STAGE=y CONFIG_SYS_RELOC_GD_ENV_ADDR=y
configs/qemu-riscv32_smode_defconfig +0 -2
@@ -7,8 +7,6 @@ CONFIG_DISTRO_DEFAULTS=y CONFIG_FIT=y CONFIG_DISPLAY_CPUINFO=y CONFIG_DISPLAY_BOARDINFO=y -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y # CONFIG_CMD_MII is not set CONFIG_OF_PRIOR_STAGE=y CONFIG_SYS_RELOC_GD_ENV_ADDR=y
configs/qemu-riscv64_defconfig +0 -2
@@ -7,8 +7,6 @@ CONFIG_DISTRO_DEFAULTS=y CONFIG_FIT=y CONFIG_DISPLAY_CPUINFO=y CONFIG_DISPLAY_BOARDINFO=y -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y # CONFIG_CMD_MII is not set CONFIG_OF_PRIOR_STAGE=y CONFIG_SYS_RELOC_GD_ENV_ADDR=y
configs/qemu-riscv64_smode_defconfig +0 -2
@@ -8,8 +8,6 @@ CONFIG_DISTRO_DEFAULTS=y CONFIG_FIT=y CONFIG_DISPLAY_CPUINFO=y CONFIG_DISPLAY_BOARDINFO=y -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y # CONFIG_CMD_MII is not set CONFIG_OF_PRIOR_STAGE=y CONFIG_SYS_RELOC_GD_ENV_ADDR=y
configs/qemu-x86_64_defconfig +0 -2
@@ -39,8 +39,6 @@ CONFIG_SPL_PCI=y CONFIG_SPL_PCH_SUPPORT=y CONFIG_SPL_RTC_SUPPORT=y CONFIG_CMD_CPU=y -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_IDE=y CONFIG_CMD_SPI=y CONFIG_CMD_USB=y
configs/qemu-x86_defconfig +0 -2
@@ -21,8 +21,6 @@ CONFIG_DISPLAY_BOARDINFO_LATE=y CONFIG_LAST_STAGE_INIT=y CONFIG_PCI_INIT_R=y CONFIG_CMD_CPU=y -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_IDE=y CONFIG_CMD_SPI=y CONFIG_CMD_USB=y
configs/qemu_arm64_defconfig +0 -2
@@ -15,8 +15,6 @@ CONFIG_USE_PREBOOT=y # CONFIG_DISPLAY_CPUINFO is not set # CONFIG_DISPLAY_BOARDINFO is not set CONFIG_PCI_INIT_R=y -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_DFU=y CONFIG_CMD_MTD=y CONFIG_CMD_PCI=y
configs/qemu_arm_defconfig +0 -2
@@ -17,8 +17,6 @@ CONFIG_USE_PREBOOT=y # CONFIG_DISPLAY_CPUINFO is not set # CONFIG_DISPLAY_BOARDINFO is not set CONFIG_PCI_INIT_R=y -CONFIG_CMD_BOOTEFI_SELFTEST=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_DFU=y CONFIG_CMD_MTD=y CONFIG_CMD_PCI=y
configs/roc-cc-rk3308_defconfig +0 -1
@@ -73,4 +73,3 @@ CONFIG_USB_GADGET_DOWNLOAD=y CONFIG_SPL_TINY_MEMSET=y CONFIG_LZO=y CONFIG_ERRNO_STR=y -# CONFIG_EFI_LOADER is not set
configs/rock-pi-n10-rk3399pro_defconfig +0 -1
@@ -62,7 +62,6 @@ CONFIG_USB_DWC3=y CONFIG_USB_DWC3_GENERIC=y CONFIG_ROCKCHIP_USB2_PHY=y CONFIG_USB_KEYBOARD=y -# CONFIG_USB_KEYBOARD_FN_KEYS is not set CONFIG_USB_GADGET=y CONFIG_DM_VIDEO=y CONFIG_DISPLAY=y
configs/rock-pi-n8-rk3288_defconfig +0 -1
@@ -71,7 +71,6 @@ CONFIG_USB_EHCI_GENERIC=y CONFIG_USB_DWC2=y CONFIG_ROCKCHIP_USB2_PHY=y CONFIG_USB_KEYBOARD=y -# CONFIG_USB_KEYBOARD_FN_KEYS is not set CONFIG_USB_GADGET=y CONFIG_USB_GADGET_DWC2_OTG=y CONFIG_DM_VIDEO=y
configs/rpi_0_w_defconfig +0 -1
@@ -41,4 +41,3 @@ CONFIG_SYS_WHITE_ON_BLACK=y CONFIG_CONSOLE_SCROLL_LINES=10 CONFIG_PHYS_TO_BUS=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_LOADER=y
configs/rpi_defconfig +0 -1
@@ -41,4 +41,3 @@ CONFIG_SYS_WHITE_ON_BLACK=y CONFIG_CONSOLE_SCROLL_LINES=10 CONFIG_PHYS_TO_BUS=y CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_LOADER=y
configs/sama5d27_wlsom1_ek_mmc_defconfig +0 -1
@@ -106,4 +106,3 @@ CONFIG_W1_GPIO=y CONFIG_W1_EEPROM=y CONFIG_W1_EEPROM_DS24XXX=y CONFIG_OF_LIBFDT_OVERLAY=y -# CONFIG_EFI_LOADER_HII is not set
configs/sama5d27_wlsom1_ek_qspiflash_defconfig +0 -1
@@ -118,4 +118,3 @@ CONFIG_W1_GPIO=y CONFIG_W1_EEPROM=y CONFIG_W1_EEPROM_DS24XXX=y CONFIG_OF_LIBFDT_OVERLAY=y -# CONFIG_EFI_LOADER_HII is not set
configs/sama5d2_icp_mmc_defconfig +0 -1
@@ -77,4 +77,3 @@ CONFIG_TIMER=y CONFIG_SPL_TIMER=y CONFIG_ATMEL_PIT_TIMER=y CONFIG_OF_LIBFDT_OVERLAY=y -# CONFIG_EFI_LOADER_HII is not set
configs/sama7g5ek_mmc1_defconfig +0 -1
@@ -68,4 +68,3 @@ CONFIG_ATMEL_USART=y CONFIG_TIMER=y CONFIG_MCHP_PIT64B_TIMER=y CONFIG_OF_LIBFDT_OVERLAY=y -# CONFIG_EFI_LOADER_HII is not set
configs/sama7g5ek_mmc_defconfig +0 -1
@@ -68,4 +68,3 @@ CONFIG_ATMEL_USART=y CONFIG_TIMER=y CONFIG_MCHP_PIT64B_TIMER=y CONFIG_OF_LIBFDT_OVERLAY=y -# CONFIG_EFI_LOADER_HII is not set
configs/sipeed_maix_bitm_defconfig +0 -1
@@ -18,4 +18,3 @@ CONFIG_SF_DEFAULT_BUS=3 # CONFIG_DM_ETH is not set CONFIG_FS_EXT4=y CONFIG_FS_FAT=y -# CONFIG_EFI_LOADER is not set
configs/sipeed_maix_smode_defconfig +0 -1
@@ -18,4 +18,3 @@ CONFIG_SF_DEFAULT_BUS=3 # CONFIG_DM_ETH is not set CONFIG_FS_EXT4=y CONFIG_FS_FAT=y -# CONFIG_EFI_UNICODE_CAPITALIZATION is not set
configs/socfpga_de1_soc_defconfig +0 -1
@@ -50,4 +50,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_DWC2=y # CONFIG_SPL_WDT is not set -# CONFIG_EFI_LOADER is not set
configs/somlabs_visionsom_6ull_defconfig +0 -1
@@ -54,4 +54,3 @@ CONFIG_USB=y CONFIG_DM_USB=y CONFIG_USB_STORAGE=y CONFIG_LZO=y -# CONFIG_EFI_LOADER is not set
configs/stemmy_defconfig +0 -1
@@ -15,4 +15,3 @@ CONFIG_CMD_GETTIME=y CONFIG_EFI_PARTITION=y # CONFIG_NET is not set # CONFIG_MMC_HW_PARTITIONING is not set -# CONFIG_EFI_LOADER is not set
configs/stm32mp15_basic_defconfig +0 -3
@@ -35,7 +35,6 @@ CONFIG_SPL_SPI_FLASH_MTD=y CONFIG_SYS_PROMPT="STM32MP> " CONFIG_CMD_ADTIMG=y CONFIG_CMD_ERASEENV=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_MEMINFO=y CONFIG_CMD_MEMTEST=y CONFIG_CMD_UNZIP=y
@@ -52,7 +51,6 @@ CONFIG_CMD_USB=y CONFIG_CMD_USB_MASS_STORAGE=y CONFIG_CMD_BMP=y CONFIG_CMD_CACHE=y -CONFIG_CMD_EFIDEBUG=y CONFIG_CMD_TIME=y CONFIG_CMD_TIMER=y CONFIG_CMD_PMIC=y
@@ -168,7 +166,6 @@ CONFIG_BMP_32BPP=y CONFIG_WDT=y CONFIG_WDT_STM32MP=y CONFIG_ERRNO_STR=y -# CONFIG_HEXDUMP is not set CONFIG_FDT_FIXUP_PARTITIONS=y # CONFIG_LMB_USE_MAX_REGIONS is not set CONFIG_LMB_MEMORY_REGIONS=2
configs/stm32mp15_trusted_defconfig +0 -3
@@ -18,7 +18,6 @@ CONFIG_BOOTCOMMAND="run bootcmd_stm32mp" CONFIG_SYS_PROMPT="STM32MP> " CONFIG_CMD_ADTIMG=y CONFIG_CMD_ERASEENV=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_MEMINFO=y CONFIG_CMD_MEMTEST=y CONFIG_CMD_UNZIP=y
@@ -35,7 +34,6 @@ CONFIG_CMD_USB=y CONFIG_CMD_USB_MASS_STORAGE=y CONFIG_CMD_BMP=y CONFIG_CMD_CACHE=y -CONFIG_CMD_EFIDEBUG=y CONFIG_CMD_TIME=y CONFIG_CMD_TIMER=y CONFIG_CMD_PMIC=y
@@ -150,7 +148,6 @@ CONFIG_BMP_32BPP=y CONFIG_WDT=y CONFIG_WDT_STM32MP=y CONFIG_ERRNO_STR=y -# CONFIG_HEXDUMP is not set CONFIG_FDT_FIXUP_PARTITIONS=y # CONFIG_LMB_USE_MAX_REGIONS is not set CONFIG_LMB_MEMORY_REGIONS=2
configs/tbs2910_defconfig +0 -1
@@ -108,4 +108,3 @@ CONFIG_VIDEO_IPUV3=y CONFIG_VIDEO_BMP_RLE8=y # CONFIG_GZIP is not set CONFIG_OF_LIBFDT_ASSUME_MASK=0xff -# CONFIG_EFI_LOADER is not set
configs/vf610twr_defconfig +0 -1
@@ -47,4 +47,3 @@ CONFIG_PHY_MICREL_KSZ8XXX=y CONFIG_MII=y CONFIG_DM_SERIAL=y CONFIG_FSL_LPUART=y -# CONFIG_EFI_LOADER is not set
configs/vf610twr_nand_defconfig +0 -1
@@ -47,4 +47,3 @@ CONFIG_PHY_MICREL_KSZ8XXX=y CONFIG_MII=y CONFIG_DM_SERIAL=y CONFIG_FSL_LPUART=y -# CONFIG_EFI_LOADER is not set
configs/xenguest_arm64_defconfig +0 -1
@@ -36,4 +36,3 @@ CONFIG_DM=y # CONFIG_MMC is not set # CONFIG_REQUIRE_SERIAL_CONSOLE is not set CONFIG_DM_SERIAL=y -# CONFIG_EFI_LOADER is not set
configs/xilinx_versal_mini_defconfig +0 -1
@@ -56,4 +56,3 @@ CONFIG_SYS_RELOC_GD_ENV_ADDR=y # CONFIG_MMC is not set CONFIG_ARM_DCC=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_versal_mini_emmc0_defconfig +0 -1
@@ -55,4 +55,3 @@ CONFIG_MMC_SDHCI_ZYNQ=y CONFIG_ARM_DCC=y CONFIG_FAT_WRITE=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_versal_mini_emmc1_defconfig +0 -1
@@ -55,4 +55,3 @@ CONFIG_MMC_SDHCI_ZYNQ=y CONFIG_ARM_DCC=y CONFIG_FAT_WRITE=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_versal_virt_defconfig +0 -2
@@ -19,7 +19,6 @@ CONFIG_USE_PREBOOT=y CONFIG_BOARD_EARLY_INIT_R=y CONFIG_SYS_PROMPT="Versal> " CONFIG_CMD_BOOTMENU=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_MEMTEST=y CONFIG_SYS_ALT_MEMTEST=y CONFIG_CMD_CLK=y
@@ -34,7 +33,6 @@ CONFIG_CMD_SF_TEST=y CONFIG_CMD_USB=y CONFIG_CMD_TFTPPUT=y CONFIG_CMD_CACHE=y -CONFIG_CMD_EFIDEBUG=y CONFIG_CMD_TIME=y CONFIG_CMD_TIMER=y CONFIG_CMD_EXT4_WRITE=y
configs/xilinx_zynqmp_mini_defconfig +0 -1
@@ -56,4 +56,3 @@ CONFIG_SYS_RELOC_GD_ENV_ADDR=y CONFIG_ARM_DCC=y CONFIG_PANIC_HANG=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_zynqmp_mini_emmc0_defconfig +0 -1
@@ -60,4 +60,3 @@ CONFIG_MMC_SDHCI_ZYNQ=y CONFIG_ARM_DCC=y CONFIG_PANIC_HANG=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_zynqmp_mini_emmc1_defconfig +0 -1
@@ -60,4 +60,3 @@ CONFIG_MMC_SDHCI_ZYNQ=y CONFIG_ARM_DCC=y CONFIG_PANIC_HANG=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_zynqmp_mini_nand_defconfig +0 -1
@@ -56,4 +56,3 @@ CONFIG_SYS_NAND_MAX_CHIPS=2 CONFIG_ARM_DCC=y CONFIG_PANIC_HANG=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_zynqmp_mini_nand_single_defconfig +0 -1
@@ -55,4 +55,3 @@ CONFIG_NAND_ARASAN=y CONFIG_ARM_DCC=y CONFIG_PANIC_HANG=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_zynqmp_mini_qspi_defconfig +0 -1
@@ -65,4 +65,3 @@ CONFIG_SPI=y CONFIG_ZYNQMP_GQSPI=y CONFIG_PANIC_HANG=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/xilinx_zynqmp_r5_defconfig +0 -1
@@ -19,4 +19,3 @@ CONFIG_SYS_RELOC_GD_ENV_ADDR=y CONFIG_ZYNQ_SERIAL=y CONFIG_TIMER=y CONFIG_CADENCE_TTC_TIMER=y -# CONFIG_EFI_LOADER is not set
configs/xilinx_zynqmp_virt_defconfig +0 -9
@@ -33,7 +33,6 @@ CONFIG_SPL_ATF=y CONFIG_SPL_ATF_NO_PLATFORM_PARAM=y CONFIG_CMD_BOOTMENU=y CONFIG_CMD_THOR_DOWNLOAD=y -CONFIG_CMD_NVEDIT_EFI=y CONFIG_CMD_MEMTEST=y CONFIG_SYS_ALT_MEMTEST=y CONFIG_CMD_BIND=y
@@ -57,7 +56,6 @@ CONFIG_CMD_USB_MASS_STORAGE=y CONFIG_CMD_TFTPPUT=y CONFIG_CMD_BMP=y CONFIG_CMD_CACHE=y -CONFIG_CMD_EFIDEBUG=y CONFIG_CMD_TIME=y CONFIG_CMD_GETTIME=y CONFIG_CMD_TIMER=y
@@ -183,11 +181,4 @@ CONFIG_WDT_CDNS=y CONFIG_PANIC_HANG=y CONFIG_TPM=y CONFIG_SPL_GZIP=y -# CONFIG_SPL_HEXDUMP is not set CONFIG_OF_LIBFDT_OVERLAY=y -CONFIG_EFI_SET_TIME=y -CONFIG_EFI_RUNTIME_UPDATE_CAPSULE=y -CONFIG_EFI_CAPSULE_ON_DISK=y -CONFIG_EFI_CAPSULE_ON_DISK_EARLY=y -CONFIG_EFI_CAPSULE_FIRMWARE_FIT=y -CONFIG_EFI_CAPSULE_FIRMWARE_RAW=y
configs/zynq_cse_nand_defconfig +0 -1
@@ -59,4 +59,3 @@ CONFIG_MTD_RAW_NAND=y CONFIG_NAND_ZYNQ=y CONFIG_ARM_DCC=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/zynq_cse_nor_defconfig +0 -1
@@ -62,4 +62,3 @@ CONFIG_SYS_FLASH_USE_BUFFER_WRITE=y CONFIG_SYS_FLASH_CFI=y CONFIG_ARM_DCC=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
configs/zynq_cse_qspi_defconfig +0 -1
@@ -72,4 +72,3 @@ CONFIG_SPI_FLASH_WINBOND=y CONFIG_ARM_DCC=y CONFIG_ZYNQ_QSPI=y # CONFIG_GZIP is not set -# CONFIG_EFI_LOADER is not set
lib/efi_loader/Kconfig +3 -2
@@ -1,6 +1,6 @@ config EFI_LOADER bool "Support running UEFI applications" - depends on OF_LIBFDT && ( \ + depends on OF_LIBFDT && DM && OF_CONTROL && ( \ ARM && (SYS_CPU = arm1136 || \ SYS_CPU = arm1176 || \ SYS_CPU = armv7 || \
@@ -10,7 +10,8 @@ config EFI_LOADER depends on !EFI_STUB || !X86_64 || EFI_STUB_64BIT # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT - default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8 + depends on !ARM || SYS_CPU = armv7 || SYS_CPU = armv8 + default y if SANDBOX select LIB_UUID select HAVE_BLOCK_DEVICE select REGEX
Signature · 1 line
2.32.0.93.g670b81a890-goog
Re: [PATCH 7/7] efi: Make EBBR optional
Quoted 4 lines · Simon Glass
Add a new
Kconfigoption for EBBR so that the naming is more explicit. MakeEFI_LOADERdepend on it.Avoid enabling the options (EBBR /
EFI_LOADER) by default, to save space.
NAK
Operating systems like Fedora, OpenBSD, FreeBSD require UEFI support.
See discussion in
https://lists.denx.de/pipermail/u-boot/2021-June/452947.html
Quoted 3 lines
Also add dependencies on driver model and
OF_CONTROL, since boards which have not migrated to these should not be using new features like EBBR.
Only supporting devices using the driver model in the UEFI sub-system is fine with me. CONFIG_BLK=y is another possible requirement.
We still have 101 defconfigs not using CONFIG_DM=y. Shouldn't they be eliminated?
We have a low number of boards using CONFIG_DM=y and CONFIG_OF_CONTROL=n. Shouldn't they be moved to CONFIG_OF_CONTROL=y and that symbol be eliminated?
armadillo-800eva_defconfigcm_t335_defconfigcolibri_pxa270_defconfigdevkit3250_defconfigdevkit8000_defconfigids8313_defconfigintegratorap_cm720t_defconfigintegratorap_cm920t_defconfigintegratorap_cm926ejs_defconfigintegratorap_cm946es_defconfigintegratorcp_cm1136_defconfigintegratorcp_cm920t_defconfigintegratorcp_cm926ejs_defconfigintegratorcp_cm946es_defconfigkmcoge4_defconfigkzm9g_defconfigmx6memcal_defconfignokia_rx51_defconfigsnapper9260_defconfigsnapper9g20_defconfigsniper_defconfigvexpress_aemv8a_semi_defconfigwork_92105_defconfigxtfpga_defconfig
We have 274 defconfigs with CONFIG_PARTITIONS=y and CONFIG_BLK=n. Shouldn't they be converted or eliminated?
Quoted 2 lines
The unwillingness to use driver model has resulted in the duplication of driver model code in EFI, which is part of the reason for this bloat.
Could you, please, point out where you see possibilities for code reduction.
We started with a situation where many boards were not using the driver model. It is only this year that Tom started to eliminate boards that don't adher to the driver model in some areas.
The semantics used in UEFI and in the rest of U-Boot differ in many aspects. Here are some examples:
* partitions are handled in UEFI like sub-devices but the driver model does not cover partitions yet
* to expose partitions UEFI requires fully probed block devices but U-Boot tries to probe upon first usage
* UEFI uses file handles but U-Boot does not know this concept
Best regards
Heinrich
Re: [PATCH 7/7] efi: Make EBBR optional
Quoted 4 lines · Heinrich Schuchardt
Quoted 3 lines · Simon Glass
Add a new
Kconfigoption for EBBR so that the naming is more explicit. MakeEFI_LOADERdepend on it.
Quoted 8 lines
Quoted 1 line
Avoid enabling the options (EBBR /
EFI_LOADER) by default, to save space.NAK
Operating systems like Fedora, OpenBSD, FreeBSD require UEFI support.
See discussion in
https://lists.denx.de/pipermail/u-boot/2021-June/452947.html
I also think this is taking things in the wrong direction. It was an intentional "make EFI_LOADER default when supported" when we introduced it, and I still think that's the right overall choice. There's a whole lot of non-customization going on when reference designs are converted to products or other reference designs.
Quoted 5 lines
Quoted 2 lines
Also add dependencies on driver model and
OF_CONTROL, since boards which have not migrated to these should not be using new features like EBBR.Only supporting devices using the driver model in the UEFI sub-system is fine with me.
CONFIG_BLK=yis another possible requirement.
Adding these would be good. And may allow for the code to be simplified?
Quoted 2 lines
We still have 101 defconfigs not using
CONFIG_DM=y. Shouldn't they be eliminated?
They will be eliminated after v2022.01, following the same 2 years of "the deadline has passed" that's being used by the board removals I've been posting so far this year.
Quoted 3 lines
We have a low number of boards using
CONFIG_DM=yandCONFIG_OF_CONTROL=n. Shouldn't they be moved toCONFIG_OF_CONTROL=yand that symbol be eliminated?
No, we don't require OF_CONTROL intentionally so that other smaller mechanisms can be used.
[snip]
Quoted 2 lines
We have 274 defconfigs with
CONFIG_PARTITIONS=yandCONFIG_BLK=n. Shouldn't they be converted or eliminated?
Some number of them will be, I think there's one or two actual funny cases on how that code is used however.
Signature · 1 line
Tom
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.
EFI_LOADER used to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disable EFI_LOADER completely by default, we're putting the cart before the horse.
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 6 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.
EFI_LOADER is required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely on EFI_LOADER to boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.
Quoted 3 lines
EBBR should be truly optional, enabled only by boards that use it. Most don't use it but it is enabled anyway. The default boot path should be one that makes use of the existing U-Boot support.
I don't particularly care about EBBR myself, but EFI_LOADER should be the default for as many armv7/arm64/riscv U-Boot targets as possible to give users an easy way to choose the OS they want to run on their machines. That is the best way to guarantee that vendors ship their firmware with it enabled.
Quoted 3 lines
To try to retify this situation, this series adds a new
Kconfigoption for EBBR so that the naming is more explicit. ThenEFI_LOADERis updated to depend on it.
Isn't using the the term EBBR for non-ARM platforms misleading? EFI_LOADER is used much more widely. Anyway, I disagree with this direction.
If size really matters here, we should look at reducing the EFI_LOADER feature set to reduce the amount of code i adds, and maybe introduce an EBBR option that can be enabled by those boards that desire full EBBR compliance?
Quoted 215 lines
For now, only sandbox enables EBBR. Other boards can be added as needed, presumably by distributions that require it. Another approach would be to add '
CONFIG_EBBR=y' to the .config before building, in the build system. That might be more friendly to U-Boot users.This series also fixes a minor issue noticed during testing.
Simon Glass (7):
configs: Resync with savedefconfig
Makefile: Dropinclude/asmdirectory as well as symlink
disk: Tidy up #ifdefs inpart_efi
UseLIB_UUIDwith ACPIGEN andFS_BTRFS
Allowefi_loaderheader to be included always
lib: Create a newKconfigoption for charset conversion
efi: Make EBBR optionalMakefile | 2 +- common/Kconfig.boot | 15 ++ configs/am335x_igep003x_defconfig | 1 - configs/am335x_pdu001_defconfig | 1 - configs/am64x_evm_a53_defconfig | 31 ++- configs/am64x_evm_r5_defconfig | 6 +- configs/apalis-imx8_defconfig | 1 - configs/apalis-imx8x_defconfig | 1 - configs/aristainetos2c_defconfig | 1 - configs/aristainetos2ccslb_defconfig | 1 - ...edev_cc_v1_0_ultrazedev_som_v1_0_defconfig | 1 - configs/bcm7260_defconfig | 1 - configs/bcm7445_defconfig | 1 - configs/bcm963158_ram_defconfig | 1 - configs/bcm968580xref_ram_defconfig | 1 - configs/bitmain_antminer_s9_defconfig | 1 - configs/bk4r1_defconfig | 1 - configs/brppt1_mmc_defconfig | 1 - configs/brppt1_nand_defconfig | 1 - configs/brppt1_spi_defconfig | 1 - configs/brppt2_defconfig | 1 - configs/brsmarc1_defconfig | 1 - configs/brxre1_defconfig | 1 - configs/chromebook_coral_defconfig | 1 - configs/chromebook_link_defconfig | 1 - configs/colibri-imx8x_defconfig | 1 - configs/colibri_vf_defconfig | 1 - configs/controlcenterdc_defconfig | 1 - configs/crs305-1g-4s-bit_defconfig | 1 - configs/crs305-1g-4s_defconfig | 1 - configs/crs326-24g-2s-bit_defconfig | 1 - configs/crs326-24g-2s_defconfig | 1 - configs/crs328-4c-20s-4s-bit_defconfig | 1 - configs/crs328-4c-20s-4s_defconfig | 1 - configs/deneb_defconfig | 1 - configs/draco_defconfig | 2 +- configs/dragonboard820c_defconfig | 1 - configs/efi-x86_app_defconfig | 1 - configs/etamin_defconfig | 2 +- configs/evb-ast2600_defconfig | 1 - configs/evb-px30_defconfig | 1 - configs/evb-rk3308_defconfig | 1 - configs/firefly-px30_defconfig | 1 - configs/ge_bx50v3_defconfig | 1 - configs/giedi_defconfig | 1 - configs/grpeach_defconfig | 1 - configs/imx8mm-cl-iot-gate_defconfig | 8 - configs/imx8qm_mek_defconfig | 1 - configs/imx8qm_rom7720_a1_4G_defconfig | 1 - configs/imx8qxp_mek_defconfig | 1 - configs/j7200_evm_r5_defconfig | 15 +- configs/j721e_evm_r5_defconfig | 13 +- configs/kontron_sl28_defconfig | 2 - configs/kp_imx53_defconfig | 1 - configs/ls1028aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1028aqds_tfa_defconfig | 1 - configs/ls1028aqds_tfa_lpuart_defconfig | 1 - configs/ls1028ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1028ardb_tfa_defconfig | 1 - configs/ls1043aqds_defconfig | 1 - configs/ls1043aqds_lpuart_defconfig | 1 - configs/ls1043aqds_nand_defconfig | 1 - configs/ls1043aqds_nor_ddr3_defconfig | 1 - configs/ls1043aqds_qspi_defconfig | 1 - configs/ls1043aqds_sdcard_ifc_defconfig | 1 - configs/ls1043aqds_sdcard_qspi_defconfig | 1 - configs/ls1043aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1043aqds_tfa_defconfig | 1 - configs/ls1043ardb_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_defconfig | 1 - configs/ls1043ardb_nand_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_nand_defconfig | 1 - .../ls1043ardb_sdcard_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_sdcard_defconfig | 1 - configs/ls1043ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1043ardb_tfa_defconfig | 1 - configs/ls1046aqds_SECURE_BOOT_defconfig | 1 - configs/ls1046aqds_defconfig | 1 - configs/ls1046aqds_lpuart_defconfig | 1 - configs/ls1046aqds_nand_defconfig | 1 - configs/ls1046aqds_qspi_defconfig | 1 - configs/ls1046aqds_sdcard_ifc_defconfig | 1 - configs/ls1046aqds_sdcard_qspi_defconfig | 1 - configs/ls1046aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1046aqds_tfa_defconfig | 1 - configs/ls1046ardb_emmc_defconfig | 1 - configs/ls1046ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_qspi_defconfig | 1 - configs/ls1046ardb_qspi_spl_defconfig | 1 - .../ls1046ardb_sdcard_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_sdcard_defconfig | 1 - configs/ls1046ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1046ardb_tfa_defconfig | 1 - configs/ls1088aqds_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1088aqds_qspi_defconfig | 1 - configs/ls1088aqds_tfa_defconfig | 1 - configs/ls1088ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls1088ardb_qspi_defconfig | 1 - configs/ls1088ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls1088ardb_tfa_defconfig | 1 - configs/ls2080aqds_SECURE_BOOT_defconfig | 1 - configs/ls2080aqds_defconfig | 1 - configs/ls2080aqds_nand_defconfig | 1 - configs/ls2080aqds_qspi_defconfig | 1 - configs/ls2080aqds_sdcard_defconfig | 1 - configs/ls2080ardb_SECURE_BOOT_defconfig | 1 - configs/ls2080ardb_defconfig | 1 - configs/ls2080ardb_nand_defconfig | 1 - configs/ls2081ardb_defconfig | 1 - configs/ls2088aqds_tfa_defconfig | 1 - configs/ls2088ardb_qspi_SECURE_BOOT_defconfig | 1 - configs/ls2088ardb_qspi_defconfig | 1 - configs/ls2088ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/ls2088ardb_tfa_defconfig | 1 - configs/lx2160aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2160aqds_tfa_defconfig | 1 - configs/lx2160ardb_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2160ardb_tfa_defconfig | 1 - configs/lx2160ardb_tfa_stmm_defconfig | 4 - configs/lx2162aqds_tfa_SECURE_BOOT_defconfig | 1 - configs/lx2162aqds_tfa_defconfig | 1 - .../lx2162aqds_tfa_verified_boot_defconfig | 1 - configs/mt7623n_bpir2_defconfig | 1 - configs/mt7629_rfb_defconfig | 1 - configs/mt8183_pumpkin_defconfig | 1 - configs/mt8516_pumpkin_defconfig | 1 - configs/mvebu_puzzle-m801-88f8040_defconfig | 1 - configs/mx6memcal_defconfig | 1 - configs/octeontx2_95xx_defconfig | 1 - configs/octeontx2_96xx_defconfig | 1 - configs/octeontx_81xx_defconfig | 1 - configs/octeontx_83xx_defconfig | 1 - configs/omap4_sdp4430_defconfig | 1 - configs/opos6uldev_defconfig | 1 - configs/pcm052_defconfig | 1 - configs/phycore-am335x-r2-regor_defconfig | 1 - configs/phycore-am335x-r2-wega_defconfig | 1 - configs/pxm2_defconfig | 2 +- configs/qemu-riscv32_defconfig | 2 - configs/qemu-riscv32_smode_defconfig | 2 - configs/qemu-riscv64_defconfig | 2 - configs/qemu-riscv64_smode_defconfig | 2 - configs/qemu-x86_64_defconfig | 2 - configs/qemu-x86_defconfig | 2 - configs/qemu_arm64_defconfig | 2 - configs/qemu_arm_defconfig | 2 - configs/rastaban_defconfig | 2 +- configs/roc-cc-rk3308_defconfig | 1 - configs/rock-pi-n10-rk3399pro_defconfig | 1 - configs/rock-pi-n8-rk3288_defconfig | 1 - configs/rpi_0_w_defconfig | 1 - configs/rpi_defconfig | 1 - configs/rut_defconfig | 2 +- configs/sama5d27_wlsom1_ek_mmc_defconfig | 1 - .../sama5d27_wlsom1_ek_qspiflash_defconfig | 1 - configs/sama5d2_icp_mmc_defconfig | 1 - configs/sama7g5ek_mmc1_defconfig | 1 - configs/sama7g5ek_mmc_defconfig | 1 - configs/sipeed_maix_bitm_defconfig | 1 - configs/sipeed_maix_smode_defconfig | 1 - configs/socfpga_de1_soc_defconfig | 1 - configs/somlabs_visionsom_6ull_defconfig | 1 - configs/stemmy_defconfig | 1 - configs/stm32mp15_basic_defconfig | 3 - configs/stm32mp15_trusted_defconfig | 3 - configs/tbs2910_defconfig | 1 - configs/thuban_defconfig | 2 +- configs/vf610twr_defconfig | 1 - configs/vf610twr_nand_defconfig | 1 - configs/xenguest_arm64_defconfig | 1 - configs/xilinx_versal_mini_defconfig | 1 - configs/xilinx_versal_mini_emmc0_defconfig | 1 - configs/xilinx_versal_mini_emmc1_defconfig | 1 - configs/xilinx_versal_virt_defconfig | 2 - configs/xilinx_zynqmp_mini_defconfig | 1 - configs/xilinx_zynqmp_mini_emmc0_defconfig | 1 - configs/xilinx_zynqmp_mini_emmc1_defconfig | 1 - configs/xilinx_zynqmp_mini_nand_defconfig | 1 - .../xilinx_zynqmp_mini_nand_single_defconfig | 1 - configs/xilinx_zynqmp_mini_qspi_defconfig | 1 - configs/xilinx_zynqmp_r5_defconfig | 1 - configs/xilinx_zynqmp_virt_defconfig | 9 - configs/zynq_cse_nand_defconfig | 1 - configs/zynq_cse_nor_defconfig | 1 - configs/zynq_cse_qspi_defconfig | 1 - disk/part_efi.c | 11 +- drivers/core/Kconfig | 1 + fs/btrfs/Kconfig | 1 + include/efi_loader.h | 186 +++++++++--------- lib/Kconfig | 8 + lib/Makefile | 2 +- lib/efi_loader/Kconfig | 5 +- 192 files changed, 160 insertions(+), 353 deletions(-)Signature · 1 line
2.32.0.93.g670b81a890-goog
I generally agree with Mark here.
Quoted 15 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.
Quoted 8 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
I think the basics of EFI (mostly those that EBBR requires) are sane and nice to boot multiple architectures as well.
Quoted 42 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 6 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.
Quoted 3 lines
EBBR should be truly optional, enabled only by boards that use it. Most don't use it but it is enabled anyway. The default boot path should be one that makes use of the existing U-Boot support.
I don't particularly care about EBBR myself, but
EFI_LOADERshould be the default for as many armv7/arm64/riscv U-Boot targets as possible to give users an easy way to choose the OS they want to run on their machines. That is the best way to guarantee that vendors ship their firmware with it enabled.
Quoted 5 lines
Quoted 3 lines
To try to retify this situation, this series adds a new
Kconfigoption for EBBR so that the naming is more explicit. ThenEFI_LOADERis updated to depend on it.Isn't using the the term EBBR for non-ARM platforms misleading?
Not really, we are discussing with RISC-V atm and having platfomrs being EBBR compliant. In essence we don't desire EBBR to be an Arm only thing.
Quoted 3 lines
EFI_LOADERis used much more widely. Anyway, I disagree with this direction.
Same here
Quoted 4 lines
If size really matters here, we should look at reducing the
EFI_LOADERfeature set to reduce the amount of code i adds, and maybe introduce an EBBR option that can be enabled by those boards that desire full EBBR compliance?
+1
[...]
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.
Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.
Well, I see I forgot to prefix my patch with RFC, but I hadn't found EFI_LOADER being used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.
Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling EFI_LOADER, it's perhaps just one of those "fun" things about Kconfig and anything other than "make oldconfig" for spotting new config options that default to enabled.
Quoted 7 lines
Quoted 6 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.
Quoted 6 lines
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.
To be clear here, I like EFI_LOADER. I too do wish some other technologies had become dominant for technical rather than inertia reasons, but here we are. Having played around with it on aarch64, there are some pretty nice comes-along-with parts to it.
What I hadn't seen, and am only a little skeptical of still, is how far backwards in generations it's going to be used on. The general wish is that users nor off the shelf OS groups need to rebuild U-Boot for a given board, and instead it just works. The number of new designs for 32bit parts is no where near the number of new designs for 64bit parts. So what we're seeing in U-Boot now is people updating support on their older designs, and not necessarily caring about using EFI_LOADER.
Signature · 1 line
Tom
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.
Yes it will be interesting to see what you find there. My results on nokia_rx51 were something like this:
default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0
rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0
rodata +5333.0 text +29712.0
with various other things:CONFIG_OF_LIBFDT_ASSUME_MASK=7
# CONFIG_OF_TRANSLATE is not set
# CONFIG_SIMPLE_BUS is not set
# CONFIG_TI_SYSC is not set
# CONFIG_CMD_FDT is not set
arm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata
+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.
I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
What exactly is impossible? Can you please be more specific?
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
Quoted 20 lines
Quoted 8 lines
Quoted 7 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.Quoted 6 lines
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.To be clear here, I like
EFI_LOADER. I too do wish some other technologies had become dominant for technical rather than inertia reasons, but here we are. Having played around with it on aarch64, there are some pretty nice comes-along-with parts to it.
Quoted 7 lines
What I hadn't seen, and am only a little skeptical of still, is how far backwards in generations it's going to be used on. The general wish is that users nor off the shelf OS groups need to rebuild U-Boot for a given board, and instead it just works. The number of new designs for 32bit parts is no where near the number of new designs for 64bit parts. So what we're seeing in U-Boot now is people updating support on their older designs, and not necessarily caring about using
EFI_LOADER.
In a reply to one of the patches in this series, Heinrich mentions a few problems that need resolving (devices for partitions and file handles). Both of those features should first be added to U-Boot, so EFI can then use that support. In general, EFI has tried to work beside driver model, creating its own parallel tables, etc. I have tried to influence this at various points along the way, including at the start and I'm happy to dig out those threads if it helps. But I wasn't kidding. it really needs to be addressed. I would love to see Linaro (for example) organise something here and take this on. I am very happy to help.
Regards,
Simon
Quoted 86 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64
I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Quoted 8 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
Quoted 6 lines
What exactly is impossible? Can you please be more specific?
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
FIT and EFI are orthogonal. A FIT image can contain an EFI binary. See CONFIG_BOOTM_EFI.
Quoted 29 lines
Quoted 21 lines
Quoted 9 lines
Quoted 8 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.Quoted 6 lines
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.To be clear here, I like
EFI_LOADER. I too do wish some other technologies had become dominant for technical rather than inertia reasons, but here we are. Having played around with it on aarch64, there are some pretty nice comes-along-with parts to it.Quoted 7 lines
What I hadn't seen, and am only a little skeptical of still, is how far backwards in generations it's going to be used on. The general wish is that users nor off the shelf OS groups need to rebuild U-Boot for a given board, and instead it just works. The number of new designs for 32bit parts is no where near the number of new designs for 64bit parts. So what we're seeing in U-Boot now is people updating support on their older designs, and not necessarily caring about using
EFI_LOADER.
Quoted 5 lines
In a reply to one of the patches in this series, Heinrich mentions a few problems that need resolving (devices for partitions and file handles). Both of those features should first be added to U-Boot, so EFI can then use that support. In general, EFI has tried to work beside driver model, creating its own parallel tables, etc.
Could you, please, be a bit more specific. Please, indicate which data structures you would like to integrate into the driver model.
EDK II shows that a driver model completely based on UEFI protocols is possible. As long as we don't follow that road we will have separate APIs for the UEFI world and U-Boot's internal use and we will have to keep track of objects that exist only in the one world or the other.
What we definitively need is a better integration between probing and removing U-Boot devices and their representation in the UEFI world.
Another area that might deserve attention is memory allocation. AllocatePool() consuming whole memory pages is quite a waste.
Quoted 6 lines
I have
tried to influence this at various points along the way, including at the start and I'm happy to dig out those threads if it helps. But I wasn't kidding. it really needs to be addressed. I would love to see Linaro (for example) organise something here and take this on. I am very happy to help.
The starting point would be comparing the DM and UEFI object models and pointing out which objects can be mapped and which cannot.
Best regards
Heinrich
Hi Heinrich,
Quoted 92 lines · Heinrich Schuchardt
Quoted 90 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
Quoted 8 lines
Quoted 3 lines
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Well there is no point in having two files if one will do. Also it allows for a hash / signature check.
Quoted 11 lines
Quoted 11 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
What exactly is impossible? Can you please be more specific?
Quoted 7 lines
Quoted 4 lines
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
FIT and EFI are orthogonal. A FIT image can contain an EFI binary. See
CONFIG_BOOTM_EFI.
(I think that is a different topic; FIT can of course contain any image)
Quoted 30 lines
Quoted 30 lines
Quoted 29 lines
Quoted 10 lines
Quoted 9 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.Quoted 6 lines
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.To be clear here, I like
EFI_LOADER. I too do wish some other technologies had become dominant for technical rather than inertia reasons, but here we are. Having played around with it on aarch64, there are some pretty nice comes-along-with parts to it.What I hadn't seen, and am only a little skeptical of still, is how far backwards in generations it's going to be used on. The general wish is that users nor off the shelf OS groups need to rebuild U-Boot for a given board, and instead it just works. The number of new designs for 32bit parts is no where near the number of new designs for 64bit parts. So what we're seeing in U-Boot now is people updating support on their older designs, and not necessarily caring about using
EFI_LOADER.
Quoted 8 lines
Quoted 5 lines
In a reply to one of the patches in this series, Heinrich mentions a few problems that need resolving (devices for partitions and file handles). Both of those features should first be added to U-Boot, so EFI can then use that support. In general, EFI has tried to work beside driver model, creating its own parallel tables, etc.
Could you, please, be a bit more specific. Please, indicate which data structures you would like to integrate into the driver model.
As a starting point, all the struct efi_device_path* things should be generated on the fly from DM, rather than existing in parallel.
Quoted 8 lines
EDK II shows that a driver model completely based on UEFI protocols is possible. As long as we don't follow that road we will have separate APIs for the UEFI world and U-Boot's internal use and we will have to keep track of objects that exist only in the one world or the other.
What we definitively need is a better integration between probing and removing U-Boot devices and their representation in the UEFI world.
That is the crux of the problem. The approach taken by UEFI (of duplicating everything) needs to change.
Quoted 11 lines
Another area that might deserve attention is memory allocation.
AllocatePool()consuming whole memory pages is quite a waste.Quoted 6 lines
I have
tried to influence this at various points along the way, including at the start and I'm happy to dig out those threads if it helps. But I wasn't kidding. it really needs to be addressed. I would love to see Linaro (for example) organise something here and take this on. I am very happy to help.
Quoted 2 lines
The starting point would be comparing the DM and UEFI object models and pointing out which objects can be mapped and which cannot.
In my book there are three categories:
- can
- cannot at present, but should be soon
- should not
Regards,
Simon
Quoted 164 lines · Simon Glass
Hi Heinrich,
Quoted 92 lines · Heinrich Schuchardt
Quoted 90 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
Quoted 8 lines
Quoted 3 lines
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Well there is no point in having two files if one will do. Also it allows for a hash / signature check.
Quoted 11 lines
Quoted 11 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
What exactly is impossible? Can you please be more specific?
Quoted 7 lines
Quoted 4 lines
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
FIT and EFI are orthogonal. A FIT image can contain an EFI binary. See
CONFIG_BOOTM_EFI.(I think that is a different topic; FIT can of course contain any image)
Quoted 36 lines
Quoted 35 lines
Quoted 22 lines
Quoted 10 lines
Quoted 9 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.Quoted 6 lines
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.To be clear here, I like
EFI_LOADER. I too do wish some other technologies had become dominant for technical rather than inertia reasons, but here we are. Having played around with it on aarch64, there are some pretty nice comes-along-with parts to it.Quoted 7 lines
What I hadn't seen, and am only a little skeptical of still, is how far backwards in generations it's going to be used on. The general wish is that users nor off the shelf OS groups need to rebuild U-Boot for a given board, and instead it just works. The number of new designs for 32bit parts is no where near the number of new designs for 64bit parts. So what we're seeing in U-Boot now is people updating support on their older designs, and not necessarily caring about using
EFI_LOADER.In a reply to one of the patches in this series, Heinrich mentions a few problems that need resolving (devices for partitions and file handles). Both of those features should first be added to U-Boot, so EFI can then use that support. In general, EFI has tried to work beside driver model, creating its own parallel tables, etc.
Quoted 5 lines
Quoted 2 lines
Could you, please, be a bit more specific. Please, indicate which data structures you would like to integrate into the driver model.
As a starting point, all the struct efi_device_path* things should be generated on the fly from DM, rather than existing in parallel.
The driver model does not describe all devices in the lifetime of an UEFI application. Applications like iPXE create handles and install there own device paths on them.
When a U-Boot device is probed a UEFI handle should be created and the device path protocol has to be installed. I hope this is what you mean by "on the fly".
Be aware that the device path protocol can be replaced by a UEFI application by calling ReinstallProtocol() at any time. So to be UEFI compliant you cannot create device paths upon access. Let's assume you did not mean this by "on the fly".
After all device paths have to exist as data structures in parallel to the currrent driver model. But we can change how we create them.
Quoted 24 lines
Quoted 9 lines
EDK II shows that a driver model completely based on UEFI protocols is possible. As long as we don't follow that road we will have separate APIs for the UEFI world and U-Boot's internal use and we will have to keep track of objects that exist only in the one world or the other.
What we definitively need is a better integration between probing and removing U-Boot devices and their representation in the UEFI world.
That is the crux of the problem. The approach taken by UEFI (of duplicating everything) needs to change.
Quoted 11 lines
Another area that might deserve attention is memory allocation.
AllocatePool()consuming whole memory pages is quite a waste.Quoted 6 lines
I have
tried to influence this at various points along the way, including at the start and I'm happy to dig out those threads if it helps. But I wasn't kidding. it really needs to be addressed. I would love to see Linaro (for example) organise something here and take this on. I am very happy to help.
Quoted 9 lines
Quoted 2 lines
The starting point would be comparing the DM and UEFI object models and pointing out which objects can be mapped and which cannot.
In my book there are three categories:
- can
- cannot at present, but should be soon
- should not
'Cannot be mapped because not defined in the UEFI spec' is missing in your scheme. E.g. there is a RNG protocol but no such thing as a RNG device defined in the UEFI spec.
Best regards
Heinrich
Hi Heinrich,
Quoted 169 lines · Heinrich Schuchardt
Quoted 167 lines · Simon Glass
Hi Heinrich,
Quoted 92 lines · Heinrich Schuchardt
Quoted 90 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
Quoted 8 lines
Quoted 3 lines
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Well there is no point in having two files if one will do. Also it allows for a hash / signature check.
Quoted 11 lines
Quoted 11 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
What exactly is impossible? Can you please be more specific?
Quoted 7 lines
Quoted 4 lines
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
FIT and EFI are orthogonal. A FIT image can contain an EFI binary. See
CONFIG_BOOTM_EFI.(I think that is a different topic; FIT can of course contain any image)
Quoted 38 lines
Quoted 30 lines
Quoted 22 lines
Quoted 10 lines
Quoted 9 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.Quoted 6 lines
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.To be clear here, I like
EFI_LOADER. I too do wish some other technologies had become dominant for technical rather than inertia reasons, but here we are. Having played around with it on aarch64, there are some pretty nice comes-along-with parts to it.Quoted 7 lines
What I hadn't seen, and am only a little skeptical of still, is how far backwards in generations it's going to be used on. The general wish is that users nor off the shelf OS groups need to rebuild U-Boot for a given board, and instead it just works. The number of new designs for 32bit parts is no where near the number of new designs for 64bit parts. So what we're seeing in U-Boot now is people updating support on their older designs, and not necessarily caring about using
EFI_LOADER.Quoted 5 lines
In a reply to one of the patches in this series, Heinrich mentions a few problems that need resolving (devices for partitions and file handles). Both of those features should first be added to U-Boot, so EFI can then use that support. In general, EFI has tried to work beside driver model, creating its own parallel tables, etc.
Could you, please, be a bit more specific. Please, indicate which data structures you would like to integrate into the driver model.
Quoted 6 lines
Quoted 2 lines
As a starting point, all the struct efi_device_path* things should be generated on the fly from DM, rather than existing in parallel.
The driver model does not describe all devices in the lifetime of an UEFI application. Applications like iPXE create handles and install there own device paths on them.
That is the core of the problem. If UEFI wants to create something, it should do so using core DM calls, not build parallel structures.
Quoted 5 lines
When a U-Boot device is probed a UEFI handle should be created and the device path protocol has to be installed. I hope this is what you mean by "on the fly".
Quoted 7 lines
Be aware that the device path protocol can be replaced by a UEFI application by calling
ReinstallProtocol()at any time. So to be UEFI compliant you cannot create device paths upon access. Let's assume you did not mean this by "on the fly".After all device paths have to exist as data structures in parallel to the currrent driver model. But we can change how we create them.
This is the piece that is incorrect, I think. U-Boot should encompass the required features so that there is no need for these tables, except in special situations. So rather than EFI being a bolt-on, it becomes more like another view. For example, devices can have some extra EFI information, if it truly is useless for DM (e.g. UUIDs), but it should be attached to the device, not in a parallel table.
By 'on the fly' I mean that in most cases, EFI requests can be satisfied by doing DM calls and by providing DM information in a format that EFI expects. With that as a general principle, there will be less duplication and more consistency.
But getting rid of storing struct efi_device_path... would be a good exercise to nut out the problems.
Quoted 30 lines
Quoted 30 lines
Quoted 10 lines
EDK II shows that a driver model completely based on UEFI protocols is possible. As long as we don't follow that road we will have separate APIs for the UEFI world and U-Boot's internal use and we will have to keep track of objects that exist only in the one world or the other.
What we definitively need is a better integration between probing and removing U-Boot devices and their representation in the UEFI world.
That is the crux of the problem. The approach taken by UEFI (of duplicating everything) needs to change.
Quoted 11 lines
Another area that might deserve attention is memory allocation.
AllocatePool()consuming whole memory pages is quite a waste.Quoted 6 lines
I have
tried to influence this at various points along the way, including at the start and I'm happy to dig out those threads if it helps. But I wasn't kidding. it really needs to be addressed. I would love to see Linaro (for example) organise something here and take this on. I am very happy to help.Quoted 2 lines
The starting point would be comparing the DM and UEFI object models and pointing out which objects can be mapped and which cannot.
In my book there are three categories:
Quoted 8 lines
Quoted 4 lines
- can
- cannot at present, but should be soon
- should not'Cannot be mapped because not defined in the UEFI spec' is missing in your scheme. E.g. there is a RNG protocol but no such thing as a RNG device defined in the UEFI spec.
Are you talking about things not in the spec? If so, I don't think they are relevant to this discussion. Or are you just trying to be fully complete?
Regards,
Simon
Quoted 99 lines · Simon Glass
Hi Heinrich,
Quoted 96 lines · Heinrich Schuchardt
Quoted 93 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64
Quoted 7 lines
Quoted 4 lines
I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Well there is no point in having two files if one will do. Also it allows for a hash / signature check.
The question of "how great would it be and how many problems would it have solved if FIT images had become popular" is one for another time. It will always have its use cases and users but never the broad adoption many of us felt it should have. Bringing it up in this context won't change that.
I'm saying this because I think there are some important technical questions within U-Boot to resolve because the EFI loader part of U-Boot is critical to our long term future. And DM is an important part of our internal design and we're (probably later than I should have) pulling out the parts that haven't been updated so that we can deliver on some of the overall promise of DM better, too.
Signature · 1 line
Tom
Hi Tom,
Quoted 106 lines · Tom Rini
Quoted 104 lines · Simon Glass
Hi Heinrich,
Quoted 100 lines · Heinrich Schuchardt
Quoted 86 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Quoted 8 lines
Quoted 2 lines
Well there is no point in having two files if one will do. Also it allows for a hash / signature check.
The question of "how great would it be and how many problems would it have solved if FIT images had become popular" is one for another time. It will always have its use cases and users but never the broad adoption many of us felt it should have. Bringing it up in this context won't change that.
I see Peter's reply below so will make time to dig into this and understand the problems with FIT better. I feel that EFI comes with all sorts of problems so I'm far from convinced, at this point. Sorry.
Quoted 7 lines
I'm saying this because I think there are some important technical questions within U-Boot to resolve because the EFI loader part of U-Boot is critical to our long term future. And DM is an important part of our internal design and we're (probably later than I should have) pulling out the parts that haven't been updated so that we can deliver on some of the overall promise of DM better, too.
Yes, migration has certainly been slow. As you know I mostly stopped pushing it a few years back when I saw all the reluctance. I'm very pleased you are taking that on and I think it will help a lot.
If what you say comes to pass, it is even more important that EFI is more integrated, rather than being a bolt on. Thanks largely to Heinrich, the tests are in quite good shape, so refactoring should be feasible.
Regards,
Simon
Quoted 118 lines · Simon Glass
Hi Tom,
Quoted 106 lines · Tom Rini
Quoted 104 lines · Simon Glass
Hi Heinrich,
Quoted 100 lines · Heinrich Schuchardt
Quoted 86 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Quoted 8 lines
Quoted 2 lines
Well there is no point in having two files if one will do. Also it allows for a hash / signature check.
The question of "how great would it be and how many problems would it have solved if FIT images had become popular" is one for another time. It will always have its use cases and users but never the broad adoption many of us felt it should have. Bringing it up in this context won't change that.
Quoted 3 lines
I see Peter's reply below so will make time to dig into this and understand the problems with FIT better. I feel that EFI comes with all sorts of problems so I'm far from convinced, at this point. Sorry.
It seems to me that we are discussing three different things:
- the code size increase by enabling UEFI interfaces
- how the UEFI interface be implemented on U-Boot
- The primary (or default/standard) boot mechanism in the future
I don't think they are totally independent, but we'd better distinguish them some how in the following discussions.
Quoted 11 lines
Quoted 7 lines
I'm saying this because I think there are some important technical questions within U-Boot to resolve because the EFI loader part of U-Boot is critical to our long term future. And DM is an important part of our internal design and we're (probably later than I should have) pulling out the parts that haven't been updated so that we can deliver on some of the overall promise of DM better, too.
Yes, migration has certainly been slow. As you know I mostly stopped pushing it a few years back when I saw all the reluctance. I'm very pleased you are taking that on and I think it will help a lot.
I posted this patch[1] two years ago and I thought that we had had some sort of consensus that UEFI interfaces be integrated more elegantly with DM in the future.
So I was a bit surprised with Heinrich's recent patch.
In [1], I was trying to define all the UEFI handles, including some protocols?, as DM objects.
I thought that it was the best way for unifying the two worlds even if there are no corresponding *notions* in the existing DM objects.
[1] https://lists.denx.de/pipermail/u-boot/2019-February/357923.html
-Takahiro Akashi
Quoted 7 lines
If what you say comes to pass, it is even more important that EFI is more integrated, rather than being a bolt on. Thanks largely to Heinrich, the tests are in quite good shape, so refactoring should be feasible.
Regards,
Simon
Quoted 149 lines · AKASHI Takahiro
Quoted 118 lines · Simon Glass
Hi Tom,
Quoted 106 lines · Tom Rini
Quoted 104 lines · Simon Glass
Hi Heinrich,
Quoted 100 lines · Heinrich Schuchardt
Quoted 86 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Quoted 8 lines
Quoted 2 lines
Well there is no point in having two files if one will do. Also it allows for a hash / signature check.
The question of "how great would it be and how many problems would it have solved if FIT images had become popular" is one for another time. It will always have its use cases and users but never the broad adoption many of us felt it should have. Bringing it up in this context won't change that.
Quoted 3 lines
I see Peter's reply below so will make time to dig into this and understand the problems with FIT better. I feel that EFI comes with all sorts of problems so I'm far from convinced, at this point. Sorry.
It seems to me that we are discussing three different things:
- the code size increase by enabling UEFI interfaces
- how the UEFI interface be implemented on U-Boot
- The primary (or default/standard) boot mechanism in the futureI don't think they are totally independent, but we'd better distinguish them some how in the following discussions.
Quoted 11 lines
Quoted 7 lines
I'm saying this because I think there are some important technical questions within U-Boot to resolve because the EFI loader part of U-Boot is critical to our long term future. And DM is an important part of our internal design and we're (probably later than I should have) pulling out the parts that haven't been updated so that we can deliver on some of the overall promise of DM better, too.
Yes, migration has certainly been slow. As you know I mostly stopped pushing it a few years back when I saw all the reluctance. I'm very pleased you are taking that on and I think it will help a lot.
I posted this patch[1] two years ago and I thought that we had had some sort of consensus that UEFI interfaces be integrated more elegantly with DM in the future.
So I was a bit surprised with Heinrich's recent patch.
Quoted 8 lines
In [1], I was trying to define all the UEFI handles, including some protocols?, as DM objects.
I thought that it was the best way for unifying the two worlds even if there are no corresponding *notions* in the existing DM objects.[1] https://lists.denx.de/pipermail/u-boot/2019-February/357923.html
-Takahiro Akashi
You wrote yourself: "bootefi doesn't work with this patch set yet".
Your series completely disregarded UEFI and DM logic, e.g. you defined DM devices per protocol.
You tried to integrate UEFI and DM world at an inappropriate level:
Instead of changing DM block device uclass you touched individual drivers like USB and SCSI completely disregarding all other block device classes.
Best regards
Heinrich
Quoted 9 lines
Quoted 9 lines
If what you say comes to pass, it is even more important that EFI is more integrated, rather than being a bolt on. Thanks largely to Heinrich, the tests are in quite good shape, so refactoring should be feasible.
Regards,
Simon
Quoted 161 lines · Heinrich Schuchardt
Quoted 149 lines · AKASHI Takahiro
Quoted 118 lines · Simon Glass
Hi Tom,
Quoted 106 lines · Tom Rini
Quoted 104 lines · Simon Glass
Hi Heinrich,
Quoted 100 lines · Heinrich Schuchardt
Quoted 86 lines · Simon Glass
Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Secondly, I don't expect that U-Boot itself would be in the FIT.
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I have not hit any distro that builds FIT images. All install vmlinux and initrd as separate files.
Why would you want to change that?
Quoted 8 lines
Quoted 2 lines
Well there is no point in having two files if one will do. Also it allows for a hash / signature check.
The question of "how great would it be and how many problems would it have solved if FIT images had become popular" is one for another time. It will always have its use cases and users but never the broad adoption many of us felt it should have. Bringing it up in this context won't change that.
Quoted 3 lines
I see Peter's reply below so will make time to dig into this and understand the problems with FIT better. I feel that EFI comes with all sorts of problems so I'm far from convinced, at this point. Sorry.
It seems to me that we are discussing three different things:
- the code size increase by enabling UEFI interfaces
- how the UEFI interface be implemented on U-Boot
- The primary (or default/standard) boot mechanism in the futureI don't think they are totally independent, but we'd better distinguish them some how in the following discussions.
Quoted 11 lines
Quoted 7 lines
I'm saying this because I think there are some important technical questions within U-Boot to resolve because the EFI loader part of U-Boot is critical to our long term future. And DM is an important part of our internal design and we're (probably later than I should have) pulling out the parts that haven't been updated so that we can deliver on some of the overall promise of DM better, too.
Yes, migration has certainly been slow. As you know I mostly stopped pushing it a few years back when I saw all the reluctance. I'm very pleased you are taking that on and I think it will help a lot.
I posted this patch[1] two years ago and I thought that we had had some sort of consensus that UEFI interfaces be integrated more elegantly with DM in the future.
So I was a bit surprised with Heinrich's recent patch.
Quoted 8 lines
In [1], I was trying to define all the UEFI handles, including some protocols?, as DM objects.
I thought that it was the best way for unifying the two worlds even if there are no corresponding *notions* in the existing DM objects.[1] https://lists.denx.de/pipermail/u-boot/2019-February/357923.html
-Takahiro Akashi
You wrote yourself: "bootefi doesn't work with this patch set yet".
Quoted 7 lines
Your series completely disregarded UEFI and DM logic, e.g. you defined DM devices per protocol.
You tried to integrate UEFI and DM world at an inappropriate level:
Instead of changing DM block device uclass you touched individual drivers like USB and SCSI completely disregarding all other block device classes.
Yes, I can agree, but the point is not there. Why not discuss how UEFI and DM be integrated instead as Simon suggested?
-Takahiro Akashi
Quoted 14 lines
Best regards
Heinrich
Quoted 9 lines
Quoted 9 lines
If what you say comes to pass, it is even more important that EFI is more integrated, rather than being a bolt on. Thanks largely to Heinrich, the tests are in quite good shape, so refactoring should be feasible.
Regards,
Simon
Quoted 81 lines
From: Simon Glass <[email protected]>
Date: Mon, 28 Jun 2021 08:18:26 -0600Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.
Quoted 5 lines
I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Well, you could of course rework the boot flow of other OSes such that booting them includes some sort of FIT if you really wanted to. I But an OS like OpenBSD comes with its own bootloader that is essential in the boot flow. On OpenBSD armv7/arm64/riscv64 it adds some essential properties to the device tree. Besides, the kernel itself relies on a valid EFI memory map.
Quoted 1 line
Secondly, I don't expect that U-Boot itself would be in the FIT.
So the FIT would only contain the OS kernel and other OS components? What about the FIT that is used on some arm64 platforms to combine U-Boot and TF-A? I guess you can have multiple FITs...
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64
I don't really use Linux on these platforms. But I'd expect the normal package management tools of my Linux distribution to build those as necessary and place them in the root file system where the bootloader picks them up. And as a distro developer, I'd like to have the same approach work on all Linux systems, regardless the specific firmware they're running (EDK2, U-Boot or something completely different). Ideally that wouldn't even depend on the architecture.
Now Armbian takes a different approach, and does treat most systems they provide as special snowflakes, providing flash images for each board. But that doesn't scale and makes for a fairly crappy user experience. They don't always support booting a mainline kernel. And updating the kernel often requires installing special packages.
Quoted 6 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
I guess I mean system firmware. Essentially everything that runs on the system before my OS bootloader runs. So for me, U-Boot is part of the system firmware even if it sometimes happens to live on removable media.
Quoted 1 line
What exactly is impossible? Can you please be more specific?
So let me explain what we want for OpenBSD. We really want a uniform user experience across platforms, and don't want to maintain different codepaths for special snowflake platforms that might exist for a specific architecture.
Installing OpenBSD on a machine should be as simple as dd'ing the installer to some boot media, plugging it in and powering the machine on. Now this is somewhat tricky to achieve on some hardware targetted by U-Boot as they don't come with usable system firmware on the board itself. But on those boards you can mostly get away with having U-Boot on uSD or eMMC and the OS installer on USB.
Updating the OpenBSD kernel should be as simple as copying the kernel to /bsd. Since the root filesystem uses the UFS/FFS filesystem, this means that whatever we use as a bootloader must be able to read from that filesystem. To make sure the kernel is properly seeded with entropy, the OpenBSD bootloader has some additional tricks up its sleeve. It will replace a special segment in the kernel with random data before handing control to the kernel. On platforms that support it, it will try to use a firmware-provided RNG to do this (and EFI supports this) but also mix in some random data from a file on the UFS/FFS filesystem. It will actually mark that file as "used" after reading it to throw a warning when the file is reused when the machine is rebooted (it should have been filled with fresh new entropy). So you really need to use the OpenBSD bootloader instead loading an OpenBSD kernel directly from system firmware.
Updating the OpenBSD bootloader should be as simple as running installboot(8) from within the OS.
This all works on pretty much any architecture that OpenBSD supports. And right now, thanks to EFI_LOADER support in U-Boot, we have a nearly uniform boot flow on amd64, arm64, armv7 and riscv64.
Quoted 4 lines
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
It just doesn't make sense for us to use a container just because the system firmware (U-Boot) insists on it. The kernel lives in the root UFS/FFS filesystem and the bootloader lives on an MS-DOS filesystem on the root disk.
Quoted 41 lines
Quoted 19 lines
Quoted 7 lines
Quoted 6 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.Quoted 6 lines
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.To be clear here, I like
EFI_LOADER. I too do wish some other technologies had become dominant for technical rather than inertia reasons, but here we are. Having played around with it on aarch64, there are some pretty nice comes-along-with parts to it.Quoted 7 lines
What I hadn't seen, and am only a little skeptical of still, is how far backwards in generations it's going to be used on. The general wish is that users nor off the shelf OS groups need to rebuild U-Boot for a given board, and instead it just works. The number of new designs for 32bit parts is no where near the number of new designs for 64bit parts. So what we're seeing in U-Boot now is people updating support on their older designs, and not necessarily caring about using
EFI_LOADER.In a reply to one of the patches in this series, Heinrich mentions a few problems that need resolving (devices for partitions and file handles). Both of those features should first be added to U-Boot, so EFI can then use that support. In general, EFI has tried to work beside driver model, creating its own parallel tables, etc. I have tried to influence this at various points along the way, including at the start and I'm happy to dig out those threads if it helps. But I wasn't kidding. it really needs to be addressed. I would love to see Linaro (for example) organise something here and take this on. I am very happy to help.
Regards,
Simon
Hi Mark,
Quoted 88 lines · Mark Kettenis
Quoted 82 lines
From: Simon Glass <[email protected]>
Date: Mon, 28 Jun 2021 08:18:26 -0600Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.Quoted 5 lines
I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Quoted 6 lines
Well, you could of course rework the boot flow of other OSes such that booting them includes some sort of FIT if you really wanted to. I But an OS like OpenBSD comes with its own bootloader that is essential in the boot flow. On OpenBSD armv7/arm64/riscv64 it adds some essential properties to the device tree. Besides, the kernel itself relies on a valid EFI memory map.
(thanks for taking the time to reply with so much detail)
That's news to me; when did that happen? Anyway, I doubt that adds a lot of code.
Quoted 23 lines
Quoted 2 lines
Secondly, I don't expect that U-Boot itself would be in the FIT.
So the FIT would only contain the OS kernel and other OS components? What about the FIT that is used on some arm64 platforms to combine U-Boot and TF-A? I guess you can have multiple FITs...
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I don't really use Linux on these platforms. But I'd expect the normal package management tools of my Linux distribution to build those as necessary and place them in the root file system where the bootloader picks them up. And as a distro developer, I'd like to have the same approach work on all Linux systems, regardless the specific firmware they're running (EDK2, U-Boot or something completely different). Ideally that wouldn't even depend on the architecture.
Quoted 5 lines
Now Armbian takes a different approach, and does treat most systems they provide as special snowflakes, providing flash images for each board. But that doesn't scale and makes for a fairly crappy user experience. They don't always support booting a mainline kernel. And updating the kernel often requires installing special packages.
I don't think it is a requirement that things have to be special snowflakes. There are a few common boot flows to support and we can put that support in U-Boot and in userspace, without needing to make everything special.
Quoted 42 lines
Quoted 7 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
I guess I mean system firmware. Essentially everything that runs on the system before my OS bootloader runs. So for me, U-Boot is part of the system firmware even if it sometimes happens to live on removable media.
Quoted 1 line
What exactly is impossible? Can you please be more specific?
So let me explain what we want for OpenBSD. We really want a uniform user experience across platforms, and don't want to maintain different codepaths for special snowflake platforms that might exist for a specific architecture.
Installing OpenBSD on a machine should be as simple as dd'ing the installer to some boot media, plugging it in and powering the machine on. Now this is somewhat tricky to achieve on some hardware targetted by U-Boot as they don't come with usable system firmware on the board itself. But on those boards you can mostly get away with having U-Boot on uSD or eMMC and the OS installer on USB.
Updating the OpenBSD kernel should be as simple as copying the kernel to /bsd. Since the root filesystem uses the UFS/FFS filesystem, this means that whatever we use as a bootloader must be able to read from that filesystem. To make sure the kernel is properly seeded with entropy, the OpenBSD bootloader has some additional tricks up its sleeve. It will replace a special segment in the kernel with random data before handing control to the kernel. On platforms that support it, it will try to use a firmware-provided RNG to do this (and EFI supports this) but also mix in some random data from a file on the UFS/FFS filesystem. It will actually mark that file as "used" after reading it to throw a warning when the file is reused when the machine is rebooted (it should have been filled with fresh new entropy). So you really need to use the OpenBSD bootloader instead loading an OpenBSD kernel directly from system firmware.
Quoted 6 lines
Updating the OpenBSD bootloader should be as simple as running installboot(8) from within the OS.
This all works on pretty much any architecture that OpenBSD supports. And right now, thanks to
EFI_LOADERsupport in U-Boot, we have a nearly uniform boot flow on amd64, arm64, armv7 and riscv64.
OK I see. Really it sounds like you have a pre-kernel loader. The functionality (some fresh random data) could just as easily be provided by U-Boot directly, with vastly less code and complexity. But I do understand that trying to design across a whole system is a pain, and it is attractive to try to use what hooks exist already to do what you want.
How do you do verified boot?
Quoted 10 lines
Quoted 5 lines
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
It just doesn't make sense for us to use a container just because the system firmware (U-Boot) insists on it. The kernel lives in the root UFS/FFS filesystem and the bootloader lives on an MS-DOS filesystem on the root disk.
It isn't so much that U-Boot insists on it. It is quite happy to load a kernel and initrd separately. But it does make verification harder and I assume it makes upgrades harder since there are multiple files to install. FIT is just such a nice format for packaging kernels and related things. It sounds like you could use FIT and everything you said above would still be true.
[..]
Regards,
Simon
Quoted 108 lines
From: Simon Glass <[email protected]>
Date: Fri, 2 Jul 2021 13:05:20 -0600
Cc: Tom Rini <[email protected]>, U-Boot Mailing List <[email protected]>,
Pali Rohár <[email protected]>,
Heinrich Schuchardt <[email protected]>,
Ilias Apalodimas <[email protected]>,
Alex Graf <[email protected]>,
Masahiro Yamada <[email protected]>
Content-Type: text/plain; charset="UTF-8"Hi Mark,
Quoted 88 lines · Mark Kettenis
Quoted 82 lines
From: Simon Glass <[email protected]>
Date: Mon, 28 Jun 2021 08:18:26 -0600Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.Quoted 5 lines
I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Quoted 6 lines
Well, you could of course rework the boot flow of other OSes such that booting them includes some sort of FIT if you really wanted to. I But an OS like OpenBSD comes with its own bootloader that is essential in the boot flow. On OpenBSD armv7/arm64/riscv64 it adds some essential properties to the device tree. Besides, the kernel itself relies on a valid EFI memory map.
Quoted 4 lines
(thanks for taking the time to reply with so much detail)
That's news to me; when did that happen? Anyway, I doubt that adds a lot of code.
Shortly after EFI_LOADER support was added to U-Boot and there was a clear consensus that EFI_LOADER support was going to be enabled by default on armv7 and armv8 targets. My initial commit of the EFI bootloader for armv7 is from May 2016.
Quoted 82 lines
Quoted 22 lines
Quoted 1 line
Secondly, I don't expect that U-Boot itself would be in the FIT.
So the FIT would only contain the OS kernel and other OS components? What about the FIT that is used on some arm64 platforms to combine U-Boot and TF-A? I guess you can have multiple FITs...
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I don't really use Linux on these platforms. But I'd expect the normal package management tools of my Linux distribution to build those as necessary and place them in the root file system where the bootloader picks them up. And as a distro developer, I'd like to have the same approach work on all Linux systems, regardless the specific firmware they're running (EDK2, U-Boot or something completely different). Ideally that wouldn't even depend on the architecture.
Quoted 5 lines
Now Armbian takes a different approach, and does treat most systems they provide as special snowflakes, providing flash images for each board. But that doesn't scale and makes for a fairly crappy user experience. They don't always support booting a mainline kernel. And updating the kernel often requires installing special packages.
I don't think it is a requirement that things have to be special snowflakes. There are a few common boot flows to support and we can put that support in U-Boot and in userspace, without needing to make everything special.
Quoted 42 lines
Quoted 7 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
I guess I mean system firmware. Essentially everything that runs on the system before my OS bootloader runs. So for me, U-Boot is part of the system firmware even if it sometimes happens to live on removable media.
Quoted 1 line
What exactly is impossible? Can you please be more specific?
So let me explain what we want for OpenBSD. We really want a uniform user experience across platforms, and don't want to maintain different codepaths for special snowflake platforms that might exist for a specific architecture.
Installing OpenBSD on a machine should be as simple as dd'ing the installer to some boot media, plugging it in and powering the machine on. Now this is somewhat tricky to achieve on some hardware targetted by U-Boot as they don't come with usable system firmware on the board itself. But on those boards you can mostly get away with having U-Boot on uSD or eMMC and the OS installer on USB.
Updating the OpenBSD kernel should be as simple as copying the kernel to /bsd. Since the root filesystem uses the UFS/FFS filesystem, this means that whatever we use as a bootloader must be able to read from that filesystem. To make sure the kernel is properly seeded with entropy, the OpenBSD bootloader has some additional tricks up its sleeve. It will replace a special segment in the kernel with random data before handing control to the kernel. On platforms that support it, it will try to use a firmware-provided RNG to do this (and EFI supports this) but also mix in some random data from a file on the UFS/FFS filesystem. It will actually mark that file as "used" after reading it to throw a warning when the file is reused when the machine is rebooted (it should have been filled with fresh new entropy). So you really need to use the OpenBSD bootloader instead loading an OpenBSD kernel directly from system firmware.
Quoted 6 lines
Updating the OpenBSD bootloader should be as simple as running installboot(8) from within the OS.
This all works on pretty much any architecture that OpenBSD supports. And right now, thanks to
EFI_LOADERsupport in U-Boot, we have a nearly uniform boot flow on amd64, arm64, armv7 and riscv64.
Quoted 6 lines
OK I see. Really it sounds like you have a pre-kernel loader. The functionality (some fresh random data) could just as easily be provided by U-Boot directly, with vastly less code and complexity. But I do understand that trying to design across a whole system is a pain, and it is attractive to try to use what hooks exist already to do what you want.
What do you mean with vastly less code and complexity. At this point 70% of the OpenBSD bootloader code is shared between most of the architectures OpenBSD runs on (alpha, amd64, arm64, armv7, hppa, i386, powerpc, riscv64, sh, sparc64). And on platforms that support UEFI (amd64, arm64, armv7 and riscv64) this is closer to 95% shared code.
Having OS-dependent code in U-Boot doesn't work. FreeBSD tried that with the U-Boot API. It was never enabled by default, and got broken all the time.
Quoted 1 line
How do you do verified boot?
We don't. I don't think it makes sense for an Open Source OS where people like to tinker with things. And it gets in the way of some of our own security features. OpenBSD relinks the kernel upon every boot as a defence against attacks that depend on the kernel address space layout. Doing that in an environment where all code needs to be signed is a big challenge.
There is a comany that has products based on OpenBSD. They do verified boot. I'm not really familliar with the details, but presumably they turned off address space randomization and I believe it is based on EFI_LOADER support. Patrick Wildt contributed patches to implement this to U-Boot before the Linaro folks did their full-blown UEFI-based implementation.
Quoted 10 lines
Quoted 9 lines
Quoted 4 lines
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
It just doesn't make sense for us to use a container just because the system firmware (U-Boot) insists on it. The kernel lives in the root UFS/FFS filesystem and the bootloader lives on an MS-DOS filesystem on the root disk.
Quoted 6 lines
It isn't so much that U-Boot insists on it. It is quite happy to load a kernel and initrd separately. But it does make verification harder and I assume it makes upgrades harder since there are multiple files to install. FIT is just such a nice format for packaging kernels and related things. It sounds like you could use FIT and everything you said above would still be true.
It would mean a kernel update would need to update the FIT container. But only on boards that use U-Boot to boot, since nobody else supports it. That isn't helpful to us.
Cheers,
Mark
Quoted 213 lines · Mark Kettenis
Quoted 108 lines
From: Simon Glass <[email protected]>
Date: Fri, 2 Jul 2021 13:05:20 -0600
Cc: Tom Rini <[email protected]>, U-Boot Mailing List <[email protected]>,
Pali Rohár <[email protected]>,
Heinrich Schuchardt <[email protected]>,
Ilias Apalodimas <[email protected]>,
Alex Graf <[email protected]>,
Masahiro Yamada <[email protected]>
Content-Type: text/plain; charset="UTF-8"Hi Mark,
Quoted 88 lines · Mark Kettenis
Quoted 82 lines
From: Simon Glass <[email protected]>
Date: Mon, 28 Jun 2021 08:18:26 -0600Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.Quoted 5 lines
I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Quoted 6 lines
Well, you could of course rework the boot flow of other OSes such that booting them includes some sort of FIT if you really wanted to. I But an OS like OpenBSD comes with its own bootloader that is essential in the boot flow. On OpenBSD armv7/arm64/riscv64 it adds some essential properties to the device tree. Besides, the kernel itself relies on a valid EFI memory map.
Quoted 4 lines
(thanks for taking the time to reply with so much detail)
That's news to me; when did that happen? Anyway, I doubt that adds a lot of code.
Shortly after
EFI_LOADERsupport was added to U-Boot and there was a clear consensus thatEFI_LOADERsupport was going to be enabled by default on armv7 and armv8 targets. My initial commit of the EFI bootloader for armv7 is from May 2016.Quoted 82 lines
Quoted 22 lines
Quoted 1 line
Secondly, I don't expect that U-Boot itself would be in the FIT.
So the FIT would only contain the OS kernel and other OS components? What about the FIT that is used on some arm64 platforms to combine U-Boot and TF-A? I guess you can have multiple FITs...
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I don't really use Linux on these platforms. But I'd expect the normal package management tools of my Linux distribution to build those as necessary and place them in the root file system where the bootloader picks them up. And as a distro developer, I'd like to have the same approach work on all Linux systems, regardless the specific firmware they're running (EDK2, U-Boot or something completely different). Ideally that wouldn't even depend on the architecture.
Quoted 5 lines
Now Armbian takes a different approach, and does treat most systems they provide as special snowflakes, providing flash images for each board. But that doesn't scale and makes for a fairly crappy user experience. They don't always support booting a mainline kernel. And updating the kernel often requires installing special packages.
I don't think it is a requirement that things have to be special snowflakes. There are a few common boot flows to support and we can put that support in U-Boot and in userspace, without needing to make everything special.
Quoted 42 lines
Quoted 7 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
I guess I mean system firmware. Essentially everything that runs on the system before my OS bootloader runs. So for me, U-Boot is part of the system firmware even if it sometimes happens to live on removable media.
Quoted 1 line
What exactly is impossible? Can you please be more specific?
So let me explain what we want for OpenBSD. We really want a uniform user experience across platforms, and don't want to maintain different codepaths for special snowflake platforms that might exist for a specific architecture.
Installing OpenBSD on a machine should be as simple as dd'ing the installer to some boot media, plugging it in and powering the machine on. Now this is somewhat tricky to achieve on some hardware targetted by U-Boot as they don't come with usable system firmware on the board itself. But on those boards you can mostly get away with having U-Boot on uSD or eMMC and the OS installer on USB.
Updating the OpenBSD kernel should be as simple as copying the kernel to /bsd. Since the root filesystem uses the UFS/FFS filesystem, this means that whatever we use as a bootloader must be able to read from that filesystem. To make sure the kernel is properly seeded with entropy, the OpenBSD bootloader has some additional tricks up its sleeve. It will replace a special segment in the kernel with random data before handing control to the kernel. On platforms that support it, it will try to use a firmware-provided RNG to do this (and EFI supports this) but also mix in some random data from a file on the UFS/FFS filesystem. It will actually mark that file as "used" after reading it to throw a warning when the file is reused when the machine is rebooted (it should have been filled with fresh new entropy). So you really need to use the OpenBSD bootloader instead loading an OpenBSD kernel directly from system firmware.
Quoted 6 lines
Updating the OpenBSD bootloader should be as simple as running installboot(8) from within the OS.
This all works on pretty much any architecture that OpenBSD supports. And right now, thanks to
EFI_LOADERsupport in U-Boot, we have a nearly uniform boot flow on amd64, arm64, armv7 and riscv64.Quoted 6 lines
OK I see. Really it sounds like you have a pre-kernel loader. The functionality (some fresh random data) could just as easily be provided by U-Boot directly, with vastly less code and complexity. But I do understand that trying to design across a whole system is a pain, and it is attractive to try to use what hooks exist already to do what you want.
What do you mean with vastly less code and complexity. At this point 70% of the OpenBSD bootloader code is shared between most of the architectures OpenBSD runs on (alpha, amd64, arm64, armv7, hppa, i386, powerpc, riscv64, sh, sparc64). And on platforms that support UEFI (amd64, arm64, armv7 and riscv64) this is closer to 95% shared code.
Quoted 3 lines
Having OS-dependent code in U-Boot doesn't work. FreeBSD tried that with the U-Boot API. It was never enabled by default, and got broken all the time.
That's good to know and I believe validates why we want EFI_LOADER as much as possible. This is a good example of how real world OSes can make use of this to reduce their own overhead. And I think this also shows why we need EFI_LOADER, the U-boot API was never being sufficiently tested and most importantly never had anyone championing it's usage. That's not the case with the EFI API.
Quoted 9 lines
Quoted 1 line
How do you do verified boot?
We don't. I don't think it makes sense for an Open Source OS where people like to tinker with things. And it gets in the way of some of our own security features. OpenBSD relinks the kernel upon every boot as a defence against attacks that depend on the kernel address space layout. Doing that in an environment where all code needs to be signed is a big challenge.
Quoted 6 lines
There is a comany that has products based on OpenBSD. They do verified boot. I'm not really familliar with the details, but presumably they turned off address space randomization and I believe it is based on
EFI_LOADERsupport. Patrick Wildt contributed patches to implement this to U-Boot before the Linaro folks did their full-blown UEFI-based implementation.
That's good to know, thanks!
Signature · 1 line
Tom
Hi Mark,
Quoted 111 lines · Mark Kettenis
Quoted 111 lines
From: Simon Glass <[email protected]>
Date: Fri, 2 Jul 2021 13:05:20 -0600
Cc: Tom Rini <[email protected]>, U-Boot Mailing List <[email protected]>,
Pali Rohár <[email protected]>,
Heinrich Schuchardt <[email protected]>,
Ilias Apalodimas <[email protected]>,
Alex Graf <[email protected]>,
Masahiro Yamada <[email protected]>
Content-Type: text/plain; charset="UTF-8"Hi Mark,
Quoted 88 lines · Mark Kettenis
Quoted 82 lines
From: Simon Glass <[email protected]>
Date: Mon, 28 Jun 2021 08:18:26 -0600Hi Tom, Mark,
Quoted 36 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.Quoted 9 lines
Quoted 1 line
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
Quoted 5 lines
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Yes it will be interesting to see what you find there. My results on
nokia_rx51were something like this:default
arm: (for 1/1 boards) all +129370.0 bss +1136.0 data +7399.0rodata +10989.0 text +109846.0
without ebbr
arm: (for 1/1 boards) all +38460.0 bss +1040.0 data +2375.0rodata +5333.0 text +29712.0
with various other things:
CONFIG_OF_LIBFDT_ASSUME_MASK=7
#CONFIG_OF_TRANSLATEis not set
#CONFIG_SIMPLE_BUSis not set
#CONFIG_TI_SYSCis not set
#CONFIG_CMD_FDTis not setarm: (for 1/1 boards) all +19170.0 bss -16.0 data +360.0 rodata+3274.0 text +15552.0
(Mark, in the same email:)
Quoted 9 lines
Quoted 9 lines
FIT simply isn't fit for purpose (pun intended). It only really works for booting Linux, and forces people to combine u-boot, kernel, initial ramdisk and other firmware components into a single image. That is really undesirable as:
- This makes it sigificantly harder to update individual components of such an image. Making it hard to update a kernel is obviously a serious security risk.
- This makes it impossible to build an OS install image that works om multiple boards/SoCs.Quoted 5 lines
I would really like to understand this better. The whole thing is a complete mystery to me.
Firstly I have sometimes fiddled with booting other OSes using FIT. It seemed OK. I can't see why it only works with Linux.
Quoted 6 lines
Well, you could of course rework the boot flow of other OSes such that booting them includes some sort of FIT if you really wanted to. I But an OS like OpenBSD comes with its own bootloader that is essential in the boot flow. On OpenBSD armv7/arm64/riscv64 it adds some essential properties to the device tree. Besides, the kernel itself relies on a valid EFI memory map.
(thanks for taking the time to reply with so much detail)
Quoted 7 lines
Quoted 2 lines
That's news to me; when did that happen? Anyway, I doubt that adds a lot of code.
Shortly after
EFI_LOADERsupport was added to U-Boot and there was a clear consensus thatEFI_LOADERsupport was going to be enabled by default on armv7 and armv8 targets. My initial commit of the EFI bootloader for armv7 is from May 2016.
OK, I am sad to hear that Linux is getting into a state where it cannot boot without EFI.
Quoted 90 lines
Quoted 83 lines
Quoted 23 lines
Quoted 2 lines
Secondly, I don't expect that U-Boot itself would be in the FIT.
So the FIT would only contain the OS kernel and other OS components? What about the FIT that is used on some arm64 platforms to combine U-Boot and TF-A? I guess you can have multiple FITs...
Quoted 7 lines
Thirdly, do you really want the kernel and initrd to be separate? At least in the systems I have used, they are built together, even having the same name, e.g.:
initrd.img-5.10.40-1rodete1-amd64
System.map-5.10.40-1rodete1-amd64
vmlinuz-5.10.28-1rodete2-amd64I don't really use Linux on these platforms. But I'd expect the normal package management tools of my Linux distribution to build those as necessary and place them in the root file system where the bootloader picks them up. And as a distro developer, I'd like to have the same approach work on all Linux systems, regardless the specific firmware they're running (EDK2, U-Boot or something completely different). Ideally that wouldn't even depend on the architecture.
Quoted 5 lines
Now Armbian takes a different approach, and does treat most systems they provide as special snowflakes, providing flash images for each board. But that doesn't scale and makes for a fairly crappy user experience. They don't always support booting a mainline kernel. And updating the kernel often requires installing special packages.
I don't think it is a requirement that things have to be special snowflakes. There are a few common boot flows to support and we can put that support in U-Boot and in userspace, without needing to make everything special.
Quoted 42 lines
Quoted 7 lines
Finally, for the firmware components, do you mean system firmware? If so, I would expect it to be more convenient to distribute updates to that separately, although I suppose they could be combined with the kernel if the combinatorial explosion can be contained. What is the problem, exactly? (If you mean peripheral firmware, I would expect fwupd to handle that.)
I guess I mean system firmware. Essentially everything that runs on the system before my OS bootloader runs. So for me, U-Boot is part of the system firmware even if it sometimes happens to live on removable media.
Quoted 1 line
What exactly is impossible? Can you please be more specific?
So let me explain what we want for OpenBSD. We really want a uniform user experience across platforms, and don't want to maintain different codepaths for special snowflake platforms that might exist for a specific architecture.
Installing OpenBSD on a machine should be as simple as dd'ing the installer to some boot media, plugging it in and powering the machine on. Now this is somewhat tricky to achieve on some hardware targetted by U-Boot as they don't come with usable system firmware on the board itself. But on those boards you can mostly get away with having U-Boot on uSD or eMMC and the OS installer on USB.
Updating the OpenBSD kernel should be as simple as copying the kernel to /bsd. Since the root filesystem uses the UFS/FFS filesystem, this means that whatever we use as a bootloader must be able to read from that filesystem. To make sure the kernel is properly seeded with entropy, the OpenBSD bootloader has some additional tricks up its sleeve. It will replace a special segment in the kernel with random data before handing control to the kernel. On platforms that support it, it will try to use a firmware-provided RNG to do this (and EFI supports this) but also mix in some random data from a file on the UFS/FFS filesystem. It will actually mark that file as "used" after reading it to throw a warning when the file is reused when the machine is rebooted (it should have been filled with fresh new entropy). So you really need to use the OpenBSD bootloader instead loading an OpenBSD kernel directly from system firmware.
Quoted 6 lines
Updating the OpenBSD bootloader should be as simple as running installboot(8) from within the OS.
This all works on pretty much any architecture that OpenBSD supports. And right now, thanks to
EFI_LOADERsupport in U-Boot, we have a nearly uniform boot flow on amd64, arm64, armv7 and riscv64.Quoted 6 lines
OK I see. Really it sounds like you have a pre-kernel loader. The functionality (some fresh random data) could just as easily be provided by U-Boot directly, with vastly less code and complexity. But I do understand that trying to design across a whole system is a pain, and it is attractive to try to use what hooks exist already to do what you want.
Quoted 5 lines
What do you mean with vastly less code and complexity. At this point 70% of the OpenBSD bootloader code is shared between most of the architectures OpenBSD runs on (alpha, amd64, arm64, armv7, hppa, i386, powerpc, riscv64, sh, sparc64). And on platforms that support UEFI (amd64, arm64, armv7 and riscv64) this is closer to 95% shared code.
I mean that IMO EFI makes the boot process extremely complex and non-deterministic. Perhaps no two devices are the same. It allows vendors to create arbitrary and proprietary features in between firmware and the OS with no shared code repo and no shared specifications or tests. From the side of distributions I can see what that is a nice feature, but I don't think it helps the industry.
Quoted 4 lines
Having OS-dependent code in U-Boot doesn't work. FreeBSD tried that with the U-Boot API. It was never enabled by default, and got broken all the time.
I was not suggesting OS-dependent code, just a set of defined features, so that if something is needed, it is specified and implemented, if supported. This isn't so different from EFI anyway, since what is actually supported can vary by platform. Overall I see a failure of imagination, perhaps due to the overall complexity of this space, with so many vendors, boards, etc.
Quoted 10 lines
Quoted 2 lines
How do you do verified boot?
We don't. I don't think it makes sense for an Open Source OS where people like to tinker with things. And it gets in the way of some of our own security features. OpenBSD relinks the kernel upon every boot as a defence against attacks that depend on the kernel address space layout. Doing that in an environment where all code needs to be signed is a big challenge.
Quoted 6 lines
There is a comany that has products based on OpenBSD. They do verified boot. I'm not really familliar with the details, but presumably they turned off address space randomization and I believe it is based on
EFI_LOADERsupport. Patrick Wildt contributed patches to implement this to U-Boot before the Linaro folks did their full-blown UEFI-based implementation.
OK.
Quoted 18 lines
Quoted 11 lines
Quoted 10 lines
Quoted 5 lines
FIT is just a container. It seems to have been rejected by the EFI crew at some point. Perhaps I just need to try to use it with one of the distros out there, to actually understand what is going on here. But any help is appreciated.
It just doesn't make sense for us to use a container just because the system firmware (U-Boot) insists on it. The kernel lives in the root UFS/FFS filesystem and the bootloader lives on an MS-DOS filesystem on the root disk.
Quoted 6 lines
It isn't so much that U-Boot insists on it. It is quite happy to load a kernel and initrd separately. But it does make verification harder and I assume it makes upgrades harder since there are multiple files to install. FIT is just such a nice format for packaging kernels and related things. It sounds like you could use FIT and everything you said above would still be true.
Quoted 3 lines
It would mean a kernel update would need to update the FIT container. But only on boards that use U-Boot to boot, since nobody else supports it. That isn't helpful to us.
It is ironic that we have added 90KB of EFI support to U-Boot, but we apparently cannot add 20KB of FIT support to UEFI. This seems like some sort of failure of the open-source process.
Regards,
Simon
Quoted 16 lines · Tom Rini
Quoted 8 lines · Mark Kettenis
Quoted 7 lines
From: Simon Glass <[email protected]>
Date: Sun, 27 Jun 2021 19:48:34 -0600It has come to light that
EFI_LOADERadds an extraordinary amount of code to U-Boot. For example, withnokia_rx51the size delta is about 90KB. About 170 boards explicitly disable the option, but is is clear that many more could, thus saving image size and boot time.Quoted 6 lines
EFI_LOADERused to be a lot smaller. It is great to see that over the years UEFI support has become more complete, but a lot of that new code implements features that are not at all essential for just booting an OS from storage. If that growth leads to the suggestion to disableEFI_LOADERcompletely by default, we're putting the cart before the horse.
Quoted 4 lines
Well, I see I forgot to prefix my patch with RFC, but I hadn't found
EFI_LOADERbeing used in the wild on armv7, but wasn't sure about the BSD families. I did see that Debian doesn't use it, and that Armbian doesn't even use it on aarch64.
Debian and Ubuntu both offer booting via UEFI and GRUB on armv7 (see
https://packages.ubuntu.com/hirsute/grub-efi-arm). You could use package flash-kernel instead if you wanted to avoid UEFI. It is just a question of how you decide to install your system. If you want the installer to automatically install GRUB, you must invoke it via UEFI.
Armbian uses the Debian repos. That lets you boot Armbian using UEFI and GRUB if you want to.
SUSE is another distro relying on UEFI and GRUB.
Quoted 40 lines
Quoted 10 lines
Quoted 2 lines
The current situation is affecting U-Boot's image as a svelt bootloader.
Really? I know UEFI has a bad reputation in the Open Source world, and some of its Microsoft-isms are really annoying (yay UCS-2). But it works, it provides a standardized approach across several platforms (ARMv7, AMRv8, RISC-V) and the industry seems to like it. Personally I'd wish the industry had standardized on Open Firmware instead, but that ship sailed a long time ago...
Quoted 4 lines
I find it hard to imagine that 90k is a serious amount of storage for something that is going to include a multi-MB Linux kernel. This isn't code that lives in SPL or TPL where severe size restrictions apply.
In one of those cases where I need to pop back in to the other (Nokia N900 specific) thread and see if the big size reduction really was just disabling
EFI_LOADER, it's perhaps just one of those "fun" things aboutKconfigand anything other than "make oldconfig" for spotting new config options that default to enabled.Quoted 7 lines
Quoted 6 lines
EFI_LOADERis required by EBBR, a new boot standard which aims to bring in UEFI protocols to U-Boot. But EBRR is not required for booting. U-Boot already provides support for FIT, the 'bootm' command and a suitable hand-off to Linux. EBRR has made the decision to create a parallel infrastructure, e.g. does not use FIT, nor U-Boot's signing infrastructure.Quoted 6 lines
EFI_LOADERis required to boot FreeBSD and OpenBSD on several platforms as well as generic Linux distros. For example OpenBSD/armv7, OpenBSD/arm64 and OpenBSD/riscv64 all rely onEFI_LOADERto boot and have done so for the last 4 years. The fact that ARM has embraced UEFI as an embedded boot standard and branded it EBBR really isn't all that relevant.To be clear here, I like
EFI_LOADER. I too do wish some other technologies had become dominant for technical rather than inertia reasons, but here we are. Having played around with it on aarch64, there are some pretty nice comes-along-with parts to it.
Quoted 8 lines
What I hadn't seen, and am only a little skeptical of still, is how far backwards in generations it's going to be used on. The general wish is that users nor off the shelf OS groups need to rebuild U-Boot for a given board, and instead it just works. The number of new designs for 32bit parts is no where near the number of new designs for 64bit parts. So what we're seeing in U-Boot now is people updating support on their older designs, and not necessarily caring about using
EFI_LOADER.
The question is not how old your board is. My i.MX6 Wandboard (bought 2013) and my Allwinner A20 Banana Pi (bought 2014) both run fine with UEFI and GRUB on Debian. The major divide is between systems running a distribution and IoT devices where you build your image via Yocto, OpenWRT, or the like.
Best regards
Heinrich