Hi, On Sun, 1 Dec 2024 at 09:15, Simon Glass <s...@chromium.org> wrote: > > Hi Matthew, > > On Sat, 23 Nov 2024 at 12:56, Matthew Garrett <mj...@srcf.ucam.org> wrote: > > > > We run a mixed environment including traditional embedded Linux and > > UEFI. For consistency it's valuable to use U-Boot in both cases, but on > > UEFI systems we want to run the full Linux kernel UEFI boot stub. This > > patchset adds support for that, along with various other quality-of-life > > improvements such as automatically finding partitions based on their > > GUID, automatic identification of load addresses where the firmware has > > given us constraints, support for EFI variable access in the EFI app, > > drivers for EFI network adapters and TPMs, support for embedding DTBs in > > the EFI app, and a couple of bug fixes. > > > > > > Janis Danisevskis (1): > > Fix efi_bind_block. > > > > Matthew Garrett (9): > > Add EFI handover support to bootm > > Add part_find command > > Add a command to find a load address > > Hook up EFI env variable support in the EFI app > > Add EFI network driver > > Add UEFI TPM2 driver > > Support separate DTB files with the UEFI app > > Use the correct ramdisk address > > Add command to set an environment variable to an EFI variable > > > > Makefile | 7 +- > > arch/x86/config.mk | 2 +- > > arch/x86/lib/bootm.c | 60 ++++++++---- > > arch/x86/lib/elf_x86_64_efi.lds | 4 + > > boot/bootm.c | 5 + > > cmd/Kconfig | 23 ++++- > > cmd/Makefile | 3 + > > cmd/addr_find.c | 87 +++++++++++++++++ > > cmd/efigetenv.c | 133 ++++++++++++++++++++++++++ > > cmd/part_find.c | 156 ++++++++++++++++++++++++++++++ > > drivers/net/Kconfig | 7 ++ > > drivers/net/Makefile | 1 + > > drivers/net/efi_net.c | 110 +++++++++++++++++++++ > > drivers/tpm/Kconfig | 7 ++ > > drivers/tpm/Makefile | 1 + > > drivers/tpm/tpm2_efi.c | 97 +++++++++++++++++++ > > include/asm-generic/sections.h | 1 + > > include/bootm.h | 6 ++ > > include/efi.h | 24 +++++ > > include/efi_tcg2.h | 1 + > > lib/efi/Makefile | 2 +- > > lib/efi/efi.c | 1 + > > lib/efi/efi_app.c | 41 ++++++++ > > lib/efi/efi_app_init.c | 163 ++++++++++++++++++++++++++++++-- > > lib/efi/efi_dtb.S | 6 ++ > > lib/efi/efi_vars.c | 44 +++++++++ > > lib/fdtdec.c | 3 + > > 27 files changed, 967 insertions(+), 28 deletions(-) > > create mode 100644 cmd/addr_find.c > > create mode 100644 cmd/efigetenv.c > > create mode 100644 cmd/part_find.c > > create mode 100644 drivers/net/efi_net.c > > create mode 100644 drivers/tpm/tpm2_efi.c > > create mode 100644 lib/efi/efi_dtb.S > > create mode 100644 lib/efi/efi_vars.c > > > > -- > > 2.47.0 > > > > Thank you for sending this and for your interesting talk at plumbers. > It is a novel approach and potentially provides an alternative to grub > and EFI_LOADER, etc. > > I'd like to try getting this running in CI as we don't have any such > tests for the app or payload at present.
I'm planning to pull this in as this seems like an important use case. I'm not sure how much time Matthew will have for this, and I want to avoid this series going stale. The comments which are not 'why do you want to do this?' are fairly minor, so I'll send a follow-up to resolve them next week. We should also think about what code we could move from lib/efi_loader to lib/efi so we can share it with the app. Matthew, thanks again for sending this. if you are planning to do more here in the short term, please let me know. Regards, Simon
