-
Notifications
You must be signed in to change notification settings - Fork 36
EBBR Notes 2025.09.10
Vincent Stehlé edited this page Sep 10, 2025
·
3 revisions
- Heinrich Schuchardt (Canonical)
- Ilias Apalodimas (Linaro)
- Vincent Stehlé (Arm)
- Pull requests #144 & #147 have been reviewed and will be merged.
- A couple of discussions around U-Boot EFI topics took place.
- Pull request #147: chapter3: raise RISC-V SBI requirement
- Reviewed by Atish
- -> merge it!
- Pull request #144: PFDI requirements
- Reviewed by Heinrich, Bertrand and Ilias
- -> merge it!
- Ilias: "U-Boot request" to fix something
- Multiple ESPs confusing
- Removable media with ESP and variables saved in there, can disappear
-
Patch on the ML at some point storing the EFI variables in a QSPI
- Not much worse than in a file
- Separate format from storage
- Introduce a SPI backend?
-
efi_set_variable_int()-
efi_var_to_xxx()backend()
-
-
- SPI driver in the secure world behind StMM
- Otherwise, do not store secure variables in there
- Introduce a SPI backend?
- Overall, U-Boot EFI in good shape now
- Multiple ESPs confusing
-
FWTS seeing EFI variables as r/o at runtime, even though we have the op-tee "trick"
- Heinrich:
-
efi_init_runtime_supported(-
if (IS_ENABLED(CONFIG_EFI_RT_VOLATILE_STORE- means: in-mem cache r/w with a magic variable to save
-
- In the case of eMMC RPMB w/op-tee, this is not enabled
- and this is true in fact
- EFI application calling
ExitBootServices()cannot callSetVariable()anymore -> table is correct- Too bad for FWTS...
- Propose a patch for FWTS?