Attention is currently required from: Jason Glenesk, Raul Rangel, Martin L Roth, Matt DeVillier, Fred Reitberger, Felix Held.
Karthik Ramasubramanian has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/71211 )
Change subject: soc/amd/mendocino: Pass option about compression status of BIOS ......................................................................
Patch Set 2:
(1 comment)
Commit Message:
https://review.coreboot.org/c/coreboot/+/71211/comment/249e9fbd_17a44c2d PS2, Line 10: This is done to accommodate metadata : hash anchor in bootblock so that the anchor can be updated when files : are added to CBFS during build time. Pass option to indicate that PSP : BIOS binary is uncompressed under such scenarios.
can we instead generate the bootblock hashes, add the table to the bootblock binary, then compress the bootblock and add it to the amdfw binary at the end of the build?
That way it doesn't need to be added uncompressed. Is there a problem with doing it that way that I'm not seeing?
Any file that contains the metadata hash anchor needs to be added first to coreboot.pre. Based on that we can generate amdfw.rom without the BIOS Directory entry Type 0x62. Keep the bootblock file as a separate CBFS file. Once all the files are added to CBFS, bootblock can be compressed and added to amdfw.rom. Like you mentioned that option has to be added to amdfwtool.
using an uncompressed bootblock will increase boot times, so this wouldn't be a good design choice; especially since Google usually pushes quite a bit towards improving boot times. did you measure how much of a slowdown this will cause?
No. The only option is to measure using STB traces. If PSP can put special codes in STB, that will be helpful to measure the boot time impact. Without that, it is impossible to measure the impact to boot time.