openness_score/coreboot.py: Handle FMAP layout problems

[miczyg1]

It is possible for the FMAP layout to be non-contiguous. Catch such cases and count the non-described space as closed source.

Add SI_BIOS to the list of container regions, to avoid warnings.

Catch regions that may have numbers in the name.

Do not print a warning when a coreboot config is not found in a CBFS region not expected to have a config file.

[miczyg1]

Example output:

./openness_score/openness_score.py protectli_vp66xx_v0.9.0.rom 
'protectli_vp66xx_v0.9.0.rom' detected as Dasharo image


'protectli_vp66xx_v0.9.0.rom' Dasharo image statistics:
CBFS region BOOTSPLASH:
	CBFS size: 524288
	Number of files: 1
	Open-source files size: 0
	Closed-source files size: 0
	Data size: 28
	Empty size: 524260
CBFS region FW_MAIN_A:
	CBFS size: 4587264
	Number of files: 13
	Open-source files size: 2370499
	Closed-source files size: 1675795
	Data size: 8622
	Empty size: 532348
CBFS region COREBOOT:
	CBFS size: 5210112
	Number of files: 17
	Open-source files size: 2460355
	Closed-source files size: 1675795
	Data size: 9058
	Empty size: 1064904
WARNING: FMAP layout is not contiguous.
The space between region RW_MRC_CACHE and RW_VPD is not described in the FMAP layout and will be classified as closed-source.

Dasharo image protectli_vp66xx_v0.9.0.rom:
	Image size: 16777216
	Number of regions: 25
	Number of CBFSes: 3
	Total open-source code size: 4830854
	Total closed-source code size: 8365094
	Total data size: 677420
	Total empty size: 2903848

Before it printed following warnings:

WARNING: Could not extract coreboot config
WARNING: Failed to extract LAN driver. It will not be counted as closed-source
WARNING: Failed to extract LAN driver. It will not be counted as closed-source
WARNING: Skipped SI_BIOS region, suspected to be a container
WARNING: Something went wrong.

[philipanda]

I get the same results as you. The open/closed/data/empty sizes now sum up to the total file size. I can't say anything about the new regions added to region lists, but other than that the changes look good to me.

[miczyg1]

I get the same results as you. The open/closed/data/empty sizes now sum up to the total file size. I can't say anything about the new regions added to region lists, but other than that the changes look good to me.

They were not added. Simply the regexp parsing the output of CBFSTOOL command didn't catch all regions. Now all regions are caught, that's the difference