uki: describe .efifw section

[ani-sinha]

Add description for the firmware section (.efifw) in the specs.

See also systemd/systemd#35091

[Foxboron]

My general feeling is that this is too generic? Is there no expectations to the embedded blob? Is this all an implementation detail?

And if it only an implementation detail, how often would this actually be used by others beyond the usecase in the systemd PR?

[keszybz]

I think being generic here is fine. What matters is that the firmware is in some format acceptable to the hardware. We can't really say much about it this point. If this becomes established, we can add additional details later.

[poettering]

Sorry for dropping the ball on this. Could you please rebase this? We really should get this landed.

Please follow the wording of the .dtbauto stuff regarding the matching.

[keszybz]

@ani-sinha ping

[katexochen]

I've rebased and updated this in #216, PTAL

[ani-sinha]

Sorry for dropping the ball on this. Could you please rebase this? We really should get this landed.

Please follow the wording of the .dtbauto stuff regarding the matching.

Sorry I too forgot about this. Let me try and get to this in the next few days.

[ani-sinha]

Sorry for dropping the ball on this. Could you please rebase this? We really should get this landed.

Please follow the wording of the .dtbauto stuff regarding the matching.

@poettering So I was trying to update the doc and looked at the code. I see that chid_match() is only called for DEVICE_TYPE_DEVICETREE types of device entries and not for DEVICE_TYPE_UEFI_FW types of entries. So it won't match against the hardware chid. I think there is a gap in the implementation.

[keszybz]

We now have two versions of this PR. Since this one was earlier, I'll merge it. If subsequent changes should be done, let's do that later.

[ani-sinha]

Sorry for dropping the ball on this. Could you please rebase this? We really should get this landed.
Please follow the wording of the .dtbauto stuff regarding the matching.

@poettering So I was trying to update the doc and looked at the code. I see that chid_match() is only called for DEVICE_TYPE_DEVICETREE types of device entries and not for DEVICE_TYPE_UEFI_FW types of entries. So it won't match against the hardware chid. I think there is a gap in the implementation.

I have started a PR here: systemd/systemd#42570 Still in draft, I want to see if I introduced any regressions or not.