Attention is currently required from: Hung-Te Lin, Shelley Chen, Paul Menzel, Yu-Ping Wu, Jianjun Wang. Angel Pons has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/62359 )
Change subject: soc/mediatek: PCI: Assert PERST# at bootblock stage ......................................................................
Patch Set 10:
(1 comment)
Commit Message:
https://review.coreboot.org/c/coreboot/+/62359/comment/83c847f1_89a004d9 PS9, Line 14: 100ms
Something to keep in mind is that it only works as expected if `COLLECT_TIMESTAMPS` is enabled, but it's only `default y if ARCH_X86`.
COLLECT_TIMESTAMPS is enabled for ChromeOS (src/vendorcode/google/chromeos/Kconfig).
The code in mt8195/cherry should still work even if not built for chromeos. I wonder if we have some common approach for sharing data across stages?
I agree. Maybe we could enable `COLLECT_TIMESTAMPS` by default on Mediatek platforms. As this option is user-controllable (it has a prompt), I wouldn't use a `select` statement but `default y` instead.
As for sharing data across stages, I think the CBMEM_INIT hook stuff is the only common thing we have.
I think an implementation of `timestamp_get()` is needed for this to work.
I agree, although we cannot use the name "timestamp_get" because it already exists for getting the timestamp. Maybe timestamp_find(id)? Alternatively, even better we could rename the current timestamp_get() to timestamp_now().
I meant to say that the `timestamp_get()` function has a __weak definition, which might have to be overriden. However, if `HAVE_MONOTONIC_TIMER` is enabled, it should work.
For the new function, I'd also use the `timestamp_find(id)` name.