Patrick Rudolph has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/32090 )
Change subject: smbios: Add and adapt fields for version 3.2 ......................................................................
Patch Set 6:
Patch Set 5:
Repost as gerrit messed up the message:
There's one file making use of struct smbios_entry:
Of course you would add a second struct for the 64bit entry point
and keep the 32bit one.
If the callers should provide the value in the correct unit the
SMBIOS spec defines, you should update all callers instead.
I don't see how chromium patches relate to coreboot and why you
would mention them in the commit message.
|> I don't see how chromium patches relate to coreboot Because these SMBIOS information are ultimatly consumed in chromium, more specifically in mosys, and smbios related changes are going there.
I don't see how it can be the second patch, if the first never will be in this repository.
|> why you would mention them in the commit message. Please see the earlier request/comment from Paul.
|> There's one file making use of struct smbios_entry: src/arch/x86/smbios.c See my comment above. That information is consumed outside of coreboot and the smbios_entry struct is also defined at the caller's side and elsewhere: platform/mosys/include/lib/smbios.h platform/mosys/lib/smbios/smbios.c platform/mosys/include/lib/smbios_tables.h third_party/u-boot/files/include/smbios.h Those definitions have to be consistent, otherwise the system becomes unstable.
I don't care about 3rdparty software and it's totally unrelated.
|> Of course you would add a second struct for the 64bit entry point and keep the 32bit one. |> If the callers should provide the value in the correct unit the SMBIOS spec defines, you should update all callers instead. Maybe I am missing something, but not sure I understand the 2 sentences above. If you have 2 formats of smbios_entry which one would you return? The caller's only passes one variable into which this function "concatenate" the fields info, and the caller retrieves the information using the same standard definition of smbios_entry. Why would you need 2 definitions?
One struct for each entrypoint. It's not returned, so it wouldn't matter for the caller. The whole SMBIOS table including the entry point is returned.
Are you asking to change all the functions through which this information passes through? And if so, how would you do that in this commit? Since those files are not include here. Even if that is your request it conflict's with Nico's who wants to reduce the amount of changes going in here. In short, having one definition of smbios_entry on one side of the system and a different one on the end, would break the whole flow. Please let me know if I am missing something
For the type 17 CLK units, you should provide the value in the correct unit the SMBIOS spec defines, as it was also the case before the update. That means you should update all providers of CBMEM_ID_MEMINFO.