stand: Support (and prefer) the SMBIOS 64-bit Entry Point Structure

[allanjude]

Summary

Continue to set the old hint to avoid breaking older kernels

Sponsored by: Ampere Computing LLC
Submitted by: Klara Inc.

[allanjude]

stand/libsa/smbios.c

108

In the original review, @andrew indicated we would want to keep doing both loops here: https://reviews.freebsd.org/D28707#inline-178708

[scottph]

stand/efi/loader/main.c

1163–1164

because smbios_detect() "locks in" on the first smbios entry point it's given, doing smbios_detect as either table type is found prefers the earlier table, and not smbios3 over smbios. Maybe this whole loop could be dropped for:

void *smbios_ep, *smbios3_ep;
bool smbios_found;
 
smbios_found = false;
smbios_ep = efi_get_table(&smbios);
smbios3_ep = efi_get_table(&smbios3);
if (smbios3_ep != NULL) {
  setenv("hint.smbios3.0.mem", ...);
  smbios_detect(smbios3_ep);
}
if (smbios_ep != NULL) {
  setenv("hint.smbios.0.mem", ...);
  smbios_detect(smbios_ep);
}

1163–1164

we wouldn't want to set hint.smbios.0.mem to the address of the smbios3 entry point, as it's a different format.

[manu]

stand/efi/loader/main.c

1164

Doing both memcmp here and do another one after doesn't seems the best things to do.

[allanjude]

Switch to Scott's newer suggested code that is much cleaner

[allanjude]

Fix a typo

[dan.kotowski_a9development.com]

Ping - is there any work left to do here or can this be merged and closed?

[allanjude]

In D28737#729430, @dan.kotowski_a9development.com wrote:

Ping - is there any work left to do here or can this be merged and closed?

This specific piece is ready to land, but many others in the stack are not accepted yet.
I need to rebase them all and re-test it on the target hardware.

If this specific bit is somehow meaningful for you, I can land it by itself if that makes sense.

[dan.kotowski_a9development.com]

@allanjude not necessary, but "nice to have" as it resolves an outstanding bug in sysutils/hw-probe on ACPI-based arm64 systems

[andrew]

It's also useful for me for hypervisor ID changes I have for https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=258550

[gallatin]

It seems like this was obsoleted by:

commit ee97f198b42d50437f87aa4111d478eca2a5be16
Author: John-Mark Gurney <[e-mail]>
Date: Wed Feb 22 02:27:37 2023 +0000

Support SMBIOS v3 for 64-bit entry systems

Summary:
Under QEMU on arm64 systems, the smbios table is above 4GB
requiring a 64-bit address to access.

Reviewers: manu

Subscribers: imp, bcran, dab

Differential Revision: https://reviews.freebsd.org/D38721

[imp]

In D28737#970427, @gallatin wrote:

It seems like this was obsoleted by:

commit ee97f198b42d50437f87aa4111d478eca2a5be16
Author: John-Mark Gurney <[e-mail]>
Date: Wed Feb 22 02:27:37 2023 +0000

Support SMBIOS v3 for 64-bit entry systems

Summary:
Under QEMU on arm64 systems, the smbios table is above 4GB
requiring a 64-bit address to access.

Reviewers: manu

Subscribers: imp, bcran, dab

Differential Revision: https://reviews.freebsd.org/D38721

This commit covers all the things this diff does, so we can close it. If by odd chance, I've missed something, create a new differential and tag me and I'll look at it.