bhyve: add config option to load ACPI tables into memory

[corvink]

Summary

For backward compatibility, the ACPI tables are loaded into the guest
memory. Windows scans the memory, finds the ACPI tables and uses them.
It ignores the ACPI tables provided by the UEFI. We are patching the
ACPI tables in the guest memory, so that's mostly fine. However, Windows
will break when the ACPI tables become to large or when we are adding
fwcfg items like the tpm log to an table which can't be patched by
bhyve.

By default ACPI tables are still loaded into guest memory for backward
compatibility. The new acpi_tables_in_memory config value can be set to
false to avoid this behaviour.

[corvink]

Do you think we should add this option now or should we wait until someone has some issues or a feature requires it?

It might be helpful for debugging purposes too.

[rew]

In D39979#910118, @corvink wrote:

Do you think we should add this option now or should we wait until someone has some issues or a feature requires it?

It might be helpful for debugging purposes too.

When do you suspect this will become an issue for users?

Sounds like you have features that are queued up that will use this?

[corvink]

In D39979#910184, @rew wrote:

In D39979#910118, @corvink wrote:

Do you think we should add this option now or should we wait until someone has some issues or a feature requires it?

It might be helpful for debugging purposes too.

When do you suspect this will become an issue for users?

There are two cases I'm aware of yet:

1. Our memory approach is size limited. We can only use the range [ 0xf2400, 0x100000]. Don't know what happens if we exceed it.

2. A TPM log is adding a pointer to ACPI tables which needs to be patched by the guest.
   • see https://github.com/qemu/qemu/blob/792f77f376adef944f9a03e601f6ad90c2f891b2/hw/acpi/aml-build.c#L2262-L2264
   • see https://www.kernel.org/doc/html/next/security/tpm/tpm_event_log.html
   • The in memory tables are created at startup but the log address is allocated by the guest firmware at runtime. So, it's impossible for our in memory tables to contain a valid address for the TPM log.

Sounds like you have features that are queued up that will use this?

My TPM emulation D32961 will add the TPM log. It's a long overdue task of mine to update this revision. As all ACPI related pre work is merged now, I'm going to update it soon.

[corvink]

• rebase onto main

[markj]

How can this behaviour be requested from the command line? -o acpi_tables_in_memory=false?

usr.sbin/bhyve/bhyve_config.5

126

127

128

130

[corvink]

In D39979#945140, @markj wrote:

How can this behaviour be requested from the command line? -o acpi_tables_in_memory=false?

Yes, you can request this behaviour with -o acpi_tables_in_memory=false from cmdline.

[corvink]

• update man page as suggested

[markj]

usr.sbin/bhyve/bhyve_config.5

129

My suggestion was to delete this sentence.

130

This line is too long.