Samuel, Thanks for the feedback on V1 patch and the pointer to host_info(). This is a much lighter way to implement what was needed.
Patch 1 provides the gnumach changes to capture the EFI system table address and a new flavour for host_info() to access it. There is nothing here that actually couldn't be added to hurd_i386 (if MULTIBOOT2_TAG_TYPE_EFI32 was in the header), but I couldn't see the point of bloating i386 with functionality that would not be needed, so guarded all these changes with '#ifdef __x86_64__'. Patch 2 is the libacpica change to use hostinfo() to find the EFI system table and then the ACPI RSDP. This is also wrapped in '#ifdef __x86_64__' but that is definitely required because I didn't add code to support 32 bit variants of the table headers. The previous patch version looked up the RSDP from BIOS first and if that failed attempted to use the UEFI approach. That however produces a (harmless) error message from libacpica which I thought looked misleading so I reversed the order of lookup which is now quiet in both UEFI and BIOS. Thanks, Mike.
