]> git.ipfire.org Git - thirdparty/linux.git/commit
x86/hyperv/vtl: Setup the 64-bit trampoline for TDX guests
authorYunhong Jiang <yunhong.jiang@linux.intel.com>
Wed, 4 Mar 2026 23:41:18 +0000 (15:41 -0800)
committerDexuan Cui <decui@microsoft.com>
Thu, 28 May 2026 20:01:25 +0000 (20:01 +0000)
commitaa7719039bd9e378c8c134b8ce5fa5f4e13cda73
treece3a45e1e154b84941e06401f455ec94af7ae4b7
parenta7ac1ea1f06314f6690ac772e9a31d6e2db977fa
x86/hyperv/vtl: Setup the 64-bit trampoline for TDX guests

The hypervisor is an untrusted entity for TDX guests. It cannot be used
to boot secondary CPUs - neither via hypercalls nor the INIT assert,
de-assert, plus Start-Up IPI messages.

Instead, the platform virtual firmware boots the secondary CPUs and
puts them in a state to transfer control to the kernel. This mechanism uses
the wakeup mailbox described in the Multiprocessor Wakeup Structure of the
ACPI specification. The entry point to the kernel is trampoline_start64.

Allocate and setup the trampoline using the default x86_platform callbacks.

The platform firmware configures the secondary CPUs in long mode. It is no
longer necessary to locate the trampoline under 1MB memory. After handoff
from firmware, the trampoline code switches briefly to 32-bit addressing
mode, which has an addressing limit of 4GB. Set the upper bound of the
trampoline memory accordingly.

Reviewed-by: Dexuan Cui <decui@microsoft.com>
Reviewed-by: Michael Kelley <mhklinux@outlook.com>
Signed-off-by: Yunhong Jiang <yunhong.jiang@linux.intel.com>
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Signed-off-by: Dexuan Cui <dexuan@kernel.org>
arch/x86/hyperv/hv_vtl.c