[PATCH v3 7/7] x86/cpuid: Advertise XEN_HVM_CPUID_EXT_DEST_ID when device model opts in

["Julian Vetter" <julian.vetter@xxxxxxxxxx>]

Set the XEN_HVM_CPUID_EXT_DEST_ID bit in the HVM hypervisor CPUID leaf
only when the device model has called XEN_DMOP_enable_ext_dest_id,
signalling it will use XEN_DMOP_bind_pt_msi_irq for passthrough MSIs.
This guarantees that Xen will decode the extended destination bits from
the raw MSI address internally, so the device model does not need to
understand the encoding itself. This ensures that an old device model
that still uses XEN_DOMCTL_bind_pt_irq will not trigger extended ID
usage in the guest, preserving backwards compatibility. However a device
model that calls XEN_DMOP_enable_ext_dest_id then has to use
XEN_DMOP_{bind,unbind}_pt_msi_irq, and enables support for x2APIC
destination IDs above 255.

Signed-off-by: Julian Vetter <julian.vetter@xxxxxxxxxx>
---
Changes in V3:
- Addressed feedback from Jan and Roger -> Don't advertise the
  XEN_HVM_CPUID_EXT_DEST_ID unconditionally
---
 xen/arch/x86/cpuid.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/xen/arch/x86/cpuid.c b/xen/arch/x86/cpuid.c
index 5decfad8cd..c1f82f83ea 100644
--- a/xen/arch/x86/cpuid.c
+++ b/xen/arch/x86/cpuid.c
@@ -148,6 +148,18 @@ static void cpuid_hypervisor_leaves(const struct vcpu *v, 
uint32_t leaf,
         res->a |= XEN_HVM_CPUID_DOMID_PRESENT;
         res->c = d->domain_id;
 
+        /*
+         * Advertise extended destination ID support only when the device model
+         * has opted in via XEN_DMOP_enable_ext_dest_id, making sure it will
+         * use XEN_DMOP_bind_pt_msi_irq for passthrough MSI (passing raw
+         * addr/data so Xen can decode extended bits). This allows guests to
+         * use bits 11:5 of the MSI address and bits 55:49 of the IO-APIC RTE
+         * for additional destination ID bits, expanding the addressable APIC
+         * ID range from 8 to 15 bits.
+         */
+        if ( d->arch.hvm.ext_dest_id_enabled )
+            res->a |= XEN_HVM_CPUID_EXT_DEST_ID;
+
         /*
          * Per-vCPU event channel upcalls are implemented and work
          * correctly with PIRQs routed over event channels.
-- 
2.51.0



--
Julian Vetter | Vates Hypervisor & Kernel Developer

XCP-ng & Xen Orchestra - Vates solutions

web: https://vates.tech