Mailing List Archive

commit 295bf24af77c0556fe023ce727734fd27226b37c
Author: Roger Pau Monne <roger.pau@citrix.com>
AuthorDate: Thu Feb 15 17:46:53 2024 +0100
Commit: Andrew Cooper <andrew.cooper3@citrix.com>
CommitDate: Tue Apr 9 17:10:28 2024 +0100

x86/vmx: Add support for virtualize SPEC_CTRL

The feature is defined in the tertiary exec control, and is available starting
from Sapphire Rapids and Alder Lake CPUs.

When enabled, two extra VMCS fields are used: SPEC_CTRL mask and shadow. Bits
set in mask are not allowed to be toggled by the guest (either set or clear)
and the value in the shadow field is the value the guest expects to be in the
SPEC_CTRL register.

By using it the hypervisor can force the value of SPEC_CTRL bits behind the
guest back without having to trap all accesses to SPEC_CTRL, note that no bits
are forced into the guest as part of this patch. It also allows getting rid of
SPEC_CTRL in the guest MSR load list, since the value in the shadow field will
be loaded by the hardware on vmentry.

Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
(cherry picked from commit 97c5b8b657e41a6645de9d40713b881234417b49)
---
xen/arch/x86/hvm/vmx/vmcs.c | 12 ++++++++-
xen/arch/x86/hvm/vmx/vmx.c | 55 ++++++++++++++++++++++++++++++++------
xen/include/asm-x86/hvm/vmx/vmcs.h | 5 ++++
xen/include/asm-x86/msr.h | 7 +++--
4 files changed, 68 insertions(+), 11 deletions(-)

diff --git a/xen/arch/x86/hvm/vmx/vmcs.c b/xen/arch/x86/hvm/vmx/vmcs.c
index c5ef01f936..806604358b 100644
--- a/xen/arch/x86/hvm/vmx/vmcs.c
+++ b/xen/arch/x86/hvm/vmx/vmcs.c
@@ -210,6 +210,7 @@ static void __init vmx_display_features(void)
P(cpu_has_vmx_virt_exceptions, "Virtualisation Exceptions");
P(cpu_has_vmx_pml, "Page Modification Logging");
P(cpu_has_vmx_tsc_scaling, "TSC Scaling");
+ P(cpu_has_vmx_virt_spec_ctrl, "Virtualize SPEC_CTRL");
#undef P

if ( !printed )
@@ -370,7 +371,7 @@ static int vmx_init_vmcs_config(bool bsp)

if ( _vmx_cpu_based_exec_control & CPU_BASED_ACTIVATE_TERTIARY_CONTROLS )
{
- uint64_t opt = 0;
+ uint64_t opt = TERTIARY_EXEC_VIRT_SPEC_CTRL;

_vmx_tertiary_exec_control = adjust_vmx_controls2(
"Tertiary Exec Control", 0, opt,
@@ -1379,6 +1380,12 @@ static int construct_vmcs(struct vcpu *v)
if ( cpu_has_vmx_tsc_scaling )
__vmwrite(TSC_MULTIPLIER, d->arch.hvm.tsc_scaling_ratio);

+ if ( cpu_has_vmx_virt_spec_ctrl )
+ {
+ __vmwrite(SPEC_CTRL_MASK, 0);
+ __vmwrite(SPEC_CTRL_SHADOW, 0);
+ }
+
/* will update HOST & GUEST_CR3 as reqd */
paging_update_paging_modes(v);

@@ -2089,6 +2096,9 @@ void vmcs_dump_vcpu(struct vcpu *v)
if ( v->arch.hvm.vmx.secondary_exec_control &
SECONDARY_EXEC_VIRTUAL_INTR_DELIVERY )
printk("InterruptStatus = %04x\n", vmr16(GUEST_INTR_STATUS));
+ if ( cpu_has_vmx_virt_spec_ctrl )
+ printk("SPEC_CTRL mask = 0x%016lx shadow = 0x%016lx\n",
+ vmr(SPEC_CTRL_MASK), vmr(SPEC_CTRL_SHADOW));

printk("*** Host State ***\n");
printk("RIP = 0x%016lx (%ps) RSP = 0x%016lx\n",
diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
index 2cbed4b21e..c86aeeafb2 100644
--- a/xen/arch/x86/hvm/vmx/vmx.c
+++ b/xen/arch/x86/hvm/vmx/vmx.c
@@ -757,23 +757,28 @@ static void vmx_cpuid_policy_changed(struct vcpu *v)
/*
* We can safely pass MSR_SPEC_CTRL through to the guest, even if STIBP
* isn't enumerated in hardware, as SPEC_CTRL_STIBP is ignored.
+ *
+ * If VMX_VIRT_SPEC_CTRL is available, it is activated by default and the
+ * guest MSR_SPEC_CTRL value lives in the VMCS. Otherwise, it lives in
+ * the MSR load/save list.
*/
if ( cp->feat.ibrsb )
{
vmx_clear_msr_intercept(v, MSR_SPEC_CTRL, VMX_MSR_RW);

- rc = vmx_add_guest_msr(v, MSR_SPEC_CTRL, 0);
- if ( rc )
- goto out;
+ if ( !cpu_has_vmx_virt_spec_ctrl )
+ {
+ rc = vmx_add_guest_msr(v, MSR_SPEC_CTRL, 0);
+ if ( rc )
+ goto out;
+ }
}
else
{
vmx_set_msr_intercept(v, MSR_SPEC_CTRL, VMX_MSR_RW);

- rc = vmx_del_msr(v, MSR_SPEC_CTRL, VMX_MSR_GUEST);
- if ( rc && rc != -ESRCH )
- goto out;
- rc = 0; /* Tolerate -ESRCH */
+ if ( !cpu_has_vmx_virt_spec_ctrl )
+ vmx_del_msr(v, MSR_SPEC_CTRL, VMX_MSR_GUEST);
}

/* MSR_PRED_CMD is safe to pass through if the guest knows about it. */
@@ -2583,9 +2588,14 @@ static uint64_t vmx_get_reg(struct vcpu *v, unsigned int reg)
uint64_t val = 0;
int rc;

+ /* Logic which doesn't require remote VMCS acquisition. */
switch ( reg )
{
case MSR_SPEC_CTRL:
+ if ( cpu_has_vmx_virt_spec_ctrl )
+ /* Guest value in VMCS - fetched below. */
+ break;
+
rc = vmx_read_guest_msr(v, reg, &val);
if ( rc )
{
@@ -2594,13 +2604,26 @@ static uint64_t vmx_get_reg(struct vcpu *v, unsigned int reg)
domain_crash(d);
}
return val;
+ }
+
+ /* Logic which maybe requires remote VMCS acquisition. */
+ vmx_vmcs_enter(v);
+ switch ( reg )
+ {
+ case MSR_SPEC_CTRL:
+ ASSERT(cpu_has_vmx_virt_spec_ctrl);
+ __vmread(SPEC_CTRL_SHADOW, &val);
+ break;

default:
printk(XENLOG_G_ERR "%s(%pv, 0x%08x) Bad register\n",
__func__, v, reg);
domain_crash(d);
- return 0;
+ break;
}
+ vmx_vmcs_exit(v);
+
+ return val;
}

static void vmx_set_reg(struct vcpu *v, unsigned int reg, uint64_t val)
@@ -2608,9 +2631,14 @@ static void vmx_set_reg(struct vcpu *v, unsigned int reg, uint64_t val)
struct domain *d = v->domain;
int rc;

+ /* Logic which doesn't require remote VMCS acquisition. */
switch ( reg )
{
case MSR_SPEC_CTRL:
+ if ( cpu_has_vmx_virt_spec_ctrl )
+ /* Guest value in VMCS - set below. */
+ break;
+
rc = vmx_write_guest_msr(v, reg, val);
if ( rc )
{
@@ -2619,12 +2647,23 @@ static void vmx_set_reg(struct vcpu *v, unsigned int reg, uint64_t val)
domain_crash(d);
}
break;
+ }
+
+ /* Logic which maybe requires remote VMCS acquisition. */
+ vmx_vmcs_enter(v);
+ switch ( reg )
+ {
+ case MSR_SPEC_CTRL:
+ ASSERT(cpu_has_vmx_virt_spec_ctrl);
+ __vmwrite(SPEC_CTRL_SHADOW, val);
+ break;

default:
printk(XENLOG_G_ERR "%s(%pv, 0x%08x, 0x%016"PRIx64") Bad register\n",
__func__, v, reg, val);
domain_crash(d);
}
+ vmx_vmcs_exit(v);
}

static struct hvm_function_table __initdata vmx_function_table = {
diff --git a/xen/include/asm-x86/hvm/vmx/vmcs.h b/xen/include/asm-x86/hvm/vmx/vmcs.h
index 9b246b802b..2b4de99ed4 100644
--- a/xen/include/asm-x86/hvm/vmx/vmcs.h
+++ b/xen/include/asm-x86/hvm/vmx/vmcs.h
@@ -279,6 +279,9 @@ extern u32 vmx_secondary_exec_control;
#define TERTIARY_EXEC_VIRT_SPEC_CTRL BIT(7, UL)
extern uint64_t vmx_tertiary_exec_control;

+#define cpu_has_vmx_virt_spec_ctrl \
+ (vmx_tertiary_exec_control & TERTIARY_EXEC_VIRT_SPEC_CTRL)
+
#define VMX_EPT_EXEC_ONLY_SUPPORTED 0x00000001
#define VMX_EPT_WALK_LENGTH_4_SUPPORTED 0x00000040
#define VMX_EPT_MEMORY_TYPE_UC 0x00000100
@@ -436,6 +439,8 @@ enum vmcs_field {
XSS_EXIT_BITMAP = 0x0000202c,
TSC_MULTIPLIER = 0x00002032,
TERTIARY_VM_EXEC_CONTROL = 0x00002034,
+ SPEC_CTRL_MASK = 0x0000204a,
+ SPEC_CTRL_SHADOW = 0x0000204c,
GUEST_PHYSICAL_ADDRESS = 0x00002400,
VMCS_LINK_POINTER = 0x00002800,
GUEST_IA32_DEBUGCTL = 0x00002802,
diff --git a/xen/include/asm-x86/msr.h b/xen/include/asm-x86/msr.h
index c37b9e771b..a7ad9bd189 100644
--- a/xen/include/asm-x86/msr.h
+++ b/xen/include/asm-x86/msr.h
@@ -289,8 +289,11 @@ struct vcpu_msrs
* For PV guests, this holds the guest kernel value. It is accessed on
* every entry/exit path.
*
- * For VT-x guests, the guest value is held in the MSR guest load/save
- * list.
+ * For VT-x guests, one of two situations exist:
+ *
+ * - If hardware supports virtualized MSR_SPEC_CTRL, it is active by
+ * default and the guest value lives in the VMCS.
+ * - Otherwise, the guest value is held in the MSR load/save list.
*
* For SVM, the guest value lives in the VMCB, and hardware saves/restores
* the host value automatically. However, guests run with the OR of the
--
generated by git-patchbot for /home/xen/git/xen.git#staging-4.16