From: Xiaoyao Li <xiaoyao.li@intel.com>
To: Sean Christopherson <seanjc@google.com>
Cc: Rick P Edgecombe <rick.p.edgecombe@intel.com>,
"sashiko-reviews@lists.linux.dev"
<sashiko-reviews@lists.linux.dev>,
"pbonzini@redhat.com" <pbonzini@redhat.com>,
"kvm@vger.kernel.org" <kvm@vger.kernel.org>
Subject: Re: [PATCH v3 3/4] KVM: TDX: Don't assume exit_reason[31:16] as all-0 in tdx_to_vmx_exit_reason()
Date: Mon, 17 Aug 2026 10:49:01 +0800 [thread overview]
Message-ID: <d67eb097-1223-4b68-92d4-1510d3d7be7c@intel.com> (raw)
In-Reply-To: <an8wUfZ9JxvEhtHS@google.com>
On 8/14/2026 11:12 PM, Sean Christopherson wrote:
> On Fri, Aug 14, 2026, Xiaoyao Li wrote:
>> On 8/14/2026 7:44 AM, Sean Christopherson wrote:
>>> On Wed, Aug 12, 2026, Rick P Edgecombe wrote:
>>>> Side note. I really dislike how tangled this area is for something that seems
>>>> like it should be much more straightforward. Deriving partially I think from the
>>>> overloading of the TDVMCALL leafs with the exit reasons. So we have things like:
>>>> ...
>>>> case EXIT_REASON_EPT_VIOLATION:
>>>> return EXIT_REASON_EPT_MISCONFIG;
>>>> ...
>>>
>>> I peeked at that code again, and FWIW I still think swizzling the exit_reason for
>>> TDVMCALL is the least awful solution. If we don't do that, then we'll have to
>>> update every single use of the exit_reason to demux TDVMCALL into the "real" exit
>>> reason, which will be a mess.
>>
>> I'm not sure if you read my idea[1]?
>>
>> I think there is only one place KVM cares about the exit_reason TDVMCALL,
>> just the
>>
>> case EXIT_REASON_TDCALL:
>
> No, the massaged exit_reason is also subtley consumed via trace_kvm_exit().
yeah. This is exactly the one I don't like, it looks like
trace_kvm_exit() is tracing the wrong exit reason, though it is by
intentional.
> It's also consumed by tdx_complete_emulated_msr():
>
> if (vmx_get_exit_reason(vcpu).basic == EXIT_REASON_MSR_READ)
>
> and by tdx_interrupt_allowed()
>
> return vmx_get_exit_reason(vcpu).basic != EXIT_REASON_HLT ||
> !to_tdx(vcpu)->vp_enter_args.r12;
>
> and by tdx_protected_apic_has_interrupt():
>
> if (vmx_get_exit_reason(vcpu).basic != EXIT_REASON_HLT ||
> to_tdx(vcpu)->vp_enter_args.r12)
> return false;
Ah, I really should have checked the code more carefully.
>> in tdx_handle_exit().
>> [1]
>> https://lore.kernel.org/all/646f9595-459f-4224-b4e5-4ec2eecc0bc6@intel.com/
>>> And once we track the exit_reason separately from vp_enter_ret, IMO it all becomes
>>> more logical and easier to follow. vp_enter_ret holds the information about why
>>> VP.ENTER returned/exited, while exit_reason holds information about why the _guest_
>>> exited. Obviously it's imperfect since we're still fudging EXIT_REASON_EPT_MISCONFIG,
>>> but again, I think that's a better alternative than demuxing exit_reason in multiple
>>> locations.
>>
>> The question do we really need to swizzle EXIT_REASON_TDCALL to other exit
>> reasons ahead? why cannot them just be handled in the central handler for
>> EXIT_REASON_TDCALL?
>
> Because as above, it's not as central as you think. If we want to not swizzle
> the exit_reason, then IMO the only sane way to do that is to not track exit_reason
> for TDX vCPUs, i.e. move vcpu_vt.exit_reason back to vcpu_vmx and force TDX to
> always demux vp_enter_ret every time.
How about adding a specific field to track the TDVMCALL leaf? Full diff
as below (the EPT MISCONFIG part can be split into a separate one)
-----8<--------
diff --git a/arch/x86/include/asm/shared/tdx.h
b/arch/x86/include/asm/shared/tdx.h
index f20e91d7ac35..61d00100a2b1 100644
--- a/arch/x86/include/asm/shared/tdx.h
+++ b/arch/x86/include/asm/shared/tdx.h
@@ -69,6 +69,13 @@
#define TD_CTLS_LOCK BIT_ULL(TD_CTLS_LOCK_BIT)
/* TDX hypercall Leaf IDs */
+#define TDVMCALL_CPUID 10 /* EXIT_REASON_CPUID */
+#define TDVMCALL_HLT 12 /* EXIT_REASON_HLT */
+#define TDVMCALL_IO_INSTRUCTION 30 /*
EXIT_REASON_IO_INSTRUCTION */
+#define TDVMCALL_MSR_READ 31 /* EXIT_REASON_MSR_READ */
+#define TDVMCALL_MSR_WRITE 32 /* EXIT_REASON_MSR_WRITE */
+#define TDVMCALL_MMIO 48 /* EXIT_REASON_EPT_VIOLATION */
+
#define TDVMCALL_GET_TD_VM_CALL_INFO 0x10000
#define TDVMCALL_MAP_GPA 0x10001
#define TDVMCALL_GET_QUOTE 0x10002
diff --git a/arch/x86/kvm/vmx/tdx.c b/arch/x86/kvm/vmx/tdx.c
index 545b03d9d10b..113bcf301ec9 100644
--- a/arch/x86/kvm/vmx/tdx.c
+++ b/arch/x86/kvm/vmx/tdx.c
@@ -734,6 +734,12 @@ void tdx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
local_irq_enable();
}
+static bool tdx_is_tdvmcall(struct kvm_vcpu *vcpu, u64
tdvmcall_exit_reason)
+{
+ return vmx_get_exit_reason(vcpu).basic == EXIT_REASON_TDCALL &&
+ to_tdx(vcpu)->tdvmcall_exit_reason == tdvmcall_exit_reason;
+}
+
bool tdx_interrupt_allowed(struct kvm_vcpu *vcpu)
{
/*
@@ -741,7 +747,7 @@ bool tdx_interrupt_allowed(struct kvm_vcpu *vcpu)
* interrupt is always allowed unless TDX guest calls TDVMCALL
with HLT,
* which passes the interrupt blocked flag.
*/
- return vmx_get_exit_reason(vcpu).basic != EXIT_REASON_HLT ||
+ return !tdx_is_tdvmcall(vcpu, TDVMCALL_HLT) ||
!to_tdx(vcpu)->vp_enter_args.r12;
}
@@ -759,7 +765,7 @@ static bool tdx_protected_apic_has_interrupt(struct
kvm_vcpu *vcpu)
* otherwise the interrupt would have been serviced at the
instruction
* boundary.
*/
- if (vmx_get_exit_reason(vcpu).basic != EXIT_REASON_HLT ||
+ if (!tdx_is_tdvmcall(vcpu, TDVMCALL_HLT) ||
to_tdx(vcpu)->vp_enter_args.r12)
return false;
@@ -906,28 +912,9 @@ int tdx_vcpu_pre_run(struct kvm_vcpu *vcpu)
return 1;
}
-static __always_inline u32 tdcall_to_vmx_exit_reason(struct kvm_vcpu *vcpu)
-{
- switch (tdvmcall_leaf(vcpu)) {
- case EXIT_REASON_CPUID:
- case EXIT_REASON_HLT:
- case EXIT_REASON_IO_INSTRUCTION:
- case EXIT_REASON_MSR_READ:
- case EXIT_REASON_MSR_WRITE:
- return tdvmcall_leaf(vcpu);
- case EXIT_REASON_EPT_VIOLATION:
- return EXIT_REASON_EPT_MISCONFIG;
- default:
- break;
- }
-
- return EXIT_REASON_TDCALL;
-}
-
static __always_inline u32 tdx_to_vmx_exit_reason(struct kvm_vcpu *vcpu)
{
struct vcpu_tdx *tdx = to_tdx(vcpu);
- u32 exit_reason;
switch (tdx->vp_enter_ret & TDX_SEAMCALL_STATUS_MASK) {
case TDX_SUCCESS:
@@ -935,30 +922,10 @@ static __always_inline u32
tdx_to_vmx_exit_reason(struct kvm_vcpu *vcpu)
case TDX_NON_RECOVERABLE_TD:
case TDX_NON_RECOVERABLE_TD_NON_ACCESSIBLE:
case TDX_NON_RECOVERABLE_TD_WRONG_APIC_MODE:
- break;
+ return (u32)tdx->vp_enter_ret;
default:
return -1u;
}
-
- exit_reason = tdx->vp_enter_ret;
-
- switch (exit_reason) {
- case EXIT_REASON_TDCALL:
- if (tdvmcall_exit_type(vcpu))
- return EXIT_REASON_VMCALL;
-
- return tdcall_to_vmx_exit_reason(vcpu);
- case EXIT_REASON_EPT_MISCONFIG:
- /*
- * Defer KVM_BUG_ON() until tdx_handle_exit() because
this is in
- * non-instrumentable code with interrupts disabled.
- */
- return -1u;
- default:
- break;
- }
-
- return exit_reason;
}
static noinstr void tdx_vcpu_enter_exit(struct kvm_vcpu *vcpu)
@@ -1096,9 +1063,6 @@ fastpath_t tdx_vcpu_run(struct kvm_vcpu *vcpu, u64
run_flags)
kvm_clear_available_registers(vcpu, ~TDX_REGS_AVAIL_SET);
- if (unlikely(tdx->vp_enter_ret == EXIT_REASON_EPT_MISCONFIG))
- return EXIT_FASTPATH_NONE;
-
if (unlikely((tdx->vp_enter_ret & TDX_SW_ERROR) == TDX_SW_ERROR))
return EXIT_FASTPATH_NONE;
@@ -1589,27 +1553,6 @@ static int
tdx_setup_event_notify_interrupt(struct kvm_vcpu *vcpu)
return 0;
}
-static int handle_tdvmcall(struct kvm_vcpu *vcpu)
-{
- switch (tdvmcall_leaf(vcpu)) {
- case TDVMCALL_MAP_GPA:
- return tdx_map_gpa(vcpu);
- case TDVMCALL_REPORT_FATAL_ERROR:
- return tdx_report_fatal_error(vcpu);
- case TDVMCALL_GET_TD_VM_CALL_INFO:
- return tdx_get_td_vm_call_info(vcpu);
- case TDVMCALL_GET_QUOTE:
- return tdx_get_quote(vcpu);
- case TDVMCALL_SETUP_EVENT_NOTIFY_INTERRUPT:
- return tdx_setup_event_notify_interrupt(vcpu);
- default:
- break;
- }
-
- tdvmcall_set_return_code(vcpu, TDVMCALL_STATUS_SUBFUNC_UNSUPPORTED);
- return 1;
-}
-
void tdx_load_mmu_pgd(struct kvm_vcpu *vcpu, hpa_t root_hpa, int
pgd_level)
{
u64 shared_bit = (pgd_level == 5) ? TDX_SHARED_BIT_PWL_5 :
@@ -2024,12 +1967,56 @@ int tdx_complete_emulated_msr(struct kvm_vcpu
*vcpu, int err)
return 1;
}
- if (vmx_get_exit_reason(vcpu).basic == EXIT_REASON_MSR_READ)
+ if (tdx_is_tdvmcall(vcpu, TDVMCALL_MSR_READ))
tdvmcall_set_return_val(vcpu, kvm_read_edx_eax(vcpu));
return 1;
}
+static int handle_tdcall(struct kvm_vcpu *vcpu)
+{
+ struct vcpu_tdx *tdx = to_tdx(vcpu);
+
+ if (tdvmcall_exit_type(vcpu)) {
+ tdx->tdvmcall_exit_reason = EXIT_REASON_VMCALL;
+ return tdx_emulate_vmcall(vcpu);
+ }
+
+ tdx->tdvmcall_exit_reason = tdvmcall_leaf(vcpu);
+ switch (tdvmcall_leaf(vcpu)) {
+ case TDVMCALL_CPUID:
+ return tdx_emulate_cpuid(vcpu);
+ case TDVMCALL_HLT:
+ return kvm_emulate_halt_noskip(vcpu);
+ case TDVMCALL_IO_INSTRUCTION:
+ return tdx_emulate_io(vcpu);
+ case TDVMCALL_MSR_READ:
+ kvm_ecx_write(vcpu, tdx->vp_enter_args.r12);
+ return kvm_emulate_rdmsr(vcpu);
+ case TDVMCALL_MSR_WRITE:
+ kvm_ecx_write(vcpu, tdx->vp_enter_args.r12);
+ kvm_eax_write(vcpu, tdx->vp_enter_args.r13);
+ kvm_edx_write(vcpu, tdx->vp_enter_args.r13 >> 32);
+ return kvm_emulate_wrmsr(vcpu);
+ case TDVMCALL_MMIO:
+ return tdx_emulate_mmio(vcpu);
+ case TDVMCALL_MAP_GPA:
+ return tdx_map_gpa(vcpu);
+ case TDVMCALL_REPORT_FATAL_ERROR:
+ return tdx_report_fatal_error(vcpu);
+ case TDVMCALL_GET_TD_VM_CALL_INFO:
+ return tdx_get_td_vm_call_info(vcpu);
+ case TDVMCALL_GET_QUOTE:
+ return tdx_get_quote(vcpu);
+ case TDVMCALL_SETUP_EVENT_NOTIFY_INTERRUPT:
+ return tdx_setup_event_notify_interrupt(vcpu);
+ default:
+ break;
+ }
+
+ tdvmcall_set_return_code(vcpu, TDVMCALL_STATUS_SUBFUNC_UNSUPPORTED);
+ return 1;
+}
int tdx_handle_exit(struct kvm_vcpu *vcpu, fastpath_t fastpath)
{
@@ -2040,11 +2027,6 @@ int tdx_handle_exit(struct kvm_vcpu *vcpu,
fastpath_t fastpath)
if (fastpath != EXIT_FASTPATH_NONE)
return 1;
- if (unlikely(vp_enter_ret == EXIT_REASON_EPT_MISCONFIG)) {
- KVM_BUG_ON(1, vcpu->kvm);
- return -EIO;
- }
-
/*
* Handle TDX SW errors, including TDX_SEAMCALL_UD,
TDX_SEAMCALL_GP and
* TDX_SEAMCALL_VMFAILINVALID.
@@ -2086,26 +2068,12 @@ int tdx_handle_exit(struct kvm_vcpu *vcpu,
fastpath_t fastpath)
case EXIT_REASON_EXTERNAL_INTERRUPT:
++vcpu->stat.irq_exits;
return 1;
- case EXIT_REASON_CPUID:
- return tdx_emulate_cpuid(vcpu);
- case EXIT_REASON_HLT:
- return kvm_emulate_halt_noskip(vcpu);
case EXIT_REASON_TDCALL:
- return handle_tdvmcall(vcpu);
- case EXIT_REASON_VMCALL:
- return tdx_emulate_vmcall(vcpu);
- case EXIT_REASON_IO_INSTRUCTION:
- return tdx_emulate_io(vcpu);
- case EXIT_REASON_MSR_READ:
- kvm_ecx_write(vcpu, tdx->vp_enter_args.r12);
- return kvm_emulate_rdmsr(vcpu);
- case EXIT_REASON_MSR_WRITE:
- kvm_ecx_write(vcpu, tdx->vp_enter_args.r12);
- kvm_eax_write(vcpu, tdx->vp_enter_args.r13);
- kvm_edx_write(vcpu, tdx->vp_enter_args.r13 >> 32);
- return kvm_emulate_wrmsr(vcpu);
+ return handle_tdcall(vcpu);
case EXIT_REASON_EPT_MISCONFIG:
- return tdx_emulate_mmio(vcpu);
+ /* EPT MISCONFIGs are *always* KVM/kernel bugs. */
+ KVM_BUG_ON(1, vcpu->kvm);
+ return -EIO;
case EXIT_REASON_EPT_VIOLATION:
return tdx_handle_ept_violation(vcpu);
case EXIT_REASON_OTHER_SMI:
diff --git a/arch/x86/kvm/vmx/tdx.h b/arch/x86/kvm/vmx/tdx.h
index ac8323a68b16..fc9c0c462f77 100644
--- a/arch/x86/kvm/vmx/tdx.h
+++ b/arch/x86/kvm/vmx/tdx.h
@@ -68,6 +68,22 @@ struct vcpu_tdx {
u64 vp_enter_ret;
+ /*
+ * Valid only when exit_reason is EXIT_REASON_TDCALL.
+ *
+ * When r10 of TDCALL is 0. It means the TDCALL is invoked for a
+ * TDVMCALL leaf that is defined by the GHCI spec. In this case,
+ * tdvmcall_exit_reason is set to the value of r11.
+ *
+ * When r10 of TDCALL is not 0. KVM treats it as a KVM hypercall.
+ * In this case, tdvmcall_exit_reason is set to EXIT_REASON_VMCALL,
+ * as what to invoke KVM hypercall in VMX. If GHCI defines the
+ * number of EXIT_REASON_VMCALL to a new TDVMCALL leaf in the future
+ * and KVM is going to support it, KVM will need to choose a new
value
+ * for tdvmcall_exit_reason to distinguish it from the KVM
hypercall.
+ */
+ u64 tdvmcall_exit_reason;
+
enum vcpu_tdx_state state;
u64 map_gpa_next;
next prev parent reply other threads:[~2026-08-17 2:49 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-12 8:02 [PATCH v3 0/4] KVM: TDX: Enable VM-DoS Prevention Features for TDX Xiaoyao Li
2026-08-12 8:02 ` [PATCH v3 1/4] KVM: TDX: Enable Notify VM exit Xiaoyao Li
2026-08-12 8:02 ` [PATCH v3 2/4] KVM: TDX: Set bits 31:16 to 0 for the synthesized Exit Reason Xiaoyao Li
2026-08-12 21:58 ` Edgecombe, Rick P
2026-08-12 8:02 ` [PATCH v3 3/4] KVM: TDX: Don't assume exit_reason[31:16] as all-0 in tdx_to_vmx_exit_reason() Xiaoyao Li
2026-08-12 8:23 ` sashiko-bot
2026-08-12 9:07 ` Xiaoyao Li
2026-08-12 22:55 ` Edgecombe, Rick P
2026-08-13 23:44 ` Sean Christopherson
2026-08-14 1:02 ` Xiaoyao Li
2026-08-14 15:12 ` Sean Christopherson
2026-08-17 2:49 ` Xiaoyao Li [this message]
2026-08-17 18:19 ` Sean Christopherson
2026-08-18 8:14 ` Xiaoyao Li
2026-08-18 20:18 ` Sean Christopherson
2026-08-13 0:14 ` Sean Christopherson
2026-08-13 8:42 ` Xiaoyao Li
2026-08-12 8:02 ` [PATCH v3 4/4] KVM: TDX: Enable Bus Lock VM exit Xiaoyao Li
2026-08-12 8:20 ` sashiko-bot
2026-08-12 9:11 ` Xiaoyao Li
2026-08-13 0:06 ` Sean Christopherson
2026-08-12 22:58 ` Edgecombe, Rick P
2026-08-13 11:17 ` Xiaoyao Li
2026-08-13 14:43 ` Sean Christopherson
2026-08-14 0:49 ` Xiaoyao Li
2026-08-14 15:46 ` Sean Christopherson
2026-08-15 13:28 ` Xiaoyao Li
2026-08-17 18:06 ` Sean Christopherson
2026-08-18 7:47 ` Xiaoyao Li
2026-08-18 16:17 ` Sean Christopherson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=d67eb097-1223-4b68-92d4-1510d3d7be7c@intel.com \
--to=xiaoyao.li@intel.com \
--cc=kvm@vger.kernel.org \
--cc=pbonzini@redhat.com \
--cc=rick.p.edgecombe@intel.com \
--cc=sashiko-reviews@lists.linux.dev \
--cc=seanjc@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox