[PATCH] KVM: x86: Revert kvm_x86_ops.mem_enc_ioctl() back to an OPTIONAL hook

[PATCH] KVM: x86: Revert kvm_x86_ops.mem_enc_ioctl() back to an OPTIONAL hook

[Sean Christopherson]

Restore KVM's handling of a NULL kvm_x86_ops.mem_enc_ioctl, as the hook is
NULL on SVM when CONFIG_KVM_AMD_SEV=n, and TDX will soon follow suit.

  ------------[ cut here ]------------
  WARNING: CPU: 0 PID: 1 at arch/x86/include/asm/kvm-x86-ops.h:130 kvm_x86_vendor_init+0x178b/0x18e0
  Modules linked in:
  CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.15.0-rc2-dc1aead1a985-sink-vm #2 NONE
  Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
  RIP: 0010:kvm_x86_vendor_init+0x178b/0x18e0
  Call Trace:
   <TASK>
   svm_init+0x2e/0x60
   do_one_initcall+0x56/0x290
   kernel_init_freeable+0x192/0x1e0
   kernel_init+0x16/0x130
   ret_from_fork+0x30/0x50
   ret_from_fork_asm+0x1a/0x30
   </TASK>
  ---[ end trace 0000000000000000 ]---

Opportunistically drop the superfluous curly braces.

Link: https://lore.kernel.org/all/20250318-vverma7-cleanup_x86_ops-v2-4-701e82d6b779@intel.com
Fixes: b2aaf38ced69 ("KVM: TDX: Add place holder for TDX VM specific mem_enc_op ioctl")
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 arch/x86/include/asm/kvm-x86-ops.h | 2 +-
 arch/x86/kvm/x86.c                 | 7 +++++--
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/arch/x86/include/asm/kvm-x86-ops.h b/arch/x86/include/asm/kvm-x86-ops.h
index 79406bf07a1c..8d50e3e0a19b 100644
--- a/arch/x86/include/asm/kvm-x86-ops.h
+++ b/arch/x86/include/asm/kvm-x86-ops.h
@@ -127,7 +127,7 @@ KVM_X86_OP(leave_smm)
 KVM_X86_OP(enable_smi_window)
 #endif
 KVM_X86_OP_OPTIONAL(dev_get_attr)
-KVM_X86_OP(mem_enc_ioctl)
+KVM_X86_OP_OPTIONAL(mem_enc_ioctl)
 KVM_X86_OP_OPTIONAL(vcpu_mem_enc_ioctl)
 KVM_X86_OP_OPTIONAL(mem_enc_register_region)
 KVM_X86_OP_OPTIONAL(mem_enc_unregister_region)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index f6ce044b090a..247d54de101a 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -7324,10 +7324,13 @@ int kvm_arch_vm_ioctl(struct file *filp, unsigned int ioctl, unsigned long arg)
 		r = READ_ONCE(kvm->arch.default_tsc_khz);
 		goto out;
 	}
-	case KVM_MEMORY_ENCRYPT_OP: {
+	case KVM_MEMORY_ENCRYPT_OP:
+		r = -ENOTTY;
+		if (!kvm_x86_ops.mem_enc_ioctl)
+			goto out;
+
 		r = kvm_x86_call(mem_enc_ioctl)(kvm, argp);
 		break;
-	}
 	case KVM_MEMORY_ENCRYPT_REG_REGION: {
 		struct kvm_enc_region region;
 

base-commit: 45eb29140e68ffe8e93a5471006858a018480a45
-- 
2.49.0.906.g1f30a19c02-goog

Re: [PATCH] KVM: x86: Revert kvm_x86_ops.mem_enc_ioctl() back to an OPTIONAL hook

[Sean Christopherson]

On Fri, 02 May 2025 13:34:21 -0700, Sean Christopherson wrote:
> Restore KVM's handling of a NULL kvm_x86_ops.mem_enc_ioctl, as the hook is
> NULL on SVM when CONFIG_KVM_AMD_SEV=n, and TDX will soon follow suit.
> 
>   ------------[ cut here ]------------
>   WARNING: CPU: 0 PID: 1 at arch/x86/include/asm/kvm-x86-ops.h:130 kvm_x86_vendor_init+0x178b/0x18e0
>   Modules linked in:
>   CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.15.0-rc2-dc1aead1a985-sink-vm #2 NONE
>   Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
>   RIP: 0010:kvm_x86_vendor_init+0x178b/0x18e0
>   Call Trace:
>    <TASK>
>    svm_init+0x2e/0x60
>    do_one_initcall+0x56/0x290
>    kernel_init_freeable+0x192/0x1e0
>    kernel_init+0x16/0x130
>    ret_from_fork+0x30/0x50
>    ret_from_fork_asm+0x1a/0x30
>    </TASK>
>   ---[ end trace 0000000000000000 ]---
> 
> [...]

Applied very quickly to kvm-x86 vmx, as I want to get the TDX macro cleanups
applied.  I'll force push if someone finds an issue.

[1/1] KVM: x86: Revert kvm_x86_ops.mem_enc_ioctl() back to an OPTIONAL hook
      https://github.com/kvm-x86/linux/commit/f2d7993314a3

--
https://github.com/kvm-x86/linux/tree/next