Re: [GIT PULL] KVM/x86 changes for Linux 6.12

[Paolo Bonzini <pbonzini@redhat.com>]

On Sun, Sep 29, 2024 at 7:36 PM Linus Torvalds <torvalds@linux-foundation.org> wrote:
> The culprit is commit 590b09b1d88e ("KVM: x86: Register "emergency
> disable" callbacks when virt is enabled"), and the reason seems to be
> this:
>
>   #if IS_ENABLED(CONFIG_KVM_INTEL) || IS_ENABLED(CONFIG_KVM_AMD)
>   void cpu_emergency_register_virt_callback(cpu_emergency_virt_cb *callback);
>   ...
>
> ie if you have a config with KVM enabled, but neither KVM_INTEL nor
> KVM_AMD set, you don't get that callback thing.
>
> The fix may be something like the attached.

Yeah, there was an attempt in commit 6d55a94222db ("x86/reboot:
Unconditionally define cpu_emergency_virt_cb typedef") but that only
covers the headers and the !CONFIG_KVM case; not the !CONFIG_KVM_INTEL
&& !CONFIG_KVM_AMD one that you stumbled upon.

Your fix is not wrong, but there's no point in compiling kvm.ko if
nobody is using it.

This is what I'll test more and submit:

------------------ 8< ------------------
From: Paolo Bonzini <pbonzini@redhat.com>
Subject: [PATCH] KVM: x86: leave kvm.ko out of the build if no vendor module is requested
     
kvm.ko is nothing but library code shared by kvm-intel.ko and kvm-amd.ko.
It provides no functionality on its own and it is unnecessary unless one
of the vendor-specific module is compiled.  In particular, /dev/kvm is
not created until one of kvm-intel.ko or kvm-amd.ko is loaded.
     
Use CONFIG_KVM to decide if it is built-in or a module, but use the
vendor-specific modules for the actual decision on whether to build it.
     
This also fixes a build failure when CONFIG_KVM_INTEL and CONFIG_KVM_AMD
are both disabled.  The cpu_emergency_register_virt_callback() function
is called from kvm.ko, but it is only defined if at least one of
CONFIG_KVM_INTEL and CONFIG_KVM_AMD is provided.

Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
index 4287a8071a3a..aee054a91031 100644
--- a/arch/x86/kvm/Kconfig
+++ b/arch/x86/kvm/Kconfig
@@ -17,8 +17,8 @@ menuconfig VIRTUALIZATION
  
  if VIRTUALIZATION
  
-config KVM
-	tristate "Kernel-based Virtual Machine (KVM) support"
+config KVM_X86_COMMON
+	def_tristate KVM if KVM_INTEL || KVM_AMD
  	depends on HIGH_RES_TIMERS
  	depends on X86_LOCAL_APIC
  	select KVM_COMMON
@@ -46,6 +47,9 @@ config KVM
  	select KVM_GENERIC_HARDWARE_ENABLING
  	select KVM_GENERIC_PRE_FAULT_MEMORY
  	select KVM_WERROR if WERROR
+
+config KVM
+	tristate "Kernel-based Virtual Machine (KVM) support"
  	help
  	  Support hosting fully virtualized guest machines using hardware
  	  virtualization extensions.  You will need a fairly recent
diff --git a/arch/x86/kvm/Makefile b/arch/x86/kvm/Makefile
index 5494669a055a..4304c89d6b64 100644
--- a/arch/x86/kvm/Makefile
+++ b/arch/x86/kvm/Makefile
@@ -32,7 +32,7 @@ kvm-intel-y		+= vmx/vmx_onhyperv.o vmx/hyperv_evmcs.o
  kvm-amd-y		+= svm/svm_onhyperv.o
  endif
  
-obj-$(CONFIG_KVM)	+= kvm.o
+obj-$(CONFIG_KVM_X86_COMMON) += kvm.o
  obj-$(CONFIG_KVM_INTEL)	+= kvm-intel.o
  obj-$(CONFIG_KVM_AMD)	+= kvm-amd.o
  
------------------ 8< ------------------

On top of this, the CONFIG_KVM #ifdefs could be changed to either
CONFIG_KVM_X86_COMMON or (most of them) to CONFIG_KVM_INTEL; I started
cleaning up the Kconfigs a few months ago and it's time to finish it
off for 6.13.

Paolo