]> git.ipfire.org Git - thirdparty/kernel/linux.git/commit
KVM: x86: Always complete hypercall via function callback
authorSean Christopherson <seanjc@google.com>
Thu, 28 Nov 2024 00:43:43 +0000 (16:43 -0800)
committerPaolo Bonzini <pbonzini@redhat.com>
Sun, 22 Dec 2024 18:00:25 +0000 (13:00 -0500)
commitd9eb86a6f43d74f08ee3b6eb99ad7eb2a7d7fce0
tree0e7150b316f860344a24600e67fbb342bc947a5a
parent05a518b49dd6f674cd0b1fe1eb6c8f9c3953b63d
KVM: x86: Always complete hypercall via function callback

Finish "emulation" of KVM hypercalls by function callback, even when the
hypercall is handled entirely within KVM, i.e. doesn't require an exit to
userspace, and refactor __kvm_emulate_hypercall()'s return value to *only*
communicate whether or not KVM should exit to userspace or resume the
guest.

(Ab)Use vcpu->run->hypercall.ret to propagate the return value to the
callback, purely to avoid having to add a trampoline for every completion
callback.

Using the function return value for KVM's control flow eliminates the
multiplexed return value, where '0' for KVM_HC_MAP_GPA_RANGE (and only
that hypercall) means "exit to userspace".

Note, the unnecessary extra indirect call and thus potential retpoline
will be eliminated in the near future by converting the intermediate layer
to a macro.

Suggested-by: Binbin Wu <binbin.wu@linux.intel.com>
Suggested-by: Kai Huang <kai.huang@intel.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Kai Huang <kai.huang@intel.com>
Message-ID: <20241128004344.4072099-6-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/x86.c
arch/x86/kvm/x86.h