On Wed, May 26, 2021, Wanpeng Li wrote: > From: Wanpeng Li <wanpengli@xxxxxxxxxxx> > > Reported by syzkaller: > > WARNING: CPU: 7 PID: 10526 at /home/kernel/ssd/linux/arch/x86/kvm//x86.c:7621 x86_emulate_instruction+0x41b/0x510 [kvm] "/home/kernel/ssd/linux/" can be omitted to make the line length a bit shorter. checkpatch also complains about using absolute path instead of relative path. > RIP: 0010:x86_emulate_instruction+0x41b/0x510 [kvm] > Call Trace: > kvm_mmu_page_fault+0x126/0x8f0 [kvm] > vmx_handle_exit+0x11e/0x680 [kvm_intel] > vcpu_enter_guest+0xd95/0x1b40 [kvm] > kvm_arch_vcpu_ioctl_run+0x377/0x6a0 [kvm] > kvm_vcpu_ioctl+0x389/0x630 [kvm] > __x64_sys_ioctl+0x8e/0xd0 > do_syscall_64+0x3c/0xb0 > entry_SYSCALL_64_after_hwframe+0x44/0xae > > Commit 4a1e10d5b5d8c (KVM: x86: handle hardware breakpoints during emulation()) > adds hardware breakpoints check before emulation the instruction and parts of > emulation context initialization, actually we don't have the EMULTYPE_NO_DECODE flag > here and the emulation context will not be reused. Commit c8848cee74ff (KVM: x86: > set ctxt->have_exception in x86_decode_insn()) triggers the warning because it > catches the stale emulation context has #UD, however, it is not during instruction > decoding which should result in EMULATION_FAILED. This patch fixes it by moving > the second part emulation context initialization into init_emulate_ctxt() and > before hardware breakpoints check. > > syzkaller source: https://syzkaller.appspot.com/x/repro.c?x=134683fdd00000 > > Reported-by: syzbot+71271244f206d17f6441@xxxxxxxxxxxxxxxxxxxxxxxxx > Fixes: 4a1e10d5b5d8 (KVM: x86: handle hardware breakpoints during emulation) > Signed-off-by: Wanpeng Li <wanpengli@xxxxxxxxxxx> > --- > v2 -> v3: > * squash ctxt->ud > v1 -> v2: > * move the second part emulation context initialization into init_emulate_ctxt() > > arch/x86/kvm/x86.c | 14 +++++++------- > 1 file changed, 7 insertions(+), 7 deletions(-) > > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index bbc4e04..ae47b19 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -7226,6 +7226,13 @@ static void init_emulate_ctxt(struct kvm_vcpu *vcpu) > BUILD_BUG_ON(HF_SMM_MASK != X86EMUL_SMM_MASK); > BUILD_BUG_ON(HF_SMM_INSIDE_NMI_MASK != X86EMUL_SMM_INSIDE_NMI_MASK); > > + ctxt->interruptibility = 0; > + ctxt->have_exception = false; > + ctxt->exception.vector = -1; > + ctxt->perm_ok = false; > + > + ctxt->ud = emulation_type & EMULTYPE_TRAP_UD; "ctxt->ud" should be left where it is in patch 01. "emulation_type" isn't passed to init_emulate_ctxt(), and I don't see any reason to add it to the params since ctxt->ud is only consumed by x86_decode_insn(), i.e. moving ctxt->ud isn't necessary to fix the bug. arch/x86/kvm/x86.c: In function ‘init_emulate_ctxt’: arch/x86/kvm/x86.c:7236:13: error: ‘emulation_type’ undeclared (first use in this function); did you mean ‘exception_type’? 7236 | ctxt->ud = emulation_type & EMULTYPE_TRAP_UD; | ^~~~~~~~~~~~~~ | exception_type arch/x86/kvm/x86.c:7236:13: note: each undeclared identifier is reported only once for each function it appears in > + > init_decode_cache(ctxt); > vcpu->arch.emulate_regs_need_sync_from_vcpu = false; > } > @@ -7561,13 +7568,6 @@ int x86_decode_emulated_instruction(struct kvm_vcpu *vcpu, int emulation_type, > kvm_vcpu_check_breakpoint(vcpu, &r)) > return r; > > - ctxt->interruptibility = 0; > - ctxt->have_exception = false; > - ctxt->exception.vector = -1; > - ctxt->perm_ok = false; > - > - ctxt->ud = emulation_type & EMULTYPE_TRAP_UD; > - > r = x86_decode_insn(ctxt, insn, insn_len); > > trace_kvm_emulate_insn_start(vcpu); > -- > 2.7.4 >
