KVM: X86: Fix warning caused by stale emulation context
PATCH v4 1/2 KVM: X86: Fix warning caused by stale emulation context
From: Wanpeng Li <[email protected]>
Reported by syzkaller:
WARNING: CPU: 7 PID: 10526 at linux/arch/x86/kvm//x86.c:7621 x86_emulate_instruction+0x41b/0x510 [kvm] 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 4a1e10d5b5d8 ("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. The ctxt->ud will be dropped by a follow-up patch.
syzkaller source: https://syzkaller.appspot.com/x/repro.c?x=134683fdd00000
---
v3 -> v4:
* warning relative path
* not move ctxt->ud
v2 -> v3:
* squash ctxt->ud
v1 -> v2:
* move the second part emulation context initialization into init_emulate_ctxt()
arch/x86/kvm/x86.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)
arch/x86/kvm/x86.c +5 -5
@@ -7226,6 +7226,11 @@ 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; +init_decode_cache (ctxt); vcpu->arch.emulate_regs_need_sync_from_vcpu = false; }
@@ -7561,11 +7566,6 @@ 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);
Signature · 1 line
2.7.4
PATCH v4 2/2 KVM: X86: Kill off ctxt->ud
From: Wanpeng Li <[email protected]>
ctxt->ud is consumed only by x86_decode_insn(), we can kill it off by passing emulation_type to x86_decode_insn() and dropping ctxt->ud altogether. Tracking that info in ctxt for literally one call is silly.
--- arch/x86/kvm/emulate.c | 5 +++-- arch/x86/kvm/kvm_emulate.h | 3 +-- arch/x86/kvm/x86.c | 4 +--- 3 files changed, 5 insertions(+), 7 deletions(-)
arch/x86/kvm/emulate.c +3 -2
@@ -5111,7 +5111,7 @@ static int decode_operand(struct x86_emulate_ctxt *ctxt, struct operand *op,return rc; } -int x86_decode_insn (struct x86_emulate_ctxt *ctxt,void *insn,int insn_len) +int x86_decode_insn (struct x86_emulate_ctxt *ctxt,void *insn,int insn_len,int emulation_type) {int rc =X86EMUL_CONTINUE ;int mode = ctxt->mode;
@@ -5322,7 +5322,8 @@ int x86_decode_insn(struct x86_emulate_ctxt *ctxt, void *insn, int insn_len) ctxt->execute = opcode.u.execute; -if (unlikely (ctxt->ud) &&likely (!(ctxt->d & EmulateOnUD))) +if (unlikely (emulation_type &EMULTYPE_TRAP_UD ) && +likely (!(ctxt->d & EmulateOnUD)))return EMULATION_FAILED ; if (unlikely(ctxt->d &
arch/x86/kvm/kvm_emulate.h +1 -2
@@ -314,7 +314,6 @@ struct x86_emulate_ctxt {int interruptibility;bool perm_ok;/* do not check permissions if true */ -bool ud;/* inject an #UD if host doesn't support insn */ bool tf;/* TF value before instruction (after for syscall/sysret) */ bool have_exception;
@@ -491,7 +490,7 @@ enum x86_intercept {#define X86EMUL_MODE_HOST X86EMUL_MODE_PROT64#endif -int x86_decode_insn (struct x86_emulate_ctxt *ctxt,void *insn,int insn_len); +int x86_decode_insn (struct x86_emulate_ctxt *ctxt,void *insn,int insn_len,int emulation_type);bool x86_page_table_writing_insn (struct x86_emulate_ctxt *ctxt);#define EMULATION_FAILED -1#define EMULATION_OK 0
arch/x86/kvm/x86.c +1 -3
@@ -7566,9 +7566,7 @@ int x86_decode_emulated_instruction(struct kvm_vcpu *vcpu, int emulation_type, kvm_vcpu_check_breakpoint(vcpu, &r))return r; - ctxt->ud = emulation_type &EMULTYPE_TRAP_UD ; - - r =x86_decode_insn (ctxt, insn, insn_len); + r =x86_decode_insn (ctxt, insn, insn_len, emulation_type);trace_kvm_emulate_insn_start (vcpu); ++vcpu->stat.insn_emulation;
Signature · 1 line
2.7.4
Re: [PATCH v4 2/2] KVM: X86: Kill off ctxt->ud
Quoted 9 lines · Wanpeng Li
From: Wanpeng Li <[email protected]>
ctxt->udis consumed only byx86_decode_insn(), we can kill it off by passingemulation_typetox86_decode_insn()and droppingctxt->udaltogether. Tracking that info in ctxt for literally one call is silly.Suggested-by Sean Christopherson <[email protected]>Signed-off-by Wanpeng Li <[email protected]>---
Quoted 27 lines · Wanpeng Li
From: Wanpeng Li <[email protected]>
Reported by syzkaller:
WARNING: CPU: 7 PID: 10526 at linux/arch/x86/kvm//x86.c:7621 x86_emulate_instruction+0x41b/0x510 [kvm] 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+0x44/0xaeCommit
4a1e10d5b5d8("KVM: x86: handle hardware breakpoints duringemulation()) adds hardware breakpoints check before emulation the instruction and parts of emulation context initialization, actually we don't have theEMULTYPE_NO_DECODEflag here and the emulation context will not be reused. Commitc8848cee74ff("KVM: x86:
setctxt->have_exceptioninx86_decode_insn()) triggers the warning because it catches the stale emulation context has #UD, however, it is not during instruction decoding which should result inEMULATION_FAILED. This patch fixes it by moving the second part emulation context initialization intoinit_emulate_ctxt()and before hardware breakpoints check. Thectxt->udwill be dropped by a follow-up patch.
Quoted 6 lines
syzkaller source: https://syzkaller.appspot.com/x/repro.c?x=134683fdd00000
Reported-by [email protected]Fixes 4a1e10d5b5d8 (KVM: x86: handle hardware breakpoints during emulation)Signed-off-by Wanpeng Li <[email protected]>---