Re: [PATCH v4] KVM: Fix stack-out-of-bounds read in write_mmio

From: Marc Zyngier
Date: Fri Dec 15 2017 - 06:07:10 EST


On 15/12/17 01:40, Wanpeng Li wrote:
> From: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
>
> Reported by syzkaller:
>
> BUG: KASAN: stack-out-of-bounds in write_mmio+0x11e/0x270 [kvm]
> Read of size 8 at addr ffff8803259df7f8 by task syz-executor/32298
>
> CPU: 6 PID: 32298 Comm: syz-executor Tainted: G OE 4.15.0-rc2+ #18
> Hardware name: LENOVO ThinkCentre M8500t-N000/SHARKBAY, BIOS FBKTC1AUS 02/16/2016
> Call Trace:
> dump_stack+0xab/0xe1
> print_address_description+0x6b/0x290
> kasan_report+0x28a/0x370
> write_mmio+0x11e/0x270 [kvm]
> emulator_read_write_onepage+0x311/0x600 [kvm]
> emulator_read_write+0xef/0x240 [kvm]
> emulator_fix_hypercall+0x105/0x150 [kvm]
> em_hypercall+0x2b/0x80 [kvm]
> x86_emulate_insn+0x2b1/0x1640 [kvm]
> x86_emulate_instruction+0x39a/0xb90 [kvm]
> handle_exception+0x1b4/0x4d0 [kvm_intel]
> vcpu_enter_guest+0x15a0/0x2640 [kvm]
> kvm_arch_vcpu_ioctl_run+0x549/0x7d0 [kvm]
> kvm_vcpu_ioctl+0x479/0x880 [kvm]
> do_vfs_ioctl+0x142/0x9a0
> SyS_ioctl+0x74/0x80
> entry_SYSCALL_64_fastpath+0x23/0x9a
>
> The path of patched vmmcall will patch 3 bytes opcode 0F 01 C1(vmcall)
> to the guest memory, however, write_mmio tracepoint always prints 8 bytes
> through *(u64 *)val since kvm splits the mmio access into 8 bytes. This
> can result in stack-out-of-bounds read due to access the extra 5 bytes.
> This patch fixes it by just accessing the bytes which we operates on.
>
> Before patch:
>
> syz-executor-5567 [007] .... 51370.561696: kvm_mmio: mmio write len 3 gpa 0x10 val 0x1ffff10077c1010f
>
> After patch:
>
> syz-executor-13416 [002] .... 51302.299573: kvm_mmio: mmio write len 3 gpa 0x10 val 0xc1010f
>
> Reported-by: Dmitry Vyukov <dvyukov@xxxxxxxxxx>
> Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> Cc: Radim KrÄmÃÅ <rkrcmar@xxxxxxxxxx>
> Cc: Marc Zyngier <marc.zyngier@xxxxxxx>
> Cc: Christoffer Dall <christoffer.dall@xxxxxxxxxx>
> Signed-off-by: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
> ---
> v3 -> v4:
> * fix the arm tracepoint
> v2 -> v3:
> * fix sparse warning
> v1 -> v2:
> * do the memcpy in kvm_mmio tracepoint
>
> arch/x86/kvm/x86.c | 8 ++++----
> include/trace/events/kvm.h | 6 ++++--
> virt/kvm/arm/mmio.c | 6 +++---
> 3 files changed, 11 insertions(+), 9 deletions(-)
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 0f82e2c..c7071e7 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -4456,7 +4456,7 @@ static int vcpu_mmio_read(struct kvm_vcpu *vcpu, gpa_t addr, int len, void *v)
> addr, n, v))
> && kvm_io_bus_read(vcpu, KVM_MMIO_BUS, addr, n, v))
> break;
> - trace_kvm_mmio(KVM_TRACE_MMIO_READ, n, addr, *(u64 *)v);
> + trace_kvm_mmio(KVM_TRACE_MMIO_READ, n, addr, v);
> handled += n;
> addr += n;
> len -= n;
> @@ -4715,7 +4715,7 @@ static int read_prepare(struct kvm_vcpu *vcpu, void *val, int bytes)
> {
> if (vcpu->mmio_read_completed) {
> trace_kvm_mmio(KVM_TRACE_MMIO_READ, bytes,
> - vcpu->mmio_fragments[0].gpa, *(u64 *)val);
> + vcpu->mmio_fragments[0].gpa, val);
> vcpu->mmio_read_completed = 0;
> return 1;
> }
> @@ -4737,14 +4737,14 @@ static int write_emulate(struct kvm_vcpu *vcpu, gpa_t gpa,
>
> static int write_mmio(struct kvm_vcpu *vcpu, gpa_t gpa, int bytes, void *val)
> {
> - trace_kvm_mmio(KVM_TRACE_MMIO_WRITE, bytes, gpa, *(u64 *)val);
> + trace_kvm_mmio(KVM_TRACE_MMIO_WRITE, bytes, gpa, val);
> return vcpu_mmio_write(vcpu, gpa, bytes, val);
> }
>
> static int read_exit_mmio(struct kvm_vcpu *vcpu, gpa_t gpa,
> void *val, int bytes)
> {
> - trace_kvm_mmio(KVM_TRACE_MMIO_READ_UNSATISFIED, bytes, gpa, 0);
> + trace_kvm_mmio(KVM_TRACE_MMIO_READ_UNSATISFIED, bytes, gpa, NULL);
> return X86EMUL_IO_NEEDED;
> }
>
> diff --git a/include/trace/events/kvm.h b/include/trace/events/kvm.h
> index e4b0b8e..dfd2170 100644
> --- a/include/trace/events/kvm.h
> +++ b/include/trace/events/kvm.h
> @@ -211,7 +211,7 @@ TRACE_EVENT(kvm_ack_irq,
> { KVM_TRACE_MMIO_WRITE, "write" }
>
> TRACE_EVENT(kvm_mmio,
> - TP_PROTO(int type, int len, u64 gpa, u64 val),
> + TP_PROTO(int type, int len, u64 gpa, void *val),
> TP_ARGS(type, len, gpa, val),
>
> TP_STRUCT__entry(
> @@ -225,7 +225,9 @@ TRACE_EVENT(kvm_mmio,
> __entry->type = type;
> __entry->len = len;
> __entry->gpa = gpa;
> - __entry->val = val;
> + __entry->val = 0;
> + if (val)
> + memcpy(&__entry->val, val, min(8, len));

I'm not fond of hardcoded sizes; they tend to break as people change the
code around. Consider using sizeof(__entry->val) instead of 8.

> ),
>
> TP_printk("mmio %s len %u gpa 0x%llx val 0x%llx",
> diff --git a/virt/kvm/arm/mmio.c b/virt/kvm/arm/mmio.c
> index b6e715f..dac7ceb 100644
> --- a/virt/kvm/arm/mmio.c
> +++ b/virt/kvm/arm/mmio.c
> @@ -112,7 +112,7 @@ int kvm_handle_mmio_return(struct kvm_vcpu *vcpu, struct kvm_run *run)
> }
>
> trace_kvm_mmio(KVM_TRACE_MMIO_READ, len, run->mmio.phys_addr,
> - data);
> + &data);
> data = vcpu_data_host_to_guest(vcpu, data, len);
> vcpu_set_reg(vcpu, vcpu->arch.mmio_decode.rt, data);
> }
> @@ -182,14 +182,14 @@ int io_mem_abort(struct kvm_vcpu *vcpu, struct kvm_run *run,
> data = vcpu_data_guest_to_host(vcpu, vcpu_get_reg(vcpu, rt),
> len);
>
> - trace_kvm_mmio(KVM_TRACE_MMIO_WRITE, len, fault_ipa, data);
> + trace_kvm_mmio(KVM_TRACE_MMIO_WRITE, len, fault_ipa, &data);
> kvm_mmio_write_buf(data_buf, len, data);
>
> ret = kvm_io_bus_write(vcpu, KVM_MMIO_BUS, fault_ipa, len,
> data_buf);
> } else {
> trace_kvm_mmio(KVM_TRACE_MMIO_READ_UNSATISFIED, len,
> - fault_ipa, 0);
> + fault_ipa, NULL);
>
> ret = kvm_io_bus_read(vcpu, KVM_MMIO_BUS, fault_ipa, len,
> data_buf);
>

I've given it a go on an arm64 platform, which works as expected.
Assuming you address the above:

Reviewed-by: Marc Zyngier <marc.zyngier@xxxxxxx>
Tested-by: Marc Zyngier <marc.zyngier@xxxxxxx>

Thanks,

M.
--
Jazz is not dead. It just smells funny...