Re: [syzbot] [kvm-x86?] WARNING in kvm_apic_accept_events (2)

[Sean Christopherson <seanjc@google.com>]

On Thu, Jan 22, 2026, Alexander Potapenko wrote:
> On Wed, Nov 26, 2025 at 12:10 AM 'Sean Christopherson' via
> syzkaller-bugs <syzkaller-bugs@googlegroups.com> wrote:
> >
> > On Tue, Nov 25, 2025, syzbot wrote:
> > > syzbot has found a reproducer for the following issue on:
> > >
> > > HEAD commit:    8a2bcda5e139 Merge tag 'for-6.18/dm-fixes' of git://git.ke..
> > > git tree:       upstream
> > > console output: https://syzkaller.appspot.com/x/log.txt?x=1604f8b4580000
> > > kernel config:  https://syzkaller.appspot.com/x/.config?x=a1db0fea040c2a9f
> > > dashboard link: https://syzkaller.appspot.com/bug?extid=59f2c3a3fc4f6c09b8cd
> > > compiler:       Debian clang version 20.1.8 (++20250708063551+0c9f909b7976-1~exp1~20250708183702.136), Debian LLD 20.1.8
> > > syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=13ecf612580000
> > > C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=13d9cf42580000
> > >
> > > Downloadable assets:
> > > disk image (non-bootable): https://storage.googleapis.com/syzbot-assets/d900f083ada3/non_bootable_disk-8a2bcda5.raw.xz
> > > vmlinux: https://storage.googleapis.com/syzbot-assets/fc3f96645396/vmlinux-8a2bcda5.xz
> > > kernel image: https://storage.googleapis.com/syzbot-assets/e20aa7be5d33/bzImage-8a2bcda5.xz
> > >
> > > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > > Reported-by: syzbot+59f2c3a3fc4f6c09b8cd@syzkaller.appspotmail.com
> > >
> > > ------------[ cut here ]------------
> > > WARNING: CPU: 0 PID: 5495 at arch/x86/kvm/lapic.c:3483 kvm_apic_accept_events+0x341/0x490 arch/x86/kvm/lapic.c:3483
> > > Modules linked in:
> > > CPU: 0 UID: 0 PID: 5495 Comm: syz.0.17 Not tainted syzkaller #0 PREEMPT(full)
> > > Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2~bpo12+1 04/01/2014
> > > RIP: 0010:kvm_apic_accept_events+0x341/0x490 arch/x86/kvm/lapic.c:3483
> > > Code: eb 0c e8 32 da 71 00 eb 05 e8 2b da 71 00 45 31 ff 44 89 f8 5b 41 5c 41 5d 41 5e 41 5f 5d c3 cc cc cc cc cc e8 10 da 71 00 90 <0f> 0b 90 e9 ec fd ff ff 44 89 f9 80 e1 07 80 c1 03 38 c1 0f 8c 4f
> > > RSP: 0018:ffffc90002b2fbf0 EFLAGS: 00010293
> > > RAX: ffffffff814e3940 RBX: 0000000000000002 RCX: ffff88801f8ca480
> > > RDX: 0000000000000000 RSI: 0000000000000002 RDI: 0000000000000002
> > > RBP: 0000000000000000 R08: 0000000000000000 R09: ffffffff814689b6
> > > R10: dffffc0000000000 R11: ffffed1002268008 R12: 0000000000000002
> > > R13: dffffc0000000000 R14: ffff888042c95c00 R15: ffff8880113402d8
> > > FS:  000055558ab60500(0000) GS:ffff88808d72f000(0000) knlGS:0000000000000000
> > > CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> > > CR2: 0000200000002000 CR3: 0000000058d0b000 CR4: 0000000000352ef0
> > > Call Trace:
> > >  <TASK>
> > >  kvm_arch_vcpu_ioctl_get_mpstate+0x128/0x480 arch/x86/kvm/x86.c:12147
> > >  kvm_vcpu_ioctl+0x625/0xe90 virt/kvm/kvm_main.c:4539
> > >  vfs_ioctl fs/ioctl.c:51 [inline]
> > >  __do_sys_ioctl fs/ioctl.c:597 [inline]
> > >  __se_sys_ioctl+0xfc/0x170 fs/ioctl.c:583
> > >  do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
> > >  do_syscall_64+0xfa/0xfa0 arch/x86/entry/syscall_64.c:94
> > >  entry_SYSCALL_64_after_hwframe+0x77/0x7f
> > > RIP: 0033:0x7f918bd8f749
> > > Code: ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 a8 ff ff ff f7 d8 64 89 01 48
> > > RSP: 002b:00007ffc74a3c2a8 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
> > > RAX: ffffffffffffffda RBX: 00007f918bfe5fa0 RCX: 00007f918bd8f749
> > > RDX: 0000000000000000 RSI: 000000008004ae98 RDI: 0000000000000005
> > > RBP: 00007f918be13f91 R08: 0000000000000000 R09: 0000000000000000
> > > R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
> > > R13: 00007f918bfe5fa0 R14: 00007f918bfe5fa0 R15: 0000000000000003
> > >  </TASK>
> >
> > Syzbot outsmarted me once again.  I thought I had made this impossible in commit
> > 0fe3e8d804fd ("KVM: x86: Move INIT_RECEIVED vs. INIT/SIPI blocked check to KVM_RUN"),
> > but now syzkaller is triggering the WARN by swapping the order and stuffing VMXON
> > after INIT (ignore the EINVAL, KVM gets through enter_vmx_operation() before detecting
> > bad guest state).
> >
> >   ioctl(5, KVM_SET_MP_STATE, 0x200000000000) = 0
> >   ioctl(5, KVM_SET_NESTED_STATE, 0x200000000a80) = -1 EINVAL (Invalid argument)
> >   ioctl(5, KVM_GET_MP_STATE, 0)           = -1 EFAULT (Bad address)
> >
> > At this point, I'm leaning strongly towards dropping the WARN as it's not helping,
> > and userspace doing odd things is completely uninteresting.
> >
> > diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
> > index 7b1b8f450f4c..df2a69da11b7 100644
> > --- a/arch/x86/kvm/lapic.c
> > +++ b/arch/x86/kvm/lapic.c
> > @@ -3521,7 +3521,6 @@ int kvm_apic_accept_events(struct kvm_vcpu *vcpu)
> >          * wait-for-SIPI (WFS).
> >          */
> >         if (!kvm_apic_init_sipi_allowed(vcpu)) {
> > -               WARN_ON_ONCE(vcpu->arch.mp_state == KVM_MP_STATE_INIT_RECEIVED);
> >                 clear_bit(KVM_APIC_SIPI, &apic->pending_events);
> >                 return 0;
> >         }
> 
> Hi Sean, this is still biting us.
> Have you made up your mind on the above patch?

I'll properly post and apply the above.  I have some ideas on how to preserve
the basic gist of the sanity checks, though I hope I wrote them down somewhere,
because I can't remember what they were...