WARNING in kvm_mmu_uninit_tdp_mmu

WARNING in kvm_mmu_uninit_tdp_mmu

[Sabri N. Ferreiro]

Hi,

When using Syzkaller to fuzz the Linux kernel, it triggers the following crash.

HEAD commit: a7904a538933 Linux 5.16-rc6
git tree: upstream
console output: https://pastebin.com/raw/keWCUeJ2
kernel config: https://docs.google.com/document/d/1w94kqQ4ZSIE6BW-5WIhqp4_Zh7XTPH57L5OF2Xb6O6o/view
C reproducer: https://pastebin.com/raw/kSxa6Yit
Syzlang reproducer: https://pastebin.com/raw/2RMu8p6E

If you fix this issue, please add the following tag to the commit:
Reported-by: Yuheng Shen mosesfonscqf75@gmail.com

------------[ cut here ]------------
WARNING: CPU: 5 PID: 29657 at arch/x86/kvm/mmu/tdp_mmu.c:46
kvm_mmu_uninit_tdp_mmu+0xb9/0xf0
Modules linked in:
CPU: 5 PID: 29657 Comm: syz-executor.5 Not tainted 5.16.0-rc8+ #10
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
1.13.0-1ubuntu1.1 04/01/2014
RIP: 0010:kvm_mmu_uninit_tdp_mmu+0xb9/0xf0
Code: ea 03 80 3c 02 00 75 39 48 8b 83 e8 ae 00 00 48 39 c5 75 1a e8
48 86 5a 00 e8 e3 bf 46 00 5b 5d e9 3c 86 5a 00 e8 37 86 5a 00 <0f> 0b
eb b7 e8 2e 86 5a 00 0f 0b eb dd e8 a5 38 a1 00 e9 60 ff ff
RSP: 0018:ffffc90016057b30 EFLAGS: 00010246
RAX: 0000000000000000 RBX: ffffc90015f89000 RCX: ffff888130d18000
RDX: 0000000000000000 RSI: ffff888130d18000 RDI: 0000000000000002
RBP: ffffc90015f93ef8 R08: ffffffff811cc8f9 R09: 0000000000000000
R10: 0000000000000001 R11: fffffbfff20e793a R12: ffffc90015f8b1c8
R13: 0000000000000003 R14: ffffc90015f8b1e8 R15: dffffc0000000000
FS:  00007f7f240cb700(0000) GS:ffff888135d00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000055f76caf3498 CR3: 0000000123d22000 CR4: 0000000000350ee0
Call Trace:
 <TASK>
 kvm_arch_destroy_vm+0x42b/0x5b0
 kvm_put_kvm+0x4e9/0xbd0
 kvm_vcpu_release+0x4d/0x70
 __fput+0x286/0x9f0
 task_work_run+0xe0/0x1a0
 get_signal+0x1fb5/0x25e0
 arch_do_signal_or_restart+0x2ed/0x1c40
 exit_to_user_mode_prepare+0x192/0x2a0
 syscall_exit_to_user_mode+0x19/0x60
 do_syscall_64+0x42/0xb0
 entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7f7f257dd89d
Code: 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 f3 0f 1e fa 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 bc ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f7f240cac28 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
RAX: 0000000000000000 RBX: 00007f7f258fd2a0 RCX: 00007f7f257dd89d
RDX: 0000000000000000 RSI: 000000000000ae9a RDI: 0000000000000005
RBP: 00007f7f2584a00d R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 00007ffdf486b75f R14: 00007f7f258fd2a0 R15: 00007f7f240cadc0
 </TASK>

Re: WARNING in kvm_mmu_uninit_tdp_mmu

[Sean Christopherson]

On Sun, Jan 09, 2022, Sabri N. Ferreiro wrote:
> Hi,
> 
> When using Syzkaller to fuzz the Linux kernel, it triggers the following crash.
> 
> HEAD commit: a7904a538933 Linux 5.16-rc6
> git tree: upstream
> console output: https://pastebin.com/raw/keWCUeJ2
> kernel config: https://docs.google.com/document/d/1w94kqQ4ZSIE6BW-5WIhqp4_Zh7XTPH57L5OF2Xb6O6o/view
> C reproducer: https://pastebin.com/raw/kSxa6Yit
> Syzlang reproducer: https://pastebin.com/raw/2RMu8p6E
> 
> If you fix this issue, please add the following tag to the commit:
> Reported-by: Yuheng Shen mosesfonscqf75@gmail.com
> 
> ------------[ cut here ]------------
> WARNING: CPU: 5 PID: 29657 at arch/x86/kvm/mmu/tdp_mmu.c:46
> kvm_mmu_uninit_tdp_mmu+0xb9/0xf0

I wasn't able to reproduce to confirm, but this is more than likely fixed by
commit 3a0f64de479c ("KVM: x86/mmu: Don't advance iterator after restart due to yielding"),
which didn't land until 5.16-rc7.