kvm: GPF in kvm_irq_map_gsi

57 views
Skip to first unread message

Dmitry Vyukov

unread,
Feb 15, 2016, 8:30:54 AM2/15/16
to Gleb Natapov, Paolo Bonzini, Thomas Gleixner, Ingo Molnar, H. Peter Anvin, x...@kernel.org, k...@vger.kernel.org, LKML, Eric Northup, Andrew Honig, Steve Rutherford, Robert Swiecki, Kostya Serebryany, Alexander Potapenko, syzkaller
Hello,

The following program triggers GPF in kvm_irq_map_gsi:

general protection fault: 0000 [#49] SMP DEBUG_PAGEALLOC KASAN
Modules linked in:
CPU: 0 PID: 5314 Comm: kvm-pit/4950 Tainted: G D 4.5.0-rc3+ #326
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
task: ffff88005b848000 ti: ffff880056b70000 task.ti: ffff880056b70000
RIP: 0010:[<ffffffff81029221>] [<ffffffff81029221>] kvm_irq_map_gsi+0x81/0x430
RSP: 0018:ffff880056b77b40 EFLAGS: 00010206
RAX: dffffc0000000000 RBX: 0000000000000000 RCX: 0000000000000002
RDX: 0000000000000024 RSI: ffff880056b77bd0 RDI: 0000000000000120
RBP: ffff880056b77b70 R08: 0000000000000000 R09: 0000000000000000
R10: ffff88005b848000 R11: ffff880055fd0390 R12: 0000000000000000
R13: ffff880055fd0000 R14: ffff880056b77bd0 R15: 0000000000000000
FS: 0000000000000000(0000) GS:ffff88003ec00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 00007f87947bbd90 CR3: 000000002a2ed000 CR4: 00000000000026f0
Stack:
ffff880055fd0390 ffff880055fd0240 ffff880055fd0390 ffff880056b77bd0
0000000000000000 0000000000000000 ffff880056b77cd8 ffffffff81029a7c
ffffffff810299fe ffff88005b848000 1ffff1000ad6ef76 0000000200000001
Call Trace:
[<ffffffff81029a7c>] kvm_set_irq+0x15c/0x440
arch/x86/kvm/../../../virt/kvm/irqchip.c:94
[<ffffffff810dbd39>] pit_do_work+0x109/0x350 arch/x86/kvm/i8254.c:297
[<ffffffff813b8309>] kthread_worker_fn+0x2d9/0x660 kernel/kthread.c:595
[<ffffffff813b7edf>] kthread+0x23f/0x2d0 drivers/block/aoe/aoecmd.c:1303
[<ffffffff866617af>] ret_from_fork+0x3f/0x70 arch/x86/entry/entry_64.S:468
Code: 80 3d 59 0f 7f 07 00 0f 84 d8 02 00 00 e8 a8 29 54 00 49 8d bc
24 20 01 00 00 48 b8 00 00 00 00 00 fc ff df 48 89 fa 48 c1 ea 03 <0f>
b6 04 02 84 c0 74 08 3c 03 0f 8e 8e 03 00 00 41 39 9c 24 20
RIP [<ffffffff81029221>] kvm_irq_map_gsi+0x81/0x430
arch/x86/kvm/../../../virt/kvm/irqchip.c:43
RSP <ffff880056b77b40>
---[ end trace a9bcc650712efe6c ]---


// autogenerated by syzkaller (http://github.com/google/syzkaller)
#include <pthread.h>
#include <stdint.h>
#include <string.h>
#include <sys/syscall.h>
#include <unistd.h>

long r[74];

void* thr(void* arg)
{
switch ((long)arg) {
case 0:
r[0] = syscall(SYS_mmap, 0x20000000ul, 0xe000ul, 0x3ul, 0x32ul,
0xfffffffffffffffful, 0x0ul);
break;
case 1:
memcpy((void*)0x2000d000, "\x2f\x64\x65\x76\x2f\x6b\x76\x6d\x00",
9);
r[2] = syscall(SYS_open, "/dev/kvm", 0x2000ul, 0, 0, 0);
break;
case 2:
r[3] = syscall(SYS_ioctl, r[2], 0xae01ul, 0x0ul, 0, 0, 0);
break;
case 3:
r[4] = syscall(SYS_ioctl, r[3], 0xae60ul, 0, 0, 0, 0);
break;
case 4:
*(uint32_t*)0x20009000 = (uint32_t)0x4;
*(uint32_t*)0x20009004 = (uint32_t)0x0;
*(uint32_t*)0x20009008 = (uint32_t)0x0;
*(uint32_t*)0x2000900c = (uint32_t)0x0;
*(uint32_t*)0x20009010 = (uint32_t)0x0;
*(uint32_t*)0x20009014 = (uint32_t)0x0;
*(uint32_t*)0x20009018 = (uint32_t)0x0;
*(uint32_t*)0x2000901c = (uint32_t)0x0;
*(uint32_t*)0x20009020 = (uint32_t)0x0;
*(uint32_t*)0x20009024 = (uint32_t)0x0;
*(uint32_t*)0x20009028 = (uint32_t)0x0;
*(uint32_t*)0x2000902c = (uint32_t)0x0;
*(uint32_t*)0x20009030 = (uint32_t)0x0;
*(uint32_t*)0x20009034 = (uint32_t)0x0;
*(uint32_t*)0x20009038 = (uint32_t)0x0;
*(uint32_t*)0x2000903c = (uint32_t)0x0;
r[21] =
syscall(SYS_ioctl, r[3], 0x4040ae77ul, 0x20009000ul, 0, 0, 0);
break;
case 5:
*(uint32_t*)0x2000a6b9 = (uint32_t)0x3e;
*(uint16_t*)0x2000a6bd = (uint16_t)0x8;
*(uint8_t*)0x2000a6bf = (uint8_t)0x8d4;
*(uint8_t*)0x2000a6c0 = (uint8_t)0xffffffffffff5fe9;
*(uint8_t*)0x2000a6c1 = (uint8_t)0x80000001;
*(uint8_t*)0x2000a6c2 = (uint8_t)0x0;
*(uint8_t*)0x2000a6c3 = (uint8_t)0xbe2;
*(uint8_t*)0x2000a6c4 = (uint8_t)0x9;
*(uint8_t*)0x2000a6c5 = (uint8_t)0x7ff;
*(uint8_t*)0x2000a6c6 = (uint8_t)0x1;
*(uint8_t*)0x2000a6c7 = (uint8_t)0x1f;
*(uint8_t*)0x2000a6c8 = (uint8_t)0x1d8;
*(uint16_t*)0x2000a6c9 = (uint16_t)0x8;
*(uint32_t*)0x2000a6cd = (uint32_t)0x736d;
*(uint16_t*)0x2000a6d1 = (uint16_t)0x3;
*(uint8_t*)0x2000a6d3 = (uint8_t)0xff;
*(uint8_t*)0x2000a6d4 = (uint8_t)0x3;
*(uint8_t*)0x2000a6d5 = (uint8_t)0xffffffffffff8000;
*(uint8_t*)0x2000a6d6 = (uint8_t)0xc20;
*(uint8_t*)0x2000a6d7 = (uint8_t)0x6;
*(uint8_t*)0x2000a6d8 = (uint8_t)0x2;
*(uint8_t*)0x2000a6d9 = (uint8_t)0x6;
*(uint8_t*)0x2000a6da = (uint8_t)0x8;
*(uint8_t*)0x2000a6db = (uint8_t)0x3;
*(uint8_t*)0x2000a6dc = (uint8_t)0x1;
*(uint16_t*)0x2000a6dd = (uint16_t)0xce;
*(uint32_t*)0x2000a6e1 = (uint32_t)0xab85;
*(uint16_t*)0x2000a6e5 = (uint16_t)0x0;
*(uint8_t*)0x2000a6e7 = (uint8_t)0xa0e3;
*(uint8_t*)0x2000a6e8 = (uint8_t)0x100000001;
*(uint8_t*)0x2000a6e9 = (uint8_t)0x3;
*(uint8_t*)0x2000a6ea = (uint8_t)0x1;
*(uint8_t*)0x2000a6eb = (uint8_t)0x2;
*(uint8_t*)0x2000a6ec = (uint8_t)0x1;
*(uint8_t*)0x2000a6ed = (uint8_t)0x7ff;
*(uint8_t*)0x2000a6ee = (uint8_t)0x2;
*(uint8_t*)0x2000a6ef = (uint8_t)0x8a;
*(uint8_t*)0x2000a6f0 = (uint8_t)0xca6;
*(uint16_t*)0x2000a6f1 = (uint16_t)0x1;
*(uint32_t*)0x2000a6f5 = (uint32_t)0x401;
*(uint32_t*)0x2000a6f9 = (uint32_t)0x0;
*(uint32_t*)0x2000a6fd = (uint32_t)0x0;
*(uint32_t*)0x2000a701 = (uint32_t)0x0;
*(uint32_t*)0x2000a705 = (uint32_t)0x0;
*(uint32_t*)0x2000a709 = (uint32_t)0x0;
*(uint32_t*)0x2000a70d = (uint32_t)0x0;
*(uint32_t*)0x2000a711 = (uint32_t)0x0;
*(uint32_t*)0x2000a715 = (uint32_t)0x0;
*(uint32_t*)0x2000a719 = (uint32_t)0x0;
r[71] =
syscall(SYS_ioctl, r[3], 0x4070aea0ul, 0x2000a6b9ul, 0, 0, 0);
break;
case 6:
r[72] = syscall(SYS_mmap, 0x2000e000ul, 0x1000ul, 0x3ul, 0x32ul,
0xfffffffffffffffful, 0x0ul);
break;
case 7:
r[73] = syscall(SYS_ioctl, r[2], 0x5424ul, 0x2000e630ul, 0, 0, 0);
break;
}
return 0;
}

int main()
{
long i;
pthread_t th[8];

srand(getpid());
memset(r, -1, sizeof(r));
for (i = 0; i < 8; i++) {
pthread_create(&th[i], 0, thr, (void*)i);
usleep(rand()%10000);
}
for (i = 0; i < 8; i++)
pthread_join(th[i], 0);
for (i = 0; i < 8; i++) {
pthread_create(&th[i], 0, thr, (void*)i);
if (rand()%2)
usleep(rand()%10000);
}
for (i = 0; i < 8; i++)
pthread_join(th[i], 0);
return 0;
}


On commit 388f7b1d6e8ca06762e2454d28d6c3c55ad0fe95 (4.5-rc3)

Paolo Bonzini

unread,
May 31, 2016, 5:56:15 AM5/31/16
to Dmitry Vyukov, Gleb Natapov, Thomas Gleixner, Ingo Molnar, H. Peter Anvin, x...@kernel.org, k...@vger.kernel.org, LKML, Eric Northup, Andrew Honig, Steve Rutherford, Robert Swiecki, Kostya Serebryany, Alexander Potapenko, syzkaller


On 15/02/2016 14:30, Dmitry Vyukov wrote:
> *(uint32_t*)0x2000a6b9 = (uint32_t)0x3e;
> *(uint16_t*)0x2000a6bd = (uint16_t)0x8;
> *(uint8_t*)0x2000a6bf = (uint8_t)0x8d4;
> *(uint8_t*)0x2000a6c0 = (uint8_t)0xffffffffffff5fe9;
> *(uint8_t*)0x2000a6c1 = (uint8_t)0x80000001;
> *(uint8_t*)0x2000a6c2 = (uint8_t)0x0;
> *(uint8_t*)0x2000a6c3 = (uint8_t)0xbe2;
> *(uint8_t*)0x2000a6c4 = (uint8_t)0x9;
> *(uint8_t*)0x2000a6c5 = (uint8_t)0x7ff;
> *(uint8_t*)0x2000a6c6 = (uint8_t)0x1;
> *(uint8_t*)0x2000a6c7 = (uint8_t)0x1f;
> *(uint8_t*)0x2000a6c8 = (uint8_t)0x1d8;
> *(uint16_t*)0x2000a6c9 = (uint16_t)0x8;

This field (.channels[0].count_load_time) should be uint64_t. I
understand that it's all random, but it makes it even harder to follow
what's going on...

Thanks,

Paolo

Dmitry Vyukov

unread,
Jun 21, 2016, 2:14:09 PM6/21/16
to Paolo Bonzini, Gleb Natapov, Thomas Gleixner, Ingo Molnar, H. Peter Anvin, x...@kernel.org, KVM list, LKML, Eric Northup, Andrew Honig, Steve Rutherford, Robert Swiecki, Kostya Serebryany, Alexander Potapenko, syzkaller
On Tue, May 31, 2016 at 11:56 AM, Paolo Bonzini <pbon...@redhat.com> wrote:
>
>
> On 15/02/2016 14:30, Dmitry Vyukov wrote:
>> *(uint32_t*)0x2000a6b9 = (uint32_t)0x3e;
>> *(uint16_t*)0x2000a6bd = (uint16_t)0x8;
>> *(uint8_t*)0x2000a6bf = (uint8_t)0x8d4;
>> *(uint8_t*)0x2000a6c0 = (uint8_t)0xffffffffffff5fe9;
>> *(uint8_t*)0x2000a6c1 = (uint8_t)0x80000001;
>> *(uint8_t*)0x2000a6c2 = (uint8_t)0x0;
>> *(uint8_t*)0x2000a6c3 = (uint8_t)0xbe2;
>> *(uint8_t*)0x2000a6c4 = (uint8_t)0x9;
>> *(uint8_t*)0x2000a6c5 = (uint8_t)0x7ff;
>> *(uint8_t*)0x2000a6c6 = (uint8_t)0x1;
>> *(uint8_t*)0x2000a6c7 = (uint8_t)0x1f;
>> *(uint8_t*)0x2000a6c8 = (uint8_t)0x1d8;
>> *(uint16_t*)0x2000a6c9 = (uint16_t)0x8;
>
> This field (.channels[0].count_load_time) should be uint64_t. I
> understand that it's all random, but it makes it even harder to follow
> what's going on...

Thanks! Fixed in:
https://github.com/google/syzkaller/commit/1816c21f2bed9b6015f1af1c5423779d2c6e58ad
Reply all
Reply to author
Forward
0 new messages