Re: UBSAN: Undefined behaviour in drivers/pps/pps.c

[Dmitry Torokhov]

On Wed, Jan 09, 2019 at 10:20:50AM +0100, Rodolfo Giometti wrote:
> On 08/01/2019 21:24, Kyungtae Kim wrote:
> > We report a bug in linux-4.20: "UBSAN: Undefined behaviour in drivers/pps/pps.c"
> > 
> > kernel config: https://kt0755.github.io/etc/config_v4.20_stable
> > repro: https://kt0755.github.io/etc/repro.a6372.c
> > 
> > pps_cdev_pps_fetch() lacks the bounds checking for computing
> > fdata->timeout.sec * HZ, that causes such integer overflow when the result
> > is larger than the boundary.
> > The  patch below checks the possibility of overflow right before the
> > multiplication.
> > 
> > =========================================
> > UBSAN: Undefined behaviour in drivers/pps/pps.c:82:30
> > signed integer overflow:
> > -7557201428062104791 * 100 cannot be represented in type 'long long int'
> > CPU: 0 PID: 10159 Comm: syz-executor6 Not tainted 4.20.0 #1
> > Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
> > Call Trace:
> >   __dump_stack lib/dump_stack.c:77 [inline]
> >   dump_stack+0xb1/0x118 lib/dump_stack.c:113
> >   ubsan_epilogue+0x12/0x94 lib/ubsan.c:159
> >   handle_overflow+0x1cf/0x21a lib/ubsan.c:190
> >   __ubsan_handle_mul_overflow+0x2a/0x35 lib/ubsan.c:214
> >   pps_cdev_pps_fetch+0x575/0x5b0 drivers/pps/pps.c:82
> >   pps_cdev_ioctl+0x567/0x910 drivers/pps/pps.c:191
> >   vfs_ioctl fs/ioctl.c:46 [inline]
> >   do_vfs_ioctl+0x1aa/0x1160 fs/ioctl.c:698
> >   ksys_ioctl+0x9e/0xb0 fs/ioctl.c:713
> >   __do_sys_ioctl fs/ioctl.c:720 [inline]
> >   __se_sys_ioctl fs/ioctl.c:718 [inline]
> >   __x64_sys_ioctl+0x7e/0xc0 fs/ioctl.c:718
> >   do_syscall_64+0xbe/0x4f0 arch/x86/entry/common.c:290
> >   entry_SYSCALL_64_after_hwframe+0x49/0xbe
> > RIP: 0033:0x4497b9
> > Code: e8 8c 9f 02 00 48 83 c4 18 c3 0f 1f 80 00 00 00 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 0f 83 9b 6b fc ff c3 66 2e 0f 1f 84 00 00 00 00
> > RSP: 002b:00007f8cf875bc68 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
> > RAX: ffffffffffffffda RBX: 00007f8cf875c6cc RCX: 00000000004497b9
> > RDX: 0000000020000240 RSI: 00000000c00870a4 RDI: 0000000000000014
> > RBP: 000000000071bea0 R08: 0000000000000000 R09: 0000000000000000
> > R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff
> > R13: 0000000000005c10 R14: 00000000006eecb0 R15: 00007f8cf875c700
> > =========================================
> > 
> > ---
> >   drivers/pps/pps.c | 2 ++
> >   1 file changed, 2 insertions(+)
> > 
> > diff --git a/drivers/pps/pps.c b/drivers/pps/pps.c
> > index 8febacb..66002e1 100644
> > --- a/drivers/pps/pps.c
> > +++ b/drivers/pps/pps.c
> > @@ -79,6 +79,8 @@ static int pps_cdev_pps_fetch(struct pps_device
> > *pps, struct pps_fdata *fdata)
> >                  dev_dbg(pps->dev, "timeout %lld.%09d\n",
> >                                  (long long) fdata->timeout.sec,
> >                                  fdata->timeout.nsec);
> > +               if (fdata->timeout.sec > S64_MAX / HZ)
> > +                               return -EINVAL;
> >                  ticks = fdata->timeout.sec * HZ;
> >                  ticks += fdata->timeout.nsec / (NSEC_PER_SEC / HZ);
> 
> It looks good to me. Do you think is better adding a check for timeout.nsec also?

Another option is to use check_mul_overflow().

> 
> Now you have to produce a patch according to
> linux/Documentation/process/submitting-patches.rst and then submitting it!
> :-)
> 

Thanks.

-- 
Dmitry