int overflow in io_getevents

From: Dmitry Vyukov
Date: Mon Dec 07 2015 - 05:27:33 EST


Hello,

While running syzkaller fuzzer on commit
31ade3b83e1821da5fbb2f11b5b3d4ab2ec39db8, I've hit the following UBSAN
warning. I think it can lead to an unexpected active wait loop, if
user-space expects such io_getevents to wait for a long duration but
instead it returns immediately, so user-space reissues the same call
again and again. Andrey suggested that read_events should validate
timeout with timespec_valid_strict before using it.


UBSAN: Undefined behaviour in include/linux/ktime.h:55:49
signed integer overflow:
1449363382000000000 + 8584381026499825158 cannot be represented in
type 'long long int'
CPU: 0 PID: 27992 Comm: syzkaller_execu Not tainted 4.4.0-rc3+ #150
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
0000000000000000 ffff880062ab7ae0 ffffffff82c6f2a8 0000000041b58ab3
ffffffff8788bf8d ffffffff82c6f1f6 ffff880062ab7aa8 ffffffff88479680
ffff880062ab7be0 7721d90fc5200e06 0000000000000001 ffff880062ab7af0
Call Trace:
[< inline >] __dump_stack lib/dump_stack.c:15
[<ffffffff82c6f2a8>] dump_stack+0xb2/0xfa lib/dump_stack.c:50
[<ffffffff82d622e7>] ubsan_epilogue+0x12/0x8f lib/ubsan.c:160
[<ffffffff82d63e58>] handle_overflow+0x22f/0x276 lib/ubsan.c:191
[<ffffffff82d63ec9>] __ubsan_handle_add_overflow+0x2a/0x31 lib/ubsan.c:199
[< inline >] ktime_set include/linux/ktime.h:55
[< inline >] timespec_to_ktime include/linux/ktime.h:83
[<ffffffff8190ca77>] read_events+0x4b7/0x560 fs/aio.c:1273
[< inline >] SYSC_io_getevents fs/aio.c:1737
[<ffffffff81913567>] SyS_io_getevents+0xc7/0x340 fs/aio.c:1726
[<ffffffff868dae76>] entry_SYSCALL_64_fastpath+0x16/0x7a
arch/x86/entry/entry_64.S:185

Thank you
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/