time: Prevent undefined behaviour in timespec64_to_ns()
authorZeng Tao <prime.zeng@hisilicon.com>
Tue, 1 Sep 2020 09:30:13 +0000 (17:30 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 18 Nov 2020 17:27:52 +0000 (18:27 +0100)
[ Upstream commit cb47755725da7b90fecbb2aa82ac3b24a7adb89b ]

UBSAN reports:

Undefined behaviour in ./include/linux/time64.h:127:27
signed integer overflow:
17179869187 * 1000000000 cannot be represented in type 'long long int'
Call Trace:
 timespec64_to_ns include/linux/time64.h:127 [inline]
 set_cpu_itimer+0x65c/0x880 kernel/time/itimer.c:180
 do_setitimer+0x8e/0x740 kernel/time/itimer.c:245
 __x64_sys_setitimer+0x14c/0x2c0 kernel/time/itimer.c:336
 do_syscall_64+0xa1/0x540 arch/x86/entry/common.c:295

Commit bd40a175769d ("y2038: itimer: change implementation to timespec64")
replaced the original conversion which handled time clamping correctly with
timespec64_to_ns() which has no overflow protection.

Fix it in timespec64_to_ns() as this is not necessarily limited to the
usage in itimers.

[ tglx: Added comment and adjusted the fixes tag ]

Fixes: 361a3bf00582 ("time64: Add time64.h header and define struct timespec64")
Signed-off-by: Zeng Tao <prime.zeng@hisilicon.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/1598952616-6416-1-git-send-email-prime.zeng@hisilicon.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
include/linux/time64.h
kernel/time/itimer.c

index ad33260618f766e8c2814530dc177961c8be485e..99ab4a686c3019c6b5bbb3d2ac32a83d67068b60 100644 (file)
@@ -189,6 +189,10 @@ static inline bool timespec64_valid_strict(const struct timespec64 *ts)
  */
 static inline s64 timespec64_to_ns(const struct timespec64 *ts)
 {
+       /* Prevent multiplication overflow */
+       if ((unsigned long long)ts->tv_sec >= KTIME_SEC_MAX)
+               return KTIME_MAX;
+
        return ((s64) ts->tv_sec * NSEC_PER_SEC) + ts->tv_nsec;
 }
 
index f26acef5d7b485ad397c21bb5466f0a332006087..7f718565507b5d158a789c553ab72fc10b8997c6 100644 (file)
@@ -153,10 +153,6 @@ static void set_cpu_itimer(struct task_struct *tsk, unsigned int clock_id,
        u64 oval, nval, ointerval, ninterval;
        struct cpu_itimer *it = &tsk->signal->it[clock_id];
 
-       /*
-        * Use the to_ktime conversion because that clamps the maximum
-        * value to KTIME_MAX and avoid multiplication overflows.
-        */
        nval = ktime_to_ns(timeval_to_ktime(value->it_value));
        ninterval = ktime_to_ns(timeval_to_ktime(value->it_interval));