kdb: use ktime_get_mono_fast_ns() instead of ktime_get_ts()
authorBaolin Wang <baolin.wang@linaro.org>
Mon, 29 Jan 2018 02:22:51 +0000 (10:22 +0800)
committerJason Wessel <jason.wessel@windriver.com>
Thu, 1 Feb 2018 03:31:09 +0000 (21:31 -0600)
The kdb code will print the monotonic time by ktime_get_ts(), but
the ktime_get_ts() will be protected by a sequence lock, that will
introduce one deadlock risk if the lock was already held in the
context from which we entered the debugger.

Thus we can use the ktime_get_mono_fast_ns() to get the monotonic
time, which is NMI safe access to clock monotonic. Moreover we can
remove the 'struct timespec', which is not y2038 safe.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Reviewed-by: Daniel Thompson <daniel.thompson@linaro.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
kernel/debug/kdb/kdb_main.c

index 6055231..16140d1 100644 (file)
@@ -2512,10 +2512,10 @@ static int kdb_kill(int argc, const char **argv)
  */
 static void kdb_sysinfo(struct sysinfo *val)
 {
-       struct timespec uptime;
-       ktime_get_ts(&uptime);
+       u64 uptime = ktime_get_mono_fast_ns();
+
        memset(val, 0, sizeof(*val));
-       val->uptime = uptime.tv_sec;
+       val->uptime = div_u64(uptime, NSEC_PER_SEC);
        val->loads[0] = avenrun[0];
        val->loads[1] = avenrun[1];
        val->loads[2] = avenrun[2];