[PATCH] kdb: use __ktime_get_real_seconds instead of __current_kernel_time

From: Arnd Bergmann
Date: Thu Oct 12 2017 - 10:07:39 EST


kdb is the only user of the __current_kernel_time() interface, which is
not y2038 safe and should be removed at some point.

The kdb code also goes to great lengths to print the time in a
human-readable format from 'struct timespec', again using a non-y2038-safe
re-implementation of the generic time_to_tm() code.

Using __current_kernel_time() here is necessary since the regular
accessors that require a sequence lock might hang when called during the
xtime update. However, this is safe in the particular case since kdb is
only interested in the tv_sec field that is updated atomically.

In order to make this y2038-safe, I'm converting the code to the generic
time64_to_tm helper, but that introduces the problem that we have no
interface like __current_kernel_time() that provides a 64-bit timestamp
in a lockless, safe and architecture-independent way. I have multiple
ideas for how to solve that:

- __ktime_get_real_seconds() is lockless, but can return
incorrect results on 32-bit architectures in the special case that
we are in the process of changing the time across the epoch, either
during the timer tick that overflows the seconds in 2038, or while
calling settimeofday.

- ktime_get_real_fast_ns() would work in this context, but does
require a call into the clocksource driver to return a high-resolution
timestamp. This may have undesired side-effects in the debugger,
since we want to limit the interactions with the rest of the kernel.

- Adding a ktime_get_real_fast_seconds() based on tk_fast_mono
plus tkr->base_real without the tk_clock_read() delta. Not sure about
the value of adding yet another interface here.

- Changing the existing ktime_get_real_seconds() to use
tk_fast_mono on 32-bit architectures rather than xtime_sec. I think
this could work, but am not entirely sure if this is an improvement.

I picked the first of those for simplicity here. It's technically
not correct but probably good enough as the time is only used for the
debugging output and the race will likely never be hit in practice.
Another downside is having to move the declaration into a public header
file.

Let me know if anyone has a different preference.

Cc: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
Link: https://patchwork.kernel.org/patch/9775309/
Signed-off-by: Arnd Bergmann <arnd@xxxxxxxx>
---
include/linux/timekeeping.h | 1 +
kernel/debug/kdb/kdb_main.c | 45 +++++---------------------------------
kernel/time/timekeeping_internal.h | 2 --
3 files changed, 6 insertions(+), 42 deletions(-)

diff --git a/include/linux/timekeeping.h b/include/linux/timekeeping.h
index eb98cbdbb323..9b59473556fe 100644
--- a/include/linux/timekeeping.h
+++ b/include/linux/timekeeping.h
@@ -41,6 +41,7 @@ struct timespec64 get_monotonic_coarse64(void);
extern void getrawmonotonic64(struct timespec64 *ts);
extern void ktime_get_ts64(struct timespec64 *ts);
extern time64_t ktime_get_seconds(void);
+extern time64_t __ktime_get_real_seconds(void);
extern time64_t ktime_get_real_seconds(void);

extern int __getnstimeofday64(struct timespec64 *tv);
diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c
index c8146d53ca67..69e70f4021fe 100644
--- a/kernel/debug/kdb/kdb_main.c
+++ b/kernel/debug/kdb/kdb_main.c
@@ -2479,41 +2479,6 @@ static int kdb_kill(int argc, const char **argv)
return 0;
}

-struct kdb_tm {
- int tm_sec; /* seconds */
- int tm_min; /* minutes */
- int tm_hour; /* hours */
- int tm_mday; /* day of the month */
- int tm_mon; /* month */
- int tm_year; /* year */
-};
-
-static void kdb_gmtime(struct timespec *tv, struct kdb_tm *tm)
-{
- /* This will work from 1970-2099, 2100 is not a leap year */
- static int mon_day[] = { 31, 29, 31, 30, 31, 30, 31,
- 31, 30, 31, 30, 31 };
- memset(tm, 0, sizeof(*tm));
- tm->tm_sec = tv->tv_sec % (24 * 60 * 60);
- tm->tm_mday = tv->tv_sec / (24 * 60 * 60) +
- (2 * 365 + 1); /* shift base from 1970 to 1968 */
- tm->tm_min = tm->tm_sec / 60 % 60;
- tm->tm_hour = tm->tm_sec / 60 / 60;
- tm->tm_sec = tm->tm_sec % 60;
- tm->tm_year = 68 + 4*(tm->tm_mday / (4*365+1));
- tm->tm_mday %= (4*365+1);
- mon_day[1] = 29;
- while (tm->tm_mday >= mon_day[tm->tm_mon]) {
- tm->tm_mday -= mon_day[tm->tm_mon];
- if (++tm->tm_mon == 12) {
- tm->tm_mon = 0;
- ++tm->tm_year;
- mon_day[1] = 28;
- }
- }
- ++tm->tm_mday;
-}
-
/*
* Most of this code has been lifted from kernel/timer.c::sys_sysinfo().
* I cannot call that code directly from kdb, it has an unconditional
@@ -2539,8 +2504,8 @@ static void kdb_sysinfo(struct sysinfo *val)
*/
static int kdb_summary(int argc, const char **argv)
{
- struct timespec now;
- struct kdb_tm tm;
+ time64_t now;
+ struct tm tm;
struct sysinfo val;

if (argc)
@@ -2554,9 +2519,9 @@ static int kdb_summary(int argc, const char **argv)
kdb_printf("domainname %s\n", init_uts_ns.name.domainname);
kdb_printf("ccversion %s\n", __stringify(CCVERSION));

- now = __current_kernel_time();
- kdb_gmtime(&now, &tm);
- kdb_printf("date %04d-%02d-%02d %02d:%02d:%02d "
+ now = __ktime_get_real_seconds();
+ time64_to_tm(now, 0, &tm);
+ kdb_printf("date %04ld-%02d-%02d %02d:%02d:%02d "
"tz_minuteswest %d\n",
1900+tm.tm_year, tm.tm_mon+1, tm.tm_mday,
tm.tm_hour, tm.tm_min, tm.tm_sec,
diff --git a/kernel/time/timekeeping_internal.h b/kernel/time/timekeeping_internal.h
index 9a18f121f399..58e79485de1b 100644
--- a/kernel/time/timekeeping_internal.h
+++ b/kernel/time/timekeeping_internal.h
@@ -30,6 +30,4 @@ static inline u64 clocksource_delta(u64 now, u64 last, u64 mask)
}
#endif

-extern time64_t __ktime_get_real_seconds(void);
-
#endif /* _TIMEKEEPING_INTERNAL_H */
--
2.9.0