Re: [PATCH 2/3] tools: timer: add test to check y2038/2106 bug

From: Alexandre Belloni
Date: Fri Jun 16 2017 - 11:52:25 EST

On 16/06/2017 at 16:03:52 +0200, Benjamin Gaignard wrote:
> The goal of this test is to check if a RTC device correctly support
> dates after years 2038 or 2106.
> It set a date (1-1-2200) on RTC and read it back to be sure that the
> driver is working well.
> The same thing is done on alarm.
>
> Signed-off-by: Benjamin Gaignard <benjamin.gaignard@xxxxxxxxxx>
> ---
> tools/testing/selftests/timers/Makefile | 4 +-
> tools/testing/selftests/timers/rtctest-2038.c | 135 ++++++++++++++++++++++++++
> 2 files changed, 138 insertions(+), 1 deletion(-)
> create mode 100644 tools/testing/selftests/timers/rtctest-2038.c
>
> diff --git a/tools/testing/selftests/timers/Makefile b/tools/testing/selftests/timers/Makefile
> index 54481f1..7791f79 100644
> --- a/tools/testing/selftests/timers/Makefile
> +++ b/tools/testing/selftests/timers/Makefile
> @@ -9,7 +9,8 @@ TEST_GEN_PROGS = posix_timers nanosleep nsleep-lat set-timer-lat mqueue-lat \
>
> TEST_GEN_PROGS_EXTENDED = alarmtimer-suspend valid-adjtimex adjtick change_skew \
> skew_consistency clocksource-switch leap-a-day \
> - leapcrash set-tai set-2038 set-tz rtctest_setdate
> + leapcrash set-tai set-2038 set-tz rtctest_setdate \
> + rtctest-2038
>
>
> include ../lib.mk
> @@ -29,4 +30,5 @@ run_destructive_tests: run_tests
> ./set-tz
> ./set-tai
> ./set-2038
> + ./rtctest-2038
>
> diff --git a/tools/testing/selftests/timers/rtctest-2038.c b/tools/testing/selftests/timers/rtctest-2038.c
> new file mode 100644
> index 0000000..213b7ee
> --- /dev/null
> +++ b/tools/testing/selftests/timers/rtctest-2038.c
> @@ -0,0 +1,135 @@
> +/* Real Time Clock Driver Test
> + * by: Benjamin Gaignard (benjamin.gaignard@xxxxxxxxxx)
> + *
> + * To build
> + * gcc rtctest-2038.c -o rtctest-2038
> + *
> + * This program is free software: you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation, either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + */
> +
> +#include <stdio.h>
> +#include <linux/rtc.h>
> +#include <sys/ioctl.h>
> +#include <sys/time.h>
> +#include <sys/types.h>
> +#include <fcntl.h>
> +#include <unistd.h>
> +#include <stdlib.h>
> +#include <errno.h>
> +
> +static const char default_rtc[] = "/dev/rtc0";
> +
> +int main(int argc, char **argv)
> +{
> + int fd, retval;
> + struct rtc_time new, current;
> + const char *rtc = default_rtc;
> +
> + switch (argc) {
> + case 2:
> + rtc = argv[1];
> + /* FALLTHROUGH */
> + case 1:
> + break;
> + default:
> + fprintf(stderr, "usage: rtctest-2038 [rtcdev]\n");
> + return 1;
> + }
> +
> + fprintf(stderr, "\nTest if RTC is robust for date after y2038/2106\n\n");
> +
> + fd = open(rtc, O_RDONLY);
> + if (fd == -1) {
> + perror(rtc);
> + exit(errno);
> + }
> +
> + new.tm_year = 300; /* 2200 - 1900 */

It would be nice to have tests for 1900, 1901, 1969, 1970, 2037,
2038, 2099, 2100, 2105, 2106, 2261 and 2262.

Those are the various cutoff dates that exist and RTC supporting up to
2099 are certainly more common that RTC handling dates after 2200 else,
most RTC will fail the test.

> + new.tm_mon = 0;
> + new.tm_mday = 1;
> + new.tm_hour = 0;
> + new.tm_min = 0;
> + new.tm_sec = 0;
> +
> + fprintf(stderr, "Test will set RTC date/time to %d-%d-%d, %02d:%02d:%02d.\n",
> + new.tm_mday, new.tm_mon + 1, new.tm_year + 1900,
> + new.tm_hour, new.tm_min, new.tm_sec);
> +
> + /* Write the new date in RTC */
> + retval = ioctl(fd, RTC_SET_TIME, &new);
> + if (retval == -1) {
> + perror("RTC_SET_TIME ioctl");
> + close(fd);
> + exit(errno);
> + }
> +
> + /* Read back */
> + retval = ioctl(fd, RTC_RD_TIME, &current);
> + if (retval == -1) {
> + perror("RTC_RD_TIME ioctl");
> + exit(errno);
> + }
> +
> + fprintf(stderr, "RTC date/time is %d-%d-%d, %02d:%02d:%02d.\n",
> + current.tm_mday, current.tm_mon + 1, current.tm_year + 1900,
> + current.tm_hour, current.tm_min, current.tm_sec);
> +
> + if (new.tm_year != current.tm_year ||
> + new.tm_mon != current.tm_mon ||
> + new.tm_mday != current.tm_mday ||
> + new.tm_hour != current.tm_hour ||
> + new.tm_min != current.tm_min ||
> + new.tm_sec != current.tm_sec) {
This may fail because nothing guarantees that the RTC will reset its
sub seconds counter to 0 when setting the time so new.tm_sec could be
current.tm_sec + 1.

Maybe you could factorize the comparison in a function.

Finally, I won't insist on that but it could be interesting to integrate
it in rtctest.c, hidden behind a option warning that it is "destructive"
(i.e. the current time and date will be lost).

> + fprintf(stderr, "\n\nSet Time test failed\n");
> + close(fd);
> + return 1;
> + }
> +
> + new.tm_sec += 5;
> +
> + fprintf(stderr, "\nTest will set RTC alarm to %d-%d-%d, %02d:%02d:%02d.\n",
> + new.tm_mday, new.tm_mon + 1, new.tm_year + 1900,
> + new.tm_hour, new.tm_min, new.tm_sec);
> +
> + /* Write the new alarm in RTC */
> + retval = ioctl(fd, RTC_ALM_SET, &new);
> + if (retval == -1) {
> + perror("RTC_ALM_SET ioctl");
> + close(fd);
> + exit(errno);
> + }
> +
> + /* Read back */
> + retval = ioctl(fd, RTC_ALM_READ, &current);
> + if (retval == -1) {
> + perror("RTC_ALM_READ ioctl");
> + exit(errno);
> + }
> +
> + fprintf(stderr, "RTC alarm is %d-%d-%d, %02d:%02d:%02d.\n",
> + current.tm_mday, current.tm_mon + 1, current.tm_year + 1900,
> + current.tm_hour, current.tm_min, current.tm_sec);
> +
> + if (new.tm_year != current.tm_year ||
> + new.tm_mon != current.tm_mon ||
> + new.tm_mday != current.tm_mday ||
> + new.tm_hour != current.tm_hour ||
> + new.tm_min != current.tm_min ||
> + new.tm_sec != current.tm_sec) {
> + fprintf(stderr, "\n\nSet alarm test failed\n");
> + close(fd);
> + return 1;
> + }
> +
> + fprintf(stderr, "\nTest complete\n");
> + close(fd);
> + return 0;
> +}
> --
> 1.9.1
>

--
Alexandre Belloni, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com