ping: use CLOCK_REALTIME for ICMP Originate Timestamp RFC 792 defines the ICMP Originate Timestamp field as milliseconds since midnight UTC. However, ping(8) currently derives this value from CLOCK_MONOTONIC, which represents time since an unspecified starting point and is not related to UTC. The issue was introduced by commit 1ad76f1b6047, which replaced gettimeofday(2) with clock_gettime(CLOCK_MONOTONIC) for timekeeping in ping(8). Fix this by using CLOCK_REALTIME when generating the ICMP originate timestamp. Before: $ ping -Mt -c1 127.0.0.1 ICMP_TSTAMP PING 127.0.0.1 (127.0.0.1): 56 data bytes <...> time=0.061 ms tso=16:50:31 tsr=17:38:28 tst=17:38:28 (note the tso is off) After: $ ping -Mt -c1 127.0.0.1 ICMP_TSTAMP PING 127.0.0.1 (127.0.0.1): 56 data bytes <...> time=0.038 ms tso=17:42:09 tsr=17:42:09 tst=17:42:09 Reviewed by: XXX Fixes: 1ad76f1b6047 MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D###
Details
Details
Diff Detail
Diff Detail
- Repository
- rG FreeBSD src repository
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
Comment Actions
Thanks for catching this. Would it be possible to add an ATF test to catch the problem?
Comment Actions
OK, I added a small shell snippet as a possible test for the issue.
Before:
$ kyua test ping_test:timestamp_origin
ping_test:timestamp_origin -> failed: tso (04:54:52) differs from tsr (23:46:59) by 67927 seconds [0.035s]
Results file id is usr_tests_sbin_ping.20260501-234659-604330
Results saved to /home/maxim/.kyua/store/results.usr_tests_sbin_ping.20260501-234659-604330.db
0/1 passed (0 broken, 1 failed, 0 skipped)
After:
$ kyua test ping_test:timestamp_origin
ping_test:timestamp_origin -> passed [0.030s]
Results file id is usr_tests_sbin_ping.20260501-234827-033957
Results saved to /home/maxim/.kyua/store/results.usr_tests_sbin_ping.20260501-234827-033957.db
1/1 passed (0 broken, 0 failed, 0 skipped)
All ping_test tests:
kyua test ping_test
ping_test:inject_opts -> passed [2.470s]
ping_test:inject_pip -> passed [3.431s]
ping_test:inject_reply -> passed [2.411s]
ping_test:ping4_nohost -> passed [0.029s]
ping_test:ping6_4 -> passed [0.046s]
ping_test:ping6_c1_s8_t1 -> passed [0.080s]
ping_test:ping6_c1t4 -> passed [0.040s]
ping_test:ping6_nohost -> passed [0.034s]
ping_test:ping_46 -> passed [0.043s]
ping_test:ping_64 -> passed [0.040s]
ping_test:ping_6_c1_s8_t1 -> passed [0.056s]
ping_test:ping_c1_s56_t1 -> passed [0.053s]
ping_test:ping_c1_s56_t1_S127 -> passed [0.050s]
ping_test:ping_c1_s8_t1_S1 -> passed [0.047s]
ping_test:ping_c1t6 -> passed [0.043s]
ping_test:ping_nohost -> passed [0.024s]
ping_test:timestamp_origin -> passed [0.039s]
Results file id is usr_tests_sbin_ping.20260501-235651-527128
Results saved to /root/.kyua/store/results.usr_tests_sbin_ping.20260501-235651-527128.db
17/17 passed (0 broken, 0 failed, 0 skipped)
| sbin/ping/tests/ping_test.sh | ||
|---|---|---|
| 274–282 | How about this to be a little bit simpler? tso_s=`date -jf %H:%M:%S $tso` tsr_s=`date -jf %H:%M:%S $tsr` | |
| sbin/ping/tests/ping_test.sh | ||
|---|---|---|
| 274–282 | Oh yeah, I completely forgot about this feature of our date(1). | |
Comment Actions
I think this is a quite minor issue and doubt that a lot of people know about and use icmp timestamp these days. That was pretty much a fun exercise.
Thank you for the review, Alan!
| sbin/ping/tests/ping_test.sh | ||
|---|---|---|
| 274–282 | This doesn't work: $ ping -Mt -c1 127.0.0.1 | sed -n 's/.*tsr=\([0-9:]*\).*/\1/p' 18:29:42 $ date -jf %H:%M:%S 18:29:42 Sun May 3 18:29:42 UTC 2026 So tso_s and tsr_s are not integers and can't be subtracted, and the test is failing. Could you please take a look? | |
| sbin/ping/tests/ping_test.sh | ||
|---|---|---|
| 274–282 | Probably needs a "+%s" at the end of the date command. | |
| sbin/ping/tests/ping_test.sh | ||
|---|---|---|
| 274–282 | Yep. I'll commit that change in a few minutes if no one beats me to it. | |