calendarspec: fix calculation of timespec iterations that fall onto a DST change #13872
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
poettering
approved these changes
Oct 30, 2019
Member
poettering
left a comment
There was a problem hiding this comment.
hmm, i don't oversee the full effect of this. but looks superficially ok. maybe should have a test case?
maybe beef up "systemd-analyze calendar" to take a new parameter --current-time= or so, which does what faketime does internally? Then compare the byte-by-byte output of the tool with what we expect?
src/shared/calendarspec.c
Outdated
|
|
||
| t = *tm; | ||
| if (mktime_or_timegm(&t, utc) < 0) | ||
| return -EINVAL; |
Member
There was a problem hiding this comment.
not that it matters much, but it's slightly ugly that we make up our own error here, even though mktime() and timegm() set errno. maybe fix this to propagate the error properly?
89aad24 to
79fb2af
Compare
… DST change
If we tested a candidate time that would fall onto the DST change, and we
realized that it is now a valid time ('cause the given "hour" is missing),
we would jump to to beginning of the next bigger time period, i.e. the next
day.
mktime_or_timegm() already tells us what the next valid time is, so let's reuse
this, and continue the calculations at this point. This should allow us to
correctly jump over DST changes, but also leap seconds and similar. It should
be OK even multiple days were removed from calendar, similarly to the
Gregorian-Julian transition. By reusing the information from normalization, we
don't have to make assumptions what the next valid time is.
Fixes systemd#13745.
$ TZ=Australia/Sydney faketime '2019-10-06 01:50' build/systemd-analyze calendar 0/1:0/1 --iterations 20 | grep Iter
Iter. #2: Sun 2019-10-06 01:52:00 AEST
Iter. #3: Sun 2019-10-06 01:53:00 AEST
Iter. #4: Sun 2019-10-06 01:54:00 AEST
Iter. #5: Sun 2019-10-06 01:55:00 AEST
Iter. #6: Sun 2019-10-06 01:56:00 AEST
Iter. #7: Sun 2019-10-06 01:57:00 AEST
Iter. #8: Sun 2019-10-06 01:58:00 AEST
Iter. #9: Sun 2019-10-06 01:59:00 AEST
Iter. #10: Sun 2019-10-06 03:00:00 AEDT
Iter. #11: Sun 2019-10-06 03:01:00 AEDT
Iter. #12: Sun 2019-10-06 03:02:00 AEDT
Iter. #13: Sun 2019-10-06 03:03:00 AEDT
Iter. #14: Sun 2019-10-06 03:04:00 AEDT
Iter. #15: Sun 2019-10-06 03:05:00 AEDT
Iter. #16: Sun 2019-10-06 03:06:00 AEDT
Iter. #17: Sun 2019-10-06 03:07:00 AEDT
Iter. #18: Sun 2019-10-06 03:08:00 AEDT
Iter. #19: Sun 2019-10-06 03:09:00 AEDT
Iter. #20: Sun 2019-10-06 03:10:00 AEDT
$ TZ=Australia/Sydney faketime 2019-10-06 build/systemd-analyze calendar 2/4:30 --iterations=3
Original form: 2/4:30
Normalized form: *-*-* 02/4:30:00
Next elapse: Sun 2019-10-06 06:30:00 AEDT
(in UTC): Sat 2019-10-05 19:30:00 UTC
From now: 5h 29min left
Iter. #2: Sun 2019-10-06 10:30:00 AEDT
(in UTC): Sat 2019-10-05 23:30:00 UTC
From now: 9h left
Iter. #3: Sun 2019-10-06 14:30:00 AEDT
(in UTC): Sun 2019-10-06 03:30:00 UTC
From now: 13h left
79fb2af to
ad0fa6d
Compare
Member
Author
A test case wouldn't be bad. I was too lazy to add one.
Sounds good. I now pushed a new version that propagates errno, no other changes. |
2 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
If we tested a candidate time that would fall onto the DST change, and we
realized that it is now a valid time ('cause the given "hour" is missing),
we would jump to to beginning of the next bigger time period, i.e. the next
day.
mktime_or_timegm() already tells us what the next valid time is, so let's reuse
this, and continue the calculations at this point. This should allow us to
correctly jump over DST changes, but also leap seconds and similar. It should
be OK even multiple days were removed from calendar, similarly to the
Gregorian-Julian transition. By reusing the information from normalization, we
don't have to make assumptions what the next valid time is.
Fixes #13745.
$ TZ=Australia/Sydney faketime '2019-10-06 01:50' build/systemd-analyze calendar 0/1:0/1 --iterations 20 | grep Iter
Iter. #2: Sun 2019-10-06 01:52:00 AEST
Iter. #3: Sun 2019-10-06 01:53:00 AEST
Iter. #4: Sun 2019-10-06 01:54:00 AEST
Iter. #5: Sun 2019-10-06 01:55:00 AEST
Iter. #6: Sun 2019-10-06 01:56:00 AEST
Iter. #7: Sun 2019-10-06 01:57:00 AEST
Iter. #8: Sun 2019-10-06 01:58:00 AEST
Iter. #9: Sun 2019-10-06 01:59:00 AEST
Iter. #10: Sun 2019-10-06 03:00:00 AEDT
Iter. #11: Sun 2019-10-06 03:01:00 AEDT
Iter. #12: Sun 2019-10-06 03:02:00 AEDT
Iter. #13: Sun 2019-10-06 03:03:00 AEDT
Iter. #14: Sun 2019-10-06 03:04:00 AEDT
Iter. #15: Sun 2019-10-06 03:05:00 AEDT
Iter. #16: Sun 2019-10-06 03:06:00 AEDT
Iter. #17: Sun 2019-10-06 03:07:00 AEDT
Iter. #18: Sun 2019-10-06 03:08:00 AEDT
Iter. #19: Sun 2019-10-06 03:09:00 AEDT
Iter. #20: Sun 2019-10-06 03:10:00 AEDT
$ TZ=Australia/Sydney faketime 2019-10-06 build/systemd-analyze calendar 2/4:30 --iterations=3
Original form: 2/4:30
Normalized form: --* 02/4:30:00
Next elapse: Sun 2019-10-06 06:30:00 AEDT
(in UTC): Sat 2019-10-05 19:30:00 UTC
From now: 5h 29min left
Iter. #2: Sun 2019-10-06 10:30:00 AEDT
(in UTC): Sat 2019-10-05 23:30:00 UTC
From now: 9h left
Iter. #3: Sun 2019-10-06 14:30:00 AEDT
(in UTC): Sun 2019-10-06 03:30:00 UTC
From now: 13h left