EngineerZone
All Places
Processors and DSP
Software and Development Tools
Linux Distribution for Blackfin
Discussions
Hi All,
We use 2010R1RC5 in our products and now I discovered a bug. We are in the GMT+1 timezone and we use DST. Our etc/TZ file looks like this:
LST-1LDT,M3.5.0/2:00:00,M10.5.0/3:00:00
The DST to summer time was in last Sunday but the system clock still shows the winter time. I checked that the DST will change to summer time on the next Sunday. It means in this week our clock lates one hour.
I know it is Thursday now and I have to wait some days only. Our customers haven't discovered this bug yet.
I checked the changing in the future dates and it seems OK.
Maybe the problem caused by the leap year?
You will probably have to update the tables yourself.
Basically the problem is that the exact date of the change over changes now more frequently,
Example: In the US it use to be always the last Sunday in October now it is the first Sunday in November.
http://www.cisco.com/en/US/tech/tk648/tk362/technologies_tech_note09186a00807ca437.shtml
Some countries/areas are frequently changing the rules... so,
You may want to start just using GMT/UTC time instead.
I don't agree. In Europe the rules are stable since decades. My rules in TZ file says the changeover is on last Sunday of March.
If there is no leap year, the last Sunday were 31st of March. But the leap day inserted in February that Sunday moved to 1st of April, so the last Sunday was in 25th of March.
Then it may be a bug in the libraries. I don't think it has anything to do with the leap-year... but, more to do with the fact 31st of March is on a Saturday.
James
I found your answer here:
The bug needs three things to line up for the problem.
1) Must be a leap year.
2) Must use the method you have for the time-zone information.
3) Last day of the month must fall on a Saturday (Sunday - 1) in your case also.
I think it may be a one time deal for the moment, but you should be able to fix before the problem happens again.
Thank you James. From today morning all of our devices show the correct time. Only one of out customer complained about the time, but it is OK now.
The bug fixed in 2011. I plan to upgrade our system to 2011R1 soon. I hope this fix already included in our libc.
No, I just put a post on the tools list to address this.