Project

General

Profile

Bug #3627

TVHeadend locks up and spins a CPU after the Spring DST change

Added by Michael Marley over 8 years ago. Updated about 8 years ago.

Status:
Fixed
Priority:
Normal
Assignee:
-
Category:
Crashes
Target version:
-
Start date:
2016-03-13
Due date:
% Done:

100%

Estimated time:
Found in version:
4.1-1670~g5a94400~wily
Affected Versions:

Description

Last night when the time changed for Daylight Savings Time, both of the TVHeadend instances I operate became unresponsive and starting spinning one CPU core at 100%. After restarting, they resumed normal operation. I have attached full GDB backtraces from both of them.


Files

gdb.txt (39.7 KB) gdb.txt Michael Marley, 2016-03-13 12:40
gdb2.txt (72.7 KB) gdb2.txt Michael Marley, 2016-03-13 12:40

History

#1

Updated by Jaroslav Kysela over 8 years ago

Unfortunately, it really difficult to track what's wrong there - I need to know where it loops, some analysis:

now=1457849040, zone=EDT

TZ=/usr/share/zoneinfo/US/Eastern  ./a.out "4 14 * * *" 1457849040
TIME = 1457849040 - 2016-03-13 01:04 (DST 0) (ZONE EST)
min  = 000010000000000000000000000000000000000000000000000000000000
hour = 000000000000001000000000
mday = 1111111111111111111111111111111
mon  = 111111111111
wday = 1111111
NOW: 2016-03-13 01:04
NXT: 1457892240 - 2016-03-13 14:04 (DST 1) (ZONE EDT)

but...

#6  0x0000000000482749 in cron_multi_next (cm=0x2148990, now=1457849040, ret=ret@entry=0x7ffe9be92dc0) at src/cron.c:341
        i = 1
        r = 1457849040
        t = 1457849040

... if r is equal to now and it's passed back to the caller - where's the source ?

ok.. ota epg has also second time...

TZ=/usr/share/zoneinfo/US/Eastern  ./a.out "4 2 * * *" 1457849040
TIME = 1457849040 - 2016-03-13 01:04 (DST 0) (ZONE EST)
min  = 000010000000000000000000000000000000000000000000000000000000
hour = 001000000000000000000000
mday = 1111111111111111111111111111111
mon  = 111111111111
wday = 1111111
NOW: 2016-03-13 01:04
NXT: 1457849040 - 2016-03-13 01:04 (DST 0) (ZONE EST)

and we're there..

#2

Updated by Michael Marley over 8 years ago

I don't really understand what you posted. Is there something else I can do to help provide more data?

#3

Updated by Jaroslav Kysela over 8 years ago

  • Status changed from New to Fixed
  • % Done changed from 0 to 100

Applied in changeset commit:tvheadend|3927788bb71ad0041a00d0daa35fae1cb6355b55.

#4

Updated by Jaroslav Kysela over 8 years ago

Fixed in v4.1-1672-g3927788 .

#5

Updated by C K over 8 years ago

sry perexg to open it again, it's not fixed, see #3661

#6

Updated by gzu gzu about 8 years ago

I still think this is an issue.
tvheadend was unresponsive after October DST change in Australia.

#7

Updated by Jaroslav Kysela about 8 years ago

Thats bad. Which version exactly?

Also available in: Atom PDF