Active Topics

 



Notices


Reply
Thread Tools
Posts: 1,100 | Thanked: 2,797 times | Joined on Apr 2011 @ Netherlands
#111
Originally Posted by freemangordon View Post
You (or whoever) can create separate package with only debug symbols, in such a way object code will be release, but in case of a crash, debug symbols could be loaded in gdb (along with coredump) for "hardcore" hackers to be able to find the reason for crash.
It is not totally clear to me.
Compiled with debug -> no crash
Running within gdb -> no crash

So how can I get a useful coredump? Do I have to use other debug flags, or use another approach?
 
Posts: 3,074 | Thanked: 12,960 times | Joined on Mar 2010 @ Sofia,Bulgaria
#112
@ade - how do you run it with gdb? AFAIK the best way is to attach gdb to a running process, not to start it from gdb. Othewise you could run the process under incorrect user and all stuff like that.

In order to have coredump you have to install sp-rich-core (to capture) and sp-rich-core-postproc(to extract) coredumps.

To get a post-mortem backtrace from coredump you have to:

1. attach gdb to a running process
2. load coredump file
 

The Following 2 Users Say Thank You to freemangordon For This Useful Post:
Posts: 1,100 | Thanked: 2,797 times | Joined on Apr 2011 @ Netherlands
#113
I tried both (running in gdb and attaching to the process). But no crashes with the debug info version, so I must use a version without debug info.
I could install the packages you named (don't know them, but also never debugged before this way ), but I am afraid I won't have enough info without the debug info attached.

But I will give it a try later.

Thanks for your time.
 

The Following User Says Thank You to ade For This Useful Post:
Posts: 838 | Thanked: 3,384 times | Joined on Mar 2009
#114
I have run adeclock under valgrind (it is 40 times slower) and haven't found anything suspicious. Some underlying libraries reports some leaking (not even sure are these real or false alerts). Also I haven't yet saw any crashes.
 

The Following 2 Users Say Thank You to AapoRantalainen For This Useful Post:
Posts: 3,074 | Thanked: 12,960 times | Joined on Mar 2010 @ Sofia,Bulgaria
#115
Originally Posted by AapoRantalainen View Post
I have run adeclock under valgrind (it is 40 times slower) and haven't found anything suspicious. Some underlying libraries reports some leaking (not even sure are these real or false alerts). Also I haven't yet saw any crashes.
It is compiled in scratchbox against Qt 4.7.4?
 
Estel's Avatar
Posts: 5,028 | Thanked: 8,613 times | Joined on Mar 2011
#116
@freemangordon
The problem with this bugger, is that it refuses to crash when kin any debug mode. It just don't crash, no matter what. Same, when shell script is wrapped around it. It fails only stand-alone, and in rare, random situations.

/Estel
__________________
N900's aluminum backcover / body replacement
-
N900's HDMI-Out
-
Camera cover MOD
-
Measure battery's real capacity on-device
-
TrueCrypt 7.1 | ereswap | bnf
-
Hardware's mods research is costly. To support my work, please consider donating. Thank You!
 
Posts: 838 | Thanked: 3,384 times | Joined on Mar 2009
#117
Originally Posted by freemangordon View Post
It is compiled in scratchbox against Qt 4.7.4?
Yes. Actually under scratchbox2.
 
Posts: 1,100 | Thanked: 2,797 times | Joined on Apr 2011 @ Netherlands
#118
@AapoRantalainen
Are you using a version without debug info?
And if you use something "valgrind --leak-check=yes worldclock" I can imagine it works as a wrapper, preventing it from crashing (like starting it in gdb)
 
Posts: 3,074 | Thanked: 12,960 times | Joined on Mar 2010 @ Sofia,Bulgaria
#119
@ade, better join #maemo-ssu, there is an ongoing discussion on the matter
 
Posts: 1,100 | Thanked: 2,797 times | Joined on Apr 2011 @ Netherlands
#120
Thanks for the advice, but I can't use IRC at the moment

Maybe in an hour or so, else I just have to find the logs.
 
Reply

Tags
clock-ui, replacement, worldclock


 
Forum Jump


All times are GMT. The time now is 21:51.