| Summary: | Crash on writing some Code (C++, cmake, nogui) | ||
|---|---|---|---|
| Product: | [Applications] kdevelop | Reporter: | Bernhard Friedreich <friesoft> |
| Component: | general | Assignee: | kdevelop-bugs-null |
| Status: | RESOLVED UPSTREAM | ||
| Severity: | crash | CC: | aleixpol |
| Priority: | NOR | ||
| Version First Reported In: | unspecified | ||
| Target Milestone: | --- | ||
| Platform: | Unlisted Binaries | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | ||
| Sentry Crash Report: | |||
|
Description
Bernhard Friedreich
2009-08-18 15:17:11 UTC
libc? Sounds again like MALLOC_CHECK_=1 should be used as a work around. @reporter: http://linux.die.net/man/3/malloc <-- search this for an explanation. @apaku / other devs: Should we reassign these bug reports to kdelibs? Or - in this case - to upstream since it seems to be something in Qt? Just wanted to report that setting MALLOC_CHECK_=1 fixes all sorts of crashes on this system :-) (ranging from dolphin crahes when using tabs to this kdevelop crash) Thx for the info.. but in the end those are still bugs? - as stated on the page you pointed me to one of the causes could be the multiple use of free() - which you shouldn't - so in theory those are bugs - now is this just on my system that MALLOC_CHECK is not set "correctly" or will those bugs also appear on other systems in the future? I'm using an up2date archlinux which sometimes results in funny effects which other distros don't encounter YET because of older versions.. my glibc version: glibc 2.10.1-3 What would you propose to set this flag permanently? Is exporting it in the bashrc sufficient? IMHO re-assign/close-as-upstream unless its our code that crashes. (In reply to comment #2) > Thx for the info.. but in the end those are still bugs? There's definetly a bug somewhere, the question is where. See below. > now is this just on my system > that MALLOC_CHECK is not set "correctly" or will those bugs also appear on > other systems in the future? KDE release tarballs will have MALLOC_CHECK set to 1 in startkde. If you're building kdebase from trunk/branches then you have to set it yourself. > my glibc version: > glibc 2.10.1-3 The problem also relates to that particular libc version as it seems to be far more restrictive on what it allows to go "through" the checker without aborting. libc 2.9 doesn't cause any of the crash reports I've seen so far. Hence I'm not 100% convinced these are app-bugs, it could just as well be a bug in libc 2.10. > What would you propose to set this flag permanently? Is exporting it in the > bashrc sufficient? Yeap thats sufficient. |