Summary: | error appears during upgrade using adept | ||
---|---|---|---|
Product: | [Frameworks and Libraries] kdelibs | Reporter: | Fernando Díaz <fdiazuy> |
Component: | kded | Assignee: | David Faure <faure> |
Status: | RESOLVED DUPLICATE | ||
Severity: | crash | CC: | finex, frank78ac, rignacioII |
Priority: | NOR | ||
Version: | 4.1 | ||
Target Milestone: | --- | ||
Platform: | Ubuntu | ||
OS: | Linux | ||
Latest Commit: | Version Fixed In: |
Description
Fernando Díaz
2008-11-13 05:40:03 UTC
Hi! Could you explain how to reproduce the crash or does it happens randomly? Moreover you should read this page for the instructions about posting an useful backtrace: http://techbase.kde.org/Development/Tutorials/Debugging/How_to_create_useful_crash_reports Thanks! I'd say it's probably another duplicate of bug 165548. Frames starting from #11 seem to match with the backtraces in that bug. It crashes in QFile::encodeName (), so it looks like it got a bit further in KDirWatchPrivate::inotifyEventReceived(). In most of the other backtraces, the crash occurred in the line if (sub_entry->path == e->path + '/' + path) break; but this looks like it arrived a few lines further down at QByteArray tpath = QFile::encodeName(path); and crashed in some function called from there. Anyway, it looks like something about the 'path' is going wrong. I don't actually have a clue about KDirWatch, but maybe one could add an assert or some debugging output before the critical line is reached to have a better idea of what goes wrong? *** Bug 175779 has been marked as a duplicate of this bug. *** SVN commit 1058169 by cfeck: Fix broken reading in inotifyEventReceived Multiple bugs: * MAX_PATH is 4096 on linux, but inotify_event needs some bytes as well * the "pending" variable was incorrectly abused to mean both pending bytes in buffer as well as pending bytes in stream * the offset was not reset to zero when the next chunk was reading * when an event was larger than what was available in buffer, it was reading beyond the buffer boundaries Many thanks to David for review and help with improving clarity of the code. David, please test for regressions, maybe add a test which changes some 1000 files in one go. CCMAIL:faure@kde.org CCBUG:165548 CCBUG:173696 CCBUG:175005 CCBUG:198692 CCBUG:204178 M +20 -9 kdirwatch.cpp WebSVN link: http://websvn.kde.org/?view=rev&revision=1058169 *** This bug has been marked as a duplicate of bug 165548 *** |