Summary: | konsole5 crashes right after login | ||
---|---|---|---|
Product: | [Applications] konsole | Reporter: | Hendrik Woltersdorf <hendrikw> |
Component: | general | Assignee: | Konsole Developer <konsole-devel> |
Status: | RESOLVED DOWNSTREAM | ||
Severity: | crash | CC: | auxsvr, bernd_schmidt_sh, bulbigood, carl.erwin, elchevive, hrvoje.senjan, jfnhs1, kollix, luizluca, Markus.Elfring, mrmazda, rainer, rftennis, skip, thomas.luebking, wbauer1 |
Priority: | NOR | Keywords: | drkonqi |
Version: | 3.0.1 | ||
Target Milestone: | --- | ||
Platform: | openSUSE | ||
OS: | Linux | ||
Latest Commit: | Version Fixed In: | ||
Sentry Crash Report: |
Description
Hendrik Woltersdorf
2015-02-15 07:28:50 UTC
*** Bug 345117 has been marked as a duplicate of this bug. *** *** Bug 340075 has been marked as a duplicate of this bug. *** The same crash happens here on every login. I'm using the latest versions, Konsole 3.0.1 from KDE Applications 14.12.3, KDE Frameworks 5.8.0, and Qt 5.4.1. *** Bug 344523 has been marked as a duplicate of this bug. *** I can avoid this only either by using Fedora 22 or Rawhide (host big41, ATI Cedar), instead of openSUSE Tumbleweed (host gx745, I), or by closing Konsole before logging out. 64 bit TW w/ plasma5-workspace-5.2.2-3.1 & konsole5-14.12.3-1.1. *** Bug 346830 has been marked as a duplicate of this bug. *** *** Bug 347108 has been marked as a duplicate of this bug. *** *** Bug 347477 has been marked as a duplicate of this bug. *** Sorry for the troubles, this is a openSUSE specific crash, as we're applying https://git.reviewboard.kde.org/r/119593/ without upstream consent... @Thomas, any ideas about this one? From *very* limited testing, seems now the the rb patch is no longer needed with Qt 5.4.x (?) The restored client restores its window size, that (due to the patch) processes events what triggers a state save (apparently posted there by the SM) what requires _viewSplitter->activeContainer(), which simply isn't present at this moment (konsole just starts) I don't know enough about the konsole code, but assume scheduling the connection to QGuiApplication::saveStateRequest() is not an option?! Checking Qt5.4 to see whether they forward early QWidget resizes to the platform window. "nope" - the Qt code changed a bit (instead of updateGeometry there's now updateSize and updatePosition) but the geneal issue remains, https://bugreports.qt.io/browse/QTBUG-40584 is still open and the testcase attached there still reliably fails on vanilla Qt. Testcase: https://bugreports.qt.io/secure/attachment/45495/qwindowsize.cpp The problem dealt with by this patch is some sort of cornercase - it happens when you resize a widget before the eventloop is up, resp. before calling "show". Most prominent cases are nested eventloops by modal dialogs. i see. somehow i feel users will find crashes a bit more annoying than incorrect size (?), i shall retract the kconfig patch for now. (In reply to Hrvoje Senjan from comment #12) > i shall retract the kconfig patch for now. Yes. Leaving aside the crash (I'd here say that konsole should harden the function, because many things could processEvents() in an early phase), the patch was ineffective w/o the fix to the Qt bug (why I never pushed it itfp) As I just figured, that patch doesn't work w/ Qt 5.4 anymore anyway - sorry :-( *** Bug 347955 has been marked as a duplicate of this bug. *** *** Bug 347904 has been marked as a duplicate of this bug. *** *** Bug 347998 has been marked as a duplicate of this bug. *** *** Bug 348007 has been marked as a duplicate of this bug. *** |