Created attachment 468622 [details] patch against endless loop crafted for dev-qt/qtdeclarative-opensource-src-5.6.2 Hello all, this is a may-be-bugfix for a bug which hunted me for the last 2 years. I'm going to post this for documentation and for other people which may have the same issue. If the bug doesn't reappear in the next 3 months I consider this a fix. What happened: Every now and then the plasmashell (from kde-plasma/plasma-workspace) freezes due to an endless loop. Most of the (seldom) occurrences happen during a popup window from plasmashell. I had to kill -15 the plasmashell and start it up again from an open terminal shell. This behaviour has been described in https://bugzilla.redhat.com/show_bug.cgi?id=1193742#c40 and https://bugzilla.redhat.com/show_bug.cgi?id=1237269 But different from the announcement in show_bug.cgi?id=1237269 the patch hasn't made it into dev-qt/qtdeclarative-5.6.2, redhat wanted to include the patch in qt-5.5.1, upstream seems not. I crafted the patch to fit to the current versions of dev-qt/qtdeclarative. 2 patches attached. Version info: Currently I use kde-plasma/plasma-workspace-5.8.5 and dev-qt/qtdeclarative-5.6.2, but the bug appeared within the last ~20 months. best regards
Created attachment 468624 [details, diff] patch against endless loop crafted for dev-qt/qtdeclarative-opensource-src-5.7.1
Has this ever been reported to upstream? In both linked redhat bugs I don't find a link to Qt.
(In reply to j.habenicht from comment #1) > Created attachment 468624 [details, diff] [details, diff] > patch against endless loop crafted for > dev-qt/qtdeclarative-opensource-src-5.7.1 ping - can you provide a link where this is submitted to Qt upstream?
https://bugzilla.redhat.com/show_bug.cgi?id=1237269#c8 mentions a fix https://bugs.kde.org/show_bug.cgi?id=348385 Only they fixed plasmashell and not Qt unlike the previous bug report. But they analysed the endless loop and found the cause, an integer overflow. So the test for values >=0 in the Qt patch seems to work around that error, too.
Neither of those is an upstream bug link. And I've checked 5.9.0, this change is not in it. If this was fixed in plasmashell, can you even reproduce the issue at this point?
In 5.9.0, Fedora is still applying this patch, but for some reason it is not listed as upstreamable.
(In reply to j.habenicht from comment #0) > Version info: > Currently I use kde-plasma/plasma-workspace-5.8.5 and > dev-qt/qtdeclarative-5.6.2, but the bug appeared within the last ~20 months. Was it ever reproduced in Plasma 5.8.x? Please test current stable Plasma-5.10.5 without the non-upstream patch to establish if there actually still is an issue. Otherwise the story remains too thin.
sorry for the late reply. I lost track of this report for a long time. Currently I run kde-plasma/plasma-workspace-5.16.5 and this bug did not reappear within the last months. So I conclude this bug seems to be fixed in the last versions. Please close this report. thank you