On Tue, Dec 29, 2015 at 1:17 PM, Genes Lists lists@sapience.com wrote:

Fully updated from testing repo.

Since update to plasma 5.5 I am getting periodic crashes of plasmashell -trace back below. Seems to be connected with dri/swrast_dri.so.

i915 graphics so not sure why swrast_dri is called - but then again not sure what that does exactly (sounds like software rendering).

Anyone got suggestions or seen similar.

plasma restarts itself and things go normally for a while until the next crash.

gene

--- trace back

Application: Plasma (plasmashell), signal: Segmentation fault Using host libthread_db library "/usr/lib/libthread_db.so.1". [Current thread is 1 (Thread 0x7f7c26ff0800 (LWP 2502))]

...

Thread 1 (Thread 0x7f7c26ff0800 (LWP 2502)): [KCrash Handler] #5 0x00007f7c34c52b22 in __memmove_avx_unaligned () from /usr/lib/libc.so.6 #6 0x00007f7c1763dc8d in ?? () from /usr/lib/xorg/modules/dri/swrast_dri.so #7 0x00007f7c1763c645 in ?? () from /usr/lib/xorg/modules/dri/swrast_dri.so #8 0x00007f7c1db1e8fa in ?? () from /usr/lib/qt/qml/org/kde/plasma/core/libcorebindingsplugin.so #9 0x00007f7c1db1f3f9 in ?? () from /usr/lib/qt/qml/org/kde/plasma/core/libcorebindingsplugin.so #10 0x00007f7c1db1f954 in ?? () from /usr/lib/qt/qml/org/kde/plasma/core/libcorebindingsplugin.so #11 0x00007f7c38652ac0 in QQuickWindowPrivate::updateDirtyNode(QQuickItem*) () from /usr/lib/libQt5Quick.so.5 #12 0x00007f7c3865331b in QQuickWindowPrivate::updateDirtyNodes() () from /usr/lib/libQt5Quick.so.5 #13 0x00007f7c38653490 in QQuickWindowPrivate::syncSceneGraph() () from /usr/lib/libQt5Quick.so.5 #14 0x00007f7c3862219c in ?? () from /usr/lib/libQt5Quick.so.5

...

For anyone having this kind of problem some additional useful info might be obtained by running the suggested commands in the page at http://dri.freedesktop.org/wiki/glxinfo/ - and getting at some 3d rendering info specifically when the system has a problem indicated by the fact that the rendering is falling back to the swrast_dri driver. Any chance this report relates to failed graphics drivers which may originate in bugs in the mesa/plasma/X etc code - but at least some initial info showing where there is a problem may point to where to look for more problems so narrow down where the initial failure occurs?