Le 12/06/2017 à 11:00, Darac Marjal a écrit : > On Sun, Jun 11, 2017 at 10:50:17PM +0200, Pétùr wrote: >> I have a serious bug when using flashplayer in debian sid. >> >> This bug freezes everything and I have to do a hard reboot. It happens >> with flashplugin and pepperflash when I play a video with flash (both >> tested with firefox and chromium). Today I am using pepperflash, >> installed by the package browser-plugin-freshplayer-pepperflash. >> >> The bug could occur after playing 2s or 2h. When I play a flash video, I >> have ~50% chances to be affected. > > Have you tried switching to a VT and back when this happens? There was a > bug in... I *think* it was something like opengl ... recently, whereby a > bunch of unrelated programs (chrome/chromium included) could freeze the > display. Switching away from X and back caused the display to refresh > and things would work again (for a time).
Yes I tried to switch to TTY1 with Ctrl+Alt+F1 or to kill X server. The freeze is global and nothing works at all except a hard, electrical, reboot. >> In /var/log/daemon.log I have some inconsistencies at 21:25:28 >> represented by: >> >> ^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@... > > This is less's representation of null characters and is an artefact of > journaling file systems. Syslog has told the file system "I'm about to > write X bytes to /var/log/daemon.log" and then, at some point during the > writing of that, you've killed the power. During either fsck or mount > (depending on the file system), the journal is replayed - that is, there > is a log saying "X bytes were being writen to /var/log/daemon.log". Now, > most of those bytes never made it to the disk, but the record is still > there. So the file grows by X bytes of nothing. Now the file system is > consistent (the metadate, at least, the idea being that it's better to > have SOME of the file still there, rather than a broken filesystem) Thanks for the explanation.
