Thanks for your reply Andreas. > What else do you expect after putting the system under > extreme memory pressure
I expect, intel driver bug-or-not ASIDE, the linux to kill the chrome and keep the X. tmpfs is another story, because there's nothing huge to kill except X. But again i expect the kernel to kill chrome first, X later. > PS: and you would be perfectly fine if X wouldn't crash but would be > killed by the oomkiller - with the same situation afterwards ? Now that you're asking it, yes! because that i could fix with Martijn's idea by setting chrome's oom_score_adj to a low value which would kill chrome first and enough memory would free up to keep the X. > Compare the versions of xserver-xorg-core and xserver-xorg-video-intel > between the Ubuntu version and the Debian version you used. Yes sir, xserver-xorg-video-intel debian: 2:2.19.0-6 ubuntu: 2:2.14.0-4 > Could you try the vesa (xserver-xorg-video-vesa) driver instead and > see what happens under extreme memory pressure? Excellent idea, dont know why i havent done this already. That's why i turned to you for help. btw this is my first time on debianbts, so i'm sorry if i do something out of etiquette - my bad english does not help either. I uninstalled xserver-xorg-video-intel as i'm dumb and dont know how to disable a driver (tried blacklist but dont know the module name, googled, nothing, so), i rebooted, tried to crash the system, opened 7 youtube videos, chrome got killed, X survived. Just like on my other systems and OSes. So it seems proven that the intel driver is the problem. Maybe because vesa driver doesnt use dynamic video memory only a fixed size small one? > The ret value is not returned to the > caller (X), so that does not know about the failure. Any chance to fix this? > I don't think its running out of *video memory*. It's just running out > of normal RAM. and > To me this looks like the intel driver cannot do a task because it > cannot allocate memory So it cannot allocate shared "video memory" ;) > this does not neccessarily trigger the kernel's OOM killer Yes, because that's "video memory", not system memory. The problem is exactly what you wrote: the oom-killer cannot step in to free memory before the intel driver crashes. Can i somehow allocate the video memory from the system memory fixed/permanently? > And since I suspect xserver-xorg-video-intel, you might want try a > newer version of that driver. Okay.. # echo 'APT::Default-release "stable";' > /etc/apt/apt.conf.d/00default # echo -e 'Package: *\nPin: release a=stable\nPin-Priority: 700\n\nPackage: *\nPin: release a=testing\nPin-Priority: 650\n\nPackage: *\nPin: release a=unstable\nPin-Priority: 600\n' > /etc/apt/preferences.d/pinning # echo -e '\ndeb http://ftp.debian.org/debian/ jessie main contrib non-free\ndeb-src http://ftp.debian.org/debian/ jessie main contrib non-free\n' >> /etc/apt/sources.list # apt-get update # apt-get install xserver-xorg-video-intel/jessie but apt-pinning did not work, since it requires update of libc. So, i did: # apt-get install dpkg-dev # apt-get build-dep xserver-xorg-video-intel/jessie # apt-get -b source xserver-xorg-video-intel/jessie # dpkg -i xserver-xorg-video-intel_2.21.15-1_amd64.deb Guys, did i do this right? This building thing? I rebooted and ran the test, it did not crash. As i started to rejoice, i realized, i'm probably using the vesa driver, so still not knowing how to specify or even list the driver to use, i removed the xserver-xorg-video-vesa package and rebooted. It crashed just like with the old driver. You see my problem is in this current state the system is broken: it is wrong my grandma opens 10 browser tabs and the next thing she sees is the lightdm login screen - she doesnt know her password, she doesnt even know she HAS a username! (autologin) It would be totally fine if there would be only 1024-128=896 MB free ram to the system, hell i wouldnt even care if there would be 256 MB and she could just open 1 browser tab, but her aisleriot solitaire and mahjongg games should stay intact. Very important. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
