Hi, When the 3.14.x kernel first hit Debian testing, I tried booting without blacklisting nouveau, and got the same systemd's emergency prompt caused by disk corruption.
The recent comments related to the issue, at the github page for bumblebee's project ( https://github.com/Bumblebee-Project/bbswitch/issues/78 ), mention that the issue goes away by adding acpi_osi="Windows 2013" to the kernel command line. That certainly worked for me, and right now I have bumblebee-nvidia installed and working properly, and the laptop's battery consumption is down by 40-50%. As far as I know, the acpi_osi trick doesn't solve the issue in all the cases ---e.g., it didn't work for Mr. Smith. The bumblebee package installs its own bumblebee.conf into /etc/modprobe.d, which effectively blacklists nouveau and nvidia*, so it seems that for the NVIDIA Optimus card those modules shouldn't be loaded at all by default. -- John. On Sun, 2014-08-10 at 20:21 +0100, Ben Hutchings wrote: > Sorry we didn't respond to this earlier. > > On Tue, 2014-04-22 at 06:30 -0500, John M. wrote: > > Hi, > > > > My workaround was to (boot in recovery mode and) blacklist nouveau, as > > in: > > > > # echo 'blacklist nouveau' > /etc/modprobe.d/blacklist-nouveau.conf > > > > After that, reboot in normal mode. > > > > It seems that the 3.13.x kernel is trying to load drivers for both the > > Intel and NVIDIA cards, and there's a BIOS-related bug that makes the > > NVIDIA card misbehave when on Linux. Previous versions of the kernel > > didn't do that. > > > > I hope that helps. > > Linux is supposed to be able to switch between the Intel and Nvidia GPUs > at run-time, so loading both drivers is right. > > The failure may be related to run-time power management and might be > fixed in a later version. Have you tested a recent kernel version with > nouveau re-enabled? There is currently Linux 3.14 in testing/unstable > and 3.16 in experimental. > > Ben. >
