On Thu, Apr 14, 2016 at 1:23 AM, Knurpht - Gertjan Lettink
Op woensdag 13 april 2016 17:10:17 CEST schreef Chris Murphy:
On Wed, Apr 13, 2016 at 3:50 PM, James Knott
wrote: On 04/13/2016 04:50 PM, Per Jessen wrote:
Umm, except when a system has been hibernated.
Since when do you get to choose another OS when coming out of hibernation?
I've always been greeted by the exact same GRUB boot menu options whether cold boot or resuming from suspend to disk. So... that's presumably how I could choose the wrong thing. That means your GRUB is controlled by "the other OS".
Yes and for mortal users "GRUB is controlled by the other OS" is total nonsense. It means nothing to a regular user who is the target market of a GUI installer, using automatic partitioning. So you need to design things accordingly, or accept that your OS is potentially dangerous for mortal users who aren't experts. I don't know the details of how a Linux hibernation image is properly restored. Can someone explain that? The hibernation image is written to the swap partition, I get that part, and there is a resume= hint as a boot parameter. So the system basically cold boots up, the bootloader is executed, and runs the proper boot menu entry that includes that resume= parameter, but also includes loading the kernel and the initramfs. Now what? The kernel knows to look for the hibernation image at the resume= defined swap location? And it ignores the initramfs? Does the kernel to any kind of sanity checking to make sure it's restoring the correct hibernation file? While this is not sufficient, at the very least I'd like to think that the kernel can compare its signing key to that of the one inside the hibernation image before committing to resume. If they don't match, for sure the kernel should panic before trying to write anything to disk. -- Chris Murphy -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org