Robert Lange <rcl24@drexel.edu> (2014-10-19): > I agree with the original poster and argue for increasing the > priority of this bug, because under certain circumstances it may > make a Debian system appear to be unbootable. > > As it stands, if something causes the computer's EFI NVRAM to get > wiped (e.g., user error, firmware bug, firmware upgrade), many > computers will appear to be unbootable because their EFI > implementations provide no way to scan for non-default bootloaders > in the EFI System Partition. For a technically-oriented user, a boot > disc such as rEFInd can be used to fix this, but less-savvy users > will simply think that Debian broke. > > For reference, the default bootloader exists in the EFI System > Partition at \EFI\BOOT\bootx64.efi (/boot/efi/EFI/BOOT/bootx64.efi > as mounted in Debian) on a x86-64 architecture machine [UEFI > Specification Version 2.4 (Errata B) Section 3.4.1]. If no NVRAM > bootloader entries are applicable on a computer, the system will > boot from the first default bootloader it finds on the first ESP > partition it finds. > > To increase robustness of installations against firmware issues, the > Debian installer should prompt the user to install a copy of the > bootloader into the default bootloader location of the ESP. I would > recommend that the default value of this prompt be Yes if no default > bootloader currently exists, and No if one currently exists (along > with the requisite warning about overwriting). Steve, what's your take on this topic? Mraw, KiBi.
Attachment:
signature.asc
Description: Digital signature