| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
i.e. in \EFI\BOOT instead of in \EFI\mageia. This should be used if
installing on a removable device, and provides a workaround for buggy
BIOSs that don't respect the EFI NVRAM settings.
|
|
|
|
|
| |
This bug was introduced by the change from renaming to copying grub.cfg.
renamef doesn't fail if the file doesn't exist. cp_af does.
|
|
|
|
|
|
| |
fix typo x86 -> x64 for matching 64bit grub2-efi(x86-64) (martinw)
(bug introduced in commit: e540363478de809a2a9769c5794b3adab60b63eb)
|
|
|
|
|
|
|
| |
Mageia mounts the ESP as /boot/EFI and thes grub2 looks at /boot/EFI/EFI
But some windows installation end in having /boot/EFI/efi (lowercase)
Rename the directory as upercase if needed so that grub2-install doesn't
fail (mga#22395)
|
| |
|
| |
|
|
|
|
|
| |
rationale: update-grub2 (or more likely os-prober) 's failure can leave
the system in an unbootable state
|
|
|
|
|
|
|
|
| |
Partially tested on Amazon EC2, instance still didn't boot (as it
needed it to be installed in removable mode (EFI/boot/bootaa64.efi)
to be found) but at least drakboot --boot worked (except for
failing to enabled acpid service) and things seem to be in the
expected place.
|
| |
|
| |
|
|
|
|
|
|
| |
Although we could install a working rEFInd boot manager, it uses
the kernel EFI stub loader to boot the O/S, and that won't work if
the EFI types don't match.
|
| |
|
| |
|
|
|
|
|
| |
should have been done with:
commit 5583c3f9e8de83b614f0707c91ee40227ceec604
|
|
|
|
|
|
|
| |
in automatic mode, we would potentially suggest a wrong disk if first
disc has a GPT layout _without_ a BIOS boot partition.
rationale:
GPT disks w/o a BIOS boot partition do not have free space for grub2 to embed
|
|
|
|
| |
needed for next commit
|
|
|
|
|
|
|
| |
Fixes a couple of issues found when reading a manually generated file:
- allow for extra white space between label and append options
- don't simplify the label when writing - rEFInd allows any text
string here
|
| |
|
|
|
|
|
| |
A parsing bug in bootloader::read_grub2() stopped the main menuentry
being added to $bootloader{entries}.
|
| |
|
|
|
|
|
|
|
|
|
| |
elevator=deadline is supposed to give better performance on flash
storage, and used by default on most Raspberry Pi distributions.
Ideally, this should not be set here for specific boards, but
automatically for all architectures depending on the storage type.
Archlinux appears to use a /etc/tmpfiles.d/set_IO_scheduler.conf file for this.
|
|
|
|
|
|
|
|
|
|
| |
console=ttyS0,115200 will work on Raspberry Pi 3.
Raspberry Pi 2 does not boot without console=tty1 on cmdline.
To have console output on both serial port and HDMI display,
8250.nr_uarts=1 is needed.
On Raspberry Pi 2, though, it is required to edit /boot/extlinux/extlinux.conf
to use ttyAMA0 instead of ttyS0 in the console parameter.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Read and write an extlinux.conf file in /boot/extlinux, it will be
automatically found by U-boot if located on an active partition.
Keep the "uboot" backend name, since we may do some U-Boot
installation steps later on.
Also, the extlinux.conf code could be used to support other
bootloaders from the syslinux family.
fixup extlinux
|
|
|
|
| |
Also save it as fdtdir
|
|
|
|
| |
It is needed for root devices with LABEL or UUID.
|
|
|
|
| |
This is the only missing piece for supporting 32-bit UEFI boot.
|
|
|
|
|
| |
This enables bootloader::read_grub2 to recognise the initrd16 keyword,
which is used by default on x86 machines.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The video mode needs to be extracted from and written back to the
GRUB_CMDLINE_LINUX_DEFAULT parameter in /etc/defaults/grub2.
This fix creates a new "default_vga" key in the $bootloader hash to
hold the current selection. Using the existing "vga" key would cause
the default video mode to be added to any bootloader entries that
don't have a specific "vga" value, e.g. the "failsafe" entry.
The $bootloader "vga" value doesn't appear to be set anywhere else,
so may in fact be obsolete.
|
|
|
|
|
| |
This reverts commit 7bb01fb038e0a882e755b963110bda381336b748 and
commit 4c1cb5006360ccb780926000bd57f150e0fc8fbe
|
|
|
|
|
| |
This allows changes to the append options to propagate to the grub2
configuration file, thus fixing mga#21263 and mga#21250.
|
| |
|
|
|
|
| |
needed anyway after last commit
|
|
|
|
|
|
|
|
|
|
|
| |
we were wrongly assuming we entered a new entry when reading eg:
"${xen_rm_opts}"
(which is why most people never see that error and why I wasn't able to
reproduce it, one would need Xen or similar to be installed)
Then drakboot would crash with:
drakboot crashed ("undefined value for mandatory argument 'text' encountered at ugtk3.pm"
Theorically, it could happen with DrakX too when updating a system with Xen.
|
|
|
|
| |
thus we got more data
|
|
|
|
|
| |
Fixes a corner case bug where GRUB_DEFAULT and GRUB_SAVEDEFAULT would be
overwritten if previously set to a value that evaluates as false.
|
| |
|
|
|
|
| |
Resolves: mga#20089
|
| |
|
|
|
|
|
| |
those are no more generated for more than 10 years...
also make kernel server default if installed
|
| |
|
|
|
|
|
|
|
|
| |
rationale: write_*() takes (bootloader, all_hd, ext) as parameters (when
called from installkernel -> bootloader-config and then only when
migrating to uuids
as we don't care for 2 first for grub2, just ignore the 3rd too
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Keep default append value (aka /etc/default/grub's
GRUB_CMDLINE_LINUX_DEFAULT into {default_append} variable)
Always use the value passed as now write_grub2_sysconfig() is only
called after having offered to display the default cmdline value
On first run (during installer) or when migrating from grub-legacy or
lilo, we get the first entry's parameters
Thus resolving mga#18952
|
| |
|
|
|
|
| |
tested under both UEFI & legacy BIOS
|
| |
|
| |
|
| |
|
|
|
|
| |
only when running drakboot (mga#18783)
|
|
|
|
| |
needed for next commit
|
|
|
|
| |
needed for next commit
|