On Mon Jan 13, 2025 at 9:36 AM CET, Pascal Hambourg wrote: > On 13/01/2025 at 07:48, Holger Wansing wrote: > > Am 12. Januar 2025 23:44:27 MEZ schrieb Pascal Hambourg > > <pas...@plouf.fr.eu.org>: > >> Indeed we must make a trade-off > >> between robustness (ext4 vs ext2), convenience (/boot in root filesystem > >> vs small separate /boot) and compatibility with boot loaders. And I > >> believe we can do it right at least for use cases we know about. You > >> addressed the case of guided partitioning without LVM on ppc64el, why > >> not do the same for guided partitioning without LVM ? Do you think that > >> the robustness or convenience gain of not having a separate ext2 /boot > >> partition is worth the loss of compatibility with Petitboot ? > > > > I cannot judge on this, because it depends on, how much Petitboot > > is used on that arch. > > If it's only some few percent, you would force much other people to > > a ext2 /boot without a reason. > > I agree. Petitboot is mentioned on IBM's website [1] so I suspect it is > rather standard on IBM POWER machines. > > [1] > <https://www.ibm.com/docs/en/linux-on-systems?topic=systems-petitboot-bootloader> > > > And the same for arm64, if we roll back to ext2 there too. > > AFAIK the only implementation of Petitboot on arm64 (Odroid) supports > ext4 so there is no need to rollback to ext2. I did not mean to roll > back to ext2 but to create a separate /boot partition only with LVM, as > I do not see the benefit of a separate ext4 /boot partition without LVM. > > > I had hoped that the OP would give some statistics about this... > > So do I...
FWIW, I hadn't heard about Petitboot before this bug either, so I looked on their Github repo and found "and ARM64 with ACPI" ... and I think the "with ACPI" is significant. So going back to ext2 on arm64 just for Petitboot users seems a bit heavy handed to me (but I may not fully understand it all). My 0.02
signature.asc
Description: PGP signature
