On Fri, Oct 18, 2024 at 10:37:54AM -0700, Carl E. Thompson wrote: > > > On 2024-10-18 1:17 AM PDT Christopher Snowhill <[email protected]> wrote: > > > ... > > > FYI, 6.9 was first released in May, and is already EOL now. In fact, so > > is 6.10, at 6.10.14. > > I'm on 6.9.4 which was released in June. It's not really by choice; there are > some nasty stability issues in the amdgpu driver in 6.10 and 6.11 which cause > my laptop to hang multiple times per day if I use a graphical environment > under them. > > > There are no LTS kernels with bcachefs support yet > > at this point. All further development is going into 6.12 now, with bug > > fixes being backported into 6.11. > > True. I understand that and accept it. I am not at all suggesting that Kent > should make fixes for unsupported kernels. I'm just reporting a bug. But I > also think that this sort of issue is one that really, **REALLY** should not > happen in the first place. A filesystem silently modifying itself so that it > no longer works on earlier kernels without warning is a very bad thing in my > opinion. There should be processes in place to catch that sort of problem > **before** a new kernel is released. And remember this isn't a one-time > thing. It's happened before. > > I'd also say that a filesystem design that requires that an older > driver be able to successfully automatically discover and undo random > on-disk modifications automatically made by a newer driver is probably > a bad design (my personal opinion). It's sounds great until it doesn't > work (as in this case). And I mean that as constructive criticism and > not something that Kent should take personally. I think that > particular design decision should be reevaluated.
It sounds like you should go back to ZFS then. > It's just dumb luck that in this case the older kernel is no longer > supported so it doesn't need to be fixed. > > As for my particular case, I should probably see if I can get the > older, stable amdgpu driver to compile under the current kernel. > Barring that, I should see if I can get the current bcachefs driver to > compile under the older kernel. I guess I'll find time to do that this > evening. I have been hearing a lot of people complain about regressions in amdgpu. Have you taken that up with them? The capability of doing seamless, automatic upgrades and downgrades, to the extent that bcachefs can, is something new that other filesystems don't have. And it's been incredibly useful. Without it, rolling out the disk accounting rewrite wouldn't have been possible - and that got us per snapshot ID accounting, compression type accounting, per-btree accounting, and per-inode fragmentation accounting (not all of these are exposed to the user yet, but they're there). While bcachefs is still marked experimental, doing these upgrades automatically makes sense because it gets us better test coverage of codepaths that will need to be rock solid later, and it drastically reduces the amount of compat code I have to carry around. I'm hoping to get several more on disk format features done while we're still in the experimental phase - for inode allocation improvements, fsck performance improvements, and assorted other odds and ends.
