On Sun, Dec 31, 2023 at 5:14 PM Qu Wenruo <wqu@xxxxxxxx> wrote: > > > > On 2024/1/1 00:42, Neal Gompa wrote: > > On Sat, Dec 30, 2023 at 6:37 PM Qu Wenruo <wqu@xxxxxxxx> wrote: > >> > >> Test case btrfs/131 is a quick tests for v1/v2 free space related > >> behavior, including the mount time conversion and disabling of v2 space > >> cache. > >> > >> However there are two problems, mostly related to the v2 cache clearing. > >> > >> - There are some features with hard dependency on v2 free space cache > >> Including: > >> * block-group-tree > >> * extent-tree-v2 > >> * subpage support > >> > >> Note those features may even not support clearing v2 cache. > >> > >> - The v1 free space cache is going to be deprecated > >> Since v5.15 the default mkfs is already going v2 cache instead. > >> It won't be long before we mark v1 cache deprecated and force to > >> go v2 cache. > >> > >> This makes the test case to fail unnecessarily, the false failure would > >> only grow with new features relying on v2 cache. > >> > >> So here let's removing the test case completely. > >> > > > > Can we pair this change with a corresponding change in btrfs-progs > > that blocks using v1? I don't think it's actually worth splitting this > > change up in phases, especially when we're explicitly dropping the > > tests around it. > > That sounds pretty reasonable. > > I'll craft one to deprecate v1 cache in progs too. > As part of the deprecation, it should probably be difficult to use v1 cache and if we're not already automatically migrating people to v2, we should probably start doing that. -- 真実はいつも一つ！/ Always, there's only one truth!
