summaryrefslogtreecommitdiffstats
path: root/fs/mnt_idmapping.c
Commit message (Collapse)AuthorAgeFilesLines
* fs/mnt_idmapping.c: Return -EINVAL when no map is writtenTaylor Jackson2024-02-081-1/+1
| | | | | | | | | | | | | | Currently, it is possible to create an idmapped mount using a user namespace without any mappings. However, this yields an idmapped mount that doesn't actually map the ids. With the following change, it will no longer be possible to create an idmapped mount when using a user namespace with no mappings, and will instead return EINVAL, an “invalid argument” error code. Reviewed-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Taylor Jackson <taylor.a.jackson@me.com> Link: https://lore.kernel.org/r/20240208-mnt-idmap-inval-v2-1-58ef26d194e0@me.com Signed-off-by: Christian Brauner <brauner@kernel.org>
* mnt_idmapping: decouple from namespacesChristian Brauner2023-11-281-15/+91
| | | | | | | | | | | | | | | There's no reason we need to couple mnt idmapping to namespaces in the way we currently do. Copy the idmapping when an idmapped mount is created and don't take any reference on the namespace at all. We also can't easily refcount struct uid_gid_map because it needs to stay the size of a cacheline otherwise we risk performance regressions (Ignoring for a second that right now struct uid_gid_map isn't actually 64 byte but 72 but that's a fix for another patch series.). Link: https://lore.kernel.org/r/20231122-vfs-mnt_idmap-v1-3-dae4abdde5bd@kernel.org Reviewed-by: Josef Bacik <josef@toxicpanda.com> Signed-off-by: Christian Brauner <brauner@kernel.org>
* mnt_idmapping: remove nop checkChristian Brauner2023-11-281-28/+8
| | | | | | | | | | | All mounts default to nop_mnt_idmap and we don't allow creating idmapped mounts that reuse the idmapping of the filesystem. So unless someone passes a non-superblock namespace to these helpers this check will always be false. Remove it and replace it with a simple check for nop_mnt_idmap. Link: https://lore.kernel.org/r/20231122-vfs-mnt_idmap-v1-2-dae4abdde5bd@kernel.org Signed-off-by: Christian Brauner <brauner@kernel.org>
* mnt_idmapping: remove check_fsmapping()Christian Brauner2023-11-281-15/+2
| | | | | | | The helper is a bit pointless. Just open-code the check. Link: https://lore.kernel.org/r/20231122-vfs-mnt_idmap-v1-1-dae4abdde5bd@kernel.org Signed-off-by: Christian Brauner <brauner@kernel.org>
* fs: export mnt_idmap_get/mnt_idmap_putAlexander Mikhalitsyn2023-11-031-0/+2
| | | | | | | | | These helpers are required to support idmapped mounts in CephFS. Cc: linux-fsdevel@vger.kernel.org Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com> Reviewed-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
* fs: move mnt_idmapChristian Brauner2023-01-191-0/+273
Now that we converted everything to just rely on struct mnt_idmap move it all into a separate file. This ensure that no code can poke around in struct mnt_idmap without any dedicated helpers and makes it easier to extend it in the future. Filesystems will now not be able to conflate mount and filesystem idmappings as they are two distinct types and require distinct helpers that cannot be used interchangeably. We are now also able to extend struct mnt_idmap as we see fit. Acked-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>