On Fri, Jan 15, 2021 at 07:43:34AM +1100, Dave Chinner wrote:

quoted

That sounds neat.  AFAICT, the VFS passes the filesystem a mount userns
structure, which is then carried down the call stack to whatever
functions actually care about mapping kernel [ug]ids to their ondisk
versions?

Does quota still work after this patchset is applied?  There isn't any
mention of that in the cover letter and I don't see a code patch, so
does that mean everything just works?  I'm particularly curious about
whether there can exist processes with CAP_SYS_ADMIN and an idmapped
mount?  Syscalls like bulkstat and quotactl present file [ug]ids to
programs, but afaict there won't be any translating going on?

bulkstat is not allowed inside user namespaces. It's an init
namespace only thing because it provides unchecked/unbounded access
to all inodes in the filesystem, not just those contained within a
specific mount container.

Hence I don't think bulkstat output (and other initns+root only
filesystem introspection APIs) should be subject to or concerned
about idmapping.

That is what the capabilities are designed for and we already check
for them.