Re: [git pull] vfs.git mount.part1

From: Al Viro
Date: Sat Jan 05 2019 - 15:46:00 EST


On Sat, Jan 05, 2019 at 01:31:21PM -0600, Eric W. Biederman wrote:

> Not having had a chance to review this code I can't really comment on
> the quality of this code. What I do know from a glance is that
> you have not removed FS_BINARY_MOUNTDATA. Which is the root cause
> of some of the crazy security mount option processing, and is an if
> not greater mess than what the security options have been doing with
> mount options.
>
> The FS_BINARY_MOUNTDATA flag is only relevant for coda and for nfs
> backwards compatiblity. The FS_BINARY_MOUNTDATA flag is only set on
> btrfs to allow calling mount_subtree.

... and thus it can't be killed without having dragged the NFS pile
into the entire thing.

> I have a set of patches that is finally reasonablly stable and cleans up
> all of the mess in the current internal mount apis that should allow
> implementing the new mount api to be much less error prone.

Quick question: how do you deal with the differences in quoting for selinux
options and for everything else?

I've no problem with working with you, now that you've resurfaced.
Fair warning: no promises of accepting your solutions. Along with
a promise to reject anything that breaks existing setups, which your
earlier proposals did. With NFS among the victims, IIRC.