Re: [git pull] new mount API

[Al Viro <viro@ZenIV.linux.org.uk>]

On Thu, Aug 23, 2018 at 07:36:15PM -0700, Andy Lutomirski wrote:

> > For fuck sake, mount(2) is a permanently supported option!  
> 
> Exactly. mount(2) has broken semantics and it’s permanently supported.
> 
> If this merge request gets pulled, then FSCONFIG_CMD_CREATE will *also* be a permanently supported API with broken semantics.

Oh no - mount(2) behaviour can be expressed that way!  The horror...

Andy, this is bullshit.  You are saying that dealing with mount(2)
mess of ABI (badly unorthogonal set of operations, overloading from
hell, etc.) must be tied with massive rework of fs drivers.  Why?

It's a simple enough question and pardon me, but your "it's broken"
doesn't inspire any confidence that you even understand what the
current behaviour is and what its problems (they are real enough)
are.  I have seen a lot of handwaving from you in these threads,
combined with "surely, it must work thus", followed with "it's
a special case/class/whatnot" or "surely, it must be broken" every
time you find something that doesn't "work thus".

Bugger it; explain why we must combine untangling the existing ABI
(and we *will* have to keep the existing semantics possible to
express for sys_mount() sake) with this, or with anything else,
for that matter.