Re: [patch] pipe: Don't oops when pipe filesystem isn't mounted

[Andrew Morton <akpm@osdl.org>]

On Mon, 11 Dec 2006 10:22:07 +0000
Al Viro <viro@ftp.linux.org.uk> wrote:

> On Mon, Dec 11, 2006 at 02:17:18AM -0800, Andrew Morton wrote:
> > I think we should aim to have as many subsystems ready to go as possible -
> > ideally all of them.  Right now we can potentially run userspace before
> > AIO, posix-timers, message-queues, BIO, networking, etc are ready to run.
> > 
> > It looks to be pretty easy to fix...
> > 
> > > As for that example, I'd love to see specifics - which driver triggers
> > > hotplug?  Presumably it happens from an initcall, so we also have something
> > > fishy here...
> > 
> > I don't know in this case - but firmware loading from a statically-linked
> > driver is a legit thing to do.
> 
> Umm... statically linked driver that might want firmware shouldn't precede
> the subsystems unless something is seriously wrong with priorities...

There are plenty of drivers in there using subsys_initcall, arch_initcall,
postcore_initcall, core_initcall and even one pure_initcall.

Heaven knows why.  They're drivers :(

> IOW, I still wonder what's really going on - pipes are fs_initcall() and
> any hardware stuff ought to be simple module_init().  So something fishy
> is going on, regardless of anything else.

A heck of a lot of things can trigger an /sbin/hotplug run.  It could well
be that Andrew's driver didn't want to run hotplug at all, but the kernel
did it anwyay.  But as soon as the script appeared at /sbin/hotplug, and it
happened to use foo|bar: boom.