Re: ehci: fix error handling?

[Maxime Villard <max%m00nbsd.net@localhost>]

Le 25/05/2019 à 08:53, Iain Hibbert a écrit :
> On Sat, 25 May 2019, Maxime Villard wrote:
>
> > To fix this problem, we need to fix the error handling in ehci_pci_attach(),
> > so that goto fail does not leave 'sc' in an inconsistent state.
> >
> > What is the best way to do that? Should I somehow forcibly unregister the
> > device, or should I add a flag in 'sc' to indicate whether it is initialized?
>
> the latter
>
> > It looks like there is no clean way to handle attach failures.
>
> attach always 'succeeds' in the sense that after attach has been called,
> detach will always be called. The detach routine should tear down
> everything that needs tearing down and not do things that will fail.
> Perhaps the init could simply be done before the attach routine gets the
> chance to fail?
>
> iain

Ok, thanks, I've fixed it and the machine shuts down correctly now.

However there seem to be other problems in the attach/detach. ehci_init()
and ehci_detach() are not symmetrical, the former initializes more stuff
than the latter deinitializes. ehci_pci.c deinitializes the extra stuff
correctly, but the others (cardbus, arm, mips) don't.