On 12/04/17 14:45, Geert Uytterhoeven wrote:
> Hi Rob,
>
> On Mon, Dec 4, 2017 at 8:35 PM, Rob Herring <[email protected]> wrote:
>> On Mon, Dec 4, 2017 at 9:47 AM, Geert Uytterhoeven
>> <[email protected]> wrote:
>>> The special overlay mutex is taken first, hence it should be released
>>> last in the error path.
>>>
>>> Move "mutex_lock(&of_mutex)" up, as suggested by Frank, as
>>> free_overlay_changeset() should be called with that mutex held if any
>>> non-trivial cleanup is to be done.
>>
>> Not holding the of_mutex for of_resolve_phandles is just wrong.
>> Without it, a node and new phandle could be added via of_attach_node
>> making the max phandle wrong.
>
> After my patch it's held, so what's the problem?
>
>> Now, with the 2 mutexes adjacent, what is the point of even having the
>> of_overlay_mutex? Seems like we should just drop it.
>
> Frank?
__of_changeset_apply_notify(), which is called by __of_changeset_apply()
unlocks of_mutex, then does notifications then locks of_mutex. So the
mutex get released in the middle of of_overlay_apply()
I have never been comfortable with the unlock/lock there, but don't have
an alternative yet.
>> I also don't think we really need to hold the mutex during post-apply
>> notifiers. It also seems like some steps could be moved outside the
>> mutex(es) like init_overlay_changeset().
>
> Perhaps.
>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 --
> [email protected]
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like
> that.
> -- Linus Torvalds
>
