Re: [PATCH 1/2] Bluetooth: fix debugfs registration

[Johan Hovold <johan@kernel.org>]

On Wed, May 03, 2023 at 10:34:06AM -0700, Luiz Augusto von Dentz wrote:
> Hi Johan,
> 
> On Wed, May 3, 2023 at 12:04 AM Johan Hovold <johan@kernel.org> wrote:
> >
> > On Tue, May 02, 2023 at 04:37:51PM -0700, Luiz Augusto von Dentz wrote:
> > > Hi Johan,
> > >
> > > On Mon, Apr 24, 2023 at 5:50 AM Johan Hovold <johan+linaro@kernel.org> wrote:
> > > >
> > > > Since commit ec6cef9cd98d ("Bluetooth: Fix SMP channel registration for
> > > > unconfigured controllers") the debugfs interface for unconfigured
> > > > controllers will be created when the controller is configured.
> > > >
> > > > There is however currently nothing preventing a controller from being
> > > > configured multiple time (e.g. setting the device address using btmgmt)
> > > > which results in failed attempts to register the already registered
> > > > debugfs entries:
> > > >
> > > >         debugfs: File 'features' in directory 'hci0' already present!
> > > >         debugfs: File 'manufacturer' in directory 'hci0' already present!
> > > >         debugfs: File 'hci_version' in directory 'hci0' already present!
> > > >         ...
> > > >         debugfs: File 'quirk_simultaneous_discovery' in directory 'hci0' already present!
> > > >
> > > > Add a controller flag to avoid trying to register the debugfs interface
> > > > more than once.
> > > >
> > > > Fixes: ec6cef9cd98d ("Bluetooth: Fix SMP channel registration for unconfigured controllers")
> > > > Cc: stable@vger.kernel.org      # 4.0
> > > > Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
> > > > ---
> >
> > > > diff --git a/net/bluetooth/hci_sync.c b/net/bluetooth/hci_sync.c
> > > > index 632be1267288..a8785126df75 100644
> > > > --- a/net/bluetooth/hci_sync.c
> > > > +++ b/net/bluetooth/hci_sync.c
> > > > @@ -4501,6 +4501,9 @@ static int hci_init_sync(struct hci_dev *hdev)
> > > >             !hci_dev_test_flag(hdev, HCI_CONFIG))
> > > >                 return 0;
> > > >
> > > > +       if (hci_dev_test_and_set_flag(hdev, HCI_DEBUGFS_CREATED))
> > > > +               return 0;
> > >
> > > Can't we just use HCI_SETUP like we do with in create_basic:
> > >
> > >     if (hci_dev_test_flag(hdev, HCI_SETUP))
> > >         hci_debugfs_create_basic(hdev);
> > >
> > > Actually we might as well move these checks directly inside the
> > > hci_debugfs function to make sure these only take effect during the
> > > setup/first init.
> >
> > The problem is that commit ec6cef9cd98d ("Bluetooth: Fix SMP channel
> > registration for unconfigured controllers") started deferring creation
> > of most parts of the debugfs interface until the controller is
> > configured (e.g. as some information is not available until then).
> >
> > Moving everything back to setup-time would effectively revert that.
> 
> Not moving back but just doing something like:
> 
> diff --git a/net/bluetooth/hci_debugfs.c b/net/bluetooth/hci_debugfs.c
> index ec0df2f9188e..a6e94c29fc5a 100644
> --- a/net/bluetooth/hci_debugfs.c
> +++ b/net/bluetooth/hci_debugfs.c
> @@ -310,6 +310,9 @@ DEFINE_INFO_ATTRIBUTE(firmware_info, fw_info);
> 
>  void hci_debugfs_create_common(struct hci_dev *hdev)
>  {
> +       if (!hci_dev_test_flag(hdev, HCI_SETUP))
> +               return;
> +
>         debugfs_create_file("features", 0444, hdev->debugfs, hdev,
>                             &features_fops);
>         debugfs_create_u16("manufacturer", 0444, hdev->debugfs,
> 

What I tried to explain above is that doing this would always create
the attributes as setup-time rather than at config-time, which
effectively reverts commit ec6cef9cd98d ("Bluetooth: Fix SMP channel
registration for unconfigured controllers"). And doing so looks like it
would amount to a regression.

Johan