Received-SPF: pass (google.com: best guess record for domain of linux-bluetooth-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Fri, 13 Dec 2019 10:18:09 +0100
From:   Johan Hovold <johan@kernel.org>
To:     Marcel Holtmann <marcel@holtmann.org>
Cc:     Nicolas Saenz Julienne <nsaenzjulienne@suse.de>,
        Guillaume La Roque <glaroque@baylibre.com>,
        Johan Hedberg <johan.hedberg@gmail.com>,
        BlueZ <linux-bluetooth@vger.kernel.org>,
        lkml <linux-kernel@vger.kernel.org>, khilman@baylibre.com,
        linux-rpi-kernel <linux-rpi-kernel@lists.infradead.org>
Subject: Re: [PATCH v3] bluetooth: hci_bcm: enable IRQ capability from node
Message-ID: <20191213091809.GW10631@localhost>
References: <20191211094923.20220-1-glaroque@baylibre.com>
 <cf77eec5df92b1845f0bf7cc8eb53edd4af9e1bf.camel@suse.de>
 <0CF02341-CF69-4680-B61F-DC5C0702F1A2@holtmann.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <0CF02341-CF69-4680-B61F-DC5C0702F1A2@holtmann.org>
Sender: linux-bluetooth-owner@vger.kernel.org
Precedence: bulk

On Thu, Dec 12, 2019 at 09:46:32PM +0100, Marcel Holtmann wrote:
> Hi Nicolas,
> 
> >> Actually IRQ can be found from GPIO but all platforms don't support
> >> gpiod_to_irq, it's the case on amlogic chip.
> >> so to have possibility to use interrupt mode we need to add interrupts
> >> field in node and support it in driver.
> >> 
> >> Signed-off-by: Guillaume La Roque <glaroque@baylibre.com>
> >> ---
> >> drivers/bluetooth/hci_bcm.c | 3 +++
> >> 1 file changed, 3 insertions(+)
> > 
> > This triggers the following panic on Raspberry Pi 4:
> > 
> > [    6.634507] Unable to handle kernel NULL pointer dereference at virtual
> > address 0000000000000018

> >> diff --git a/drivers/bluetooth/hci_bcm.c b/drivers/bluetooth/hci_bcm.c
> >> index f8f5c593a05c..9f52d57c56de 100644
> >> --- a/drivers/bluetooth/hci_bcm.c
> >> +++ b/drivers/bluetooth/hci_bcm.c
> >> @@ -1409,6 +1409,7 @@ static int bcm_serdev_probe(struct serdev_device
> >> *serdev)
> >> {
> >> 	struct bcm_device *bcmdev;
> >> 	const struct bcm_device_data *data;
> >> +	struct platform_device *pdev;
> >> 	int err;
> >> 
> >> 	bcmdev = devm_kzalloc(&serdev->dev, sizeof(*bcmdev), GFP_KERNEL);
> >> @@ -1421,6 +1422,8 @@ static int bcm_serdev_probe(struct serdev_device
> >> *serdev)
> >> #endif
> >> 	bcmdev->serdev_hu.serdev = serdev;
> >> 	serdev_device_set_drvdata(serdev, bcmdev);
> >> +	pdev = to_platform_device(bcmdev->dev);
> > 
> > Ultimately bcmdev->dev here comes from a serdev device not a platform device,
> > right?
> 
> I was afraid of this, but then nobody spoke up. Can we fix this or
> should I just revert the patch?

Just revert it, the patch is plain broken and makes no sense.

As Nicolas pointed out, bcmdev->dev is a member of struct serdev_device
so cannot be cast to a platform device.

Johan