On 05/12/2022 14:13, gaochao wrote:
> From: Gao Chao <gaochao49@xxxxxxxxxx>
>
> commit a8782f669c35 ("V4L/DVB (9796): drivers/media/video/cx88/cx88-alsa.c:
> Adjust error-handling code") added fail path. However, when cx88_core_get()
> returns null with chip->pci has not been setted, snd_cx88_dev_free() won't
> call a valid pci_disable_device() due to invalid chip->pci. Move the
> assignment of chip->pci forward to ensure snd_cx88_dev_free() can get
> valid chip->pci.
>
> Fixes: b7f355d23c34 ("V4L/DVB (3293): Added digital support for cx88 (cx88-alsa)")
> Signed-off-by: Gao Chao <gaochao49@xxxxxxxxxx>
>
> v2:
> - move the assignment of chip->pci after chip = card->private_data
> ---
> drivers/media/pci/cx88/cx88-alsa.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> --
> 2.17.1
>
> diff --git a/drivers/media/pci/cx88/cx88-alsa.c b/drivers/media/pci/cx88/cx88-alsa.c
> index 29fb1311e443..22fe8c48a853 100644
> --- a/drivers/media/pci/cx88/cx88-alsa.c
> +++ b/drivers/media/pci/cx88/cx88-alsa.c
> @@ -861,6 +861,7 @@ static int snd_cx88_create(struct snd_card *card, struct pci_dev *pci,
> pci_set_master(pci);
>
> chip = card->private_data;
> + chip->pci = pci;
This isn't sufficient. If the pci_enable_device fails, then this function
also returns an error and chip->pci is still not set.
The same problem actually occurs with chip->core: this too is used in snd_cx88_dev_free,
and it is set much later.
Regards,
Hans
>
> core = cx88_core_get(pci);
> if (!core) {
> @@ -877,7 +878,6 @@ static int snd_cx88_create(struct snd_card *card, struct pci_dev *pci,
>
> /* pci init */
> chip->card = card;
> - chip->pci = pci;
> chip->irq = -1;
> spin_lock_init(&chip->reg_lock);
>
