Hi,
On Sun, Oct 23, 2011 at 07:55:33PM -0700, Kuninori Morimoto wrote:
> Since renesas_usbhs mod_host didn't use
> struct completion as static object, the warning of lockdep came out.
> This patch fixup this issue.
>
> Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@xxxxxxxxxxx>
> ---
> drivers/usb/renesas_usbhs/mod_host.c | 22 +++++-----------------
> 1 files changed, 5 insertions(+), 17 deletions(-)
>
> diff --git a/drivers/usb/renesas_usbhs/mod_host.c b/drivers/usb/renesas_usbhs/mod_host.c
> index 1a7208a..0d42015 100644
> --- a/drivers/usb/renesas_usbhs/mod_host.c
> +++ b/drivers/usb/renesas_usbhs/mod_host.c
> @@ -103,8 +103,6 @@ struct usbhsh_hpriv {
>
> u32 port_stat; /* USB_PORT_STAT_xxx */
>
> - struct completion *done;
> -
> /* see usbhsh_req_alloc/free */
> struct list_head ureq_link_active;
> struct list_head ureq_link_free;
> @@ -522,6 +520,7 @@ static int usbhsh_queue_push(struct usb_hcd *hcd,
> /*
> * DCP setup stage
> */
> +static DECLARE_COMPLETION(setup_ack_done);
are you sure this is wise to do ? What happens if you ever end up with
two instances of this core on one SoC ?
I have been fighting a lot against static globals because that's a real
issue. We never thought there would ever be a device with multiple ports
but OTG3.0 has dropped the requirement of one port only and we have
devices (at least one from TI) with two device-side ports (well, those
ports are OTG-capable, in fact).
--
balbi
Attachment:
signature.asc
Description: Digital signature
