On Tue, Sep 2, 2008 at 1:11 AM, Anton Vorontsov
<avorontsov@ru.mvista.com> wrote:
quoted text
> On Thu, Aug 28, 2008 at 05:43:33PM +0800, Li Yang wrote:
>> Some of Freescale SoC chips have a QE or CPM co-processor which
>> supports full speed USB.  The driver adds device mode support
>> of both QE and CPM USB controller to Linux USB gadget.  The
>> driver is tested with MPC8360 and MPC8272, and should work with
>> other models having QE/CPM given minor tweaks.
>>
>> Signed-off-by: Xie Xiaobo <X.Xie@freescale.com>
>> Signed-off-by: Li Yang <leoli@freescale.com>
>> ---
>> This is the second submission of the driver.  This version addressed:
>> Comments from Anton Vorontsov.
>> A lot of cosmetic problem.
>> Sparse and various kernel DEBUG warnings.
>
> Just caught this:
>
> g_ether gadget: high speed config #1: CDC Ethernet (ECM)
> BUG: sleeping function called from invalid context at mm/page_alloc.c:1450
> in_atomic():1, irqs_disabled():1
> Call Trace:
> [c03a9c30] [c0008a90] show_stack+0x4c/0x16c (unreliable)
> [c03a9c70] [c002cd28] __might_sleep+0xe4/0x114
> [c03a9c80] [c006e550] __alloc_pages_internal+0x328/0x42c
> [c03a9d00] [c006e67c] __get_free_pages+0x28/0x68
> [c03a9d10] [c0090efc] __kmalloc+0xc0/0xe4
> [c03a9d30] [c01d8efc] qe_ep_rxbd_update+0x98/0x1e0
> [c03a9d50] [c01d90f0] qe_ep_init+0xac/0x37c
> [c03a9d70] [c01d9458] qe_ep_enable+0x98/0xb8
> [c03a9d80] [c01db268] gether_connect+0xa0/0x178
> [c03a9da0] [c01dbba8] ecm_set_alt+0x108/0x12c
> [c03a9db0] [c01dc988] composite_setup+0x2ec/0x3a8
> [c03a9de0] [c01d8bec] setup_received_handle+0x1e4/0x26c
> [c03a9e00] [c01d8ce0] ep0_setup_handle+0x6c/0x74
> [c03a9e10] [c01d8e54] qe_ep0_rx+0x16c/0x17c
> [c03a9e30] [c01d9db8] rx_irq+0x88/0xc0
> [c03a9e40] [c01da32c] qe_udc_irq+0x10c/0x134
> [c03a9e60] [c00631f4] handle_IRQ_event+0x5c/0xb0
> [c03a9e80] [c006526c] handle_level_irq+0xa8/0x144
> [c03a9ea0] [c002a5d0] qe_ic_cascade_low_ipic+0x58/0x70
> [c03a9eb0] [c0006820] do_IRQ+0xa4/0xc8
> [c03a9ec0] [c0013e14] ret_from_except+0x0/0x14
> --- Exception: 501 at cpu_idle+0xa0/0x104
>    LR = cpu_idle+0xa0/0x104
> [c03a9f80] [c0009d88] cpu_idle+0x50/0x104 (unreliable)
> [c03a9fa0] [c029307c] _etext+0x7c/0x90
> [c03a9fc0] [c035491c] start_kernel+0x1ac/0x230
> [c03a9ff0] [00003438] 0x3438
>
> The workaround is obvious (below), but should we really allocate
> things in interrupts? Can we just allocate everything in probe()
> and free in remove()?

It's possible but wasteful.  We don't know which ep will be used and
what's the buffer size needed at probe time.

quoted text
>
>
> diff --git a/drivers/usb/gadget/fsl_qe_udc.c b/drivers/usb/gadget/fsl_qe_udc.c
> index c1a0beb..c54863b 100644
> --- a/drivers/usb/gadget/fsl_qe_udc.c
> +++ b/drivers/usb/gadget/fsl_qe_udc.c
> @@ -433,7 +433,7 @@ static int qe_ep_rxbd_update(struct qe_ep *ep)
>
>        bd = ep->rxbase;
>
> -       ep->rxframe = kmalloc(sizeof(*ep->rxframe), GFP_KERNEL);
> +       ep->rxframe = kmalloc(sizeof(*ep->rxframe), GFP_ATOMIC);
>        if (ep->rxframe == NULL) {
>                dev_err(ep->udc->dev, "malloc rxframe failed\n");
>                return -ENOMEM;
> @@ -447,7 +447,7 @@ static int qe_ep_rxbd_update(struct qe_ep *ep)
>                bdring_len = USB_BDRING_LEN;
>
>        size = (ep->ep.maxpacket + USB_CRC_SIZE + 2) * (bdring_len + 1);
> -       ep->rxbuffer = kzalloc(size, GFP_KERNEL);
> +       ep->rxbuffer = kzalloc(size, GFP_ATOMIC);
>        if (ep->rxbuffer == NULL) {
>                dev_err(ep->udc->dev, "malloc rxbuffer failed,size=%d\n",
>                                size);
> @@ -680,7 +680,7 @@ static int qe_ep_init(struct qe_udc *udc,
>        }
>
>        if ((ep->tm == USBP_TM_CTL) || (ep->dir == USB_DIR_IN)) {
> -               ep->txframe = kmalloc(sizeof(*ep->txframe), GFP_KERNEL);
> +               ep->txframe = kmalloc(sizeof(*ep->txframe), GFP_ATOMIC);
>                if (ep->txframe == NULL) {
>                        dev_err(udc->dev, "malloc txframe failed\n");
>                        goto en_done;
> @@ -1930,7 +1930,9 @@ static int reset_queues(struct qe_udc *udc)
>                udc_reset_ep_queue(udc, pipe);
>
>        /* report disconnect; the driver is already quiesced */
> +       spin_unlock(&udc->lock);
>        udc->driver->disconnect(&udc->gadget);
> +       spin_lock(&udc->lock);

Good catch.

quoted text
>
>        return 0;
>  }
> --
> To unsubscribe from this list: send the line "unsubscribe linux-usb" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/