On Sat, 24 May 2014, Pratyush Anand wrote:
> OTG3 and EH Compliance Plan 1.0 talks about Super Speed OTG Verification
> system (SS-OVS) which consists of an excersizer and analyzer.
>
> USB Compliance Suite from Lecroy or Ellisys can act as such SS-OVS for
> Link Layer Validation (LVS).
>
> Some modifications are needed for an embedded Linux USB host to pass all
> these tests. Most of these tests require just Link to be in U0. They do
> not work with default Linux USB stack since, default stack does port
> reset and then starts sending setup packet, which is not expected by
> Link Layer Validation (LVS) device of Lecroy Compliance Suit. Then,
> There are many Link Layer Tests which need host to generate specific
> traffic.
>
> This patch supports specific traffic generation cases. As of now all the
> host Lecroy Link Layer-USBIF tests (except TD7.26) passes
> with this patch for single run using Lecroy USB Compliance Suite
> Version 1.98 Build 239 and Lecroy USB Protocol Analyzer version 4.80
> Build 1603. Therefore patch seems to be a good candidate for inclusion.
> Further modification can be done on top of it.
Acked-by: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
There are a couple of things you should be aware of...
> +static void lvs_rh_work(struct work_struct *work)
> +{
> + struct lvs_rh *lvs = container_of(work, struct lvs_rh, rh_work);
> + struct usb_interface *intf = lvs->intf;
> + struct usb_device *hdev = interface_to_usbdev(intf);
> + struct usb_hcd *hcd = bus_to_hcd(hdev->bus);
> + struct usb_hub_descriptor *descriptor = &lvs->descriptor;
> + struct usb_port_status *port_status;
> + int i, ret = 0;
> +
> + port_status = kmalloc(sizeof(*port_status), GFP_KERNEL);
> + if (!port_status) {
> + dev_err(&intf->dev, "failed to allocate port_status memory\n");
> + return;
> + }
If you allocated the port_status buffer just once and stored a pointer
to it in the lvs_rh structure then you wouldn't need to allocate a
buffer each time this routine runs.
> + /* Examine each root port */
> + for (i = 1; i <= descriptor->bNbrPorts; i++) {
> + ret = usb_control_msg(hdev, usb_rcvctrlpipe(hdev, 0),
> + USB_REQ_GET_STATUS, USB_DIR_IN | USB_RT_PORT, 0, i,
> + port_status, sizeof(*port_status), 1000);
> + if (ret < 4)
> + continue;
> + /* handle only connection change notification */
> + if (!(port_status->wPortChange & USB_PORT_STAT_C_CONNECTION))
> + continue;
There will be other bits set in the wPortChange word from time to time,
such as USB_PORT_STAT_C_RESET and maybe USB_PORT_STAT_C_ENABLE. By not
ever clearing them, you will cause this routine to be polled 4 times
per second instead of only when there's a connect change. This isn't
terribly important but you should realize what you are doing.
Alan Stern
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
