Re: USB2: ugd round bug? including patch to solve.

On Thursday 22 January 2009, Daichi GOTO wrote:
> Hi USB2 folks ;-)
>
> First I should show you my big respect. Thank you very much for your
> great usb2 work. That works very well.
>
> Using usb2, I have found a issue around uhid.  ugd will be zero-cleared
> when ioctl(fd, USB_GET_REPORT_DESC, &ugd) is called.
>
> For example, follow ioctl are used to get information of report descriptor.
>
>      memset(&ugd, 0, sizeof(ugd));
>      ugd.ugd_data = malloc(len);
>      ugd.ugd_maxlen = len;
>
>      ioctl(fd, USB_GET_REPORT_DESC, &ugd)
>
> The information of report descriptor should be copied to ugd, but ugd is
> zero-cleared.
>
>  From my research, perphaps follow patch is good solution to solve this
> problem. How do you usb2 guys make of this?  Please check and solve
> this issue. Thanks :)

Hi Daichi!

Your patch is absolutely correct! Committed to P4. Will reach -current early 
next week!

http://perforce.freebsd.org/chv.cgi?CH=156515

Thanks for testing and finding bugs!

--HPS

> --- sys/dev/usb2/include/usb2_ioctl.h.orig      2009-01-22
> 20:11:13.000000000 +0900 +++ sys/dev/usb2/include/usb2_ioctl.h   2009-01-22
> 20:09:35.000000000 +0900 _at__at_ -223,7 +223,7 _at__at_
>   #define        USB_DEVICEENUMERATE     _IOW ('U', 6, int)
>
>   /* Generic HID device */
> -#define        USB_GET_REPORT_DESC     _IOR ('U', 21, struct
> usb2_gen_descriptor) +#define        USB_GET_REPORT_DESC     _IOWR('U', 21,
> struct usb2_gen_descriptor) #define        USB_SET_IMMED           _IOW
> ('U', 22, int)
>   #define        USB_GET_REPORT          _IOWR('U', 23, struct
> usb2_gen_descriptor) #define        USB_SET_REPORT          _IOW ('U', 24,
> struct usb2_gen_descriptor)