Hi,
Here's an updated version of the patch. It tests kmalloc return
and comments that the ioctls must obey the size macro convention
(basically, don't depend on more than 256 bytes :-). Plus it's
got a couple minor error path tweaks, uses C comments, and calls
the interface lookup routine Randy pointed out. Basically, it's
addressing all the comments I got!
Against ac14, one patch is offset by 10 lines (hub.c updates).
- Dave
> Here's the stuff I promised: an API letting a userspace client
> issue ioctls to kernel space drivers. That means you can send
> and receive ioctl-sized (max 16K) chunks of data to drivers for
> any purpose you need, through usbdevfs. Define your own ioctls.
>
> The ioctl provided (see sample "ioctl.c") talks to the hub driver
> and finds out what children the hub has, by USB address. It's
> behaved quite nicely and correctly for me so far. This was one
> of the key remaining bits of usb/devices text that was unavailable
> without writing text parsing code.
>
> This sort of framework should work nicely for the case where a
> user mode driver (maybe some sort of manager) needs to interact
> with a kernel mode driver to control something. Also, such a
> kernel driver might help "safely" issue control messages.
>
> It ought to work for any USB driver that may not have another
> presence on the file system ... hubs, network interfaces, and
> so forth. Against 2.4.0-test1-ac10 (maybe ac8+).
ifioctl3.patch
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
