Aaron Young wrote:
>> On Fri, 30 Nov 2007 08:43:07 -0800 (PST)
>> Aaron Young <ayoung@xxxxxxxxxxxxxxxxxxxx> wrote:
>>
>>>>
>>>>
>>>> Looking better! More comments:
>>>>
>>>> 1. I don't understand the need for the call to
>>>> kdb_uhci_keyboard_clear(). It's only called if
>>>> kdb_usb_keyboard_attach() fails and in that case we
>>>> didn't add the urb to kdb_usb_kbds[]. So, I don't
>>>> see the need to clear it out of kdb_usb_kbds[]...
>>>> (I'm probably missing something here).
>> - Now I call kdb_uhci_keyboard_clear() if usb_submit_urb() failed, that is
>> called after kdb_usb_keyboard_attach().
>
> Why not just call kdb_usb_keyboard_detach()?
>
>>>> 2. I probably would have made kdb_uhci_submit_urb() return
>>>> the newly created kdb_urb (extra arg) on success and not call
>>>> kdb_usb_keyboard_attach() directly from within. Then you can
>>>> just make a single call to kdb_usb_keyboard_attach() for all three
>>>> cases (OHCI, EHCI and UHCI) out of hid_probe(). But, not
>>>> a *BIG* deal...
>> - We should first fill kdb_usb_kbds[] before call of usb_submit_urb()
>> because uhci_urb_enqueue() checks kdb_usb_kbds[] to find out if KDB URB is
>> enqueued. So kdb_usb_keyboard_attach() should be called in the "middle" of
>> the function kdb_uhci_submit_urb().
>
> OK..
>
>>>> 3. We'll have to get updates to ia64 kdba_io.c as well, otherwise
>>>> it will result in compile errors on ia64. i.e.
>>>> kdb_usb_keyboard_attach() will have a number of args mismatch and
>>>> all the routines you added to kdba_io_32.c and kdba_io_64.c will
>>>> not be there.
>> - Can't find any ia64 file...
>
> linux/arch/ia64/kdb/kdba_io.c
>
> Maybe it's part of a different KDB patch or something? Jay?
Hi Konstantin,
It is in the ia64 kdb patch, e.g. kdb-v4.4-2.6.24-rc3-ia64-1.
You can get the one corresponding to the version of your x86 and
common patches.
Thanks,
- jay
>
>>>> Does hotplug/hotunplug of the keyboards work?
>>>> Does it correctly remove the KDB URBs on hotunplug?
>> - kdb_usb_keyboard_detach() is never called on my PC. So I tested only 1
>> case: attaching keyboard after kernel boot - works OK.
>
> Should get called out of hid_disconnect() when a
> keyboard is unplugged. If you are using a usb1.1
> hub, it may not work - I've seen this problem with
> USB1.1 hub. MIght want to try plugging/unplugging the keyboard
> directly into the chassis port if so.
>
> BTW - do you have a USB2.0 hub? The EHCI KDB code has yet to
> be tested on x86 and I'd be curious if it worked or not (hope it does)...
>
> -Aaron
> ---------------------------
> Use http://oss.sgi.com/ecartis to modify your settings or to unsubscribe.
---------------------------
Use http://oss.sgi.com/ecartis to modify your settings or to unsubscribe.
|