Garrett D'Amore wrote
> I believe that Solaris uses keyboards (and maybe mice too) in what is
> called "BOOT Protocol" mode.
That's true for the USB keyboards.
An USB mouse is used in report mode, though (I think
otherwise we couldn't have USB mouse wheel support).
> This is a simplistic report format that
> was originally intended to allow simple translation to BIOS-compatible
> keyboard support. (I.e. so you don't have to have a full HID parser in
> your BIOS or other embedded environments if you only want basic
> keyboard/mouse support.)
Yes.
> Apple doesn't have BIOS, and I *suspect* doesn't support the BOOT
> protocol in their latest products.
That can't be true, because Solaris' GRUB needs a BIOS,
and GRUB is working just fine on the MacBook Pro,
including the internal USB keyboard.
Old Apple iMacs didn't have a BIOS, only EFI firmware. But
at some point a firmware update named "BootCamp" was offered
by Apple that added the missing BIOS support (to allow installing
Windows - or Solaris / Linux / ....).
> Having a full HID parser would probably solve the problem.
Hmm, usbkbm is already using the HID parser.
http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/uts/common/io/usb/clients/usbkbm/usbkbm.c#548
And I suspect that's the problem! Since it's using
the keyboard in boot protocol mode, it shouldn't
try to find out the keyboard's packet size from
looking at / parsing the HID descriptor. I suspect
that would only be a valid strategy if the boot
protocol is disabled.
I suspect that when an usb keyboard is used in boot
protocol mode, it should just use an expected keyboard
data packet size of USB_KBD_DEFAULT_PACKET_SIZE (=8)
bytes (without looking at the HID descriptors)!
> That might be difficult to get into something the likes of GRUB....
No, keyboard in GRUB works just fine.
> A question: do the keyboard/mouse work in X11 after the system has
> booted (assuming you used an external keyboard to get that far)?
Yes:
The external USB keyboard works in X11. The internal USB trackpad
works in X11. The internal USB keyboard doesn't work in X11, either.
dtracing in usbkbm_rput() reveals that the MacBook Pro USB
keyboard is sending standard 8 bytes boot protocol packets,
but they are dropped on the floor because the usbkbm module
thinks that the correct packets size is 0 (zero!) bytes - apparently
that size of 0 bytes doesn't make any sense at all.
All I have to do is patch the usbkbmd->usbkbm_packet_size
variable from 0 to 8, and the internal usb MacBook Pro keyboard
starts to work.
So, I think for the MacBook USB keyboard we either need:
replace this block:
548 if
(hidparser_get_packet_size(usbkbmd->usbkbm_report_descr,
549 0, HIDPARSER_ITEM_INPUT, (uint32_t *)&packet_size)
==
550 HIDPARSER_FAILURE) {
551
552 USB_DPRINTF_L3(PRINT_MASK_OPEN,
553 usbkbm_log_handle, "get_packet_size failed"
554 "setting default packet size(8)");
555
556 /* Setting to default packet size = 8 */
557 usbkbmd->usbkbm_packet_size =
558 USB_KBD_DEFAULT_PACKET_SIZE;
559 } else {
560 usbkbmd->usbkbm_packet_size = packet_size/8;
561 }
with
usbkbmd->usbkbm_packet_size = USB_KBD_DEFAULT_PACKET_SIZE;
Or we could at least sanity check the packet size value returned by parsing the
usb keyboard's HID descriptors after line 560. usbkbmd->usbkbm_packet_size == 0
doesn't make sense and should be replaced by USB_KBD_DEFAULT_PACKET_SIZE.
--
This message posted from opensolaris.org
