On Wed, Dec 10, 2014 at 07:47:05PM +0100, Werner Zeh wrote:
> Hi Kevin.
>
> Yes, USB2.0 hub on the same port woks just fine!
>
> Werner
>
> Am 10.12.2014 um 18:02 schrieb Kevin O'Connor:
> >Thanks. I'm forwarding to the seabios mailing list.
> >
> >Do you know if USB2 hubs work? (IE, is this issue isolated to USB3
> >hubs?)
> >
> >-Kevin
Thanks. Can your forward the output of both "lsusb -t" and "lsusb -v"
after booting with the hub present?
BTW, please CC the seabios mailing list.
-Kevin
On Wed, Dec 10, 2014 at 09:44:06PM -0800, Julius Werner wrote:
> This is a complicated case most firmware USB stacks don't support (I
> don't have experience with SeaBIOS, but I know that both U-Boot and
> libpayload won't work with that either). A USB 3.0 hub is essentially
> two completely separate logical hubs connecting to two completely
> separate logical ports on the host. The USB 3.0 part of the hub uses a
> slightly different protocol (e.g. some bits in the port status have
> different meanings) and from a quick glance the SeaBIOS code doesn't
> seem to take that into account. I'm not sure where exactly it would
> break, but one possibility is the speed detection (in USB 2.0 the port
> status bits 9 and 10 indicate the speed, but in USB 3.0 it's implied
> to always be SuperSpeed and those two bits indicate the link state and
> the port power status instead). The speed is written in the XHCI Slot
> Context for the device and using the wrong value there might cause
> problems.
Right, SeaBIOS does not currently have USB3 hub specific code. The
USB3 spec has close to 100 pages on hubs, but much of it resembles the
USB2 hub details and it's not immiediately clear which parts differ.
BTW, please CC the seabios mailing list on emails.
-Kevin
On Fr, 2014-11-21 at 20:55 +0000, Pandey, Sunil K wrote:
> > Can you confirm the patch below on top of rel-1.7.5.1 fixes the issue for you?
>
> Yes, it works. I applied this patch on latest seabios-1.7.5.1 sources and it compile successfully with Intel Compiler as well as with GCC4.9.1.
Added to 1.7.5-stable branch. Anything else worth including in 1.7.5.2?
cheers,
Gerd
2014-12-03 11:16 GMT+01:00 Gerd Hoffmann <kraxel(a)redhat.com>:
>
> I suspect this isn't going to fly as each makefile line is run
> separately.
Correct. We discussed that on IRC. That's probably one of make's less
intuitive features. I also proposed to use a global export, but had no
opportunity to test it right then.
> How about this: [export LC_ALL globally in Makefile]
Tested-By: Patrick Georgi <pgeorgi(a)google.com>
Thanks,
Patrick
--
Google Germany GmbH
ABC-Str. 19
20354 Hamburg
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg
Geschäftsführer: Graham Law, Christine Elizabeth Flores