Hi Sarah.
On 12/18/13 17:23, Chris Clayton wrote:
>
>
> On 12/17/13 21:39, Sarah Sharp wrote:
>> On Sat, Dec 14, 2013 at 11:59:15AM +0000, Chris Clayton wrote:
>>> On 12/13/13 16:38, Sarah Sharp wrote:
>>>> On Fri, Dec 13, 2013 at 02:31:47PM +0000, Chris Clayton wrote:
>>>>> Hi,
>>>>
>>>> Hi Chris,
>>>>
>>>
>>> Hi Sarah and thanks for the reply.
>>>
>>>> Thanks for the bug report!
>>>>
>>>>> Firstly, I'm not subscribed, so please cc me on any replies.
>>>>>
>>>>> I see the problem I describe below on 3.12.[0..5] and on the current 3.13 development kernel, including a kernel pulled
>>>>> from Linus' tree just a few minutes ago. The diagnostics below and the config file attached are from 3.12.5. I can
>>>>> easily repeat on 3.13 if that would be more useful.
>>>>>
>>>>> My Fujitsu Lifebook AH531 laptop has an expresscard slot and I bought an expresscard USB3.0 card.
>>
>> So, after looking at the two different dmesg outputs, I think either
>> your host controller or the device is just not working at USB 3.0
>> speeds. When you hot-plug the device, the host doesn't give a port
>> status event at all, so there's really nothing the xHCI driver can do.
>> Even when you plug in the host with the hard drive attached, it only
>> detects it as a USB 2.0 device.
>>
>> It really feels like the hardware is just broken. Have you tried this
>> host and device combination under Windows? I hate to say it, but
>> perhaps you want to try a different vendor for either the host or
>> device?
>>
I think your feeling that hardware is broken is probably right. I've abandoned that card and bought one with a different
an NEC/Renesas chipset. That works fine - udev events are raised when drives are {un,}plugged and the performance
improvement over USB2 is very noticeable.
This message should help people who in the future search looking for information on the linux compatibility of
expresscard USB3 adapters. The three-port adapters (in this case manufactured by AKE and) based on the Fresco Logic
chipset do not, at the time of writing, work with the linux XHCI driver.
Thanks for your analysis.
> Once I got the drivers installed, Windows 7 reports the card as "Fresco Logic XHCI (USB3) Controller FL1100 Series" and
> "Fresco Logic xHCI (USB3) Root Hub" and detects the drive when I plug that in. However, it reports that the drive could
> transfer data faster if it where "... connected to a Super Speed USB 3.0 port".
>
>> If you plug in other USB 2.0 or 1.1 devices into the xHCI expresscard,
>> do those work? E.g. a mouse, keyboard, hub, etc. If they do, I would
>> lean towards trying to find a replacement USB 3.0 hard drive. If they
>> don't, I would try to replace the xHCI host. Or if the host is cheaper
>> to replace, maybe try replacing that first?
>>
> My USB 2.0 mouse and USB 1.1 speakers are detected and work fine. Oddly enough though, when I plugged in the HDD
> alongside the mouse and speakers it was detected. usbview reports it as USB 2.1. I then detached all three devices and
> unplugged the card. When I re-attached the mouse and the speakers, they were detected. The drive, however, was not
> detected - the "no hotplug settings from platform" message I've reported previously was produced
>
> So, it would appear that Windows 7 sees the card as not actually supporting USB 3.0, but the drive as preferring USB
> 3.0. Linux, on the other hand, sees the card as supporting USB 3.0, but the drive as USB 2.1.
>
> I've cut and paste the relevant data from usbview into a file which is attached, as is the output from lsusb -v with the
> devices attached.
>
>> Sorry I can't be of more help, I just don't know how to fix this. The
>> only thing I could possibly think would help is enabling the compliance
>> mode polling. Maybe the USB 3.0 port is stuck in compliance mode, and
>> can't give the xHCI host a port status change event? Can you send me
>> the output of `sudo lspci -vvv -n` and `sudo lspci`?
>>
> That output is in the attached file lspci.data
>
> Thanks for looking into this, Sarah.
>
> Chris
>> Sarah Sharp
>>
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
