Re: [RFC] xhci: Let completion handlers run before rings are restarted.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Jun 6, 2012 at 10:26 AM, Xiaofan Chen <xiaofanc@xxxxxxxxx> wrote:
> On Wed, Jun 6, 2012 at 3:32 AM, Sarah Sharp
> <sarah.a.sharp@xxxxxxxxxxxxxxx> wrote:
>> On Tue, Jun 05, 2012 at 01:12:44PM -0400, Alan Stern wrote:
>>> On Tue, 5 Jun 2012, Sarah Sharp wrote:
>>>
>>> > Or just change libusb to not break up large transfers, and get rid of
>>> > the BULK_CONTINUATION flag all together.  libusb should submit
>>> > the whole transfer to usbfs with an iovec.  usbfs can then:
>>> >
>>> >  a) verify each iovec entry is not too long
>>> >  b) verify the total length of the iovec isn't too big
>>> >  c) translate each iovec entry into one sglist vector
>>> >  d) submit one URB with a populated sglist to the USB core
>>>
>>> I think this would work.  But it wouldn't help people trying to use
>>> libusb with xHCI on older kernels.
>>
>> No, it wouldn't help them.  At some point though, users just need to
>> upgrade their kernels.  And if enterprise distro kernels decide they
>> care about those particular users, they can pick up the patches.  I
>> think Hans was already volunteering to try and push the necessary kernel
>> changes for RHEL.
>
> I think this is fair. xHCI on older kernels may not work well
> anyway. BTW, It is similar under Windows that xHCI may or
> may not work well on Windows XP/Vista/7 where there is no
> official Microsoft driver for them.
>
> On the other hand, libusb/libusbx should probably need
> to put a note that it may not work with USB 3.0 ports prior to kernel
> 3.x (where 3.x is the first kernel to solve this issue).

Or if there will no kernel changes due to the complexity required, then we
need to put a note that prior to libusb-1.0.10 / libusbx-1.0.13 (if that is the
first version to follow Sarah's change), libusb/libusbx may not work with
USB 3.0 ports.

>> It might be worth it to ensure that the old usbfs calls without iovecs
>> can co-exist with the new calls, so that we don't break older API.  That
>> will make it easier to convince distro kernels to pick up the usbfs
>> changes.


-- 
Xiaofan
--
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


[Index of Archives]     [Linux Media]     [Linux Input]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Old Linux USB Devel Archive]

  Powered by Linux