Re: [Qemu-devel] Mixed USB 1.1 and USB 2.0 on the same port

[Hans de Goede <hdegoede@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1951 bytes --]



On 01/04/2012 05:52 PM, Erik Rull wrote:
> erik.rull@rdsoftware.de wrote:
>>> On 12/31/11 13:11, Erik Rull wrote:
>>>> Hi all,
>>>>
>>>> how can I use a USB 1.1 device on the USB 2.0 bus? Currently the EHCI
>>>> implementation complains that the device is mismatches the USB version.
>>>
>>> -readconfig docs/ich9-ehci-uhci.cfg
>>>
>>> cheers,
>>> Gerd
>>
>> Thanks for the hint.
>>
>> It looks better now. But some things are still a bit strange.Hi,
>>
>> Sequence:
>> device_add usb-host,bus=ehci.0,hostbus=2,hostport=1.4
>> Plug in a USB 2.0 printer
>> (gets detected by the guest, printing is possible, no bluescreen, it just
>> works)
>> Remove the USB 2.0 printer
>> Plug in a USB 1.1 dongle
>> Gets detected, etc., fine
>> Remove the USB 1.1 dongle
>> Plug in the USB 2.0 printer again
>> Guest complains now, that a USB 2.0 device was plugged into a USB 1.1 port
>> => printer is now 1.1 and does not work as if EHCI is missing now
>> => reboot guest, everything is fine again??
>>
>> Any idea what could have happened here?
>> Same behavior when using a 2.0 USB key and the USB 1.1 dongle - also on
>> other ports - the transfer rate is horrible after having removed the USB
>> 1.1 device and reconnected the 2.0 device.
>> Exchanging the two USB 2.0 devices on the same port without having the 1.1
>> device plugged in is fine!
>>
>> Any hint what is wrong here would be great.
>>
>> Best regards,
>>
>> Erik
>>
>
> Additional Information:
> This behavior is present on a Linux guest as well! After having removed the 1.1 Dongle and plugged in the printer, the Linux guest detects the hardware via the UHCI kernel drivers and tells me to use a faster hub for max. performance.
> It looks as if the speed downgrade by the 1.1 device cannot be reversed at runtime.

I've managed to reproduce this and written a fix for it. I've just send
a patch for this to the list. I've also attached this patch here for
your convenience.

Regards,

Hans

[-- Attachment #2: 0001-usb-ehci-Clear-the-portstatus-powner-bit-on-device-d.patch --]
[-- Type: text/x-patch, Size: 1186 bytes --]

>From 0b5059a8c0ae45c66caf5947f66b5c5cae81f622 Mon Sep 17 00:00:00 2001
From: Hans de Goede <hdegoede@redhat.com>
Date: Fri, 13 Jan 2012 14:26:26 +0100
Subject: [PATCH] usb-ehci: Clear the portstatus powner bit on device
 disconnect

According to the EHCI spec port ownerhsip should revert to the EHCI controller
on device disconnect. This fixes the problem of a port getting stuck on USB 1
when using redirection and plugging in a USB 2 device after a USB 1 device
has been redirected.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
 hw/usb-ehci.c |    5 +++++
 1 files changed, 5 insertions(+), 0 deletions(-)

diff --git a/hw/usb-ehci.c b/hw/usb-ehci.c
index a946e1d..69bcc4b 100644
--- a/hw/usb-ehci.c
+++ b/hw/usb-ehci.c
@@ -764,6 +764,11 @@ static void ehci_detach(USBPort *port)
         USBPort *companion = s->companion_ports[port->index];
         companion->ops->detach(companion);
         companion->dev = NULL;
+        /*
+         * EHCI spec 4.2.2: "When a disconnect occurs... On the event,
+         * the port ownership is returned immediately to the EHCI controller."
+         */
+        *portsc &= ~PORTSC_POWNER;
         return;
     }
 
-- 
1.7.7.4