Re: [Qemu-devel] Large USB patch

[Lonnie Mendez <lmendez19@austin.rr.com>]

nix.wie.weg@gmx.de wrote:

>I have fixed this issue, also I have found the segfault on usb_del.
>Patch is attached.
>Next problem:
>Linux does not recognize it, if I add a "tablet" while linux is allready
>running. The attach is not delivered to the operating system.
>  
>
   Hm.  There also seems to be some unchecked bounds when adding devices:

(qemu) usb_add host:001:021
Could not add USB Device 'host:001:021'
(qemu) info usb
     Controller 001: uhci
          001:001 = mouse
          001:002 = tablet
          001:001 = host:001:021
     Summary: 1 USB Controller, 3 USB Devices
(qemu)

   Also I've a question on where the emulated hub comes into play.  
Specifically how do you currently add new devices to it?  The ,addto 
syntax doesn't seem to apply to the emulated hub as it takes a bus 
address (the controller) and an address on the root hub.  But the hub 
itself would be centrally connected to a bus number identifying a 
controller and not have its own.  It seems we've presently traded a hub 
with dynamic ports for 2 static ones.  I'm glad there is a lot more 
internal tracking here as with the previous system you could usb_add the 
same host device several times and it wouldn't care.  This should also 
make it easier to reconnect the ps2 mouse when the hid mouse is dettached.

   Adding a device by Vendor id and Product id doesn't seem to work.  If 
this is intentional then perhaps the functionality should be restored.  
qemu users will find this convenient as it is a fixed/static 
identification for a single usb device which is most likely going to be 
the common usage.

   In regards to the linux guest problem I'm not seeing it.  On an older 
debian SID install which is currently using linux kernel 2.6.14 all 
seems well.  The code here is merely the first patch combined with CVS 
along with the already submitted patches.  I'll try to find some 
variation to test with.  In which distro is this happening?