Re: usb2 stack.

On Monday 10 November 2008, Charles DeBardeleben wrote:
> I have upgraded to the new usb2 stack and am having several problems.
>
> First, I too could not prevent the old usb stack from getting loaded. I
> decided to take the approach of static loading of the usb2 stack,
> and discovered that I could not include usb2_bluetooth, and
> usb2_bluetooth_ng whithouk adding more of the netgraph stack,
> to the statuc kernel, so I added
>
> options         NETGRAPH_BLUETOOTH              # ng_bluetooth(4)
> options         NETGRAPH_BLUETOOTH_BT3C         # ng_bt3c(4)
> options         NETGRAPH_BLUETOOTH_HCI          # ng_hci(4)
> options         NETGRAPH_BLUETOOTH_L2CAP        # ng_l2cap(4)
> options         NETGRAPH_BLUETOOTH_SOCKET       # ng_btsocket(4)
>
> to my config. Now I bind with the new stack. However, many
> usb devices still are not working correctly. 

> First, many devices 
> are not making entries in /dev, even though they seam to bind
> to the driver. 

The entries are invisible.

> This seams to include ums, uhid, and ugen. I was 
> hoping to have my mouse bind to ums, so that I can run hid2hci
> on it so that I activate my bluetooth hub. My m-audio usb audio
> codec is binding to audio, but it is not working. As I was trying
> to debug this, I decided I needed usbconfig. I ran buildworld and
> found that none of the userland code has been added to the makefiles.
> So I edited these files to add libusb20, the header files, and
> usbconfig to the build:
>
> M      include/Makefile
> M      share/mk/bsd.libnames.mk
> M      usr.sbin/Makefile
> M      lib/Makefile
> M      etc/mtree/BSD.include.dist
> M      sys/dev/usb2/bluetooth/ng_ubt2.c

Yes, this will be fixed eventually.

>
> I then found the kdump will not build due to ioctl definition confilcts
> with the old usb stack. So I guess at this point the 2 stacks do not
> really coexist together. However the build got far enough along to build
> usbconfig.

Looks like a problem that needs to be addressed. 

>
> I have not added the patch that was posted today, but I doubt that that
> is the underlying cause of my issues.
>
> I had high hopes for this new stack as my usb often "locks up" making my
> keyboard stop.
>
> 2 more issues I had, not related to usb2. First when I did a cvs update,
> I got many conflicts on files I had not changed. I was guessing maybe
> something was wrong with the svn-cvs bridge, so I decided to try the
> svn server. However, there seam to be no ID strings if you checout the
> svn tree. This seams to confuse mergemaster. Is there a new procedure
> for upgrading a system built from the svn tree?

I don't know.

Maybe you need to resolve the code first.

Could you send me your current patchset again, and I will see if I can get 
some parts of it in.

--HPS