On Tue, May 27, 2003, Randy.Dunlap <[EMAIL PROTECTED]> wrote:
> On Fri, 23 May 2003 16:50:35 -0700 Greg KH <[EMAIL PROTECTED]> wrote:
>
> | On Thu, May 22, 2003 at 02:09:56PM -0400, Johannes Erdfelt wrote:
> | > On Thu, May 22, 2003, Randy.Dunlap <[EMAIL PROTECTED]> wrote:
> | > > As mentioned in Jan. 2001 [1] and on linux-hotplug-devel very
> | > > recently [2], lsusb produces byte-swapped output on big-endian
> | > > (e.g., PPC) systems -- for some USB descriptor fields. This is
> | > > due to the following USB core code in usb_get_device_descriptor()
> | > > [in 2.4.x and 2.5.x]:
> | > >
> | > > if (ret >= 0) {
> | > > le16_to_cpus(&dev->descriptor.bcdUSB);
> | > > le16_to_cpus(&dev->descriptor.idVendor);
> | > > le16_to_cpus(&dev->descriptor.idProduct);
> | > > le16_to_cpus(&dev->descriptor.bcdDevice);
> | > > }
> | > >
> | > > so we always make these 4 fields native-endian, but other multi-byte
> | > > fields are not modified.
> | > >
> | > > I doubt that this is a good time to change the USB core (if it should
> | > > be done at all). However, lsusb can be modified to handle this
> | > > endian-ness easily. I am attaching a patch to lsusb from usbutils
> | > > version 0.11 that handles this. Tested/verified on x86 by me and on
> | > > PPC by Alexander Kühn, who reported the Vosonic X-drive problem on
> | > > linux-hotplug-devel. Thanks, Alex.
> | > >
> | > > This isn't the prettiest patch that I can imagine, but it does fix
> | > > the problem.
> | >
> | > I've run into this problem as well and I think the fix should go into
> | > the kernel.
> |
> | I agree, I'd take a patch for this for 2.5 right now if you want to make
> | it. We can add it to 2.4 during the 2.4.22-pre series too.
>
> I agree that it's the right thing to do in 2.5.
> IMO 2.4 shouldn't be touched. That would be like changing an ABI/API
> in a stable kernel.
I disagree. It's been broken forever in 2.4 and no one has attempted to
workaround it even.
Just look at lsusb :)
I think it should change for 2.5 and 2.4.
JE
-------------------------------------------------------
This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
