[WIP] Add support for UAC3 descriptors

[ghost]

This adds a new way of dumping descriptors, which splits the knowledge of how to interpret descriptor data from the actual dumping. This has two advantages:

1. It is easy to add support for new descriptors, since they are now simple definitions that resemble the tables in the USB specifications.
2. The code for dumping descriptors is common, so the output is easy to keep consistent. It is also consistent and thorough in its handling of insufficient descriptor data buffer, and junk data at the end of a descriptor.

UAC1 and UAC2 are converted to use the new mechanism, and finally initial support for UAC3 is added.

[ghost]

I would be really grateful for any feedback on this.

For the descriptor definitions stuff, my aim was to make the definitions as readable as possible, and make them easy to write from the tables in the USB specifications.

I have not moved all the the descriptor dumping over to the new mechanism, just the ones where I've added support for UAC3 variants.

For UAC3, the changes here give much more useful output, however there are more parts of UAC3 that I have not yet added support for.

[ghost]

I've tweaked desc_dump(). Previously it took an buffer, buf, and the descriptor definition desc, and it assumed that the buffer started at the start of the descriptor (i.e. pointing at bLength), but that the first three bytes of the buffer had already been dumped. Now this assumption is removed, making the API more generally useful.

[ghost]

Added support for dumping the USB3 BOS Configuration Summary Descriptor and improved the comment for get_uac_interface_subtype().

[ghost]

Added the option to tell the descriptor dumper to read the buffer length from the first field in the descriptor definition, and made the Configuration Summary use that.

This will be simpler going forwards with USB Audio Class 3 high capability descriptors, which have a two byte wLength field instead of the one byte bLength; the descriptor dumper will be able to do the right thing itself if the definition includes these common fields.

[gregkh]

This series is getting long, how about just posting it on the linux-usb mailing list and we can review it and iterate there? That should be easier overall, as I have a lot of individual code comments which is a pain to do in github :)

[ghost]

OK, I've subscribed to the list. It seems git send-email is currently not configured right, so I'll post my series there once I've worked out this config issue.

[gregkh]

On Tue, Dec 05, 2017 at 03:38:02PM +0000, Michael Drake wrote: OK, I've subscribed to the list. It seems git send-email is currently not configured right, so I'll post my series there once I've worked out this config issue.

No need to subscribe if you don't want to, it takes anyones emails, as long as it isn't in html :)

[ghost]

Ah, thanks! I'll unsubscribe if I find it too high-volume then. :)

[ghost]

Looks like I found the right runes. :) Sent to the linux-usb list.

[ghost]

I sent v2 to the list. Should I close this merge request, or update it with what I sent to the list?

[gregkh]

On Thu, Dec 07, 2017 at 10:30:41AM +0000, Michael Drake wrote: I sent v2 to the list. Should I close this merge request, or update it with what I sent to the list?

Yeah, let's just close this, I can take the patches from the list easier, thanks! greg k-h

[ghost]

OK, suits me too. :)

Closing: Discussion has moved to linux-usb mailing list.