HID gadgets and SET/GET_REPORT

HID gadgets and SET/GET_REPORT

[Vicki Pfau]

Hello,

I've been working on a HID gadget (in userspace, using ConfigFS) that relays USB data from another device transparently, but I noticed that the HID SET_REPORT and GET_REPORT request types appear to be unusable from userspace. As best as I can tell, in `hidg_setup`, they just aren't handled. One sends empty data and the other just does nothing. Sniffing the USB traffic, the only "reply" I get is -EPIPE, which appears to be generated by the UDC driver (dwc3), and I can't access any of the data involved. The device I'm relaying requires those to work transparently.

I was wondering what would need to be done to get this to work? I've tried pivoting to using FeatureFS with ConfigFS, but there's effectively no documentation on that that I can find, and no real sample code either, so I've been groping around in the dark for the past few weeks while I previously had something that *almost* worked, save for apparently one stubbed out feature. I have no real familiarity with the intricacies of the USB subsystem, otherwise I might try to prepare a patchset myself.

Thanks,
Vicki

Re: HID gadgets and SET/GET_REPORT

[Alan Stern]

On Thu, Jun 16, 2022 at 04:58:32PM -0700, Vicki Pfau wrote:
> Hello,
> 
> I've been working on a HID gadget (in userspace, using ConfigFS) that 
> relays USB data from another device transparently, but I noticed that 
> the HID SET_REPORT and GET_REPORT request types appear to be unusable 
> from userspace. As best as I can tell, in `hidg_setup`, they just 
> aren't handled. One sends empty data and the other just does nothing. 
> Sniffing the USB traffic, the only "reply" I get is -EPIPE, which 
> appears to be generated by the UDC driver (dwc3), and I can't access 
> any of the data involved. The device I'm relaying requires those to 
> work transparently.
> 
> I was wondering what would need to be done to get this to work? I've 
> tried pivoting to using FeatureFS with ConfigFS, but there's 
> effectively no documentation on that that I can find, and no real 
> sample code either, so I've been groping around in the dark for the 
> past few weeks while I previously had something that *almost* worked, 
> save for apparently one stubbed out feature. I have no real 
> familiarity with the intricacies of the USB subsystem, otherwise I 
> might try to prepare a patchset myself.

I'm not familiar with the HID gadget, but it certainly seems that your 
best approach would be to fix up the hid function driver by adding 
support for Set-Report and Get-Report requests.

If you really don't feel up to that task, there probably are people on 
the mailing list who would be happy to help.

Alan Stern