Bug 152604

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.5) Gecko/20041107 Firefox/1.0

Description of problem:
Sending an output report to a HID device works perfectly with 2.6 kernels (I tested with 2.6.6, .9, .10, and .11) but not with 2.4.21, 2.4.22, or 2.4.29.

We have a device which is controlled by hid output reports.
There is one output report, with two usages.
Obtaining the report id and usage code, setting the values, and sending the report, works on 2.6.  On 2.4, the HIDIOCSREPORT ioctl succeeds (zero return code), but the report is in fact not sent to the device.  I was trying it first with 2.4.21 which is the kernel installed out-of-the-box on RHEL3, but I understand there was a major issue with multiple usages; so I tried with .29 but the behavior is the same.

So I put a bunch of printk()'s in various places of hid-core.c and hiddev.c to better understand what was going on.  Everything in hid_submit_report(), hid_write_report(), hid_output_report(), hid_submit_out(), all the way to hid_output_field(), looks normal.  Hid_output_field() gets called the right number of times (one for each of the two usages), with the correct arguemnts (like, correct usage and correct value to put in).  The problem seems to be with submit_urb(), which printks "ctrl urb status -32 received".  The return value of the ioctl itself, however, is 0.  (I would vaguely guess this has to do with the fact that the send is only initiated here, and not carried out synchronously, so when the actual error happens, the ioctl has already returned... would be happy if somebody explained to me how this works exactly).


Version-Release number of selected component (if applicable):
2.4.29

How reproducible:
Always

Steps to Reproduce:
A full test case code is attached.

Actual Results:  The report is not sent, the return code from the ioctl is 0, and the kernel outputs: "ctrl urb status -32 received".


Additional info: