On Sep 05 2024, Dan Carpenter wrote:
> [ Moving these functions around made them show up as new warnings. ]
Thanks for the report Dan.
>
> Hello Benjamin Tissoires,
>
> Commit 4f7153cf461e ("selftests/hid: add tests for
> bpf_hid_hw_request") from Nov 3, 2022 (linux-next), leads to the
> following Smatch static checker warning:
>
> tools/testing/selftests/hid/hid_common.h:212 uhid_event()
> warn: assigning (-5) to unsigned variable 'answer.u.get_report_reply.err'
>
> tools/testing/selftests/hid/hid_common.h
> 207 case UHID_GET_REPORT:
> 208 UHID_LOG("UHID_GET_REPORT from uhid-dev");
> 209
> 210 answer.type = UHID_GET_REPORT_REPLY;
> 211 answer.u.get_report_reply.id = ev.u.get_report.id;
> --> 212 answer.u.get_report_reply.err = ev.u.get_report.rnum == 1 ? 0 : -EIO;
>
> answer.u.get_report_reply.err is a u16. I couldn't figure out which kind of
> error codes it's supposed to store or how they're used...
It's in drivers/hid/uhid.c:262:
req = &uhid->report_buf.u.get_report_reply;
if (req->err) {
ret = -EIO;
} else {
ret = min3(count, (size_t)req->size, (size_t)UHID_DATA_MAX);
memcpy(buf, req->data, ret);
}
So any non null value triggers an error, and furthermore the hardcoded
value is -EIO, which probably explains why I did not noticed it.
We should either store EIO, either a plain 1...
Cheers,
Benjamin
>
> 213 answer.u.get_report_reply.size = sizeof(feature_data);
> 214 memcpy(answer.u.get_report_reply.data, feature_data, sizeof(feature_data));
> 215
> 216 uhid_write(_metadata, fd, &answer);
> 217
> 218 break;
> 219 case UHID_SET_REPORT:
> 220 UHID_LOG("UHID_SET_REPORT from uhid-dev");
> 221 break;
> 222 default:
>
> regards,
> dan carpenter
