Frequent data read problem

[strichter]

I have an HEM-670IT that I am trying to talk to. While I sometimes can read the data correctly, most of the time I get failures at various steps. Consider these 2 sessions:

[root@localhost build]# ./bin/omron_790IT_test
Found 1 omron 790ITs
Opened omron 790IT
omron_get_command_return: read_result_result -7 < zero
Device serial: M604
omron_get_command_return: read_result_result -7 < zero
Device version: 0010
^C

[root@localhost build]# ./bin/omron_790IT_test
Found 1 omron 790ITs
Opened omron 790IT
omron_get_command_return: read_result_result -7 < zero
Device serial: M604
omron_get_command_return: read_result_result -7 < zero
Device version: 0010
AJR data count: 2
10/08/2011 23:16:38 SYS: 120 DIA: 69 PULSE: 59
omron_get_command_return: read_result_result -7 < zero
11/08/2011 09:16:00 SYS: 0 DIA: 0 PULSE: 0
Weekly info:
Morning[0 07/08/2011] = sys:138 dia:85 pulse:59.
omron_790IT_test: /opt/zope/cipherhealth/packages/cipher.device/libomron-trunk/src/omron.c:190: omron_get_command_return: Assertion `current_read_size <= size - total_read_size' failed.
Aborted

As you can see, I frequently read less data than expected. When connecting the USB cable to Linux, I get the following output:

Aug 11 09:32:54 localhost kernel: usb 2-1.2: new full speed USB device using ehci_hcd and address 24
Aug 11 09:32:54 localhost kernel: usb 2-1.2: New USB device found, idVendor=0590, idProduct=0028
Aug 11 09:32:54 localhost kernel: usb 2-1.2: New USB device strings: Mfr=1, Product=3, SerialNumber=0
Aug 11 09:32:54 localhost kernel: usb 2-1.2: Product: HHX-CABLE-USB1
Aug 11 09:32:54 localhost kernel: usb 2-1.2: Manufacturer: OMRON Corporation
Aug 11 09:32:54 localhost kernel: generic-usb 0003:0590:0028.000E: hiddev0,hidraw0: USB HID v1.10 Device [OMRON Corporation HHX-CABLE-USB1] on usb-0000:00:1d.0-1.2/input0

But the Omron documentation states that it does not follow the standard HID protocol, so I wonder whether the Linux driver is getting in the way.

Has anyone had similar problems?

[strichter]

libomron has a bug int he data read function. It treats reads that start with 0x08 as a valid line, but it seems to be a placeholder for "data not yet available" and should simply be ignored.