To make my world a better place to live, I have downloaded the development version of the driver and tried to figure out why it was being so slow. It turns out that in the library that is used to communicate with the fingerprint-reader, there was a call to the function usb_bulk_read that was taking too long. I have tried to find some official documentation about this function, but all I could find was how not to document a function (seriously, what is the role of the parameter timeout? and is it measured in seconds? milliseconds?).

Anyway, I have changed the value of the USB_TIMEOUT constant from 5000 to 5 and the fingerprint-reader is much, much better! In detail, here is the patch:

Index: libthinkfinger/libthinkfinger.c
===================================================================
--- libthinkfinger/libthinkfinger.c (revision 118)
+++ libthinkfinger/libthinkfinger.c (working copy)
@@ -35,7 +35,7 @@

#define USB_VENDOR_ID 0x0483
#define USB_PRODUCT_ID 0x2016
-#define USB_TIMEOUT 5000
+#define USB_TIMEOUT 5
#define USB_WR_EP 0x02
#define USB_RD_EP 0x81
#define DEFAULT_BULK_SIZE 0x40

I am not sure why this works and what are the implications, but all the tests I did were satisfactory. If you know more about this problem, please let me know! I would be more than happy to understand why the USB_TIMEOUT constant seems to be acting as a time-wait delay, rather than as a timeout in the following call (line 324, libthinkfinger/libthinkfinger.c):

usb_retval = usb_bulk_read (tf->usb_dev_handle, USB_RD_EP, bytes, size, USB_TIMEOUT);

(I think it is strange that I haven’t found anyone complaining about the same problem. Nevertheless, I’ve decided to publish this post anyway, in case someone runs into the same problem.)

Update (2009/07/31): I’ve upload a patched SRC RPM: thinkfinger-0.3-9.jff.src.rpm. Please note that I have added the patch to Fedora’s SRC RPM.