As notified by "The Source" in the mailing list, if one runs fprintd-verify and get a "verify-retry-scan" message, all subsequent attempts to match will fail with that same message. Steps to reproduce: - run fprintd-enroll to enroll a finger - run fprintd-verify and quickly tap the sensor to trigger a bad scan - try to swipe normally, you should get a "verify-retry-scan" message Tested affected versions: From git repository (run on Gentoo): fprintd-gitdc504bd1 From Fedora: fprintd-0.2.0-1.fc13 My fingerprint sensor model is Upek TCS4C (USBID 147e:1000).
That's a bug in the driver, not in fprintd. The code in fprintd is correct (look for SIGNAL_VERIFY_STATUS in src/device.c), and the code in verify.c will loop until the finger is actually verified.
I think I found out what caused this bug. This is because imgdev gets stuck into action_state IMG_ACQUIRE_STATE_AWAIT_FINGER_OFF when a verify/identify operation fails with a non-terminating condition. A fix would consist in always going to state IMG_ACQUIRE_STATE_AWAIT_FINGER_ON when a bad scan occurs instead of only doing so for enrolling. (function fpi_imgdev_report_finger_status in imgdev.c)
Created attachment 39240 [details] [review] Patch implementing aforementioned fix. Modified imgdev.c to jump to correct state in case of a verify or identify operation returning a FP_VERIF
With the patch applied, a segfault occurs in nbis after an apparently random number of failed scans (verify-retry-scan) during a identify operation.
Please re-test with recent libfprint version
I have not 147e:1000 device, and it works fine on other devices. Closing this bug, because I can't reproduce it.
Blocks 73762
It's a driver bug, so please fix it in driver instead of modifying common code.
(In reply to Vasily Khoruzhick from comment #8) > It's a driver bug, so please fix it in driver instead of modifying common > code. This also occurs with 08ff:2810, AuthenTec, Inc. AES2810, which according to http://freedesktop.org/wiki/Software/fprint/libfprint/ uses a different driver to the original report.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.