[fprint] UPEK TouchStrip device variant ID 147e:2016

[Ugo Riboni]

> > I'd like to help with trying the binary drivers, at least trying
> > installing them and trying to get them to work.
> > But the link above leads me nowhere, and searching the upek drivers
> > area for Linux turns out only a BioAPI based SDK.
> > Is the binary driver inside that ?
>
> http://www.upek.com/solutions/pc_and_networking/sdks/linux/
> I think that's all you need. You can probably find some howto's on the
> thinkpad wiki or something like that.

I finally had some free time and gave this a shot.
Installed from source the linux-bioapi framework. Then downloaded and
installed the UPEK BSP binary module.
Then after fiddling a while to get it to build, i ran (as root) the
sample included with the BSP, which essentially tries to use the
bioapi framework to load the UPEK BSP, initialize it, then do some
enrollment.
But i never got to that point. The sample fails when initializing the
BSP module.

Basically it goes like this:
- at some point BioAPI_ModuleLoad is called, asking it to load the
UPEK binary library
- bioapi find the library, and dynamically loads it.
- then finds the address of the SPI Load function in the library and
tries to call it.
- the function returns an error code of 0x149D which is described in
the BSP manual as:
BioAPIERR_TFMESS_PT_FUNCTION_FAILED : Function failed

Not a very useful error message.
So i went in and tried to see if strace had anything useful to say.
It appears everything goes fine until the library tries to access the
actual USB device.
You can see from the log yourself, i have attached to this email, but
it looks like it does some searching through all the USB devices on
the system, until it settles down on this:

open("/dev/bus/usb/003/002", O_RDWR)    = 12
ioctl(12, USBDEVFS_SETCONFIGURATION, 0xbfaa09a4) = 0
ioctl(12, USBDEVFS_CLAIMINTERFACE, 0xbfaa09a4) = 0
nanosleep({0, 10000000}, NULL)          = 0
ioctl(12, USBDEVFS_CONTROL, 0xbfaa096c) = 1
gettimeofday({1197590255, 551261}, NULL) = 0
ioctl(12, USBDEVFS_SUBMITURB, 0xbfaa00f4) = 0

after this, it enters in a very long loop doing the following:

ioctl(12, USBDEVFS_REAPURBNDELAY, 0xbfaa0138) = -1 EAGAIN (Resource
temporarily unavailable)
select(13, NULL, [12], NULL, {0, 1000}) = 0 (Timeout)
gettimeofday({1197590255, 553691}, NULL) = 0

It looks to me like it sent a command to the device (possibly an
initialization command) and the device isn't responding at all.
After keeping retrying for like 10 seconds, it does bail out and
return the error message listed above.

Notice that it doesn't seem to fail because of some authentication
scheme, like i've read elsewhere.
If it was that, it would be getting a response of some kind from the
device, unless the authentication itself is just "knowing the right
initial USB command to send", which would be a pretty silly auth
scheme... but you never know how silly people can get with this stuff.

So it looks like either the BSP library isn't designed to work with
this specific model of reader, or something else is going wrong. But
being closed source stuff, there's only so much one can do.

It may be possible to trace USB messages and compare the
initialization the BSP sends with the one that windows sends. But i
don't know how to do that, and i'm doubting it would lead to anything
useful since we can't alter the binary BSP library in any way to make
it send the correct data.

> > Please let me know where i can get this binary driver from, if it does
> > exist at all, so i can try to help further.
>
> The first step is to figure out the image format. If you'd like to help,
> read the thread titled "image format challenges" I started the other day.

I will try to play with the image when i have time, but not sure i can
get anywhere with that.

Any other ideas based on information above ?
--
Ugo
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: strace_biapi_upek_bsp_fail.txt
Url: http://lists.reactivated.net/pipermail/fprint/attachments/20071214/60ff9c4c/strace_biapi_upek_bsp_fail-0001.txt