[fprint] Would you please help to create the function to update the template for "verify" and "identify"?

[Dave.Wang]

Hi Bastien,

Thanks for your detailed reply!

Unfortunately, our "match on chip" device need to load template from disk, it does not have persistent storage..

I think we will need your help as below to implement "update template":
1. The way which can transfer updated template from driver to verify/identify AP through libfprint 
2. the verify/identify AP in fprintd need to store the updated template into disk

Any suggestion or help will be really appreciate.

Best regards,
Dave

-----Original Message-----
From: Bastien Nocera [mailto:hadess at hadess.net] 
Sent: Wednesday, January 23, 2019 10:12 PM
To: Dave.Wang <dave.wang at emc.com.tw>; fprint at lists.freedesktop.org
Subject: Re: [fprint] Would you please help to create the function to update the template for "verify" and "identify"?

Hello Dave,

Sorry about the late reply.

On Wed, 2019-01-02 at 10:58 +0800, Dave.Wang wrote:
> Dear,
>  
> We want to implement the primitive driver which will use our "match on 
> chip" sensor.
> Our "match on chip" sensor has function which can update the template 
> after obtaining matched result for "verify" or "identify". It will 
> make better performance.

Is there persistent storage on the device, or would you need to "load"
some data from the disk?

> As I know, the template located in "/var/lib/fprint/.. " is only 
> stored by "fprintd-enroll" now.
> Would you please help to create the function to update the template 
> for "verify" and "identify"?

The driver does not know how to load or save data to disk.

The enrolling process will give out a "struct fp_print_data" variable that the front-end (usually fprintd) can save to disk, and which the front-end will load from disk to pass to the driver for each "fp_async_verify_start" call.

There currently isn't any way for the driver to modify that data that was saved once during enrollment, but I would imagine that it could be possible to pass that data back to fprintd via the
fp_async_verify_stop() result.

Is there any way you could try implementing your driver without that feature first, and then we can figure out how to implement the template update on disk.

Cheers