I have returned, this time with a fully working fingerprint sensor!
But TheMightyCat i hear you say, we already had a working fingerprint sensor made by nick!
And you're right, that's why i specified fully while maybe nick's one is also fully working he doesn't show enrollment or a negative control, so either way it isn't confirmed or not confirmed from his video alone.
Detecting if any finger is on the sensor is the easy part, enrolling and checking if its the right finger is the hard part.
But while it has taken longer then i would have liked it is now fully working!
And it even has a party trick!
As you can see in the video, using my index finger does a regular unlock, my middle finger unlocks and opens the camera, and my ring finger is a duress and removes the super secret file. (the removal of the file is instant, the delay comes from updating the desktop entry)
this is done trough fingerprintd, for root scripts this file is used:
/etc/fingerprintd/actions.conf
right-index-finger /etc/fingerprintd/duress.sh
While for user scripts its published on the net.catcrafts.Fingerprintd1.FingerMatched dbus
expect this to land in fp6-img somewhere this weekend, i was already fighting for my life trying to make the recording and don't have the mental energy left to publish it.
Recording something that requires 2 hands is the biggest pain in the ass ever, these were all the attempts:
ALSO I HATE TEE. THERE IS NOTHING MORE ANTI USER THEN A CHIP WHERE THE MANUFACTURER CAN GO "erm akshually ERROR_ELF_SIGNATURE_ERROR"
See this girl? This is Mia.
She has hacked this computer and jailbroken its TEE
Be like Mia.
~~hey look the sun has come up, time to go to bed.~~
News roundup
imsd issues
Thanks very much to 2 members of the community for their patience of me asking them to run scripts, i'm trying to be the least of a bother possible.
If you find any issues with imsd please let me know! the current known issues are:
No call audio on carriers that only support VoLTE (alteast that is the hypothesis for this issue)
Getting called from a landline doesn't work.
I'm working hard to fix these issues, they will hopefully be fixed somewhere next week.
catcrafts.net shop
Ive been thinking hard about this, but there is no business case to be made to sell pre flashed units on my website, its more expensive for the people buying them, it puts a huge administrative burden onto me. and for people that want to finacially support me its much better for the both of us to purchase a unit at your local retailer and then donate on the website instead of purchasing as then it isn't taxed because its a donation, while if its a sale it is taxed. so i don't think im going to do this unless anyone has a good case for it.
RISC-V phone
This project has been low priority while working on the FP6 but i think i have settled on a chip, its probably going to be the spacemiT k1, not the best, but the best for this purpose. Finding a risc-v chip with a low power consumption and good peformance is slim pickings, and while the k1 is not exaclty great its as far as i know the pest peformance per watt.
What's next
Just the camera and laser rangefinder, then the fp6 is officially 100% hardware complete
Mental health
Don't want to make this too personal but it's important, take care of yourselves out there.
Bonus question
One lemmy thing i don't understand
According to the metrics !linuxphones@lemmy.ml should be more active then here, but that place is completely deserted!
Are all those subscribers just dead accounts?
My biggest tip is to run trough the full sequence of whatever you want to make on rooted android first, and then capture everything.
Then when making it work on Linux you can compare against those captures.
Interesting. And with which tool is this possible? I would also love, if you could document your work a little bit, for new-comers🥳
I used logcat aswell as just reading form /proc/tzdbg/qsee_log for this.
the general steps go as follows:
Poke these pins to see if i can bring up the sensor, see if i can talk to it. The power rail worked fine, but the data pins were blocked behind the TEE. So started a whole new adventure.
I flashed android again as i needed more information, capturing the logs. Which i didn't really do great as i had to flash android several times over again but oh well.
Then its a ton of work comparing against the android logs, adjusting your linux code, looking at the logs that come out of that. This part is not especially hard, its just pulling threads until it works pretty much, but its really boring. What made this process 10x more annoying is that i couldn't attach a debugger because it was running in the TEE.
Then test it, record the video, make the post.
If its going to the kernel export the patch and write the messages and cover letter. Run it trough the AI review, fix any issues that comes out of that, and finally send.
This the general process, i would suggest starting with an easy part first.
For me, as long as i have a thread to pull on then its not really hard, just a ton of work.
The nightmare scenario is that it doesn't work and you have no idea why. Then it gets hard.