search

monich / harbour-foilauth

Secure OTP (One-Time Password) generator for Sailfish OS
https://openrepos.net/content/slava/foil-auth
Other
11 stars 7 forks source link

UI hang on Foil Auth startup #33

Closed direc85 closed 1 year ago

direc85 commented 1 year ago

Hi,

I've experienced this enough times that it seems issue-worthy: On random, let's say every 20th startup or so, Foil Auth hangs the phone/UI/Silica and the screen stops updating its contents. It's also impossible to turn the screen off or on by pressing the power button. It does react to it, since it vibrates when I press it. The only way out of the situation I've found is holding the power key for 8 seconds (?) until the device powers off gracefully (it seems / I hope).

I vaguely remember something about libsailfishsecrets hanging someone and using some Rust implementation of something has fixed the issue for them, but I can't recall anything more and I can't find it again... Do you think this would be a lead to start following?

Thanks for the great application triplet!

monich commented 1 year ago

Which device is that?

BTW in case if you haven't noticed there's also YubiKey OTP for calculating OTPs with a YubiKey. And that app has an option to store the hash of your YubiKey password, i.e. you don't have to type it in every time 🙂

direc85 commented 1 year ago

I have Sony Xperia 10 III running SFOS 4.4.0.72. I don't think my previous devices (X10II, XA2) had the issue, but I could be wrong though.

Thanks for the tip! I should give that a try!

direc85 commented 1 year ago

https://gitlab.com/Schmiddiii/flare/-/issues/4 This was the issue I mentioned.

direc85 commented 1 year ago

It just happened again, after several days of uptime and heavy device use... I think something essential may get OOM sniped. It happens early in the application startup, so I may be able to figure something out myself; I'll skim through some code in a day or two.

direc85 commented 1 year ago

This was actually fixed in SFOS 4.5.0.16 libhybris (server_wlegl: fix a crash when format = 0 in the changelog), so I'll go ahead and close this bug!