PB0 acts really strange after attachInterrupt

[imayoda]

Hi there guys :)
Tonight I've encountered a strange behavior.
PB0 manipulation works OK in setup, but after one INT0 interrupt (in which I equal an unsigned long variable to millis() ) the PB0 low->high manipulation results in a barely 1 volt square wave (got with oscilloscope), correct in duty cycle but very weak in voltage.
My circuit is a 1 signal to 4 multiplier with direct calculation of the high and low duty cycle period from the signal sender, and behaves very regularly if I "manually" do the PB0-2-3-4 switching with digitalWrite but before the attachInterrupt..
I think it's somewhat related (I'm not a pro) with #125 (comment) ..
Is there something I could do? Unfortunately I'm on a run, will switch to 25/45/85 line (what a waste 😿 ) so will leave this just as a "note".

Kind regards and thank you all