Regarding some "popular" bugs:

- HTB: bad diff in charge ....

 it means that when comparing current time with time of last
 class check we have got weird time difference (negative, too
 big ...). There is bug in all 2.2, 2.4 kernels regarding this.
 Part of HTB3.4 and 3.5 is fix for this one.
 If you get the error and use these new patches, let me know.

- assertion failed: delay > 0 && delay < 5000000
- HTB: suspicious delay in wait_tree ...

 some class wanted to add itself to waitlist and specified suspicious
 delay to wait. I'm not sure why yet. In 3.5 I changed reporting of
 it to include value of the delay and other info which could allow
 me to locate the error.
		
