Redrawing so we can easily see the registers feeding each other without all the crossing wires.
Oh, look it's a register feeding a register and the output of the last register is feeding the input of the first register.
Looks like a divide by 2 (as long as there is a reset, which is not shown).
Now on to your question, with the tclock_skew of 1ns it won't work, once that skew gets added/subtracted you can no longer meet the setup or hold of one of the flops as the result of any calculation will end up 0 ns. (i.e. -1 ns of slack)
If you skew the clock in the other direction the problem just shifts to the other register.