There's little use studying mathematics without actually doing mathematics. There is a plethora of exercises in any textbook worth its salt. I suppose those with some business in looking up something might hone in on what exercises to do. I have research to do as a PhD student and I still don't.
Should one do a kind of mathematical calisthenics, training daily with the mundane like doing scales for a musician or should one sink one's teeth into some impossible problem then let its foreplay and teasing guide the way?
NB: To avoid this becoming primarily opinion-based, let's concentrate on the work habits of the greats, through the mediocre, to the average Joe.