I would argue in that in the case of smoking, there is an overwhelming amount of evidence that suggests the activity harms others around you. Therefore, the public has the right to regulate the negative externality that this activity creates -- needless physical harm to others. The regulation of this activity creates virtually zero new negative externalities and is therefore justified.
This is in contrast to the other behavioral regulations you mentioned such as gay marriage, feeding the homeless, or even other drugs. The supposed negative externalities these activities produce have been viewed with more skepticism by the general public in recent years.
For example, the argument that gay marriage would somehow harm the way certain children were raised, has been mostly debunked. If sufficient evidence to the contrary would arise, evidence that was comparable to the quality of the evidence for second hand smoke, then the public would be justified in reevaluating the issue.
Fortunately, from a civil rights perspective, this is not the case.
In the case of non-smokable drugs you have to consider how much harm drug use causes vs. drug prohibition. It becomes clear that the negative externalities from the latter are much larger in scope and scale. Therefore, a conversation in ending drug prohibition is justified.
So, I would argue that whether or not a regulation on feeding the homeless is justified depends on the scope and scale of the negative externalities the activity creates, vs the scope and scale of negative externalities the regulation itself may or may not create.