Technologically, nothing. They've even been made before! (Japanese scientists even made a device that would let you taste things!)
The problem is, nobody actually wants to buy them so nobody is making them for people to not buy because that would be a waste of time and money. Knowing that death and sewers are super common in games, I can't say I would want smell-o-vision myself.
TL;DR it worked but was often considered a poor, synthetic, replacement for the real world scents. Some people liked it, but most seemed to dislike it.
"Smell-o-Vision" was predated by "AromaRama" and followed by "Odorama" and some other scratch and sniff attempts. Various motion or "4D" rides have also tried to incorporate smell, then often drop it.
Too complex, too many inputs needed, it would be a proprietary exploitation product nightmare like liquid ink paper printers, initially bringing such a product to market would make it cost a fortune, and it would need widespread adoption before the economy of scale could kick in.
It included 'odorama', a contraption that looked like a TV studio camera on dolly rails across the front of the stage, that sprayed the audience stands with about 6 different smells throughout the performance.
One of the smells was horse manure, which was a nervous shock for the audience. Yes it smelt like horse poo.
A very memorable performance, because of the smells, but also the exceptional company behind it.
For vision, we only have four different kinds of receptors, which can be stimulated by electromagnetic waves on a one-dimensional spectrum.
For smells, we have about 350 different kinds of receptors. Also, they can't easily be stimulated by electromagnetic waves, but only by molecules, which are much more difficult/costly to transport to their corresponding receptors.