Yes, it appears we have a different idea of what narrative control is. Let me try to reframe this.
When I take an opportunity attack, this is part of the normal play loop -- I have the authority to declare this action for my PC, the GM has the authority to determine it's likelihood of success, and the GM has the authority to narrate the outcome. I, as a player, do not get to say that my attack is successful or what the result is -- this is outside my authority. I will normally have more agency here, in 5e, because of how the combat rules codify how the GM is expected to determine success, but the resolution of the action -- does my attack kill or dissuade the enemy? -- is still the GM's purview.
Now look at Wall of Stone. Here the spell says what it does. The only way the GM can gainsay this is to break the expectation of the game and introduce some arbitrary reason the spell doesn't function. So, the spell functions and the result is what I, the player, say it is with the scope of the spell. The GM is pretty hamstrung in determining this.
A better example would be something similar -- let's look at a declared action to befriend an NPC vs Charm Person. The former is entirely up to the GM in all ways -- I have no ability to control the outcome here as a player, I have zero authority. But, if I cast Charm Person, suddenly the GM is faced with having to come up with a reason that it can't work if they want to stop it (assuming we haven't hit any of the issues described in the spell (or other spells)). They have to go to the saving throw, and abide by this. On a success, I have the authority to say that that NPC is now friendly with my PC. There is no GM authority to narrate a different outcome.