On Mon, Sep 25, 2017 at 11:00:11AM +0200, Jean Delvare wrote: > Then we have that in common. While reading the code and its history, I > was worried that the justification to add this warning in the first > place was technically weak. Not every coding error must automatically > translate to a patch to make the code robust against said error. > Sometimes you just have to admit that you did not pay attention as you > should have, fix your mistake, possibly document it for others, and > move on. Otherwise we end up with slow bloated code.
That WARN_ON() is a form of documentation. And if you care about performance for your code path, hide it under some CONFIG_*_DEBUG, but in general WARN_ON() isn't terribly expensive (depending entirely on the complexity of the condition of course).
