The issue is that it restricts from locking-down and securing the system with Write xor Execute memory. So it has system wide implication.
It has to do that anyway?
Which at this point most companies would rather save money and forbid JIT altogether.
Note that mainframes and micros have JIT environments that aren't at the same safety level as regular desktop PCs.
For example,
https://medium.com/@dhemanthc/ibm-i-architecture-how-timi-an...
JITs do not grant the ability to bypass any OS/system sandboxes. The lack of W^X doesn't do that, either. If a process opts out of W^X, such as to enable a JIT, it's voluntarily making itself less hardened, but at the end of the day this isn't any more meaningful than the program being allowed to be written in, say, C, which also voluntarily reduces the processes security hardening.
And my point about C is literally that even without a JIT, applications can still have arbitrary execution vulnerabilities.
A JIT intended to run untrusted code as part of a sandbox, like a browser, is a big risk. But that's because of the untrusted code part, not the JIT. By comparison, something like a Python or Java JIT is as near as makes no difference completely risk free. The JIT is working on exclusively "trusted" code. Same basic concept applies here with this database usage.
Only on operating systems that allow dynamic executable code.
I am refuting that statement by showing that not allowing dynamic executable code is not enough. It is irrelevant to that end that you can mitigate ROP by other means.
To the OS those are all identical scenarios, it's irrelevant what caused the arbitrary code execution to happen.