RISC-V JALR: Handle cases where rd==rs1, and rd!=zero

[jeanmicheldeva]

Solves #6003

[CLAassistant]

All committers have signed the CLA.

[emesare]

Adding the unresolved branch

Before:

After:

[emesare]

So half of this PR was merged with #6213. Your other change, adding the unresolved branch to JALR, is being considered. I am not entirely sure if there was a reason why the branch was left out which is why I can't merge that part now.

[emesare]

For an example this is the issue with not adding a branch and what this PR would fix.

[emesare]

Talking to @rssor

so it was intentional, and when rd is ra especially we don't want it (those are, in fact, calls)
plt stubs are always going to be super bespoke and, generally speaking, we don't want calls to them to be showing up anyways (since they can't be meaningfully analyzed)

If you look at #6064 (comment) you can see an example of the function calls.

Looking at your comment #6003 (comment) it seems you are running into a different set of issues resulting from the omitted branch info, if you can share more information about what was fixed by this change maybe their is some alternative solution.

[emesare]

I am going to close this as done, the comment above explains my reasoning for not including your unresolved branch for the JALR instruction is in the above comments. If you have anything further please comment on this PR. Thanks again.