Add __builtin_stack_address to clang to increase compatibility to gcc

[Febbe]

I want to get the stack address and gcc has the function __builtin_stack_address. It would be good, to also have this builtin in clang, it would on top increase compatibility to gcc.

[dtcxzyw]

We can reuse @llvm.sponentry https://llvm.org/docs/LangRef.html#llvm-sponentry-intrinsic. However it is only available on AArch64 and ARM.

Another way to implement this is using @llvm.read_register. It is implemented on ARM, AArch64, PowerPC and x86_64.

[llvmbot]

Hi!

This issue may be a good introductory issue for people new to working on LLVM. If you would like to work on this issue, your first steps are:

1. In the comments of the issue, request for it to be assigned to you.
2. Fix the issue locally.
3. Run the test suite locally. Remember that the subdirectories under test/ create fine-grained testing targets, so you can e.g. use make check-clang-ast to only run Clang's AST tests.
4. Create a Git commit.
5. Run git clang-format HEAD~1 to format your changes.
6. Open a pull request to the upstream repository on GitHub. Detailed instructions can be found in GitHub's documentation.

If you have any further questions about this issue, don't hesitate to ask via a comment in the thread below.

[llvmbot]

@llvm/issue-subscribers-good-first-issue

Author: Fabian Keßler (Febbe)

I want to get the stack address and gcc has the function [__builtin_stack_address](https://gcc.gnu.org/onlinedocs/gcc/Return-Address.html). It would be god, to also have this builtin in clang, it would on top increase compatibility to gcc.

[anematode]

Could I be assigned this one please?

[anematode]

@dtcxzyw Would you prefer there be an LLVM IR intrinsic llvm.stackaddress and then lower it in LLVM, or to just emit the appropriate llvm.read_register call in clang's CGBuiltins.cpp (and an error otherwise if on an unsupported architecture)? It seems that the first approach is taken for __builtin_frame_address etc., but those require complicated logic compared to getting the stack address.

Relatedly, it seems like llvm.frameaddress leads to a crash in ISel DAG->DAG for m68k: https://gcc.godbolt.org/z/rcnfEbPvY (I guess it's unimplemented), but all other targets on Godbolt seem to work.

[Febbe]

I am not a member of LLVM, but I think adding the intrinsic to LLVM would be cleaner, since it is more general.

[komalverma04]

@anematode are you still working on this?

[anematode]

@komalverma04 Ack, sorry. I think if the preferred solution is adding a new LLVM intrinsic it's probably not a good task for a newbie like myself. Could I be unassigned please?

[komalverma04]

@anematode I am also newbie. I would like to try it once. If you have worked on it, maybe we can do it together.