Add AArch64 support for RSEQ.

[avieira-arm]

Hi all,

This is a patch for initial RSEQ support for AArch64.

Tested on Ubuntu with 5.1.8+ kernel:
CC=clang bazel test -c dbg --copt "-mcpu=neoverse-n1" --linkopt "-fuse-ld=lld" --sandbox_debug --override_repository="tcmalloc=src/tcmalloc" -k //tcmalloc/..
CC=clang bazel test -c opt --copt "-mcpu=neoverse-n1" --linkopt "-fuse-ld=lld" --sandbox_debug --override_repository="tcmalloc=src/tcmalloc" -k //tcmalloc/..

On both occasions all tests ran and percpu_tcmalloc_test ran without skipping for slow path.

Is this OK?

I am currently looking at adding BTI to protect the branch targets and maybe even PAC for when we spill the LR onto the stack in START_RSEQ. Do you have any objections to this? I noticed the other ports don't use anything similar.

Another question I had is whether you might be open to re-using linux headers to specify these RSEQ sequences, like https://github.com/compudj/librseq/blob/master/include/rseq/rseq-arm64.h

I believe they lead to pretty much the same, but might help make it easier to rewrite the code in case things change in the future?

Kind regards,
Andre

[google-cla]

Thanks for your pull request. It looks like this may be your first contribution to a Google open source project (if not, look below for help). Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please visit https://cla.developers.google.com/ to sign.

Once you've signed (or fixed any issues), please reply here with @googlebot I signed it! and we'll verify it.

---

What to do if you already signed the CLA

Individual signers

• It's possible we don't have your GitHub username or you're using a different email address on your commit. Check your existing CLA data and verify that your email is set on your git commits.

Corporate signers

• Your company has a Point of Contact who decides which employees are authorized to participate. Ask your POC to be added to the group of authorized contributors. If you don't know who your Point of Contact is, direct the Google project maintainer to go/cla#troubleshoot (Public version).
• The email used to register you as an authorized contributor must be the email used for the Git commit. Check your existing CLA data and verify that your email is set on your git commits.
• The email used to register you as an authorized contributor must also be attached to your GitHub account.

ℹ️ Googlers: Go here for more info.

[resistor]

I'm not an expert on rseq, but I've highlighted a few areas where I believe the asm sequences could be tighter.

[ckennelly]

We'll also need to get the CLA fixed before merging.

[avieira-arm]

I have fixed the indentation problem pointed out by Owen. The other comments Owen provided are either blocked because of the need to not clobber argument registers in the critical section or me not having found a valid addressing mode that would improve performance.

[resistor]

LGTM

[ckennelly]

Nit: Let's note that we're using x4 for hdr

[ckennelly]

During some internal testing, I noticed a need to make sure TCMalloc is defining its own rseq-related macros (rather than requiring internal functions to follow TCMalloc's implementation's in lock-step). I should have a commit landed in the tree soon to rename those, but that'll require rebasing...

[avieira-arm]

I'll wait for that commit then.

[avieira-arm]

Had messed up the condition, now this should work, skip the test if the version < 9. For versions 9 or newer it will still check it is using the small memory model and error otherwise.