LLVM 18 no longer emits atomic load/store machine code for targets without s32c1i (LLVM-408)

[MabezDev]

Checklist

• Checked the issue tracker for similar issues to ensure this is not a duplicate
• Read the documentation to confirm the issue is not addressed there and your configuration is set correctly
• Tested with the latest version to ensure the issue hasn't been fixed

How often does this bug occurs?

often

Expected behavior

Atomic load/stores to function on the esp32s2

Actual behavior (suspected bug)

We sometimes get linker errors on the s2: https://github.com/esp-rs/esp-hal/actions/runs/11518396724/job/32065069522, in the Rust target we set max atomic width to 32, but in here it gets set to 0:

llvm-project/llvm/lib/Target/Xtensa/XtensaISelLowering.cpp

Lines 440 to 445 in 256abfe

	if (Subtarget.hasS32C1I()) {
	setMaxAtomicSizeInBitsSupported(32);
	setMinCmpXchgSizeInBits(32);
	} else {
	setMaxAtomicSizeInBitsSupported(0);
	}

.

This heavily relates to LLVM-175 and we should now take the time to implement it and follow what the RISCV backend does:

llvm-project/llvm/lib/Target/RISCV/RISCVISelLowering.cpp

Lines 631 to 638 in 256abfe

	if (Subtarget.hasStdExtA()) {
	setMaxAtomicSizeInBitsSupported(Subtarget.getXLen());
	setMinCmpXchgSizeInBits(32);
	} else if (Subtarget.hasForcedAtomics()) {
	setMaxAtomicSizeInBitsSupported(Subtarget.getXLen());
	} else {
	setMaxAtomicSizeInBitsSupported(0);
	}

Error logs or terminal output

No response

Steps to reproduce the behavior

1. Clone esp-hal
2. run cargo xtask run-example examples esp32s2 embassy_usb_serial in the root
3. Look at linker errors

Project release version

latest 18 release branch

System architecture

Intel/AMD 64-bit (modern PC, older Mac)

Operating system

Linux

Operating system version

Gentoo

Shell

ZSH

Additional context

No response