global_asm! should document that it doesn't guarantee section placement

[Qix-]

Location

https://doc.rust-lang.org/stable/core/arch/macro.global_asm.html

Summary

Ran into a surprising bug today - global_asm! stuck some executable code into .rodata under release mode. The fix was simple - just needed to add a .section .text directive, but was surprising and neither the docs nor the book seem to mention this (maybe I missed it).

I documented my findings here: https://github.com/oro-os/development-notes/blob/master/Development%20Notes/x86/Scheduler%20Refactor%20(Mar%20'25).md#9-march-2025-fun-with-global_asm-linkage

Would you accept a PR to add a note about this?

[Noratrieb]

uhhhh, I don't think it's supposed to require this. in fact, I think it's undefined behavior to add such directives there.. something else must have been wrong in your program to cause this

[Qix-]

? Why would it be undefined behavior to add section directives? Is that documented somewhere?

[Noratrieb]

Not quite undefiend behavior, but forbidden nevertheless.

https://doc.rust-lang.org/nightly/reference/inline-assembly.html#r-asm.ts-args.llvm-syntax

These targets impose an additional restriction on the assembly code: any assembler state (e.g. the current section which can be changed with .section) must be restored to its original value at the end of the asm string. Assembly code that does not conform to the GAS syntax will result in assembler-specific behavior