This was fixed with the upgrade to LLVM 11 in #73526.
It seems extremely unlikey that this exact issue will ever reoccur,
since slight modifications to the code caused the crash to stop
happening. However, it can't hurt to have a test for it.
This commit is an attempted resurrection of #70458 where LLVM bitcode
emitted by rustc into rlibs is stored into object file sections rather
than in a separate file. The main rationale for doing this is that when
rustc emits bitcode it will no longer use a custom compression scheme
which makes it both easier to interoperate with existing tools and also
cuts down on compile time since this compression isn't happening.
The blocker for this in #70458 turned out to be that native linkers
didn't handle the new sections well, causing the sections to either
trigger bugs in the linker or actually end up in the final linked
artifact. This commit attempts to address these issues by ensuring that
native linkers ignore the new sections by inserting custom flags with
module-level inline assembly.
Note that this does not currently change the API of the compiler at all.
The pre-existing `-C bitcode-in-rlib` flag is co-opted to indicate
whether the bitcode should be present in the object file or not.
Finally, note that an important consequence of this commit, which is also
one of its primary purposes, is to enable rustc's `-Clto` bitcode
loading to load rlibs produced with `-Clinker-plugin-lto`. The goal here
is that when you're building with LTO Cargo will tell rustc to skip
codegen of all intermediate crates and only generate LLVM IR. Today
rustc will generate both object code and LLVM IR, but the object code is
later simply thrown away, wastefully.
Previously, rustc mandated that cdylibs could only link against rlibs as
dependencies (not dylibs).
This commit disables that restriction and tests that it works in a
simple case.
This commit introduces more dirty span manipulation into the compiler
in order to handle the various edge cases in moving/renaming the macro
import so it is at the root of the import.
This commit adds a test demonstrating the current behaviour when a macro
defined in a module with the `#[macro_export]` is imported from the
module rather than the crate root.
