SGX: Change ELF entrypoint
This fixes [rust-sgx issue #148](https://github.com/fortanix/rust-sgx/issues/148).
A new entry point is created for the ELF file generated by `rustc`, separate from the enclave entry point. When the ELF file is executed as a Linux binary, the error message below is written to stderr.
> Error: This file is an SGX enclave which cannot be executed as a standard Linux binary.
> See the installation guide at https://edp.fortanix.com/docs/installation/guide/ on how to use 'cargo run' or follow the steps at https://edp.fortanix.com/docs/tasks/deployment/ for manual deployment.
When the ELF file is converted to an SGXS using `elf2sgxs`, the old entry point is still set as the enclave entry point. In a future pull request in the rust-sgx repository, `elf2sgxs` will be modified to remove the code in the ELF entry point, since this code is not needed in the enclave.
Fix constant propagation for scalar pairs
We now only propagate a scalar pair if the Rvalue is a tuple with two scalars. This for example avoids propagating a (u8, u8) value when Rvalue has type `((), u8, u8)` (see the regression test). While this is a correct thing to do, implementation is tricky and will be done later.
Fixes#66971Fixes#66339Fixes#67019
We now only propagate a scalar pair if the Rvalue is a tuple with two
scalars. This for example avoids propagating a (u8, u8) value when
Rvalue has type `((), u8, u8)` (see the regression test). While this is
a correct thing to do, implementation is tricky and will be done later.
Fixes#66971Fixes#66339Fixes#67019
rustc: split FnAbi's into definitions/direct calls ("of_instance") and indirect calls ("of_fn_ptr").
After this PR:
* `InstanceDef::Virtual` is only used for "direct" virtual calls, and shims around those calls use `InstanceDef::ReifyShim` (i.e. for `<dyn Trait as Trait>::f as fn(_)`)
* this could easily be done for intrinsics as well, to allow their reification, but I didn't do it
* `FnAbi::of_instance` is **always** used for declaring/defining an `fn`, and for direct calls to an `fn`
* this is great for e.g. https://github.com/rust-lang/rust/pull/65881 (`#[track_caller]`), which can introduce the "caller location" argument into "codegen signatures" by only changing `FnAbi::of_instance`, after this PR
* `FnAbi::of_fn_ptr` is used primarily for indirect calls, i.e. to `fn` pointers
* *not* virtual calls (which use `FnAbi::of_instance` with `InstanceDef::Virtual`)
* there's also a couple uses where the `rustc_codegen_llvm` needs to declare (i.e. FFI-import) an LLVM function that has no Rust declaration available at all
* at least one of them could probably be a "weak lang item" instead
As there are many steps, this PR is best reviewed commit by commit - some of which arguably should be in their own PRs, I may have gotten carried away a bit.
cc @nagisa @rkruppe @oli-obk @anp
For SGX, the relocation using the relocation table is done by
the code in rust/src/libstd/sys/sgx/abi/reloc.rs and this code
should not require relocation. Setting RelaxELFRelocations flag
if allows this to happen, hence adding a Target Option for it.
introduce a target to build the kernel of the unikernel HermitCore
We are developing the unikernel HermitCore, where the kernel is written in Rust and is already supported by the Rust Standard Library. To compile the kernel with the new build flag "-Z build-std", we introduce a new target, which avoids the usage of SSE & AVX within the kernel.
Add riscv64gc-unknown-linux-gnu target
This PR add the target, but doesn't build std on CI yet.
I have a port for `libc` crate and std which I will upstream soon after this target is added.
r? @alexcrichton
Add a proc-macro to derive HashStable in librustc dependencies
A second proc-macro is added to derive HashStable for crates librustc depends on.
This proc-macro HashStable_Generic (to bikeshed) allows to decouple code and some librustc's boilerplate.
Not everything is migrated, because `Span` and `TokenKind` require to be placed inside librustc.
Types using them stay there too.
Split out of #66279
r? @Zoxc
Drop long-section-names linker workaround for windows-gnu
If we can trust objdump Rust doesn't emit sections loaded at runtime longer than 8 characters on windows-gnu (but still does on linux-gnu), debug sections are not affected by that limit.
I've ran tests and built few crates using exactly the same mingw-w64 version as Rusts CI just fine using **x86_64** toolchain.
The motivation for this change is making LLD work (it doesn't support `--enable-long-section-names`) with this target without hacks.
Bit of history:
The behaviour of LD changed in Binutils 2.20 released on 2009-10-16 and `--enable-long-section-names` was added to return to the old non conformant behaviour. Looking at the comment I can only guess there was a bug fixed in newer versions.
This workaround was added in https://github.com/rust-lang/rust/pull/13315 half a decade ago.
Fix C aggregate-passing ABI on powerpc
The existing code (which looks like it was copied from MIPS) passes
aggregates by value in registers. This is wrong. According to the SVR4
powerpc psABI, all aggregates are passed indirectly.
See #64259 for more discussion, which addresses the ABI for the special
case of ZSTs (empty structs).
We also sever syntax's dependency on rustc_target as a result.
This should slightly improve pipe-lining.
Moreover, some cleanup is done in related code.
Add target thumbv7neon-unknown-linux-musleabihf
This is a copy of thumbv7neon-unknown-linux-gnueabihf with musl changes
merged from armv7-unknown-linux-musleabihf. This appears to have been
missed when adding the other ARMv7-A thumb targets.
The existing code (which looks like it was copied from MIPS) passes
aggregates by value in registers. This is wrong. According to the SVR4
powerpc psABI, all aggregates are passed indirectly.
See #64259 for more discussion, which addresses the ABI for the special
case of ZSTs (empty structs).
Allow specifying LLVM's MCTargetOptions::ABIName in target specification files
This addresses #65024, as it allows RISC-V target specification files to set `"llvm-abiname": "lp64d"`.
Other languages (read: C) usually expose this codegen parameter under a compiler argument like `-mabi=<XYZ>`.
This addresses #65024, as it allows RISC-V target specification
files to set "llvm-abiname": "lp64d". In general, it is useful
for the programmer to be able to set this codegen parameter,
which other languages usually expose under a compiler argument
like "-mabi=<XYZ>".
Re-enable Emscripten's exception handling support
Passes LLVM codegen and Emscripten link-time flags for exception
handling if and only if the panic strategy is `unwind`. Sets the
default panic strategy for Emscripten targets to `unwind`. Re-enables
tests that depend on unwinding support for Emscripten, including
`should_panic` tests.
r? @alexcrichton
Add new EFIAPI ABI
Fixes#54527
Adds a new ABI, "efiapi", which reflects the calling convention as specified by [the current spec UEFI spec](https://uefi.org/sites/default/files/resources/UEFI%20Spec%202_7_A%20Sept%206.pdf#G6.999903). When compiling for x86_64, we should select the `win64` ABI, while on all other architectures (Itanium, x86, ARM and ARM64 and RISC-V), we should select the `C` ABI.
Currently, this is done by just turning it into the C ABI everywhere except on x86_64, where it's turned into the win64 ABI. Should we prevent this ABI from being used on unsupported architectures, and if so, how would this be done?
