| Age | Commit message (Collapse) | Author | Lines |
|---|
|
This schema is helpful for people writing custom target spec JSON. It
can provide autocomplete in the editor, and also serves as documentation
when there are documentation comments on the structs, as `schemars` will
put them in the schema.
|
|
|
|
It is always false nowadays. ThinLTO summary writing is instead done by
llvm_optimize.
|
|
|
|
|
|
A lot of places had special handling just in case they would get an
allocator module even though most of these places could never get one or
would have a trivial implementation for the allocator module. Moving all
handling of the allocator module to a single place simplifies things a
fair bit.
|
|
Making it participate in LTO would be incorrect if you compile a crate
as both a dylib (which needs it) and rlib (which must not include it) in
the same rustc invocation. With linker plugin LTO, the allocator shim
will still participate in LTO as it is safe to do so in that case.
|
|
|
|
Ensure we emit an allocator shim when only some crate types need one
Found this while trying to write a test for https://github.com/rust-lang/rust/pull/145955.
|
|
Revert "Make `lto` and `linker-plugin-lto` work the same for `compiler_builtins`
This reverts commit cf8753e4f9c3597f04cd5d3aa261e4561d5378a6 (PR https://github.com/rust-lang/rust/pull/145368) and fix the regressions reported at rust-lang/rust#145981, rust-lang/rust#146109, and rust-lang/rust#146145.
|
|
`compiler_builtins`"
This reverts commit cf8753e4f9c3597f04cd5d3aa261e4561d5378a6 and fixes the
regressions reported.
|
|
|
|
|
|
exported
Previously it would attempt to export the allocator shim even linking
for a crate type which pulls in the allocator shim from a dylib rather
than locally defining it.
|
|
|
|
|
|
Move exporting of profiler and sanitizer symbols to the LLVM backend
Only the LLVM backend needs those specific symbols exported and it only needs them to be exported for LTO, not from cdylibs in general.
|
|
CFI: Make `lto` and `linker-plugin-lto` work the same for `compiler_builtins`
Fix rust-lang/rust#142284 by ensuring that `#![no_builtins]` crates can still emit bitcode when proper (i.e., non-rustc) LTO (i.e., -Clinker-plugin-lto) is used.
|
|
Don't export them from cdylibs. There is no need to do so and it
complicates exported_non_generic_symbols. In addition the GCC backend
likely uses different symbols and may potentially not even need us to
explicitly tell it to export the symbols it needs.
|
|
Port `#[link]` to the new attribute parsing infrastructure
Ports `link` to the new attribute parsing infrastructure for https://github.com/rust-lang/rust/issues/131229#issuecomment-2971353197
|
|
Ensure the coordinator thread terminates before its channels drop
Fixes rust-lang/rust#142949
Explanation: https://github.com/rust-lang/rust/issues/142949#issuecomment-3224573185
|
|
|
|
|
|
Fix #142284 by ensuring that `#![no_builtins]` crates can still emit bitcode
when proper (i.e., non-rustc) LTO (i.e., -Clinker-plugin-lto) is used.
|
|
add a flag to codegen fn attrs for foreign items
r? `@ghost`
refiled to rerun CI
|
|
rustc_codegen_ssa: More comprehensive RISC-V ELF flags
This change implements more conformant, more comprehensive RISC-V ELF flags handling when generating certain object files directly from rustc.
* Use `"zca"` instead of `"c"`
The "Zca" extension (a subset of "C") is the minimal configuration for compressed instructions to set `EF_RISCV_RVC` flag.
* Set TSO flag from `"ztso"`
The "Ztso" extension denotes that the program depends on the RVTSO (Total Store Ordering) memory consistency model, which is stronger than the standard RVWMO (Weak Memory Ordering) consistency model and on ELF targets, we need to set `EF_RISCV_TSO` flag.
|
|
|
|
|
|
|
|
Don't show warnings from xcrun with -Zverbose-internals
These kinds of warnings can make our test suite fail spuriously, so if we want them, we'll need a different flag.
This was introduced in https://github.com/rust-lang/rust/pull/131477. Fixes https://github.com/rust-lang/rust/issues/145543.
r? apiraino
```@bors``` rollup
|
|
Handle unwinding fatal errors in codegen workers
Also directly unwind on fatal errors at the point they are emitted inside the codegen backends.
Fixes the coordinator ICE of https://github.com/rust-lang/rust/issues/132240, https://github.com/rust-lang/rust/issues/135075 and https://github.com/rust-lang/rust/issues/145800.
|
|
This commit implements more conformant, more comprehensive RISC-V ELF
flags handling when generating certain object files directly from rustc.
* Use "zca" instead of "c"
The "Zca" extension (a subset of "C") is the minimal configuration
for compressed instructions to set `EF_RISCV_RVC` flag.
* Set TSO flag from "ztso"
The "Ztso" extension denotes that the program depends on the RVTSO
(Total Store Ordering) memory consistency model, which is stronger
than the standard RVWMO (Weak Memory Ordering) consistency model and
on ELF targets, we need to set `EF_RISCV_TSO` flag.
|
|
|
|
These kinds of warnings can make our test suites fail, so if we want
them, we'll need a different flag.
|
|
Previously it indicated a single version, regardless of their count.
Observed in: https://github.com/davidlattimore/wild/pull/1041
|
|
As opposed to passing it around through Result.
|
|
|
|
Add VEXos "linked files" support to `armv7a-vex-v5`
Third-party programs running on the VEX V5 platform need a linker script to ensure code and data are always placed in the allowed range `0x3800000-0x8000000` which is read/write/execute. However, developers can also configure the operating system (VEXos) to preload a separate file at any location between these two addresses before the program starts (as a sort of basic linking or configuration loading system). Programs have to know about this at compile time - in the linker script - to avoid placing data in a spot that overlaps where the linked file will be loaded. This is a very popular feature with existing V5 runtimes because it can be used to modify a program's behavior without re-uploading the entire binary to the robot controller.
It's important for Rust to support this because while VEXos's runtime user-exposed file system APIs may only read data from an external SD card, linked files are allowed to load data directly from the device's onboard storage.
This PR adds the `__linked_file_start` symbol to the existing VEX V5 linker script which can be used to shrink the stack and heap so that they do not overlap with a memory region containing a linked file. It expects the linked file to be loaded in the final N bytes of user RAM (this is not technically required but every existing runtime does it this way to avoid having discontinuous memory regions).
With these changes, a developer targeting VEX V5 might add a second linker script to their project by specifying `-Clink-arg=-Tcustom.ld` and creating the file `custom.ld` to configure their custom memory layout. The linker would prepend this to the builtin target linker script.
```c
/* custom.ld: Reserves 10MiB for a linked file. */
/* (0x7600000-0x8000000) */
__linked_file_length = 10M;
/* The above line is equivalent to -Clink-arg=--defsym=__linked_file_length=10M */
/* Optional: specify one or more sections that */
/* represent the developer's custom format. */
SECTIONS {
.linked_file_metadata (NOLOAD) : {
__linked_file_metadata_start = .;
. += 1M;
__linked_file_metadata_end = .;
}
.linked_file_data (NOLOAD) : {
__linked_file_data_start = .;
. += 9M;
__linked_file_data_end = .;
}
} INSERT AFTER .stack;
```
Then, using an external tool like the `vex-v5-serial` crate, they would configure the metadata of their uploaded program to specify the path of their linked file and the address where it should be loaded into memory (in the above example, `0x7600000`).
|
|
|
|
Couple of minor cleanups
|
|
|
|
|
|
Make no_mangle on foreign items explicit instead of implicit
for a followup PR I'm working on I need some foreign items to mangle. I could add a new attribute: `no_no_mangle` or something silly like that but by explicitly putting `no_mangle` in the codegen fn attrs of foreign items we can default it to `no_mangle` and then easily remove it when we don't want it.
I guess you'd know about this r? `@bjorn3.` Shouldn't be too hard to review :)
Builds on rust-lang/rust#144655 which should merge first.
|
|
|
|
The exact reasoning why we do not always pass the SDK root when linking
on macOS eludes me, but I suspect it's because we want to support
compiler drivers which do not support the `-isysroot` option.
Since we now pass the SDK root via the environment variable SDKROOT,
compiler drivers that don't support it can just ignore it.
Similarly, since we only warn when xcrun fails, users that expect their
compiler driver to provide the SDK location can do so now.
|
|
To allow using zig-cc or similar as the compiler driver.
|
|
This is more in-line with what Apple's tooling expects, and allows us to
better support custom compiler drivers (such as certain Homebrew and
Nixpkgs compilers) that prefer their own `-isysroot` flag.
Effectively, we now invoke the compiler driver as-if it was invoked as
`xcrun -sdk $sdk_name $tool`.
|
|
Reduce indentation and avoid needless checks (checking the target OS and
vendor is unnecessary).
|
|
r=wesleywiser
Add diagnostic explaining STATUS_STACK_BUFFER_OVERRUN not only being used for stack buffer overruns if link.exe exits with that exit code
`STATUS_STACK_BUFFER_OVERRUN` is also used for fast abnormal program termination, e.g. by abort(). Emit a special diagnostic to let people know that this most likely doesn't indicate a stack buffer overrun.
This doesn't look up the crash report in the event log to determine what the fast fail error code is. This is due to the way crashes are logged: When a process crash happens, the system logs an "Application Error" event, which contains the exit code and the process ID, but not the fast fail error code. A second event by Windows Error Reporting does contain that fast fail code, but not the process ID - but that event is not emitted at process exit, but when WER has dealt with it (on my system, it happens roughly two seconds later), so querying the code would have to read the `IntegratorReportId`, wait two seconds or potentially longer for the WER event with the same `ReportID`, and read out the code. (Also, that second event doesn't happen if WER is disabled.)
Fixes rust-lang/rust#100519.
|
|
|
