mk: Request -march=i686 on i686 Linux
Apparently the gcc on our dist bot is so old and/or obscure that the default
`-m32` switch doesn't think it can generate i686 code (or something like that).
The compiler-rt build system probes for the `__i686__` define in GCC to compile
for an i686 (vs i386) target, so this was failing on the bots.
This tweaks instead to pass `-march=i686` on i686-unknown-linux-gnu to C code to
ensure that we're compiling for i686 instead of i386. This should hopefully not
actually have an impact other than maybe doing some random optimization it
wasn't able to do so before. In theory this isn't making the target less
compatible as all Rust code is already compiled for i686.
Hopefully closes#34572
mk: Don't consider LLVM done until it's done
Currently if an LLVM build is interrupted *after* it creates the llvm-config
binary but before it's done it puts us in an inconsistent state where we think
LLVM is compiled but it's not actually. This tweaks our logic to only consider
LLVM done building once it's actually done building.
This should hopefully alleviate problems on the bots where if we interrupt at
the wrong time it doesn't corrupt the build directory.
rustbuild: Clean out tmp directory on `make clean`
Right now we generate error index information into this directory, but it's
never cleaned out. This means that if a build *bounces* because of something in
this directory it'll continue to cause all future builds to fail because the
relevant files are never removed.
Right now we generate error index information into this directory, but it's
never cleaned out. This means that if a build *bounces* because of something in
this directory it'll continue to cause all future builds to fail because the
relevant files are never removed.
Apparently the gcc on our dist bot is so old and/or obscure that the default
`-m32` switch doesn't think it can generate i686 code (or something like that).
The compiler-rt build system probes for the `__i686__` define in GCC to compile
for an i686 (vs i386) target, so this was failing on the bots.
This tweaks instead to pass `-march=i686` on i686-unknown-linux-gnu to C code to
ensure that we're compiling for i686 instead of i386. This should hopefully not
actually have an impact other than maybe doing some random optimization it
wasn't able to do so before. In theory this isn't making the target less
compatible as all Rust code is already compiled for i686.
Hopefully closes#34572
Currently if an LLVM build is interrupted *after* it creates the llvm-config
binary but before it's done it puts us in an inconsistent state where we think
LLVM is compiled but it's not actually. This tweaks our logic to only consider
LLVM done building once it's actually done building.
This should hopefully alleviate problems on the bots where if we interrupt at
the wrong time it doesn't corrupt the build directory.
The ./configure command in README.md's Building Documentation section was
missing the $ prefix. Add the prefix to be consistent with other commands in the
document.
This commit enhances the rustbuild support for testing Android to the same level
of parity as the makefiles. This involved:
* A new step to copy the standard library and other shared objects to the
emulator. This is injected as a dependency of all test suites for Android.
* Appropriate arguments are now passed through to compiletest to ensure that it
can run tests.
* When testing the standard library the test executables are probed for and
shipped to the emulator to run for each test.
* Fixing compilation of compiler-rt a bit
All support added here is modeled after what's found in the makefiles, just
translating one strategy to another. As an added bonus this commit adds support
for the "check" step to automatically run tests for all targets, and the
"check-target" step now runs all tests for a particular target, automatically
filtering the tests if the target is detected as a cross-compile.
Note that we don't (and probably won't) have a bot which is actually going to
exercise any of this just yet, but all tests have passed locally for me at
least.
configure: Fix cross-compiling LLVM for realz
Actually got it working this time, and it was again just a problem specifying
the llvm-tblgen binary. We need to point it at the $CFG_BUILD target's tblgen
and then we also needed to correct the path a bit.
For example root-relative links will now be rejected.
Also remove some exceptions which have since been fixed and fix a typo in
the broken redirect handling.
Actually got it working this time, and it was again just a problem specifying
the llvm-tblgen binary. We need to point it at the $CFG_BUILD target's tblgen
and then we also needed to correct the path a bit.
Disable debuginfo tests for a given blacklist of LLDB versions
Anyone having trouble with most LLDB tests failing on OSX, please report your LLDB version here so I can add it to the blacklist.
Blacklisted versions so far:
* lldb-350.*
cc @rust-lang/tools
cc @tedhorst @indutny @jonathandturner (people from the original bug report)
Fixes#32520.
Revert "skip double negation in const eval"
This reverts commit 735c018974.
fixes#34395
The original commit was based on a mis-understanding of the overflowing literal lint.
This needs to be ported to beta.
r? @eddyb
Forbid type parameters and global paths in macro invocations
Fixes#28558.
This is a [breaking-change]. For example, the following would break:
```rust
macro_rules! m { () => { () } }
fn main() {
m::<T>!(); // Type parameters are no longer allowed in macro invocations
::m!(); // Global paths are no longer allowed in macro invocations
}
```
Any breakage can be fixed by removing the type parameters or the leading `::` (respectively).
r? @eddyb
Give `ast::ExprKind::Paren` no-op expressions the same ids as their children.
Having `ast::ExprKind::Paren` expressions share ids with their children
- reduces the number of unused `NodeId`s in the hir map and
- guarantees that `tcx.map.expect_expr(ast_expr.id)` is the hir corresponding to `ast_expr`.
This fixes the bug from #34327, which was introduced in #33296 when I assumed the above guarantee.
r? @nrc
Previously, any function using a `ToSocketAddrs` input would fail if
passed a hostname that resolves to an address type different from the
ones recognized by Rust.
This also changes the `LookupHost` iterator to only include the known
address types, as a result, it doesn't have to return `Result`s any
more, which are likely misinterpreted as failed name lookups.
