rustc: Disable NEON on armv7 android.
We thought Google's ABI for arvm7 required neon, but it is
currently optional, perhaps because there is a significant
population of Tegra 2 devices still in use.
This turns off neon code generation outside #[target-feature]
blocks just like we do on armv7-unknown-linux-gnu, but unlike
most other armv7 targets. LLVM defaults to +neon for this target,
so an explicit disable is necessary.
See https://developer.android.com/ndk/guides/abis.html#v7a
for instruction set extension requirements.
Closes#38402.
llvm: backport r280651
fixes#38406
r? @alexcrichton
(I'm secretly hoping this will error in the same way as #38314. That would sort of confirm that the problem is OOM and not enabling the SPARC backend)
Don't perform span mangling when building field/tup access nodes
There are no guarantees that the two spans used to create the new one
come from the same place or are even valid.
Fixes#36081.
This not only avoids the small – and unnecessary – constant overhead for each compiler invocation,
but also helps somewhat by only having “correct” rustc processes to look for in `/proc/`.
This also makes the wrapper behave effectively as a regular exec wrapper its intended to be.
stdc++ is from base, and is an old library (GCC 4.2)
estdc++ is from ports, and is a recent library (GCC 4.9 currently)
as LLVM requires the newer version, use it if under OpenBSD.
struct field reordering and optimization
This is work in progress. The goal is to divorce the order of fields in source code from the order of fields in the LLVM IR, then optimize structs (and tuples/enum variants)by always ordering fields from least to most aligned. It does not work yet. I intend to check compiler memory usage as a benchmark, and a crater run will probably be required.
I don't know enough of the compiler to complete this work unaided. If you see places that still need updating, please mention them. The only one I know of currently is debuginfo, which I'm putting off intentionally until a bit later.
r? @eddyb
rustbuild: Fix LC_ID_DYLIB directives on OSX
Currently libraries installed by rustbuild on OSX have an incorrect
`LC_ID_DYLIB` directive located in the dynamic libraries that are
installed. The directive we expect looks like:
@rpath/libstd.dylib
Which means that if you want to find that dynamic library you should
look at the dylib's other `@rpath` directives. Typically our `@rpath`
directives look like `@loader_path/../lib` for the compiler as that's
where the installed libraries will be located. Currently, though,
rustbuild produces dylibs with the directive that looks like:
/Users/rustbuild/src/rust-buildbot/slave/nightly-dist-rustc-mac/build/build/x86_64-apple-darwin/stage1-std/x86_64-apple-darwin/release/deps/libstd-713ad88203512705.dylib
In other words, the build directory is encoded erroneously. The compiler
already [knows how] to change this directive, but it only passes that
argument when `-C rpath` is also passed. The rustbuild system, however,
explicitly [does not pass] this option explicitly and instead bakes its
own. This logic then also erroneously didn't pass `-Wl,-install_name`
like the compiler.
[knows how]: 4a008cccaa/src/librustc_trans/back/linker.rs (L210-L214)
[does not pass]: 4a008cccaa/src/bootstrap/bin/rustc.rs (L133-L158)
To fix this regression this patch introduces a new `-Z` flag, `-Z
osx-rpath-install-name` which basically just forces the compiler to take
the previous `-install_name` branch when creating a dynamic library.
Hopefully we can sort out a better rpath story in the future, but for
now this "hack" should suffice in getting our nightly builds back to the
same state as before.
Closes#38430
rustbuild: Create directories in mingw dist
Previously we accidentally relied on the mingw dist step running last, but the
step just needed to ensure the directories were created.
Currently libraries installed by rustbuild on OSX have an incorrect
`LC_ID_DYLIB` directive located in the dynamic libraries that are
installed. The directive we expect looks like:
@rpath/libstd.dylib
Which means that if you want to find that dynamic library you should
look at the dylib's other `@rpath` directives. Typically our `@rpath`
directives look like `@loader_path/../lib` for the compiler as that's
where the installed libraries will be located. Currently, though,
rustbuild produces dylibs with the directive that looks like:
/Users/rustbuild/src/rust-buildbot/slave/nightly-dist-rustc-mac/build/build/x86_64-apple-darwin/stage1-std/x86_64-apple-darwin/release/deps/libstd-713ad88203512705.dylib
In other words, the build directory is encoded erroneously. The compiler
already [knows how] to change this directive, but it only passes that
argument when `-C rpath` is also passed. The rustbuild system, however,
explicitly [does not pass] this option explicitly and instead bakes its
own. This logic then also erroneously didn't pass `-Wl,-install_name`
like the compiler.
[knows how]: 4a008cccaa/src/librustc_trans/back/linker.rs (L210-L214)
[does not pass]: 4a008cccaa/src/bootstrap/bin/rustc.rs (L133-L158)
To fix this regression this patch introduces a new `-Z` flag, `-Z
osx-rpath-install-name` which basically just forces the compiler to take
the previous `-install_name` branch when creating a dynamic library.
Hopefully we can sort out a better rpath story in the future, but for
now this "hack" should suffice in getting our nightly builds back to the
same state as before.
Closes#38430
The current beta that rustc is bootstrapping from contains a bug in Cargo that
erroneously links to OpenSSL in /usr/local, but this is fixed in the most recent
1.14 beta, so let's use that.
There was an off-by-one error discovered by @tbelaire.
So, the two invariants we are enforcing are:
1. Run lengths are decreasing.
2. Sum of lengths of any two adjacent runs is less than the length of
their predecessor.
This commit changes the comment to be clearer and have correct bounds on `i`.
travis: Fix testing 32-bit OSX target
We passed --target when we meant to pass --build, meaning we tested only the
standard library for 32-bit, not the whole compiler like we intended.
This tiny patch merely applies @bluss's suggestion for how to get a more
informative error message when the feature check can't open a file, a
matter that had briefly annoyed the present author, leading to the
filing of #38417.
