Speed up APFloat division by using short division for small divisors.
Fixes#43828 (hopefully), by not doing long division bit-by-bit for small divisors.
When parsing the ~200,000 decimal float literals in the `tuple-stress` benchmark, this change brings roughly a 5x speed increase (from `0.6s` to `0.12s`), and the hottest instructions are native `div`s.
Profile queries
This PR implements the "profile queries" debugging feature described here:
https://github.com/rust-lang-nursery/rust-forge/blob/master/profile-queries.md
In particular, it implements the debugging flag `-Z profile-queries`
FYI: This PR is my second attempt at pushing these changes. My original PR required a rebase; I have now done that rebase manually, after messing up with git's "interactive" rebase support. The original (now closed/cancelled) PR is this one: https://github.com/rust-lang/rust/issues/43156
r? @nikomatsakis
L4Re Target: Add the needed Libraries and locate them
Add the libraries and objects that have to be linked to a get working L4Re Binary using pre- and post-link-args. Additionaly some ld commands had to be passed.
* L4Re libraries and objects will be located by an environment variable.
* gcc libraries and objects will be located using a gcc call.
GCC is mandatory for this target, that might need documentation somewhere. As soon as something mandatory cannot be found, the compiler will panic. This is intended, because the functions involved don't allow the usage of a Result type. libgcc_eh is now passed using `-l` and crtbeginT.o and crtend.o are now located using `gcc -print-filename`.
Implement From<&[T]> and others for Arc/Rc (RFC 1845)
* Implements `From<`{`&[T]`, `&str`, `String`, `Box<T> where T: ?Sized`, `Vec<T>`}`>` for `Arc`/`Rc`
* Removes `rustc_private`-marked methods `Rc::__from_array` and `Rc::__from_str`, replacing their use with `Rc::from`
Tracking issue: #40475
- fixes evaluation of array length for zero-sized type referenced by
rvalue operand.
- adds test to verify fix.
Cause of the issue.
Zero-sized aggregates are handled as operands, not lvalues. Therefore while
visiting Assign statement by LocalAnalyser, mark_as_lvalue() is not called for
related Local. This behaviour is controlled by rvalue_creates_operand() method.
As result it causes error later, when rvalue operand is evaluated in
trans_rvalue_operand() while handling Rvalue::Len case. Array length evaluation
invokes trans_lvalue() which expects referenced Local to be value, not operand.
How it is fixed.
In certain cases result of Rvalue::Len can be evaluated without calling
trans_lvalue(). Method evaluate_array_len() is introduced to handle length
evaluation for zero-sized types referenced by Locals.
Clarify windows build instructions in README
The old wording made me think you were supposed to do `python x.py --build=msvc`, which is not the case. Specify that you need to use the target triple.
Mention null_mut on the pointer primitive docs.
Also adds a few mentions that both `*const` and `*mut` support functions, when only `*const` was mentioned before.
libproc_macro docs: fix brace and bracket mixup
The documentation indicates that brace is `[`.
Brace is mapped token::Brace which (expectedly) is `{`.
So the documentation is simply confusing brace and bracket there.
Even though it's just a very small issue, it can lead to quite some confusion.
Redox: correct is_absolute() and has_root()
This is awkward, but representing schemes properly in `Components` is not easily possible without breaking backwards compatibility, as discussed earlier in https://github.com/rust-lang/rust/pull/37702.
But these methods can be corrected anyway.
Point "deref coercions" links to new book
Currently the link on doc.rust-lang.org is semi-broken; it links to a page that links to the exact page in the first edition in the book, or to the index of the second edition of the book. If the second editions
is the recommended one now, we should point the links at that one. (In the mean time, the links have been updated to point directly to the first edition of the book, but that hasn't made it onto
the stable channel yet.) By the time this commit makes it onto the stable channel, the second edition of the book should be complete enough. At least the part about deref coercions is.
r? @steveklabnik
Support dynamically-linked and/or native musl targets
These changes allow native compilation on musl-based distributions and the use of dynamic libraries on linux-musl targets. This is intended to remove limitations based on past assumptions about musl targets, while maintaining existing behavior by default.
A minor related bugfix is included.
Before `#[must_use]` for functions was implemented, a `#[must_use]` attribute
on a function was a no-op. To avoid a breaking change in this behavior, we add
an option for "this-and-such feature is experimental" feature-gate messages to
be a mere warning rather than a compilation-halting failure (so old code that
used to have a useless no-op `#[must_use]` attribute now warns rather than
breaking). When we're on stable, we add a help note to clarify that the feature
isn't "on."
This is in support of #43302.
std: Respect formatting flags for str-like OsStr
Historically many `Display` and `Debug` implementations for `OsStr`-like
abstractions have gone through `String::from_utf8_lossy`, but this was updated
in #42613 to use an internal `Utf8Lossy` abstraction instead. This had the
unfortunate side effect of causing a regression (#43765) in code which relied on
these `fmt` trait implementations respecting the various formatting flags
specified.
This commit opportunistically adds back interpretation of formatting trait flags
in the "common case" where where `OsStr`-like "thing" is all valid utf-8 and can
delegate to the formatting implementation for `str`. This doesn't entirely solve
the regression as non-utf8 paths will format differently than they did before
still (in that they will not respect formatting flags), but this should solve
the regression for all "real world" use cases of paths and such. The door's also
still open for handling these flags in the future!
Closes#43765
We'll actually want a new "soft" warning-only gate to maintain
backwards-compatibility, but it's cleaner to start out with the established,
well-understood gate before implementing the alternative warn-only behavior in
a later commit.
This is in the matter of #43302.
Fix logic error in test guarding prototype MIR borrowck code.
Fix logic error in test guarding prototype MIR borrowck code.
tl;dr: 🤦
(This crept in during the shift from a transform to a query (#44009); I didn't notice because my muscle memory was still always passing `-Z mir-borrowck`, while my test cases *also* had the `#[rustc_mir_borrowck]` attribute attached to them.)
Now that musl supports dynamic libraries (although not by default)
enable the tests that now pass. Additional currently-ignored tests
will pass if rustc is built with crt_static=false in config.toml.
