Implement rotate using funnel shift on LLVM >= 7
Implement the rotate_left and rotate_right operations using
llvm.fshl and llvm.fshr if they are available (LLVM >= 7).
Originally I wanted to expose the funnel_shift_left and
funnel_shift_right intrinsics and implement rotate_left and
rotate_right on top of them. However, emulation of funnel
shifts requires emitting a conditional to check for zero shift
amount, which is not necessary for rotates. I was uncomfortable
doing that here, as I don't want to rely on LLVM to optimize
away that conditional (and for variable rotates, I'm not sure it
can). We should revisit that question when we raise our minimum
version requirement to LLVM 7 and don't need emulation code
anymore.
Fixes#52457.
Do not attempt to ascribe projections out of a ty var
If we encounter `_` ascribed to structural pattern like `(a, b)`, just skip relate_types.
Fix#55552
Bubble up an overflow error so that rustdoc can ignore it
fixes#54524
Idk how to write a test for this, other than trying to minimize the entire diesel crate. If desirable I will do that.
Note that there are many other such overflow errors hiding out there. Should we try to proactively eliminate them or do we just whack-a-mole them?
cc @GuillaumeGomez
r? @nikomatsakis
When I fixed the previous mis-optimizations, I didn't realize there were
actually two different places where we mutate `callsites` and both of
them should have the same behavior.
As a result, if a function was inlined and that function contained
virtual function calls, they were incorrectly being inlined. I also
added a test case which covers this.
Update emscripten
This updates emscripten to 1.38.15, which is based on LLVM 6.0.1 and would allow us to drop code for handling LLVM 4.
The main issue I ran into is that exporting statics through `EXPORTED_FUNCTIONS` no longer works. As far as I understand exporting non-functions doesn't really make sense under emscripten anyway, so I've modified the symbol export code to not even try.
Closes#52323.
Rollup of 17 pull requests
Successful merges:
- #55576 (Clarify error message for -C opt-level)
- #55633 (Support memcpy/memmove with differing src/dst alignment)
- #55638 (Fix ICE in msg_span_from_free_region on ReEmpty)
- #55659 (rustc: Delete grouping logic from the musl target)
- #55719 (Sidestep link error from rustfix'ed code by using a *defined* static.)
- #55736 (Elide anon lifetimes in conflicting impl note)
- #55739 (Consume optimization fuel from the MIR inliner)
- #55742 (Avoid panic when matching function call)
- #55753 (borrow_set: remove a helper function and a clone it uses)
- #55755 (Improve creation of 3 IndexVecs)
- #55758 ([regression - rust2018]: unused_mut lint false positives on nightly)
- #55760 (Remove intermediate font specs)
- #55761 (mir: remove a hacky recursive helper function)
- #55774 (wasm32-unknown-emscripten expects the rust_eh_personality symbol)
- #55777 (Use `Lit` rather than `P<Lit>` in `ast::ExprKind`.)
- #55783 (Deprecate mpsc channel selection)
- #55788 (rustc: Request ansi colors if stderr isn't a tty)
Failed merges:
r? @ghost
rustc: Request ansi colors if stderr isn't a tty
Currently Cargo will always capture the output of rustc meaning that
rustc is never hooked up to a tty. To retain colors Cargo uses the
`fwdansi` crate to ensure that ansi color codes are translated to
windows terminal methods (and ansi codes otherwise just go their natural
route on Unix).
Cargo passes `--color always` to rustc to ensure that using a pipe
doesn't trick it into not emitting colors at all. It turns out, however,
that `--color always` ends up still accidentally using the native shell
api on native windows shells.
The fix here is to instead pass `AlwaysAnsi` to `termcolor` instead of
`Always`, ensuring that when `--color always` is passed to rustc and its
output isn't a terminal, we're always generating ansi colors regardless
of the platform.
Closes#55769
