trans: don't misuse C_nil for ZSTs other than ().
`C_nil` is actually `C_null` for `()` so `TempRef::new_operand` was treating all ZSTs as `()`.
This should allow running Servo with `RUSTFLAGS=-Zorbit`, assuming there are no other bugs.
[MIR] Make scopes debuginfo-specific (visibility scopes).
Fixes#32949 by having MIR (visibility) scopes mimic the lexical structure.
Unlike #33235, this PR also removes all scopes without variable bindings.
Printing of scopes also changed, e.g. for:
```rust
fn foo(x: i32, y: i32) { let a = 0; let b = 0; let c = 0; }
```
Before my changes:
```rust
fn foo(arg0: i32, arg1: i32) -> () {
let var0: i32; // "x" in scope 1 at <anon>:1:8: 1:9
let var1: i32; // "y" in scope 1 at <anon>:1:16: 1:17
let var2: i32; // "a" in scope 3 at <anon>:1:30: 1:31
let var3: i32; // "b" in scope 6 at <anon>:1:41: 1:42
let var4: i32; // "c" in scope 9 at <anon>:1:52: 1:53
...
scope tree:
0 1 2 3 {
4 5
6 {
7 8
9 10 11
}
}
}
```
After my changes:
```rust
fn foo(arg0: i32, arg1: i32) -> () {
scope 1 {
let var0: i32; // "x" in scope 1 at <anon>:1:8: 1:9
let var1: i32; // "y" in scope 1 at <anon>:1:16: 1:17
scope 2 {
let var2: i32; // "a" in scope 2 at <anon>:1:30: 1:31
scope 3 {
let var3: i32; // "b" in scope 3 at <anon>:1:41: 1:42
scope 4 {
let var4: i32; // "c" in scope 4 at <anon>:1:52: 1:53
}
}
}
}
...
}
[MIR] Fix MIR trans edge cases that showed up on crater.
These fixes cover all of the [regressions found by crater](https://gist.github.com/nikomatsakis/88ce89ed06ef7f7f19bfd1e221d7f7ec) (for #34096).
Two of them were `Pair` edge cases (ZSTs and constants) causing LLVM assertions, the other one was causing stack overflows in debug scripts compiled in debug mode, due to the `fn_ret_cast` `alloca` ending up in a loop.
rustc: add ReErased to be used by trait selection, MIR and trans.
`ReErased` replaces `ReStatic` (i.e. `'static`) for erasing regions.
Using a distinct lifetime helps prevent accidental mix-ups between the two.
It also allows cleaner type printing (see test changes), including in symbol names:
```rust
str..pattern..CharSearcher$LT$$u27$static$GT$::drop.30560::h840c2f2afc03bbea // before
str..pattern..CharSearcher::drop.30561::h6bd31d2af614377a // after
```
Not that we should be producing symbols this way, but it's still better.
MSVC requires unwinding code to be split to a tree of *funclets*, where each funclet
can only branch to itself or to to its parent.
Luckily, the code we generates matches this pattern. Recover that structure in
an analyze pass and translate according to that.
this introduces a DropAndReplace terminator as a fix to #30380. That terminator
is suppsoed to be translated by desugaring during drop elaboration, which is
not implemented in this commit, so this breaks `-Z orbit` temporarily.
trans-collector: Assorted fixes and refactorings needed for making trans collector-driven.
As the title says. The messages on the individual commits should do a good job of explaining what they are about.
r? @nikomatsakis
[MIR trans] Optimize trans for biased switches
Currently, all switches in MIR are exhausitive, meaning that we can have
a lot of arms that all go to the same basic block, the extreme case
being an if-let expression which results in just 2 possible cases, be
might end up with hundreds of arms for large enums.
To improve this situation and give LLVM less code to chew on, we can
detect whether there's a pre-dominant target basic block in a switch
and then promote this to be the default target, not translating the
corresponding arms at all.
In combination with #33544 this makes unoptimized MIR trans of
nickel.rs as fast as using old trans and greatly improves the times for
optimized builds, which are only 30-40% slower instead of ~300%.
cc #33111
trans: Always lower to `frem`
Long ago LLVM unfortunately didn't handle the 32-bit MSVC case of `frem` where
it can't be lowered to `fmodf` because that symbol doesn't exist. That was since
fixed in http://reviews.llvm.org/D12099 (landed as r246615) and was released in
what appears to be LLVM 3.8. Now that we're using that branch of LLVM let's
remove our own hacks and help LLVM optimize a little better by giving it
knowledge about what we're doing.
Currently, all switches in MIR are exhausitive, meaning that we can have
a lot of arms that all go to the same basic block, the extreme case
being an if-let expression which results in just 2 possible cases, be
might end up with hundreds of arms for large enums.
To improve this situation and give LLVM less code to chew on, we can
detect whether there's a pre-dominant target basic block in a switch
and then promote this to be the default target, not translating the
corresponding arms at all.
In combination with #33544 this makes unoptimized MIR trans of
nickel.rs as fast as using old trans and greatly improves the times for
optimized builds, which are only 30-40% slower instead of ~300%.
cc #33111
Long ago LLVM unfortunately didn't handle the 32-bit MSVC case of `frem` where
it can't be lowered to `fmodf` because that symbol doesn't exist. That was since
fixed in http://reviews.llvm.org/D12099 (landed as r246615) and was released in
what appears to be LLVM 3.8. Now that we're using that branch of LLVM let's
remove our own hacks and help LLVM optimize a little better by giving it
knowledge about what we're doing.
