-
Notifications
You must be signed in to change notification settings - Fork 12.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Resolve FIXME in libcore/iter/mod.rs #56630
Conversation
r? @kennytm (rust_highfive has picked a reviewer for you, use r? to override) |
_ => unreachable!(), | ||
} | ||
let iter = &mut self.iter; | ||
self.peeked.get_or_insert_with(|| iter.next()).as_ref() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Seemingly LLVM cannot optimize out the _ => unreachable!
, so this improves codegen.
Diff of LLVM IR with opt-level=3
pub fn foo(iter: &mut Peekable<vec::IntoIter<i32>>) -> Option<&i32> {
iter.peek()
}
--- test.old.ll 2018-12-08 22:14:35.424253110 +0900
+++ test.ll 2018-12-08 22:14:43.778324073 +0900
@@ -9,73 +9,56 @@
%"unwind::libunwind::_Unwind_Exception" = type { [0 x i64], i64, [0 x i64], void (i32, %"unwind::libunwind::_Unwind_Exception"*)*, [0 x i64], [6 x i64], [0 x i64] }
%"unwind::libunwind::_Unwind_Context" = type { [0 x i8] }
-@0 = private unnamed_addr constant <{ [40 x i8] }> <{ [40 x i8] c"internal error: entered unreachable code" }>, align 1
-@1 = private unnamed_addr constant <{ [23 x i8] }> <{ [23 x i8] c"src/libcore/iter/mod.rs" }>, align 1
-@2 = private unnamed_addr constant <{ i8*, [8 x i8], i8*, [16 x i8] }> <{ i8* getelementptr inbounds (<{ [40 x i8] }>, <{ [40 x i8] }>* @0, i32 0, i32 0, i32 0), [8 x i8] c"(\00\00\00\00\00\00\00", i8* getelementptr inbounds (<{ [23 x i8] }>, <{ [23 x i8] }>* @1, i32 0, i32 0, i32 0), [16 x i8] c"\17\00\00\00\00\00\00\00\B4\07\00\00\12\00\00\00" }>, align 8
-
; test::foo
-; Function Attrs: nonlazybind uwtable
-define align 4 dereferenceable_or_null(4) i32* @_ZN4test3foo17h48910d518edefbc0E(%"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* dereferenceable(40) %iter) unnamed_addr #0 personality i32 (i32, i32, i64, %"unwind::libunwind::_Unwind_Exception"*, %"unwind::libunwind::_Unwind_Context"*)* @rust_eh_personality {
+; Function Attrs: norecurse nounwind nonlazybind uwtable
+define align 4 dereferenceable_or_null(4) i32* @_ZN4test3foo17h3aaa37f182f5f74fE(%"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* dereferenceable(40) %iter) unnamed_addr #0 personality i32 (i32, i32, i64, %"unwind::libunwind::_Unwind_Exception"*, %"unwind::libunwind::_Unwind_Context"*)* @rust_eh_personality {
start:
- %.idx.i = getelementptr %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 3, i32 0
- %.idx.val.i = load i32, i32* %.idx.i, align 4
- %0 = icmp eq i32 %.idx.val.i, 2
- br i1 %0, label %bb2.i, label %bb4.i
-
-bb2.i: ; preds = %start
- %1 = getelementptr inbounds %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 1, i32 7
- %2 = load i32*, i32** %1, align 8
- %3 = getelementptr inbounds %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 1, i32 9
- %4 = load i32*, i32** %3, align 8
- %5 = icmp eq i32* %2, %4
- br i1 %5, label %"_ZN80_$LT$alloc..vec..IntoIter$LT$T$GT$$u20$as$u20$core..iter..iterator..Iterator$GT$4next17hac55a8161e3fb5c3E.exit.i", label %bb2.i.i
+ %0 = getelementptr inbounds %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 3, i32 0
+ %1 = load i32, i32* %0, align 4, !range !1
+ %2 = icmp eq i32 %1, 2
+ br i1 %2, label %bb2.i.i, label %"start._ZN38_$LT$core..option..Option$LT$T$GT$$GT$18get_or_insert_with17h1b7503b69cd4dde4E.exit_crit_edge.i"
+
+"start._ZN38_$LT$core..option..Option$LT$T$GT$$GT$18get_or_insert_with17h1b7503b69cd4dde4E.exit_crit_edge.i": ; preds = %start
+ %.pre.i = getelementptr inbounds %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 3, i32 1
+ br label %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek17hd154f34ce665a0bcE.exit"
-bb2.i.i: ; preds = %bb2.i
- %6 = getelementptr inbounds i32, i32* %2, i64 1
- store i32* %6, i32** %1, align 8
- %.val.i.i = load i32, i32* %2, align 4
- br label %"_ZN80_$LT$alloc..vec..IntoIter$LT$T$GT$$u20$as$u20$core..iter..iterator..Iterator$GT$4next17hac55a8161e3fb5c3E.exit.i"
-
-"_ZN80_$LT$alloc..vec..IntoIter$LT$T$GT$$u20$as$u20$core..iter..iterator..Iterator$GT$4next17hac55a8161e3fb5c3E.exit.i": ; preds = %bb2.i.i, %bb2.i
- %_0.sroa.4.0.i.i = phi i32 [ %.val.i.i, %bb2.i.i ], [ undef, %bb2.i ]
- %_0.sroa.0.0.i.i = phi i32 [ 1, %bb2.i.i ], [ 0, %bb2.i ]
- store i32 %_0.sroa.0.0.i.i, i32* %.idx.i, align 8
- %7 = getelementptr inbounds %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 3, i32 1
- store i32 %_0.sroa.4.0.i.i, i32* %7, align 4
- br label %bb4.i
-
-bb4.i: ; preds = %"_ZN80_$LT$alloc..vec..IntoIter$LT$T$GT$$u20$as$u20$core..iter..iterator..Iterator$GT$4next17hac55a8161e3fb5c3E.exit.i", %start
- %8 = phi i32 [ %_0.sroa.0.0.i.i, %"_ZN80_$LT$alloc..vec..IntoIter$LT$T$GT$$u20$as$u20$core..iter..iterator..Iterator$GT$4next17hac55a8161e3fb5c3E.exit.i" ], [ %.idx.val.i, %start ]
- switch i32 %8, label %bb6.i [
- i32 2, label %bb5.i
- i32 0, label %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek17h89fcfd64b4417881E.exit"
- ]
-
-bb5.i: ; preds = %bb4.i
-; call core::panicking::panic
- tail call void @_ZN4core9panicking5panic17h2cd043862cc6dc4aE({ [0 x i64], { [0 x i8]*, i64 }, [0 x i64], { [0 x i8]*, i64 }, [0 x i32], i32, [0 x i32], i32, [0 x i32] }* noalias readonly dereferenceable(40) bitcast (<{ i8*, [8 x i8], i8*, [16 x i8] }>* @2 to { [0 x i64], { [0 x i8]*, i64 }, [0 x i64], { [0 x i8]*, i64 }, [0 x i32], i32, [0 x i32], i32, [0 x i32] }*))
- unreachable
-
-bb6.i: ; preds = %bb4.i
+bb2.i.i: ; preds = %start
+ %3 = getelementptr inbounds %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 1, i32 7
+ %4 = load i32*, i32** %3, align 8
+ %5 = getelementptr inbounds %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 1, i32 9
+ %6 = load i32*, i32** %5, align 8
+ %7 = icmp eq i32* %4, %6
+ br i1 %7, label %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek28_$u7b$$u7b$closure$u7d$$u7d$17h5d8c39d83231aa35E.exit.i.i", label %bb2.i.i.i.i
+
+bb2.i.i.i.i: ; preds = %bb2.i.i
+ %8 = getelementptr inbounds i32, i32* %4, i64 1
+ store i32* %8, i32** %3, align 8
+ %.val.i.i.i.i = load i32, i32* %4, align 4
+ br label %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek28_$u7b$$u7b$closure$u7d$$u7d$17h5d8c39d83231aa35E.exit.i.i"
+
+"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek28_$u7b$$u7b$closure$u7d$$u7d$17h5d8c39d83231aa35E.exit.i.i": ; preds = %bb2.i.i.i.i, %bb2.i.i
+ %_0.sroa.4.0.i.i.i.i = phi i32 [ %.val.i.i.i.i, %bb2.i.i.i.i ], [ undef, %bb2.i.i ]
+ %_0.sroa.0.0.i.i.i.i = phi i32 [ 1, %bb2.i.i.i.i ], [ 0, %bb2.i.i ]
+ store i32 %_0.sroa.0.0.i.i.i.i, i32* %0, align 4
%9 = getelementptr inbounds %"core::iter::Peekable<alloc::vec::IntoIter<i32>>", %"core::iter::Peekable<alloc::vec::IntoIter<i32>>"* %iter, i64 0, i32 3, i32 1
- br label %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek17h89fcfd64b4417881E.exit"
+ store i32 %_0.sroa.4.0.i.i.i.i, i32* %9, align 4
+ br label %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek17hd154f34ce665a0bcE.exit"
-"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek17h89fcfd64b4417881E.exit": ; preds = %bb4.i, %bb6.i
- %_0.0.i = phi i32* [ %9, %bb6.i ], [ null, %bb4.i ]
- ret i32* %_0.0.i
+"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek17hd154f34ce665a0bcE.exit": ; preds = %"start._ZN38_$LT$core..option..Option$LT$T$GT$$GT$18get_or_insert_with17h1b7503b69cd4dde4E.exit_crit_edge.i", %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek28_$u7b$$u7b$closure$u7d$$u7d$17h5d8c39d83231aa35E.exit.i.i"
+ %.pre-phi.i = phi i32* [ %.pre.i, %"start._ZN38_$LT$core..option..Option$LT$T$GT$$GT$18get_or_insert_with17h1b7503b69cd4dde4E.exit_crit_edge.i" ], [ %9, %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek28_$u7b$$u7b$closure$u7d$$u7d$17h5d8c39d83231aa35E.exit.i.i" ]
+ %10 = phi i32 [ %1, %"start._ZN38_$LT$core..option..Option$LT$T$GT$$GT$18get_or_insert_with17h1b7503b69cd4dde4E.exit_crit_edge.i" ], [ %_0.sroa.0.0.i.i.i.i, %"_ZN38_$LT$core..iter..Peekable$LT$I$GT$$GT$4peek28_$u7b$$u7b$closure$u7d$$u7d$17h5d8c39d83231aa35E.exit.i.i" ]
+ %switch.i.i = icmp eq i32 %10, 1
+ %_0.0.i.i = select i1 %switch.i.i, i32* %.pre-phi.i, i32* null
+ ret i32* %_0.0.i.i
}
; Function Attrs: nonlazybind uwtable
declare i32 @rust_eh_personality(i32, i32, i64, %"unwind::libunwind::_Unwind_Exception"*, %"unwind::libunwind::_Unwind_Context"*) unnamed_addr #1
-; core::panicking::panic
-; Function Attrs: cold noinline noreturn nonlazybind uwtable
-declare void @_ZN4core9panicking5panic17h2cd043862cc6dc4aE({ [0 x i64], { [0 x i8]*, i64 }, [0 x i64], { [0 x i8]*, i64 }, [0 x i32], i32, [0 x i32], i32, [0 x i32] }* noalias readonly dereferenceable(40)) unnamed_addr #2
-
-attributes #0 = { nonlazybind uwtable "probe-stack"="__rust_probestack" }
+attributes #0 = { norecurse nounwind nonlazybind uwtable "probe-stack"="__rust_probestack" }
attributes #1 = { nonlazybind uwtable "probe-stack"="__rust_probestack" "target-cpu"="x86-64" }
-attributes #2 = { cold noinline noreturn nonlazybind uwtable "probe-stack"="__rust_probestack" }
!llvm.module.flags = !{!0}
!0 = !{i32 2, !"RtLibUseGOT", i32 1}
+!1 = !{i32 0, i32 3}
name old ns/iter new ns/iter diff ns/iter diff % speedup iter::bench_cycle_take_ref_sum 927,152 927,194 42 0.00% x 1.00 iter::bench_cycle_take_sum 938,129 603,492 -334,637 -35.67% x 1.55
Thanks! @bors r+ rollup |
📌 Commit 5728a04 has been approved by |
Resolve FIXME in libcore/iter/mod.rs and makes a few improvements.
☀️ Test successful - status-appveyor, status-travis |
See #56883 about a bug we found, if you have time. |
and makes a few improvements.