e665e781dc
When expanding a load into two loads, use nuw for the add that computes the offset from the base of the second load, because the original load doesn't straddle the address space. It turns out there's already a dedicated helper function for doing this, `getObjectPtrOffset`. This is in target-independent code, however in practice it only seems to affact WebAssembly code, because WebAssembly load and store instructions' constant offsets don't perform wrapping, so constant folding often depends on the nuw flag being present. This was noticed in the development of #119204.
93 lines
3.8 KiB
LLVM
93 lines
3.8 KiB
LLVM
; RUN: llc < %s -mtriple=wasm32 -wasm-keep-registers | FileCheck %s --check-prefixes=WASM32
|
|
; NOTE: did not compile on wasm64 at the time the test was created!
|
|
|
|
define { i128, i8 } @muloti_test(i128 %l, i128 %r) unnamed_addr #0 {
|
|
; WASM32-LABEL: muloti_test
|
|
; WASM32: global.get $push16=, __stack_pointer
|
|
; WASM32: i32.const $push17=, 48
|
|
; WASM32: i32.sub $push38=, $pop16, $pop17
|
|
; WASM32: local.tee $push37=, 5, $pop38
|
|
; WASM32: global.set __stack_pointer, $pop37
|
|
; WASM32: local.get $push39=, 5
|
|
; WASM32: i32.const $push22=, 32
|
|
; WASM32: i32.add $push23=, $pop39, $pop22
|
|
; WASM32: local.get $push41=, 1
|
|
; WASM32: i64.const $push0=, 0
|
|
; WASM32: local.get $push40=, 3
|
|
; WASM32: i64.const $push36=, 0
|
|
; WASM32: call __multi3, $pop23, $pop41, $pop0, $pop40, $pop36
|
|
; WASM32: local.get $push42=, 5
|
|
; WASM32: i32.const $push20=, 16
|
|
; WASM32: i32.add $push21=, $pop42, $pop20
|
|
; WASM32: local.get $push44=, 4
|
|
; WASM32: i64.const $push35=, 0
|
|
; WASM32: local.get $push43=, 1
|
|
; WASM32: i64.const $push34=, 0
|
|
; WASM32: call __multi3, $pop21, $pop44, $pop35, $pop43, $pop34
|
|
; WASM32: local.get $push47=, 5
|
|
; WASM32: local.get $push46=, 2
|
|
; WASM32: i64.const $push33=, 0
|
|
; WASM32: local.get $push45=, 3
|
|
; WASM32: i64.const $push32=, 0
|
|
; WASM32: call __multi3, $pop47, $pop46, $pop33, $pop45, $pop32
|
|
; WASM32: local.get $push49=, 0
|
|
; WASM32: local.get $push48=, 5
|
|
; WASM32: i64.load $push1=, 32($pop48)
|
|
; WASM32: i64.store 0($pop49), $pop1
|
|
; WASM32: local.get $push53=, 0
|
|
; WASM32: local.get $push50=, 5
|
|
; WASM32: i64.load $push31=, 40($pop50)
|
|
; WASM32: local.tee $push30=, 3, $pop31
|
|
; WASM32: local.get $push51=, 5
|
|
; WASM32: i64.load $push3=, 0($pop51)
|
|
; WASM32: local.get $push52=, 5
|
|
; WASM32: i64.load $push2=, 16($pop52)
|
|
; WASM32: i64.add $push4=, $pop3, $pop2
|
|
; WASM32: i64.add $push29=, $pop30, $pop4
|
|
; WASM32: local.tee $push28=, 1, $pop29
|
|
; WASM32: i64.store 8($pop53), $pop28
|
|
; WASM32: local.get $push60=, 0
|
|
; WASM32: local.get $push54=, 2
|
|
; WASM32: i64.const $push27=, 0
|
|
; WASM32: i64.ne $push6=, $pop54, $pop27
|
|
; WASM32: local.get $push55=, 4
|
|
; WASM32: i64.const $push26=, 0
|
|
; WASM32: i64.ne $push5=, $pop55, $pop26
|
|
; WASM32: i32.and $push7=, $pop6, $pop5
|
|
; WASM32: local.get $push56=, 5
|
|
; WASM32: i64.load $push8=, 8($pop56)
|
|
; WASM32: i64.const $push25=, 0
|
|
; WASM32: i64.ne $push9=, $pop8, $pop25
|
|
; WASM32: i32.or $push10=, $pop7, $pop9
|
|
; WASM32: local.get $push57=, 5
|
|
; WASM32: i64.load $push11=, 24($pop57)
|
|
; WASM32: i64.const $push24=, 0
|
|
; WASM32: i64.ne $push12=, $pop11, $pop24
|
|
; WASM32: i32.or $push13=, $pop10, $pop12
|
|
; WASM32: local.get $push59=, 1
|
|
; WASM32: local.get $push58=, 3
|
|
; WASM32: i64.lt_u $push14=, $pop59, $pop58
|
|
; WASM32: i32.or $push15=, $pop13, $pop14
|
|
; WASM32: i32.store8 16($pop60), $pop15
|
|
; WASM32: local.get $push61=, 5
|
|
; WASM32: i32.const $push18=, 48
|
|
; WASM32: i32.add $push19=, $pop61, $pop18
|
|
; WASM32: global.set __stack_pointer, $pop19
|
|
|
|
start:
|
|
%0 = tail call { i128, i1 } @llvm.umul.with.overflow.i128(i128 %l, i128 %r) #2
|
|
%1 = extractvalue { i128, i1 } %0, 0
|
|
%2 = extractvalue { i128, i1 } %0, 1
|
|
%3 = zext i1 %2 to i8
|
|
%4 = insertvalue { i128, i8 } undef, i128 %1, 0
|
|
%5 = insertvalue { i128, i8 } %4, i8 %3, 1
|
|
ret { i128, i8 } %5
|
|
}
|
|
|
|
; Function Attrs: nounwind readnone speculatable
|
|
declare { i128, i1 } @llvm.umul.with.overflow.i128(i128, i128) #1
|
|
|
|
attributes #0 = { nounwind readnone uwtable }
|
|
attributes #1 = { nounwind readnone speculatable }
|
|
attributes #2 = { nounwind }
|