Skip to content

[llvm][CodeGen] Fix failure in window scheduler caused by weak dependencies #95636

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

Merged
merged 2 commits into from
Jun 15, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion llvm/lib/CodeGen/WindowScheduler.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -422,6 +422,8 @@ int WindowScheduler::calculateMaxCycle(ScheduleDAGInstrs &DAG,
int ExpectCycle = CurCycle;
// The predecessors of current MI determine its earliest issue cycle.
for (auto &Pred : SU->Preds) {
if (Pred.isWeak())
continue;
auto *PredMI = Pred.getSUnit()->getInstr();
int PredCycle = getOriCycle(PredMI);
ExpectCycle = std::max(ExpectCycle, PredCycle + (int)Pred.getLatency());
Expand Down Expand Up @@ -479,7 +481,7 @@ int WindowScheduler::calculateStallCycle(unsigned Offset, int MaxCycle) {
auto *SU = TripleDAG->getSUnit(&MI);
int DefCycle = getOriCycle(&MI);
for (auto &Succ : SU->Succs) {
if (Succ.getSUnit() == &TripleDAG->ExitSU)
if (Succ.isWeak() || Succ.getSUnit() == &TripleDAG->ExitSU)
continue;
// If the expected cycle does not exceed MaxCycle, no check is needed.
if (DefCycle + (int)Succ.getLatency() <= MaxCycle)
Expand Down
39 changes: 39 additions & 0 deletions llvm/test/CodeGen/Hexagon/swp-ws-weak-dep.mir
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# REQUIRES: asserts
# RUN: llc --march=hexagon %s -run-pass=pipeliner -debug-only=pipeliner \
# RUN: -window-sched=force -filetype=null 2>&1 | FileCheck %s

# CHECK: SU(3): Ord Latency=0 Weak
# CHECK: SU(1): Ord Latency=0 Weak
# CHECK: Window scheduling is not needed!

---
name: khazad_setkey_in_key
tracksRegLiveness: true
body: |
bb.0:
successors: %bb.1(0x80000000)
liveins: $r0, $r1

%0:intregs = COPY $r1
%1:intregs = COPY $r0
J2_loop0i %bb.1, 1, implicit-def $lc0, implicit-def $sa0, implicit-def $usr

bb.1:
successors: %bb.2(0x04000000), %bb.1(0x7c000000)

%2:doubleregs = L2_loadrd_io %1, 0
%3:intregs = COPY %2.isub_lo
S2_storeri_io %0, 0, %3
%4:intregs = S2_asl_i_r %2.isub_hi, 2
%5:intregs = L2_loadri_io %4, 0
%6:intregs = S4_andi_asl_ri 4, %3, 1
%7:intregs = L2_loadri_io %6, 0
%8:intregs = A2_xor %7, %5
S2_storeri_io %1, 0, %8
ENDLOOP0 %bb.1, implicit-def $pc, implicit-def $lc0, implicit $sa0, implicit $lc0
J2_jump %bb.2, implicit-def dead $pc

bb.2:
PS_jmpret $r31, implicit-def dead $pc

...
Loading