[llvm][GVN] Propagate trunc nuw to i1 equalities
#143273
Conversation
|
Thank you for submitting a Pull Request (PR) to the LLVM Project! This PR will be automatically labeled and the relevant teams will be notified. If you wish to, you can add reviewers by using the "Reviewers" section on this page. If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers. If you have further questions, they may be answered by the LLVM GitHub User Guide. You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums. |
|
@llvm/pr-subscribers-llvm-transforms Author: Yash Solanki (yashnator) ChangesThis patch adds to GVN's Given: We can deduce that if The patch adds logic to propagate this information via the GVN worklist. This enables further simplification opportunities downstream, such as folding redundant stores or conditionals that depend on Includes a test case in Full diff: https://github.com/llvm/llvm-project/pull/143273.diff 2 Files Affected:
diff --git a/llvm/lib/Transforms/Scalar/GVN.cpp b/llvm/lib/Transforms/Scalar/GVN.cpp
index 77ca14f88a834..046918d8cf6a2 100644
--- a/llvm/lib/Transforms/Scalar/GVN.cpp
+++ b/llvm/lib/Transforms/Scalar/GVN.cpp
@@ -2579,6 +2579,22 @@ bool GVNPass::propagateEquality(Value *LHS, Value *RHS,
continue;
}
+
+ // Propagate equality that result from truncation with no unsigned wrap
+ // like (trunc nuw i64 %v to i1) == "true" or (trunc nuw i64 %v to i1) ==
+ // "false"
+ if (auto *Trunc = dyn_cast<TruncInst>(LHS)) {
+ if (Trunc->hasNoUnsignedWrap() && Trunc->getType()->isIntegerTy(1)) {
+ Value *Input = Trunc->getOperand(0);
+ if (ConstantInt *CI = dyn_cast<ConstantInt>(RHS)) {
+ if (CI->isZero()) {
+ Worklist.push_back({Input, ConstantInt::get(Input->getType(), 0)});
+ } else if (CI->isOne()) {
+ Worklist.push_back({Input, ConstantInt::get(Input->getType(), 1)});
+ }
+ }
+ }
+ }
}
return Changed;
diff --git a/llvm/test/Transforms/GVN/trunc-nuw-equality.ll b/llvm/test/Transforms/GVN/trunc-nuw-equality.ll
new file mode 100644
index 0000000000000..03d4a16aeaf20
--- /dev/null
+++ b/llvm/test/Transforms/GVN/trunc-nuw-equality.ll
@@ -0,0 +1,23 @@
+; RUN: opt -passes=gvn -S < %s | FileCheck %s
+
+define void @test(ptr %p, i64 %v) {
+; CHECK-LABEL: @test(
+; CHECK: %tr = trunc nuw i64 %v to i1
+; CHECK-NEXT: br i1 %tr, label %ret, label %store
+
+entry:
+ %tr = trunc nuw i64 %v to i1
+ br i1 %tr, label %ret, label %store
+
+store:
+ ; CHECK-LABEL: store:
+ ; CHECK: store i64 0, ptr %p
+ store i64 %v, ptr %p
+ ret void
+
+ret:
+ ; CHECK-LABEL: ret:
+ ; CHECK: store i64 1, ptr %p
+ store i64 %v, ptr %p
+ ret void
+}
|
yashnator
force-pushed
the
gvn-trunc-nuw
branch
4 times, most recently
from
4e7a479 to
30ede86
Compare
|
✅ With the latest revision this PR passed the C/C++ code formatter. |
If the result of `trunc nuw iN` is known to be true/false, the original value of the truncated integer can be inferred. This is folded under GVN's propagateEquality. Fixes missed optimization noted in llvm#142744
|
@yashnator here is one example for the not link probably some easier way to trigger it but can not think of any now and I'm on my way out, will merge this PR when I'm back |
|
@yashnator Congratulations on having your first Pull Request (PR) merged into the LLVM Project! Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR. Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues. How to do this, and the rest of the post-merge process, is covered in detail here. If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again. If you don't get any reports, no action is required from you. Your changes are working as expected, well done! |
This patch adds to GVN's
propagateEquality()to reason about equality constraints throughtrunc nuw iN to i1.Given:
%tr = trunc nuw iN %v to i1
We can deduce that if
%tr == true, then%v == 1, and if%tr == false, then%v == 0. This is valid becausenuwguarantees that truncation didn't lose unsigned bits, so%vmust have been either 0 or 1.The patch adds logic to propagate this information via the GVN worklist. This enables further simplification opportunities downstream, such as folding redundant stores or conditionals that depend on
%v.Includes a test case in
GVN/trunc-nuw-equality.ll.Resolves #142744