[llvm] [GlobalISel] Correct comment about type vs register class (PR #116083)
Daniel Sanders via llvm-commits
llvm-commits at lists.llvm.org
Wed Nov 13 09:41:07 PST 2024
https://github.com/dsandersllvm created https://github.com/llvm/llvm-project/pull/116083
Type and register class aren't mutually exclusive in gMIR but target
instructions don't have types on their operands that have register classes.
>From 4042038d00e80bbc94f63b21a976c341ba080aaa Mon Sep 17 00:00:00 2001
From: Daniel Sanders <daniel_l_sanders at apple.com>
Date: Tue, 12 Nov 2024 18:10:14 -0800
Subject: [PATCH 1/2] [GlobalISel] Move DemandedElt's APInt size assert after
isValid() check
This prevents the assertion from wrongly triggering on invalid LLT's
---
llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp | 18 +++++++++---------
1 file changed, 9 insertions(+), 9 deletions(-)
diff --git a/llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp b/llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp
index fefa8f2ea85942..6ae6d56f2ca51b 100644
--- a/llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp
+++ b/llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp
@@ -147,6 +147,15 @@ void GISelKnownBits::computeKnownBitsImpl(Register R, KnownBits &Known,
unsigned Opcode = MI.getOpcode();
LLT DstTy = MRI.getType(R);
+ // Handle the case where this is called on a register that does not have a
+ // type constraint (i.e. it has a register class constraint instead). This is
+ // unlikely to occur except by looking through copies but it is possible for
+ // the initial register being queried to be in this state.
+ if (!DstTy.isValid()) {
+ Known = KnownBits();
+ return;
+ }
+
#ifndef NDEBUG
if (DstTy.isFixedVector()) {
assert(
@@ -158,15 +167,6 @@ void GISelKnownBits::computeKnownBitsImpl(Register R, KnownBits &Known,
}
#endif
- // Handle the case where this is called on a register that does not have a
- // type constraint (i.e. it has a register class constraint instead). This is
- // unlikely to occur except by looking through copies but it is possible for
- // the initial register being queried to be in this state.
- if (!DstTy.isValid()) {
- Known = KnownBits();
- return;
- }
-
unsigned BitWidth = DstTy.getScalarSizeInBits();
auto CacheEntry = ComputeKnownBitsCache.find(R);
if (CacheEntry != ComputeKnownBitsCache.end()) {
>From 4321b949b88ef69217f1d3f64d514efc78b09c54 Mon Sep 17 00:00:00 2001
From: Daniel Sanders <daniel_l_sanders at apple.com>
Date: Wed, 13 Nov 2024 09:35:48 -0800
Subject: [PATCH 2/2] [GlobalISel] Correct comment about type vs register class
Type and register class aren't mutually exclusive in gMIR but target
instructions don't have types on their operands that have register classes.
---
llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp b/llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp
index 6ae6d56f2ca51b..179ac9deb3d4a5 100644
--- a/llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp
+++ b/llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp
@@ -148,9 +148,10 @@ void GISelKnownBits::computeKnownBitsImpl(Register R, KnownBits &Known,
LLT DstTy = MRI.getType(R);
// Handle the case where this is called on a register that does not have a
- // type constraint (i.e. it has a register class constraint instead). This is
- // unlikely to occur except by looking through copies but it is possible for
- // the initial register being queried to be in this state.
+ // type constraint (i.e. it's a target instruction with a register class
+ // constraint instead). This is unlikely to occur except by looking through
+ // copies but it is possible for the initial register being queried to be in
+ // this state.
if (!DstTy.isValid()) {
Known = KnownBits();
return;
More information about the llvm-commits
mailing list