- Notifications
You must be signed in to change notification settings - Fork 15.3k
[ConstantRange] Fix off by 1 bugs in UIToFP and SIToFP handling. #86041
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
Conversation
| @llvm/pr-subscribers-llvm-transforms @llvm/pr-subscribers-llvm-ir Author: Craig Topper (topperc) ChangesWe were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason #79158 was reverted. CC: @AtariDreams Full diff: https://github.com/llvm/llvm-project/pull/86041.diff 2 Files Affected:
diff --git a/llvm/lib/IR/ConstantRange.cpp b/llvm/lib/IR/ConstantRange.cpp index 3394a1ec8dc476..59e7a9f5eb1114 100644 --- a/llvm/lib/IR/ConstantRange.cpp +++ b/llvm/lib/IR/ConstantRange.cpp @@ -746,7 +746,7 @@ ConstantRange ConstantRange::castOp(Instruction::CastOps CastOp, Min = Min.zext(ResultBitWidth); Max = Max.zext(ResultBitWidth); } - return ConstantRange(std::move(Min), std::move(Max)); + return getNonEmpty(std::move(Min), std::move(Max) + 1); } case Instruction::SIToFP: { // TODO: use input range if available @@ -757,7 +757,7 @@ ConstantRange ConstantRange::castOp(Instruction::CastOps CastOp, SMin = SMin.sext(ResultBitWidth); SMax = SMax.sext(ResultBitWidth); } - return ConstantRange(std::move(SMin), std::move(SMax)); + return getNonEmpty(std::move(SMin), std::move(SMax) + 1); } case Instruction::FPTrunc: case Instruction::FPExt: diff --git a/llvm/unittests/IR/ConstantRangeTest.cpp b/llvm/unittests/IR/ConstantRangeTest.cpp index 34a162a5514e95..8ec120d70e99f6 100644 --- a/llvm/unittests/IR/ConstantRangeTest.cpp +++ b/llvm/unittests/IR/ConstantRangeTest.cpp @@ -2479,6 +2479,24 @@ TEST_F(ConstantRangeTest, castOps) { ConstantRange IntToPtr = A.castOp(Instruction::IntToPtr, 64); EXPECT_EQ(64u, IntToPtr.getBitWidth()); EXPECT_TRUE(IntToPtr.isFullSet()); + + ConstantRange UIToFP = A.castOp(Instruction::UIToFP, 16); + EXPECT_EQ(16u, UIToFP.getBitWidth()); + EXPECT_TRUE(UIToFP.isFullSet()); + + ConstantRange UIToFP2 = A.castOp(Instruction::UIToFP, 64); + ConstantRange B(APInt(64, 0), APInt(64, 65536)); + EXPECT_EQ(64u, UIToFP2.getBitWidth()); + EXPECT_EQ(B, UIToFP2); + + ConstantRange SIToFP = A.castOp(Instruction::SIToFP, 16); + EXPECT_EQ(16u, SIToFP.getBitWidth()); + EXPECT_TRUE(SIToFP.isFullSet()); + + ConstantRange SIToFP2 = A.castOp(Instruction::SIToFP, 64); + ConstantRange C(APInt(64, -32768), APInt(64, 32768)); + EXPECT_EQ(64u, SIToFP2.getBitWidth()); + EXPECT_EQ(C, SIToFP2); } TEST_F(ConstantRangeTest, binaryAnd) { |
Could you please add the float2int tests while you are at it? |
3054d11 to e3c5053 Compare
nikic left a comment
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.
LGTM, nice find!
| ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4 | ||
| ; RUN: opt < %s -passes=float2int -S | FileCheck %s | ||
| | ||
| define i32 @pr79158(i32 %x.i.0.x.i.0.x.0.x.0.x.0..i) { |
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.
Please adjust the variable name :)
e3c5053 to 7a62d5f Compare We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason llvm#79158 was reverted.
7a62d5f to 83bb4a3 Compare
Error: Command failed due to missing milestone. |
Error: Command failed due to missing milestone. |
Error: Command failed due to missing milestone. |
Error: Command failed due to missing milestone. |
Error: Command failed due to missing milestone. |
| Failed to cherry-pick: release/18.x https://github.com/llvm/llvm-project/actions/runs/8378684757 Please manually backport the fix and push it to your github fork. Once this is done, please create a pull request |
| /pull-request #86153 |
…m#86041) We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason llvm#79158 was reverted. (cherry picked from commit 1283646)
…m#86041) We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason llvm#79158 was reverted.
…m#86041) We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason llvm#79158 was reverted.
…m#86041) We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason llvm#79158 was reverted.
…m#86041) We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason llvm#79158 was reverted. (cherry picked from commit 1283646)
…m#86041) We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason llvm#79158 was reverted. (cherry picked from commit 1283646)
…m#86041) We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1. We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion. I believe this fixes at least some of the reason llvm#79158 was reverted. (cherry picked from commit 1283646)
We were passing the min and max values of the range to the ConstantRange constructor, but the constructor expects the upper bound to 1 more than the max value so we need to add 1.
We also need to use getNonEmpty so that passing 0, 0 to the constructor creates a full range rather than an empty range. And passing smin, smax+1 doesn't cause an assertion.
I believe this fixes at least some of the reason #79158 was reverted.
CC: @AtariDreams