diff options
| author | Rafael Espindola <rafael.espindola@gmail.com> | 2015-04-06 22:29:07 +0000 |
|---|---|---|
| committer | Rafael Espindola <rafael.espindola@gmail.com> | 2015-04-06 22:29:07 +0000 |
| commit | d58de064b8eb041efab796c37986718311d0ee7a (patch) | |
| tree | 333a9f2c6e5cc2098124b9beb4beac1b8907ebd8 /llvm/lib/CodeGen/SelectionDAG/FastISel.cpp | |
| parent | 5bf3cdc566dfd2a653aa1342a850a7bd3d451677 (diff) | |
| download | bcm5719-llvm-d58de064b8eb041efab796c37986718311d0ee7a.tar.gz bcm5719-llvm-d58de064b8eb041efab796c37986718311d0ee7a.zip | |
Use sext in fast isel.
Fast isel used to zero extends immediates to 64 bits. This normally goes
unnoticed because the value is truncated to 32 bits for output.
Two cases were it is noticed:
* We fail to use smaller encodings.
* If the original constant was smaller than i32.
In the tests using i1 constants, codegen would change to use -1, which is fine
(and matches what regular isel does) since only the lowest bit is then used.
Instead, this patch then changes the ir to use i8 constants, which looks more
like what clang produces.
llvm-svn: 234249
Diffstat (limited to 'llvm/lib/CodeGen/SelectionDAG/FastISel.cpp')
| -rw-r--r-- | llvm/lib/CodeGen/SelectionDAG/FastISel.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/llvm/lib/CodeGen/SelectionDAG/FastISel.cpp b/llvm/lib/CodeGen/SelectionDAG/FastISel.cpp index 8f504e152d0..cf26425074a 100644 --- a/llvm/lib/CodeGen/SelectionDAG/FastISel.cpp +++ b/llvm/lib/CodeGen/SelectionDAG/FastISel.cpp @@ -425,7 +425,7 @@ bool FastISel::selectBinaryOp(const User *I, unsigned ISDOpcode) { // Check if the second operand is a constant and handle it appropriately. if (const auto *CI = dyn_cast<ConstantInt>(I->getOperand(1))) { - uint64_t Imm = CI->getZExtValue(); + uint64_t Imm = CI->getSExtValue(); // Transform "sdiv exact X, 8" -> "sra X, 3". if (ISDOpcode == ISD::SDIV && isa<BinaryOperator>(I) && |
