diff options
author | Sanjay Patel <spatel@rotateright.com> | 2022-06-19 14:13:27 -0400 |
---|---|---|
committer | Sanjay Patel <spatel@rotateright.com> | 2022-06-19 15:12:19 -0400 |
commit | 4022551a154e01ca5347629fd429ca04b36801e0 (patch) | |
tree | df62de92410fbf72475f074cde2ef66c880358ec /llvm/lib/Analysis/ValueTracking.cpp | |
parent | bfb915ec8bd057c566772b6f620a5ad3f55e9e17 (diff) | |
download | llvm-4022551a154e01ca5347629fd429ca04b36801e0.zip llvm-4022551a154e01ca5347629fd429ca04b36801e0.tar.gz llvm-4022551a154e01ca5347629fd429ca04b36801e0.tar.bz2 |
[ValueTracking] recognize sub X, (X -nuw Y) as not overflowing
This extends a similar pattern from D125500 and D127754.
If we know that operand 1 (RHS) of a subtract is itself a
non-overflowing subtract from operand 0 (LHS), then the
final/outer subtract is also non-overflowing:
https://alive2.llvm.org/ce/z/Bqan8v
InstCombine uses this analysis to trigger a narrowing
optimization, so that is what the first changed test shows.
The last test models a motivating case from issue #48013.
In that example, we determine 'nuw' on the first sub from
the urem, then we determine that the 2nd sub can be narrowed,
and that leads to eliminating both subtracts.
here are still several missing subtract narrowing optimizations
demonstrated in the tests above the diffs shown here - those
should be handled in InstCombine with another set of patches.
Diffstat (limited to 'llvm/lib/Analysis/ValueTracking.cpp')
-rw-r--r-- | llvm/lib/Analysis/ValueTracking.cpp | 13 |
1 files changed, 10 insertions, 3 deletions
diff --git a/llvm/lib/Analysis/ValueTracking.cpp b/llvm/lib/Analysis/ValueTracking.cpp index a6c256f..4ae1c6d 100644 --- a/llvm/lib/Analysis/ValueTracking.cpp +++ b/llvm/lib/Analysis/ValueTracking.cpp @@ -4955,11 +4955,18 @@ OverflowResult llvm::computeOverflowForUnsignedSub(const Value *LHS, // X - (X % ?) // The remainder of a value can't have greater magnitude than itself, // so the subtraction can't overflow. + + // X - (X -nuw ?) + // In the minimal case, this would simplify to "?", so there's no subtract + // at all. But if this analysis is used to peek through casts, for example, + // then determining no-overflow may allow other transforms. + // TODO: There are other patterns like this. // See simplifyICmpWithBinOpOnLHS() for candidates. - if (match(RHS, m_URem(m_Specific(LHS), m_Value())) && - isGuaranteedNotToBeUndefOrPoison(LHS, AC, CxtI, DT)) - return OverflowResult::NeverOverflows; + if (match(RHS, m_URem(m_Specific(LHS), m_Value())) || + match(RHS, m_NUWSub(m_Specific(LHS), m_Value()))) + if (isGuaranteedNotToBeUndefOrPoison(LHS, AC, CxtI, DT)) + return OverflowResult::NeverOverflows; // Checking for conditions implied by dominating conditions may be expensive. // Limit it to usub_with_overflow calls for now. |