This mostly copies cases that already exist in ValueTracking, although
it skips the more complex ones. Those can be filled in as needed.
Details
Details
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Unit Tests
Unit Tests
Time | Test | |
---|---|---|
230 ms | x64 debian > BOLT.X86::cfi-expr-rewrite.s |
Event Timeline
llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp | ||
---|---|---|
5063–5078 | I assume we checked operand 1 first to favor the cheap case of a non-zero constant right hand side. |
Comment Actions
Swap operand order. Fix some style nits
llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp | ||
---|---|---|
5063–5078 | Makes sense, updated that everywhere else. |
Comment Actions
Are we really hitting all these cases? We should really ensure we have test coverage for them all.
llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp | ||
---|---|---|
5078 | This looks like a bad merge? |
llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp | ||
---|---|---|
5078 | hmm? |
llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp | ||
---|---|---|
5078 | Sorry, I think I was miscounting brackets :) |
llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp | ||
---|---|---|
5078 | You where right! |
Comment Actions
ping. Dropped the assosiated setcc patch, but would like to get this through as an improvement to the helper as is.
I assume we checked operand 1 first to favor the cheap case of a non-zero constant right hand side.