[PATCH] D50789: [TableGen] Remove unnecessary TypeSetByHwMode -> ValueTypeByHwMode -> TypeSetByHwMode conversions in getPatternSize

Simon Pilgrim via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed Aug 15 10:16:42 PDT 2018


RKSimon created this revision.
RKSimon added reviewers: sdardis, kparzysz, rengolin, zturner, MatzeB, asb.

I noticed this during profiling of tablegen (PR28222) that we were calling Child->getType(0) which creates a ValueTypeByHwMode on the fly from the requested internal TypeSetByHwMode type and returns it by value, we then treat it as a TypeSetByHwMode reference which involves constructing a new TypeSetByHwMode on the stack with a large amount of std::map iterating/copying all along the way.

I am not an expert on tablegen so please check my working, but AFAICT this is all unnecessary and we should be calling Child->getExtType(0) which returns the original TypeSetByHwMode by reference.

This gives me a 90sec reduction in msvc debug builds of x86 -gen-dag-isel.


Repository:
  rL LLVM

https://reviews.llvm.org/D50789

Files:
  utils/TableGen/CodeGenDAGPatterns.cpp


Index: utils/TableGen/CodeGenDAGPatterns.cpp
===================================================================
--- utils/TableGen/CodeGenDAGPatterns.cpp
+++ utils/TableGen/CodeGenDAGPatterns.cpp
@@ -1278,7 +1278,7 @@
   for (unsigned i = 0, e = P->getNumChildren(); i != e; ++i) {
     const TreePatternNode *Child = P->getChild(i);
     if (!Child->isLeaf() && Child->getNumTypes()) {
-      const TypeSetByHwMode &T0 = Child->getType(0);
+      const TypeSetByHwMode &T0 = Child->getExtType(0);
       // At this point, all variable type sets should be simple, i.e. only
       // have a default mode.
       if (T0.getMachineValueType() != MVT::Other) {


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D50789.160840.patch
Type: text/x-patch
Size: 667 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180815/624c2a81/attachment.bin>


More information about the llvm-commits mailing list