RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-12-04 17:56:53 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

now that complexpatterns are all emitted at the end of the match

Browse Source 
sequence, just emit instruction predicates right before them.  This
exposes yet more factoring opportunitites, shrinking the X86 table 
to 79144 bytes.

llvm-svn: 97704
This commit is contained in:
Chris Lattner 2010-03-04 01:25:36 +00:00
parent 2bbca2de9e
commit 72df178091
1 changed files with 3 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
utils/TableGen/DAGISelMatcherGen.cpp
View File

@ -471,6 +471,9 @@ bool MatcherGen::EmitMatcherCode(unsigned Variant) {
if (Variant != 0) return true;
}
// Emit the matcher for the pattern structure and types.
EmitMatchCode(Pattern.getSrcPattern(), PatWithNoTypes);
// If the pattern has a predicate on it (e.g. only enabled when a subtarget
// feature is around, do the check).
// FIXME: This should get emitted after the match code below to encourage
@ -479,15 +482,11 @@ bool MatcherGen::EmitMatcherCode(unsigned Variant) {
// X86's MatchAddress.
if (!Pattern.getPredicateCheck().empty())
AddMatcher(new CheckPatternPredicateMatcher(Pattern.getPredicateCheck()));
// Emit the matcher for the pattern structure and types.
EmitMatchCode(Pattern.getSrcPattern(), PatWithNoTypes);
// Now that we've completed the structural type match, emit any ComplexPattern
// checks (e.g. addrmode matches). We emit this after the structural match
// because they are generally more expensive to evaluate and more difficult to
// factor.
// FIXME2: Can the patternpredicatematcher be moved to right before this??
for (unsigned i = 0, e = MatchedComplexPatterns.size(); i != e; ++i) {
const TreePatternNode *N = MatchedComplexPatterns[i].first;