x86 atomic: optimize a.store(reg op a.load(acquire), release)
Summary: PR24191 finds that the expected memory-register operations aren't generated when relaxed { load ; modify ; store } is used. This is similar to PR17281 which was addressed in D4796, but only for memory-immediate operations (and for memory orderings up to acquire and release). This patch also handles some floating-point operations.
Reviewers: reames, kcc, dvyukov, nadav, morisset, chandlerc, t.p.northover, pete
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D11382
llvm-svn: 244128
diff --git a/llvm/lib/Target/X86/X86MCInstLower.cpp b/llvm/lib/Target/X86/X86MCInstLower.cpp
index 68977d3..2b30ce6 100644
--- a/llvm/lib/Target/X86/X86MCInstLower.cpp
+++ b/llvm/lib/Target/X86/X86MCInstLower.cpp
@@ -598,17 +598,29 @@
case X86::RELEASE_MOV32mi: OutMI.setOpcode(X86::MOV32mi); goto ReSimplify;
case X86::RELEASE_MOV64mi32: OutMI.setOpcode(X86::MOV64mi32); goto ReSimplify;
case X86::RELEASE_ADD8mi: OutMI.setOpcode(X86::ADD8mi); goto ReSimplify;
+ case X86::RELEASE_ADD8mr: OutMI.setOpcode(X86::ADD8mr); goto ReSimplify;
case X86::RELEASE_ADD32mi: OutMI.setOpcode(X86::ADD32mi); goto ReSimplify;
+ case X86::RELEASE_ADD32mr: OutMI.setOpcode(X86::ADD32mr); goto ReSimplify;
case X86::RELEASE_ADD64mi32: OutMI.setOpcode(X86::ADD64mi32); goto ReSimplify;
+ case X86::RELEASE_ADD64mr: OutMI.setOpcode(X86::ADD64mr); goto ReSimplify;
case X86::RELEASE_AND8mi: OutMI.setOpcode(X86::AND8mi); goto ReSimplify;
+ case X86::RELEASE_AND8mr: OutMI.setOpcode(X86::AND8mr); goto ReSimplify;
case X86::RELEASE_AND32mi: OutMI.setOpcode(X86::AND32mi); goto ReSimplify;
+ case X86::RELEASE_AND32mr: OutMI.setOpcode(X86::AND32mr); goto ReSimplify;
case X86::RELEASE_AND64mi32: OutMI.setOpcode(X86::AND64mi32); goto ReSimplify;
+ case X86::RELEASE_AND64mr: OutMI.setOpcode(X86::AND64mr); goto ReSimplify;
case X86::RELEASE_OR8mi: OutMI.setOpcode(X86::OR8mi); goto ReSimplify;
+ case X86::RELEASE_OR8mr: OutMI.setOpcode(X86::OR8mr); goto ReSimplify;
case X86::RELEASE_OR32mi: OutMI.setOpcode(X86::OR32mi); goto ReSimplify;
+ case X86::RELEASE_OR32mr: OutMI.setOpcode(X86::OR32mr); goto ReSimplify;
case X86::RELEASE_OR64mi32: OutMI.setOpcode(X86::OR64mi32); goto ReSimplify;
+ case X86::RELEASE_OR64mr: OutMI.setOpcode(X86::OR64mr); goto ReSimplify;
case X86::RELEASE_XOR8mi: OutMI.setOpcode(X86::XOR8mi); goto ReSimplify;
+ case X86::RELEASE_XOR8mr: OutMI.setOpcode(X86::XOR8mr); goto ReSimplify;
case X86::RELEASE_XOR32mi: OutMI.setOpcode(X86::XOR32mi); goto ReSimplify;
+ case X86::RELEASE_XOR32mr: OutMI.setOpcode(X86::XOR32mr); goto ReSimplify;
case X86::RELEASE_XOR64mi32: OutMI.setOpcode(X86::XOR64mi32); goto ReSimplify;
+ case X86::RELEASE_XOR64mr: OutMI.setOpcode(X86::XOR64mr); goto ReSimplify;
case X86::RELEASE_INC8m: OutMI.setOpcode(X86::INC8m); goto ReSimplify;
case X86::RELEASE_INC16m: OutMI.setOpcode(X86::INC16m); goto ReSimplify;
case X86::RELEASE_INC32m: OutMI.setOpcode(X86::INC32m); goto ReSimplify;