Bill Wendling 869d60d39d Machine sink could potentially sink instructions into a block where the physical ... registers it defines then interfere with an existing preg live range. For instance, if we had something like these machine instructions: BB#0 ... = imul ... EFLAGS<imp-def,dead> test ..., EFLAGS<imp-def> jcc BB#2 EFLAGS<imp-use> BB#1 ... ; fallthrough to BB#2 BB#2 ... ; No code that defines EFLAGS jcc ... EFLAGS<imp-use> Machine sink will come along, see that imul implicitly defines EFLAGS, but because it's "dead", it assumes that it can move imul into BB#2. But when it does, imul's "dead" imp-def of EFLAGS is raised from the dead (a zombie) and messes up the condition code for the jump (and pretty much anything else which relies upon it being correct). The solution is to know which pregs are live going into a basic block. However, that information isn't calculated at this point. Nor does the LiveVariables pass take into account non-allocatable physical registers. In lieu of this, we do a *very* conservative pass through the basic block to determine if a preg is live coming out of it. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@105387 91177308-0d34-0410-b5e6-96231b3b80d8		2010-06-03 07:54:20 +00:00
..
Analysis	Add a comment to this test.	2010-04-26 21:37:43 +00:00
Archive	fix a broken archive that was breaking dejagnu only (not lit)	2010-02-04 07:11:08 +00:00
Assembler	add newlines at the end of files.	2010-04-07 22:53:17 +00:00
Bindings/Ocaml	Add a LLVMWriteBitcodeToFD that exposes the raw_fd_ostream options.	2010-03-06 00:30:06 +00:00
Bitcode	reapply 'reject forward references to functions whose type don't match'	2010-04-20 04:49:11 +00:00
BugPoint
CodeGen	Machine sink could potentially sink instructions into a block where the physical	2010-06-03 07:54:20 +00:00
DebugInfo	DwarfWrite is now smart enough to drop debug value pointing to undefined register. Update this test to avoid this.	2010-06-01 23:01:43 +00:00
ExecutionEngine	Revert r101455, which fails on the llvm-arm-linux buildbot.	2010-04-16 18:37:31 +00:00
Feature	fix PR7105 by enumerating MDNodes on all @llvm.foo	2010-05-10 20:53:17 +00:00
FrontendAda
FrontendC	Add a test for the llvm-gcc commit in r90200.	2010-05-31 20:39:10 +00:00
FrontendC++	Revert 104841, 104842, 104876 due to buildbot failures. Radar 7424645.	2010-05-28 16:41:07 +00:00
FrontendFortran
FrontendObjC	Make sure this runs in 64-bit only, 32-bit won't produce the correct stores.	2010-03-25 01:46:07 +00:00
FrontendObjC++
Integer
lib	Eliminate llvmgcc_version testing variable.	2010-02-23 07:56:28 +00:00
Linker
LLVMC	XFAIL tests from LLVMC on valgrind or valgrind+leak-checking. We	2010-03-21 08:12:46 +00:00
MC	MC/X86: Add alias for movzx.	2010-05-28 21:20:21 +00:00
Other	Add lint checks for function attributes.	2010-05-28 21:43:57 +00:00
Scripts	Revert r98892. BSD systems may not have bash installed at all.	2010-03-19 00:32:11 +00:00
TableGen	XFAIL a new tblgen test for memory leak checking.	2010-03-27 04:59:47 +00:00
Transforms	Fix PR7272: when inlining through a callsite with byval arguments,	2010-05-31 21:00:26 +00:00
Unit	Try r96559 for the third time. This time the shared library is only built if	2010-02-25 06:34:33 +00:00
Verifier	Reapply address space patch after fixing an issue in MemCopyOptimizer.	2010-04-04 03:10:48 +00:00
CMakeLists.txt	Fix make check with cmake/lit	2010-03-10 18:41:59 +00:00
lit.cfg	Add support for XFAILing valgrind runs with memory leak checking independently	2010-03-20 23:08:45 +00:00
lit.site.cfg.in
Makefile	doh, didn't mean to check in my hackaround lit sucking. :)	2010-04-17 19:04:03 +00:00
Makefile.tests	Switch .bc/.ll Makefile rules to use LLVM{CC,CXX} instead of LLVMG{CC,XX}	2010-02-23 10:28:06 +00:00
site.exp.in	Eliminate llvmgcc_version testing variable.	2010-02-23 07:56:28 +00:00
TestRunner.sh