| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
| |
patch by Micah Villmow for PR6572.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98188 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98181 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
| |
I'll get this loop right yet.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98155 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
| |
semantic instead of syntactic. This completes MCization of
darwin/x86[-64]!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98145 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98137 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98132 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
for darwin/x86 to be completely mcized.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98130 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98125 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98124 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
|
|
|
|
| |
1. Be careful with cse "cheap" expressions. e.g. constant materialization. Only cse them when the common expression is local or in a direct predecessor. We don't want cse of cheap instruction causing other expressions to be spilled.
2. Watch out for the case where the expression doesn't itself uses a virtual register. e.g. lea of frame object. If the common expression itself is used by copies (common for passing addresses to function calls), don't perform the cse. Since these expressions do not use a register, it creates a live range but doesn't close any, we want to be very careful with increasing register pressure.
Note these are heuristics so machine cse doesn't make register allocator unhappy. Once we have proper live range splitting and re-materialization support in place, these should be evaluated again.
Now machine cse is almost always a win on llvm nightly tests on x86 and x86_64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98121 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
dead IsPCRel argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98117 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
|
| |
need to be MCized, but the last debug info thing are LEB and
cygwin specific (which the MC api doesn't support yet) and
one specific form of EmitReference which I'll tackle next.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98116 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98109 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98108 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98107 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98106 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
method. With this, comments should end up on the same lines as the .byte
directives (for example) and we now get no output with:
$ llc CodeGen/X86/2009-02-12-DebugInfoVLA.ll -o - -filetype=null -asm-verbose
woot.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98105 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98104 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
off "EOL".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98102 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
| |
significant debug info testcases are now all going through
MCStreamer, though they print a lot of extraneous newlines to "O".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98101 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
|
|
| |
is preparatory to having PEI's scavenged frame index value reuse logic
properly distinguish types of frame values (e.g., whether the value is
stack-pointer relative or frame-pointer relative).
No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98086 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98048 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
| |
Now that setStartLabel takes an MCSymbol, we can de-ID'ize
beginScope and RecordSourceLine.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98047 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
failure remaining on mainline.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98046 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98045 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98044 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
|
|
| |
coalescer) handle sub-register classes.
- Add heuristics to avoid non-profitable cse. Given the current lack of live
range splitting, avoid cse when an expression has PHI use and the would be
new use is in a BB where the expression wasn't already being used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98043 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
failures.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98039 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
multibyte opcodes and in the case of multiple scopes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98036 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
| |
beginScope and RecordSourceLine.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98035 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
now that the dependence on ID is removed from MMI.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98034 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
happens, the start/end of a scope can never be the same.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98032 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
| |
and add a FIXME about how we are eventually going to zap this
lookup table once mc world domination is complete.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98031 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98025 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
physreg becomes ridiculously high.
std::upper_bound may be log(N), but for sufficiently large live intervals, it
becomes log(N)*cachemiss = a long long time.
This patch improves coalescer time by 4500x for a function with 20000
function calls. The generated code is different, but not significantly worse -
the allocator hints are almost as good as physreg coalescing anyway.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98023 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98017 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
recomputation of the labels.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98016 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98015 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
| |
the one special case into EmitSectionOffset. MCize
the non-special case in EmitSectionOffset.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98014 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98013 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
used for 'flavor'.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98012 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
we can eliminate "flavor".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98011 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
| |
MCize it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98010 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
| |
cse implicit-def for obvious performance reason.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98009 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
| |
to a real coalescer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98007 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
|
|
|
|
|
|
| |
available, the only thing this affects is that we produce
.set in one case we didn't before, which shouldn't harm
anything. Make EmitSectionOffset call EmitDifference
instead of duplicating it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98005 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98002 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97999 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97998 91177308-0d34-0410-b5e6-96231b3b80d8
|
| |
|
|
| |
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@97997 91177308-0d34-0410-b5e6-96231b3b80d8
|