Commit Graph

98360 Commits

Author SHA1 Message Date
Chris Lattner cb18bfa3d2 fix some issues Frits noticed, add AliasAnalysis as a dependency
llvm-svn: 122585
2010-12-27 18:39:08 +00:00
Rafael Espindola 1de2dd0e5e Add support for .cfi_lsda.
llvm-svn: 122584
2010-12-27 15:56:22 +00:00
Daniel Dunbar a895c69431 MC/Mach-O/Thumb: Select appropriate relocation types for Thumb.
llvm-svn: 122583
2010-12-27 14:49:49 +00:00
Cameron Zwarich 25d046ce68 Land a first cut at StrongPHIElimination. There are only 5 new test failures
when running without the verifier, and I have not yet checked them to see if
the new results are still correct. There are more verifier failures, but they
all seem to be additional occurrences of verifier failures that occur with the
existing PHIElimination pass. There are a few obvious issues with the code:

1) It doesn't properly update the register equivalence classes during copy
insertion, and instead recomputes them before merging live intervals and
renaming registers. I wanted to keep this first patch simple for debugging
purposes, but it shouldn't be very hard to do this.

2) It doesn't mix the renaming and live interval merging with the copy insertion
process, which leads to a lot of virtual register churn. Virtual registers and
live intervals are created, only to later be merged into others. The code should
be smarter and only create a new virtual register if there is no existing
register in the same congruence class.

3) In one place the code uses a DenseMap per basic block, which is unnecessary
heap allocation. There should be an inline storage version of DenseMap.

I did a quick compile-time test of running llc on 403.gcc with and without
StrongPHIElimination. It is slightly slower with StrongPHIElimination, because
the small decrease in the coalescer runtime can't beat the increase in phi
elimination runtime. Perhaps fixing the above performance issues will narrow
the gap.

I also haven't yet run any tests of the quality of the generated code.

llvm-svn: 122582
2010-12-27 10:08:19 +00:00
Cameron Zwarich b95bfe1667 Add knowledge of phi-def and phi-kill valnos to MachineVerifier's predecessor
valno verification. The "Different value live out of predecessor" check is
incorrect in the case of phi-def valnos, so just skip that check for phi-def
valnos and instead check that all of the valnos for predecessors have phi-kill.
Fixes PR8863.

llvm-svn: 122581
2010-12-27 05:17:23 +00:00
Michael J. Spencer 9e590024f6 Support/PathV1: Deprecate GetRootDirectory.
llvm-svn: 122580
2010-12-27 03:21:41 +00:00
Rafael Espindola 8fc59a682f Handle reloc_riprel_4byte_movq_load. Should make the bots happy.
llvm-svn: 122579
2010-12-27 02:03:24 +00:00
Francois Pichet b7577657cd More __uuidof validation:
1. Do not validate for uuid attribute if the type is template dependent.
2. Search every class declaration and definition for the uuid attribute.

llvm-svn: 122578
2010-12-27 01:32:00 +00:00
Rafael Espindola 2ac8355ecd Add support for the same encodings of the personality function that gnu as
supports.

llvm-svn: 122577
2010-12-27 00:36:05 +00:00
Benjamin Kramer 84bd73c527 BuildLibCalls: Nuke EmitMemCpy, EmitMemMove and EmitMemSet. They are dead and superseded by IRBuilder.
llvm-svn: 122576
2010-12-27 00:25:32 +00:00
Benjamin Kramer 7cba269dfb SimplifyLibCalls: Use IRBuilder to simplify code.
llvm-svn: 122575
2010-12-27 00:16:46 +00:00
Chris Lattner b9fe685b9a have loop-idiom nuke instructions that feed stores that get removed.
llvm-svn: 122574
2010-12-27 00:03:23 +00:00
Chris Lattner 29e14edc8d implement enough of the memset inference algorithm to recognize and insert
memsets.  This is still missing one important validity check, but this is enough
to compile stuff like this:

void test0(std::vector<char> &X) {
  for (std::vector<char>::iterator I = X.begin(), E = X.end(); I != E; ++I)
    *I = 0;
}

void test1(std::vector<int> &X) {
  for (long i = 0, e = X.size(); i != e; ++i)
    X[i] = 0x01010101;
}

With:
 $ clang t.cpp -S -o - -O2 -emit-llvm | opt -loop-idiom | opt -O3 | llc 

to:

__Z5test0RSt6vectorIcSaIcEE:            ## @_Z5test0RSt6vectorIcSaIcEE
## BB#0:                                ## %entry
	subq	$8, %rsp
	movq	(%rdi), %rax
	movq	8(%rdi), %rsi
	cmpq	%rsi, %rax
	je	LBB0_2
## BB#1:                                ## %bb.nph
	subq	%rax, %rsi
	movq	%rax, %rdi
	callq	___bzero
LBB0_2:                                 ## %for.end
	addq	$8, %rsp
	ret
...
__Z5test1RSt6vectorIiSaIiEE:            ## @_Z5test1RSt6vectorIiSaIiEE
## BB#0:                                ## %entry
	subq	$8, %rsp
	movq	(%rdi), %rax
	movq	8(%rdi), %rdx
	subq	%rax, %rdx
	cmpq	$4, %rdx
	jb	LBB1_2
## BB#1:                                ## %for.body.preheader
	andq	$-4, %rdx
	movl	$1, %esi
	movq	%rax, %rdi
	callq	_memset
LBB1_2:                                 ## %for.end
	addq	$8, %rsp
	ret

llvm-svn: 122573
2010-12-26 23:42:51 +00:00
Chris Lattner 6cf8d6cc6e start using irbuilder to make mem intrinsics in a few passes.
llvm-svn: 122572
2010-12-26 22:57:41 +00:00
Chris Lattner 143a07cfee add methods to IRBuilder to create memcpy/memset/memmove.
llvm-svn: 122571
2010-12-26 22:49:25 +00:00
Rafael Espindola 54462cd730 Fix .cfi_personality on 32 bit systems.
llvm-svn: 122570
2010-12-26 22:47:37 +00:00
David Chisnall 168b80f2c0 Add support for GNU runtime property set / get structure functions. Minor refactoring of Mac runtime (returns the same function for both, as the Mac runtimes currently only provide a single entry point for setting and getting struct properties, although this will presumably be fixed at some point).
llvm-svn: 122569
2010-12-26 22:13:16 +00:00
Rafael Espindola 9ae2d05d45 Add support for @note. Patch by Jörg Sonnenberger.
llvm-svn: 122568
2010-12-26 21:30:59 +00:00
Chris Lattner 7c5f9c35d1 sketch more of this out.
llvm-svn: 122567
2010-12-26 20:45:45 +00:00
Rafael Espindola 9141b611ad Add basic support for .cfi_personality.
llvm-svn: 122566
2010-12-26 20:20:31 +00:00
Chris Lattner 9cb1035f94 move isBytewiseValue out to ValueTracking.h/cpp
llvm-svn: 122565
2010-12-26 20:15:01 +00:00
David Chisnall 6f0a7d224b Fix for PR8695.
llvm-svn: 122564
2010-12-26 20:12:30 +00:00
Chris Lattner 81ae3f299a actually add the file...
llvm-svn: 122563
2010-12-26 19:39:38 +00:00
Chris Lattner 2ef535a4e4 Start of a pass for recognizing memset and memcpy idioms.
No functionality yet.

llvm-svn: 122562
2010-12-26 19:32:44 +00:00
Benjamin Kramer 30342fb1fd Simplify code.
llvm-svn: 122561
2010-12-26 15:23:45 +00:00
Chris Lattner f9e0a56b94 fix some sort of weird pasto
llvm-svn: 122560
2010-12-26 12:05:11 +00:00
Chris Lattner 424de3498b add a note
llvm-svn: 122559
2010-12-26 03:53:31 +00:00
Chris Lattner ad3467ee89 The -fshort-wchar option causes wchar_t to become unsigned, in addition to being
16-bits in size.  Implement this by splitting WChar into two enums, like we have
for char.  This fixes a miscompmilation of XULRunner, PR8856.

llvm-svn: 122558
2010-12-25 23:25:43 +00:00
Chris Lattner 03a102bff3 Generalize a previous change, fixing PR8855 - an valid large immediate
rejected by the mc assembler.

llvm-svn: 122557
2010-12-25 21:36:35 +00:00
Chris Lattner d729d0dcdb don't lose TD info
llvm-svn: 122556
2010-12-25 20:52:04 +00:00
Chris Lattner 20fca48341 switch the inliner alignment enforcement stuff to use the
getOrEnforceKnownAlignment function, which simplifies the code
and makes it stronger.

llvm-svn: 122555
2010-12-25 20:42:38 +00:00
Chris Lattner 6fcd32e7d7 Move getOrEnforceKnownAlignment out of instcombine into Transforms/Utils.
llvm-svn: 122554
2010-12-25 20:37:57 +00:00
Michael J. Spencer e13f1ead9b Support/PathV1: Deprecate makeAbsolute and remove Unix impl because it annoys people.
llvm-svn: 122553
2010-12-25 20:10:11 +00:00
Michael J. Spencer bc7fcc2191 Remove all uses of PathV1::GetRootDirectory.
llvm-svn: 122552
2010-12-25 20:09:27 +00:00
Eric Christopher 481de07b00 Header warning patrol.
llvm-svn: 122551
2010-12-25 02:38:01 +00:00
Benjamin Kramer b90b2f0635 Fix a thinko pointed out by Frits van Bommel: looking through global variables in isBytewiseValue is not safe.
llvm-svn: 122550
2010-12-24 22:23:59 +00:00
Rafael Espindola 0f8abeba1d Merge IsFixupFullyResolved and IsSymbolRefDifferenceFullyResolved. We now
have a single point where targets test if a relocation is needed.

llvm-svn: 122549
2010-12-24 21:22:02 +00:00
Benjamin Kramer ea9152e551 MemCpyOpt: Turn memcpys from a constant into a memset if possible.
This allows us to compile "int cst[] = {-1, -1, -1};" into
  movl  $-1, 16(%rsp)
  movq  $-1, 8(%rsp)
instead of
  movl  _cst+8(%rip), %eax
  movl  %eax, 16(%rsp)
  movq  _cst(%rip), %rax
  movq  %rax, 8(%rsp)

llvm-svn: 122548
2010-12-24 21:17:12 +00:00
Daniel Dunbar 7f2bb4dcae MC/Mach-O/ARM: Start handling some Thumb branches.
llvm-svn: 122547
2010-12-24 16:41:46 +00:00
Ted Kremenek 5614c46fcf Add basic support for pointer arithmetic in
SimpleSValBuilder.  This clears up some
false positives emitted by ArrayBoundCheckerV2
due to the lack of support for pointer arithmetic.

llvm-svn: 122546
2010-12-24 08:39:33 +00:00
Andrew Trick 5ce945ca3a Minor cleanup related to my latest scheduler changes.
llvm-svn: 122545
2010-12-24 07:10:19 +00:00
Andrew Trick c94056692a Fix a few cases where the scheduler is not checking for phys reg copies. The scheduling node may have a NULL DAG node, yuck.
llvm-svn: 122544
2010-12-24 06:46:50 +00:00
Argyrios Kyrtzidis 9d6af5328e Remove the EntoSA directories.
llvm-svn: 122543
2010-12-24 06:19:58 +00:00
Jim Grosbach 50986b5d39 Trailing whitespace.
llvm-svn: 122542
2010-12-24 05:06:32 +00:00
Andrew Trick 10ffc2b6c2 Various bits of framework needed for precise machine-level selection
DAG scheduling during isel. Most new functionality is currently
guarded by -enable-sched-cycles and -enable-sched-hazard.

Added InstrItineraryData::IssueWidth field, currently derived from
ARM itineraries, but could be initialized differently on other targets.

Added ScheduleHazardRecognizer::MaxLookAhead to indicate whether it is
active, and if so how many cycles of state it holds.

Added SchedulingPriorityQueue::HasReadyFilter to allowing gating entry
into the scheduler's available queue.

ScoreboardHazardRecognizer now accesses the ScheduleDAG in order to
get information about it's SUnits, provides RecedeCycle for bottom-up
scheduling, correctly computes scoreboard depth, tracks IssueCount, and
considers potential stall cycles when checking for hazards.

ScheduleDAGRRList now models machine cycles and hazards (under
flags). It tracks MinAvailableCycle, drives the hazard recognizer and
priority queue's ready filter, manages a new PendingQueue, properly
accounts for stall cycles, etc.

llvm-svn: 122541
2010-12-24 05:03:26 +00:00
Chris Lattner 9c47b4ac28 don't use #pragma mark, it isn't portable.
llvm-svn: 122540
2010-12-24 04:45:23 +00:00
Andrew Trick c416ba612b whitespace
llvm-svn: 122539
2010-12-24 04:28:06 +00:00
Michael J. Spencer 38e92f6734 ifndef _MSC_VER out #pragma mark on MSVC. It still tries to parse the text even
though it doesn't know what it is, and complains about invalid tokens ;/.

llvm-svn: 122538
2010-12-24 04:07:39 +00:00
Cameron Zwarich ab434079d3 Simplify a check for implicit defs and remove a FIXME.
llvm-svn: 122537
2010-12-24 03:09:36 +00:00
Argyrios Kyrtzidis 338f9aaab2 Handle locations coming from macro instantiations properly in SourceManager::isBeforeInTranslationUnit().
Fixes rdar://8790245 and http://llvm.org/PR8821.

llvm-svn: 122536
2010-12-24 02:53:53 +00:00