maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
27,780 Commits 23 Branches 0 Tags 2.2 GiB
Commit Graph

27739 Commits

Author SHA1 Message Date
Chris Lattner d12a4bf799 implement InstCombine/and-compare.ll:test1. This compiles: 
typedef struct { unsigned prefix : 4; unsigned code : 4; unsigned unsigned_p : 4; } tree_common;
int foo(tree_common *a, tree_common *b) { return a->code == b->code; }

into:

_foo:
        movl 4(%esp), %eax
        movl 8(%esp), %ecx
        movl (%eax), %eax
        xorl (%ecx), %eax
        # TRUNCATE movb %al, %al
        shrb $4, %al
        testb %al, %al
        sete %al
        movzbl %al, %eax
        ret

instead of:

_foo:
        movl 8(%esp), %eax
        movb (%eax), %al
        shrb $4, %al
        movl 4(%esp), %ecx
        movb (%ecx), %cl
        shrb $4, %cl
        cmpb %al, %cl
        sete %al
        movzbl %al, %eax
        ret

saving one cycle by eliminating a shift.

llvm-svn: 31727
 2006-11-14 06:06:06 +00:00
Chris Lattner 9adf8294ef new testcase 
llvm-svn: 31726
 2006-11-14 06:04:35 +00:00
Chris Lattner 97ff46b3cc lower "X = seteq Y, Z" to '(shr (ctlz (xor Y, Z)), 5)' instead of 
'(shr (ctlz (sub Y, Z)), 5)'.

The use of xor better exposes the operation to bit-twiddling logic in the
dag combiner.  For example, this:

typedef struct {
  unsigned prefix : 4;
  unsigned code : 4;
  unsigned unsigned_p : 4;
} tree_common;

int foo(tree_common *a, tree_common *b) {
  return a->code == b->code;
}

Now compiles to:

_foo:
        lwz r2, 0(r4)
        lwz r3, 0(r3)
        xor r2, r3, r2
        rlwinm r2, r2, 28, 28, 31
        cntlzw r2, r2
        srwi r3, r2, 5
        blr

instead of:

_foo:
        lbz r2, 3(r4)
        lbz r3, 3(r3)
        srwi r2, r2, 4
        srwi r3, r3, 4
        subf r2, r2, r3
        cntlzw r2, r2
        srwi r3, r2, 5
        blr

saving a cycle.

llvm-svn: 31725
 2006-11-14 05:28:08 +00:00
Andrew Lenharth 0ad1e3845b A shim over other AA impls to catch incorrect uses 
llvm-svn: 31724
 2006-11-14 05:21:04 +00:00
Reid Spencer 1b89a7bcc1 Discard code that supported old bytecode formats. This makes the Bytecode 
Reader code much easier to read and maintain. Backwards compatibility from
version 5 format has been retained. Older formats will produce an error.

llvm-svn: 31723
 2006-11-14 04:47:22 +00:00
Devang Patel ca1892699c Implement removeDeadPasses(). 
llvm-svn: 31722
 2006-11-14 03:05:08 +00:00
Devang Patel 3f0832a3b5 Keep track of Last user of analysis phase. 
llvm-svn: 31721
 2006-11-14 02:54:23 +00:00
Devang Patel f60b5d9676 Use std::map to map AnalysisID and Pass. 
llvm-svn: 31720
 2006-11-14 01:59:59 +00:00
Chris Lattner 6dc22332bf add a note 
llvm-svn: 31719
 2006-11-14 01:57:53 +00:00
Chris Lattner 683712583a minor tweaks, reject vector preinc. 
llvm-svn: 31717
 2006-11-14 01:38:31 +00:00
Devang Patel 050ec72ae4 Manage analysis passes during run. 
llvm-svn: 31716
 2006-11-14 01:23:29 +00:00
Tanya Lattner 9ce0379da2 Updating configure script for 2.0 
llvm-svn: 31715
 2006-11-14 00:59:52 +00:00
Devang Patel dafa4ddc46 Update comments. 
llvm-svn: 31713
 2006-11-14 00:03:04 +00:00
Evan Cheng dbd3d294e6 Matches MachineInstr changes. 
llvm-svn: 31712
 2006-11-13 23:36:35 +00:00
Evan Cheng 77af6ac5e8 - Let MachineInstr ctors add implicit def and use operands. Other operands 
will be inserted before these operands. If the opcode changes (by
  setOpcode), the implicit operands are updated as well.
- Added IsKill, IsDead fields to MachineOperand in preparation for changes
  that move kill / dead info to MachineInstr's.

llvm-svn: 31711
 2006-11-13 23:34:06 +00:00
Devang Patel 7086844634 Update PassManagerImpl_New::analysisCurrentlyAvailable to check all 
managed passmanagers.

llvm-svn: 31710
 2006-11-13 22:53:19 +00:00
Devang Patel ebba970a6d Check currently available anlysis in active managers. 
llvm-svn: 31709
 2006-11-13 22:40:09 +00:00
Chris Lattner 1121e50f7e teach the g5 hazard recognizer about update loads. This fixes 
Ptrdist/anagram among others.

llvm-svn: 31708
 2006-11-13 20:11:06 +00:00
Reid Spencer 3d1027193f For PR950: 
Changes for new cast instructions that are backwards compatible.

llvm-svn: 31707
 2006-11-13 18:00:52 +00:00
Reid Spencer 7354395590 Don't hard code test names into the test. 
llvm-svn: 31706
 2006-11-13 17:55:34 +00:00
Reid Spencer 83c7a1e710 Fix 80 cols violations. 
llvm-svn: 31705
 2006-11-13 16:21:09 +00:00
Reid Spencer 5953e4e805 For PR950: 
Use the notcast script in preparation for 12 casts.

llvm-svn: 31704
 2006-11-13 16:13:38 +00:00
Reid Spencer 6170dff047 For PR950: 
Use the notcast script in preparation for 12 cast instructions.

llvm-svn: 31703
 2006-11-13 16:12:25 +00:00
Reid Spencer 7debe5e948 Make it handle plain old cast too, for old assembly. 
llvm-svn: 31702
 2006-11-13 16:11:14 +00:00
Reid Spencer 6f406f91d2 Handle ptrtoint and inttoptr 
llvm-svn: 31701
 2006-11-13 16:09:37 +00:00
Reid Spencer cac60e8eb4 Add some output so the user is informed while they wait. 
llvm-svn: 31700
 2006-11-13 16:08:51 +00:00
Reid Spencer e473e41e59 Add a link to the GEP FAQ. 
llvm-svn: 31698
 2006-11-13 08:03:59 +00:00
Chris Lattner 5c833df633 remove redundant code 
llvm-svn: 31697
 2006-11-13 01:10:12 +00:00
Nick Lewycky 0df2ada9d4 Cute example from Chris Lattner. 
llvm-svn: 31696
 2006-11-13 00:23:28 +00:00
Bill Wendling ad75361d42 Initialize some pointers to quiet the compiler (when doing build_gcc 
builds).

llvm-svn: 31694
 2006-11-12 11:10:39 +00:00
Reid Spencer 5b9506439a Document new constant expr operators, rename bitconvert as bitcast. 
llvm-svn: 31693
 2006-11-11 23:08:07 +00:00
Reid Spencer b608daeec3 Revert premature patch. 
llvm-svn: 31692
 2006-11-11 23:06:47 +00:00
Reid Spencer 8b8913ba61 Document new constant expr operators, rename bitconvert as bitcast. 
llvm-svn: 31691
 2006-11-11 22:34:59 +00:00
Jim Laskey 1dfc6110ff Make sure stack link is set in 64-bit. 
llvm-svn: 31690
 2006-11-11 22:22:07 +00:00
Jim Laskey a3262ee8ec XFAIL All but powerpc 
llvm-svn: 31689
 2006-11-11 22:14:46 +00:00
Reid Spencer b7344ff025 Add a preview of two new conversion operators: ptrtoint and inttoptr. Update 
the defintion of bitconvert correspondingly.

llvm-svn: 31688
 2006-11-11 21:00:47 +00:00
Reid Spencer 06d8e0fa3b Remvoe a do-nothing else block. 
llvm-svn: 31687
 2006-11-11 20:27:49 +00:00
Reid Spencer 2230144a75 Make an assert comment match the tested assertion. 
llvm-svn: 31686
 2006-11-11 20:07:59 +00:00
Reid Spencer 3005d47797 Minor style fixes from review. 
llvm-svn: 31685
 2006-11-11 19:59:25 +00:00
Chris Lattner 0d550cc56c implement proper PPC64 prolog/epilog codegen. 
llvm-svn: 31684
 2006-11-11 19:05:28 +00:00
Reid Spencer 471bcb750e Add a -disable-loop-extraction option to bugpoint. 
llvm-svn: 31683
 2006-11-11 19:05:02 +00:00
Reid Spencer 75db664f44 For PR998: 
Fix an infinite loop in the Linker and a few other assorted link problems.

Patch contributed by Scott Michel. Thanks, Scott!

llvm-svn: 31680
 2006-11-11 11:54:25 +00:00
Jim Laskey ebdf6f46eb Works for PowerPC with frame pointers. 
llvm-svn: 31679
 2006-11-11 10:22:56 +00:00
Reid Spencer 2e80be8251 Add -lstdc++ to the link line for C++ programs. 
llvm-svn: 31678
 2006-11-11 10:22:48 +00:00
Jim Laskey 66c502b54a Running with frame pointers prevented debugging, external probes and 
potentially some system calls/exception handling from working.  TOS must always
link to previous frame.  This is a short term workaround until alloca scheme is
reworked.

llvm-svn: 31677
 2006-11-11 10:21:58 +00:00
Evan Cheng db04c958a5 Add implicit use / def operands to created MI's. 
llvm-svn: 31676
 2006-11-11 10:21:44 +00:00
Evan Cheng 979bbf48d5 Add methods to add implicit def use operands to a MI. 
llvm-svn: 31675
 2006-11-11 10:20:02 +00:00
Evan Cheng a36cdcfaf8 Add all implicit defs to FP_REG_KILL mi. 
llvm-svn: 31674
 2006-11-11 07:19:36 +00:00
Chris Lattner c5102bfc7c allow the offset of a preinc'd load to be the low-part of a global. This 
produces this clever code:

_millisecs:
        lis r2, ha16(_Time.1182)
        lwzu r3, lo16(_Time.1182)(r2)
        lwz r2, 4(r2)
        addic r4, r2, 1
        addze r3, r3
        blr

instead of this:

_millisecs:
        lis r2, ha16(_Time.1182)
        la r3, lo16(_Time.1182)(r2)
        lwz r2, lo16(_Time.1182)(r2)
        lwz r3, 4(r3)
        addic r4, r3, 1
        addze r3, r2
        blr

for:

long %millisecs() {
        %tmp = load long* %Time.1182            ; <long> [#uses=1]
        %tmp1 = add long %tmp, 1                ; <long> [#uses=1]
        ret long %tmp1
}

llvm-svn: 31673
 2006-11-11 04:53:30 +00:00
Chris Lattner 2ff632c54b Mark operands as symbol lo instead of imm32 so that they print lo(x) around 
globals.

llvm-svn: 31672
 2006-11-11 04:51:36 +00:00
Devang Patel 1a6eaa49fe Implement schedulePasses(). 
llvm-svn: 31671
 2006-11-11 02:22:31 +00:00
Devang Patel db789fb8f2 Implement PassManagerImpl_New::add(). 
Just add pass into the pass manager queue without processing analysis.

llvm-svn: 31670
 2006-11-11 02:06:21 +00:00
Devang Patel 90b05e0bc3 While adding pass into the manager, process Analysis only if it is 
required to do so.

llvm-svn: 31669
 2006-11-11 02:04:19 +00:00
Devang Patel 2efbad4a72 Remove dead code. 
llvm-svn: 31668
 2006-11-11 01:56:39 +00:00
Devang Patel 8cad70de14 Code refactoring. Move common code into CommonPassManagerImpl :) 
llvm-svn: 31667
 2006-11-11 01:51:02 +00:00
Devang Patel a98445962d Move CommonPassManagerImpl from PassManager.h to PassManager.cpp 
llvm-svn: 31666
 2006-11-11 01:31:05 +00:00
Devang Patel 349170f5da Remove analysis that is not preserved by the pass from AvailableAnalysis. 
llvm-svn: 31665
 2006-11-11 01:24:55 +00:00
Devang Patel 643676c1f5 Keep track if analysis made available by the pass. 
llvm-svn: 31664
 2006-11-11 01:10:19 +00:00
Reid Spencer e183ed753f Instruction name changes. 
llvm-svn: 31663
 2006-11-11 01:04:13 +00:00
Reid Spencer 279a3aefc3 Enable path completion when typing the TESTSUITE= option by allowing the 
test directory to precede the test suite name. That is, it will strip off
test/ from TESTSUITE which allows path completion from the command line.

llvm-svn: 31662
 2006-11-11 01:02:45 +00:00
Chris Lattner a0a8003f59 disallow preinc of a frameindex. This is not profitable and causes 2-addr 
pass to explode.  This fixes a bunch of llc-beta failures on ppc last night.

llvm-svn: 31661
 2006-11-11 01:00:15 +00:00
Chris Lattner eabc15c1d8 reduce indentation by using early exits. No functionality change. 
llvm-svn: 31660
 2006-11-11 00:56:29 +00:00
Devang Patel 6c9f548704 Keep track of analysis required by the passes. Force use of new pass 
manager if a pass does not preserve analysis that is used by other
passes.

llvm-svn: 31659
 2006-11-11 00:42:16 +00:00
Chris Lattner ffad2166e1 move big chunks of code out-of-line, no functionality change. 
llvm-svn: 31658
 2006-11-11 00:39:41 +00:00
Chris Lattner e15018b8b7 enable this test, check ppc64 as well. 
llvm-svn: 31657
 2006-11-11 00:13:07 +00:00
Chris Lattner b314b155ed ppc64 doesn't have lwau, don't attempt to form it. 
llvm-svn: 31656
 2006-11-11 00:08:42 +00:00
Reid Spencer d185c02007 Allow ENABLE_OPTIMIZED=0 to turn off optimization (turn on debug). 
llvm-svn: 31655
 2006-11-11 00:00:31 +00:00
Chris Lattner c9fa36d706 implement preinc support for r+i loads on ppc64 
llvm-svn: 31654
 2006-11-10 23:58:45 +00:00
Chris Lattner d4dee405cb Fix InstCombine/2006-11-10-ashr-miscompile.ll a miscompilation introduced 
by the shr -> [al]shr patch.  This was reduced from 176.gcc.

llvm-svn: 31653
 2006-11-10 23:38:52 +00:00
Chris Lattner d530a503da This testcase is miscompiled by instcombine, reduced from 176.gcc 
llvm-svn: 31652
 2006-11-10 23:37:54 +00:00
Evan Cheng 915026cdc3 Add a note. 
llvm-svn: 31650
 2006-11-10 22:09:17 +00:00
Evan Cheng 5b725a71a9 These are done. 
llvm-svn: 31649
 2006-11-10 22:03:35 +00:00
Evan Cheng 49683ba236 Don't dag combine floating point select to max and min intrinsics. Those 
take v4f32 / v2f64 operands and may end up causing larger spills / restores.
Added X86 specific nodes X86ISD::FMAX, X86ISD::FMIN instead.

This fixes PR996.

llvm-svn: 31645
 2006-11-10 21:43:37 +00:00
Chris Lattner 4eac5f59e6 Fix a dag combiner bug exposed by my recent instcombine patch. This fixes 
CodeGen/Generic/2006-11-10-DAGCombineMiscompile.ll and PPC gsm/toast

llvm-svn: 31644
 2006-11-10 21:37:15 +00:00
Chris Lattner 82a85b66a3 new testcase 
llvm-svn: 31643
 2006-11-10 21:36:07 +00:00
Devang Patel 0ed477969f s/PassManagerAnalysisHelper/CommonPassManagerImpl 
Inherit CommonPassManagerImpl from Pass.

llvm-svn: 31642
 2006-11-10 21:33:13 +00:00
Evan Cheng fb44822a98 Fix a bug in SelectScalarSSELoad. Since the load is wrapped in a 
SCALAR_TO_VECTOR, even if the hasOneUse() check pass we may end up folding
the load into two instructions. Make sure we check the SCALAR_TO_VECTOR
has only one use as well.

llvm-svn: 31641
 2006-11-10 21:23:04 +00:00
Jim Laskey bc60d34748 Must have a frame pointer argument fixed. Now fails on PowerPC. 
llvm-svn: 31639
 2006-11-10 17:56:29 +00:00
Jim Laskey 11ba2bf721 Must have a frame pointer. 
llvm-svn: 31638
 2006-11-10 17:51:25 +00:00
Chris Lattner 6c8656a6b1 dform 8/9 are identical to dform 1 
llvm-svn: 31637
 2006-11-10 17:51:02 +00:00
Reid Spencer 8488eac9d7 This is XFAILed on i[0-9]86 systems not Linux 
llvm-svn: 31636
 2006-11-10 17:47:28 +00:00
Jim Laskey 0c75372933 Shut up the warning. 
llvm-svn: 31635
 2006-11-10 14:44:12 +00:00
Evan Cheng 286357480c Fix a potential bug. 
llvm-svn: 31634
 2006-11-10 09:13:37 +00:00
Evan Cheng 8c9c6d71ed Add implicit def / use operands to MachineInstr. 
llvm-svn: 31633
 2006-11-10 08:43:01 +00:00
Evan Cheng f5bebe83a5 Add implicit def / use operands to MachineInstr. 
llvm-svn: 31632
 2006-11-10 08:32:14 +00:00
Evan Cheng 13440b025c When forming a pre-indexed store, make sure ptr isn't the same or is a pred of value being stored. It would cause a cycle. 
llvm-svn: 31631
 2006-11-10 08:28:11 +00:00
Anton Korobeynikov a1561d4fdf Adding target triple to test 
llvm-svn: 31630
 2006-11-10 06:59:53 +00:00
Tanya Lattner 1537ecce93 xfail for llvmgcc3 
llvm-svn: 31629
 2006-11-10 04:53:50 +00:00
Chris Lattner d5e604dbb2 commentate 
llvm-svn: 31627
 2006-11-10 04:41:34 +00:00
Chris Lattner ce6455489a add an initial cut at preinc loads for ppc32. This is broken for ppc64 
(because the 64-bit reg target versions aren't implemented yet), doesn't
support r+r addr modes, and doesn't handle stores, but it works otherwise. :)

This is disabled unless -enable-ppc-preinc is passed to llc for now.

llvm-svn: 31621
 2006-11-10 02:08:47 +00:00
Chris Lattner 7a82314782 make this test more interesting 
llvm-svn: 31619
 2006-11-10 02:04:52 +00:00
Chris Lattner ba7b3673f9 allow ptr_rc to explicitly appear in an instructions operand list, it doesn't 
have to be a subpart of a complex operand.

llvm-svn: 31618
 2006-11-10 02:01:40 +00:00
Chris Lattner cd436327f1 add note about ugly codegen with preinc 
llvm-svn: 31617
 2006-11-10 01:33:53 +00:00
Evan Cheng 14140059f0 Use TargetInstrInfo::getNumOperands() instead of MachineInstr::getNumOperands(). In preparation for implicit reg def/use changes. 
llvm-svn: 31616
 2006-11-10 01:28:43 +00:00
Anton Korobeynikov b9c91c265c Fixing PR990: http://llvm.org/PR990. 
This should unbreak csretcc on Linux & mingw targets. Several tests from
llvm-test should be also restored (fftbench, bigfib).

llvm-svn: 31613
 2006-11-10 00:48:11 +00:00
Chris Lattner 4e03cb1945 add a note about viterbi 
llvm-svn: 31612
 2006-11-10 00:23:26 +00:00
Chris Lattner 82928ca290 second patch to fix PR992/993. 
llvm-svn: 31610
 2006-11-09 23:36:08 +00:00
Chris Lattner 8511bd95f5 new testcases for PR992/993 
llvm-svn: 31609
 2006-11-09 23:35:01 +00:00
Chris Lattner 924f4fee8b Minimal patch to fix PR992/PR993 
llvm-svn: 31608
 2006-11-09 23:17:45 +00:00
Reid Spencer 51b07256a3 Post-review changes on conversion instructions. 
llvm-svn: 31607
 2006-11-09 23:03:26 +00:00
Reid Spencer 2e2740d46a Fix runaway <tt> element and section ordering. 
llvm-svn: 31606
 2006-11-09 21:48:10 +00:00
Reid Spencer 8a82cc6304 Remove futurisms that won't be in our future. 
llvm-svn: 31605
 2006-11-09 21:42:14 +00:00
Reid Spencer a89fb18c98 One more time for indentation. 
llvm-svn: 31604
 2006-11-09 21:18:01 +00:00
Reid Spencer 443460a318 Fix indentation problem starting at GEP. 
llvm-svn: 31603
 2006-11-09 21:15:49 +00:00
Chris Lattner 5fdbe55979 if lazy compilation is disabled, print an error message and abort if 
lazy compilation is ever attempted

llvm-svn: 31602
 2006-11-09 19:32:13 +00:00
Chris Lattner 87aee74ccb init ivar 
llvm-svn: 31601
 2006-11-09 19:31:15 +00:00
Chris Lattner 6dbe06de39 allow clients to indicate that they never want lazy compilation. 
llvm-svn: 31600
 2006-11-09 19:30:47 +00:00
Evan Cheng 36a8fbf771 PPC supports i32 / i64 pre-inc load / store. 
llvm-svn: 31599
 2006-11-09 19:11:50 +00:00
Evan Cheng 6878378390 Don't attempt expensive pre-/post- indexed dag combine if target does not support them. 
llvm-svn: 31598
 2006-11-09 19:10:46 +00:00
Evan Cheng d550248f2c Add a mechanism to specify whether a target supports a particular indexed load / store. 
llvm-svn: 31597
 2006-11-09 18:56:43 +00:00
Evan Cheng c034f14fbe Rename ISD::MemOpAddrMode to ISD::MemIndexedMode 
llvm-svn: 31596
 2006-11-09 18:44:21 +00:00
Evan Cheng b15000736c Rename ISD::MemOpAddrMode to ISD::MemIndexedMode 
llvm-svn: 31595
 2006-11-09 17:55:04 +00:00
Jim Laskey f1606b80f1 Allows debugging llc self hosted. 
llvm-svn: 31594
 2006-11-09 16:32:26 +00:00
Jim Laskey 97f89b11b3 Merging dwarf info to a single compile unit. 
llvm-svn: 31593
 2006-11-09 14:52:14 +00:00
Rafael Espindola 5f7ab1b964 implement load effective address similar to the alpha backend 
remove lea_addri and the now unused memri addressing mode

llvm-svn: 31592
 2006-11-09 13:58:55 +00:00
Jim Laskey 48b21d5024 math.h creates ambiguity 
llvm-svn: 31591
 2006-11-09 08:26:22 +00:00
Jim Laskey 31890e7e86 math.h for C's sake. 
llvm-svn: 31590
 2006-11-09 08:05:39 +00:00
Chris Lattner 7275b02ea5 completely revert patrick's enhancement to bugpoint. Though it makes bugpoint 
speed up, it sometimes makes it crash on complex bc files, which isn't very nice.


With this, bugpoint can reduce the 176.gcc failure.

llvm-svn: 31589
 2006-11-09 06:24:56 +00:00
Chris Lattner f1a1a7ae6f add a new bugpoint mode -llc-safe. This uses LLC for both halves of a 
miscompilation.  This is useful for working around GCC+CBE bugs or for handling
programs that CBE doesn't support (e.g. inline asm) when searching for
optimizer bugs.

llvm-svn: 31588
 2006-11-09 05:57:53 +00:00
Chris Lattner 1b9633d7f0 silence warnings 
llvm-svn: 31587
 2006-11-09 05:18:12 +00:00
Chris Lattner 6e2c15c158 Teach ShrinkDemandedConstant how to handle X+C. This implements: 
add.ll:test33, add.ll:test34, shift-sra.ll:test2

llvm-svn: 31586
 2006-11-09 05:12:27 +00:00
Chris Lattner 92fc0c0651 new testcases 
llvm-svn: 31585
 2006-11-09 05:11:23 +00:00
Evan Cheng b58e06bc9e getPostIndexedAddressParts change: passes in load/store instead of its loaded / stored VT. 
llvm-svn: 31584
 2006-11-09 04:29:46 +00:00
Evan Cheng 1698c2999c Remove M_2_ADDR_FLAG. 
llvm-svn: 31583
 2006-11-09 02:22:54 +00:00
Reid Spencer c6389c10cc Don't run bugpoint if we can't find a misoptimization. 
llvm-svn: 31582
 2006-11-09 01:47:04 +00:00
Reid Spencer 206d609cdd Print a usage message if too few arguments to program. 
llvm-svn: 31581
 2006-11-09 00:50:32 +00:00
Reid Spencer e73551303f Add a utility script to find a mis-optimization problem. This sometimes 
helps when bugpoint can't find the problem directly because it needs the
set of optimizations that cause the program to fail.

llvm-svn: 31580
 2006-11-09 00:26:17 +00:00
Tanya Lattner 01167a7eff Should be xfailed for llvmgcc3 
llvm-svn: 31578
 2006-11-08 23:26:16 +00:00
Tanya Lattner 8a01d7a820 Should be xfailed for llvmgcc4 and NOT *. * means all platforms regardless of what llvmgcc you use. 
llvm-svn: 31577
 2006-11-08 23:25:58 +00:00
Evan Cheng c9a4cdc9c9 Added indexed store node and patfrag's. 
llvm-svn: 31576
 2006-11-08 23:02:11 +00:00
Evan Cheng 7120b9e9c7 Divide select methods into groups by SelectionDAG node opcodes (ISD::ADD, 
X86ISD::CMP, etc.) instead of SDNode names (add, x86cmp, etc). We now allow
multiple SDNodes to map to the same SelectionDAG node (e.g. store, indexed
store).

llvm-svn: 31575
 2006-11-08 23:01:03 +00:00
Reid Spencer eacb043c0a Ensure we don't regress on reading version 6 bytecode. 
llvm-svn: 31574
 2006-11-08 21:28:30 +00:00
Reid Spencer c4ecf9ff47 Bump the bytecode version number to 7. Implement upgrade of version 6 and 
version 6 bytecode.

llvm-svn: 31573
 2006-11-08 21:27:54 +00:00
Evan Cheng 3b3b786f03 Use movl+xchgl instead of pushl+popl. 
llvm-svn: 31572
 2006-11-08 20:35:37 +00:00
Evan Cheng 6cd0909da7 Match tblegen changes. 
llvm-svn: 31571
 2006-11-08 20:34:28 +00:00
Evan Cheng d715a2cbfb Always pass the root node to ComplexPattern isel matching function. 
llvm-svn: 31570
 2006-11-08 20:31:10 +00:00
Evan Cheng 85e54223cd Match more post-indexed ops. 
llvm-svn: 31569
 2006-11-08 20:27:27 +00:00
Tanya Lattner fd6e54439a Should be xfailed for llvm-gcc3.4 
llvm-svn: 31568
 2006-11-08 20:16:05 +00:00
Reid Spencer b3aa10f933 Document correct opcodes for post 1.9 release. 
llvm-svn: 31566
 2006-11-08 20:07:40 +00:00
Reid Spencer 99b21508e6 Update the instruction opcodes for release 1.9 
llvm-svn: 31565
 2006-11-08 20:06:36 +00:00
Chris Lattner 4f218d56f5 reenable factoring of GEP expressions, being more precise about the 
case that it bad to do.

llvm-svn: 31563
 2006-11-08 19:42:28 +00:00
Chris Lattner cd62f11227 make this code more efficient by not creating a phi node we are just going to 
delete in the first place.  This also makes it simpler.

llvm-svn: 31562
 2006-11-08 19:29:23 +00:00
Jim Laskey 61feeb90f9 Remove redundant <cmath>. 
llvm-svn: 31561
 2006-11-08 19:16:44 +00:00
Chris Lattner a3acfca920 disable this factoring optzn for GEPs for now, this severely pessimizes some 
loops.

llvm-svn: 31560
 2006-11-08 18:49:31 +00:00
Jim Laskey 7b6fb8048e Make sure <cmath> comes first 
llvm-svn: 31559
 2006-11-08 18:48:14 +00:00
Jim Laskey 3caa009892 Add backup support for HUGH_VALF. 
llvm-svn: 31553
 2006-11-08 17:19:29 +00:00
Rafael Espindola 708cb60588 initial implementation of addressing mode 2 
TODO: fix lea_addri

llvm-svn: 31552
 2006-11-08 17:07:32 +00:00
Jim Laskey 050747fc8c Make it work on Darwin. 
llvm-svn: 31551
 2006-11-08 16:38:45 +00:00
John Criswell 8ea2e68493 Include llvm/Support/DataTypes.h to define intptr_t. 
This fixes the build on OpenBSD and potentially other systems.

llvm-svn: 31550
 2006-11-08 15:04:35 +00:00
Jim Laskey 62bbc9ef58 Now can re-enable debug label folding. 
llvm-svn: 31549
 2006-11-08 14:17:45 +00:00
Jim Laskey ac05d7d084 Wasn't handling case of when machine move labels were undefined. 
llvm-svn: 31548
 2006-11-08 14:16:39 +00:00
Devang Patel 4e12f86a1b Split FunctionPassManager_New into FunctionPassManager_New and FunctionPassManagerImpl_New. 
FunctionPassManagerImpl_New implements the pass manager.
FunctionPassManager_New is the public interface.

llvm-svn: 31547
 2006-11-08 10:44:40 +00:00
Devang Patel 376fefafb2 Split PassManager_New into PassManager_New and PassManagerImpl_New. 
PassManagerImpl_New implements the pass manager.
PassManager_New is the public interface.

llvm-svn: 31546
 2006-11-08 10:29:57 +00:00
Devang Patel ca58e3532f Move BasicBlockPassManager_New, FunctionPassManager_New and 
ModulePassManager_New class declarations from PassManager.h
to PassManager.cpp

llvm-svn: 31545
 2006-11-08 10:05:38 +00:00
Evan Cheng 0303cb9b33 - When performing pre-/post- indexed load/store transformation, do not worry 
about whether the new base ptr would be live below the load/store. Let two
  address pass split it back to non-indexed ops.
- Minor tweaks / fixes.

llvm-svn: 31544
 2006-11-08 08:30:28 +00:00
Evan Cheng 6072435756 Fixed a minor bug preventing some pre-indexed load / store transformation. 
llvm-svn: 31543
 2006-11-08 06:56:05 +00:00
Reid Spencer fdff938a7e For PR950: 
This patch converts the old SHR instruction into two instructions,
AShr (Arithmetic) and LShr (Logical). The Shr instructions now are not
dependent on the sign of their operands.

llvm-svn: 31542
 2006-11-08 06:47:33 +00:00
Reid Spencer 41f6c7cfb2 XFAIL this on Linux until Jim gets Darwin ship shape. 
llvm-svn: 31541
 2006-11-08 06:43:16 +00:00
Chris Lattner 71cd8cf478 now that we handle single mbb loops better, one branch is allowed 
llvm-svn: 31540
 2006-11-08 06:42:42 +00:00
Chris Lattner 6aee6f2a93 regenerate 
llvm-svn: 31539
 2006-11-08 05:58:47 +00:00
Chris Lattner 60f1f194f9 Fix a bug noticed by Emil Mikulic. 
llvm-svn: 31538
 2006-11-08 05:58:11 +00:00
Evan Cheng d48f7dd250 Fix a obscure post-indexed load / store dag combine bug. 
llvm-svn: 31537
 2006-11-08 02:38:55 +00:00
Chris Lattner a801fcedd3 Refactor all the addressing mode selection stuff into the isel lowering 
class, where it can be used for preinc formation.

llvm-svn: 31536
 2006-11-08 02:15:41 +00:00
Chris Lattner 6a5a4f85d3 correct the (currently unused) pattern for lwzu. 
llvm-svn: 31535
 2006-11-08 02:13:12 +00:00
Chris Lattner 543572716b preincrement case we miss. xfail until we catch it 
llvm-svn: 31534
 2006-11-08 01:44:35 +00:00
Devang Patel d65e9e9fd8 Beautify. 
llvm-svn: 31533
 2006-11-08 01:31:28 +00:00
Reid Spencer 97c5fa4607 Fix the section headings. 
llvm-svn: 31532
 2006-11-08 01:18:52 +00:00
Reid Spencer 59b6b7d9e4 For PR950: 
Early commit of the documentation for new conversion instructions for
easier review.

llvm-svn: 31531
 2006-11-08 01:11:31 +00:00
Chris Lattner bf3b57f221 optimize single MBB loops better. In particular, produce: 
LBB1_57:        #bb207.i
        movl 72(%esp), %ecx
        movb (%ecx,%eax), %cl
        movl 80(%esp), %edx
        movb %cl, 1(%edx,%eax)
        incl %eax
        cmpl $143, %eax
        jne LBB1_57     #bb207.i
        jmp LBB1_64     #cond_next255.i

intead of:

LBB1_57:        #bb207.i
        movl 72(%esp), %ecx
        movb (%ecx,%eax), %cl
        movl 80(%esp), %edx
        movb %cl, 1(%edx,%eax)
        incl %eax
        cmpl $143, %eax
        je LBB1_64      #cond_next255.i
        jmp LBB1_57     #bb207.i

This eliminates a branch per iteration of the loop.  This hurted PPC
particularly, because the extra branch meant another dispatch group for each
iteration of the loop.

llvm-svn: 31530
 2006-11-08 01:03:21 +00:00
Devang Patel 800596d60e Beautify. 
Clarify comments.

llvm-svn: 31529
 2006-11-08 00:19:31 +00:00
Devang Patel 3c8eb62560 Update new pass managers to use PassManagerAnalysisHelper API. 
llvm-svn: 31526
 2006-11-07 22:56:50 +00:00
Devang Patel cc85563dd9 Derive new pass managers from PassManagerAnalysisHelper. 
llvm-svn: 31525
 2006-11-07 22:44:55 +00:00
Chris Lattner 4967f6ddea scalarrepl should not split the two elements of the vsiidx array: 
int func(vFloat v0, vFloat v1) {
        int ii;
        vSInt32 vsiidx[2];
        vsiidx[0] = _mm_cvttps_epi32(v0);
        vsiidx[1] = _mm_cvttps_epi32(v1);
        ii = ((int *) vsiidx)[4];
        return ii;
}

This fixes Transforms/ScalarRepl/2006-11-07-InvalidArrayPromote.ll

llvm-svn: 31524
 2006-11-07 22:42:47 +00:00
Chris Lattner 2d7a5d9b5d scalarrepl should not split the two elements of the vsiidx array: 
int func(vFloat v0, vFloat v1) {
	int ii;
	vSInt32 vsiidx[2];
	vsiidx[0] = _mm_cvttps_epi32(v0);
	vsiidx[1] = _mm_cvttps_epi32(v1);
	ii = ((int *) vsiidx)[4];
	return ii;
}

llvm-svn: 31523
 2006-11-07 22:42:29 +00:00
Devang Patel f68a34908d Introduce PassManagerAnalysisHelper. 
llvm-svn: 31522
 2006-11-07 22:35:17 +00:00
Devang Patel c290c8a2c8 Add PassManager_New. 
llvm-svn: 31521
 2006-11-07 22:23:34 +00:00
Evan Cheng 847adeebb8 New test case. 
llvm-svn: 31520
 2006-11-07 22:18:14 +00:00
Evan Cheng 922e191116 Fixed a bug which causes x86 be to incorrectly match 
shuffle v, undef, <2, ?, 3, ?>
to movhlps
It should match to unpckhps instead.

Added proper matching code for
shuffle v, undef, <2, 3, 2, 3>

llvm-svn: 31519
 2006-11-07 22:14:24 +00:00
Devang Patel 85da9b73b7 Fix comment. 
llvm-svn: 31518
 2006-11-07 22:04:53 +00:00
Devang Patel 05e1a970f3 Add ModulePassManager_New. 
llvm-svn: 31517
 2006-11-07 22:03:15 +00:00
Jim Laskey 21d4d895ca Accidently reran commit. 
llvm-svn: 31516
 2006-11-07 21:58:55 +00:00
Devang Patel 0c2012ffb3 Add FunctionPassManager_New. 
llvm-svn: 31515
 2006-11-07 21:49:50 +00:00
Jim Laskey a919ba9abd Missed a label map, reverting till covered. 
llvm-svn: 31514
 2006-11-07 21:37:31 +00:00
Devang Patel 6e5a113923 Add BasicBlockPassManager_New. 
llvm-svn: 31513
 2006-11-07 21:31:57 +00:00
Jim Laskey 21ce53b94e Missed a label map, reverting till covered. 
llvm-svn: 31512
 2006-11-07 20:53:05 +00:00
Andrew Lenharth 30b254e48b Optionally allow comparison operations from affect DSGraphs 
llvm-svn: 31511
 2006-11-07 20:39:05 +00:00
Andrew Lenharth 1ee904e503 Allow loop detection during debug in forwarding nodes, and revert auxcall patch as it make 176.gcc untenable 
llvm-svn: 31510
 2006-11-07 20:36:02 +00:00
Andrew Lenharth cb902c4942 debug type for DSA TD 
llvm-svn: 31509
 2006-11-07 20:35:11 +00:00
Jim Laskey 6ea4fae838 1. Add a pass to fold debug label instructions so a debug info client can detect 
empty ranges.

2. Reorg how MachineDebugInfo maintains changes to debug labels.

3. Have dwarf writer use debug label info to simplify scopes and source line
coorespondence.

4. Revert the merging of compile units until I can get the bugs ironed out.

llvm-svn: 31507
 2006-11-07 19:33:46 +00:00
Chris Lattner be7033b346 add a note from viterbi 
llvm-svn: 31506
 2006-11-07 18:30:21 +00:00
Chris Lattner b18177bfab Enable improved spilling costs by default. This speeds up viterbi on x86 
by 40%, FreeBench/fourinarow by 20%, and many other programs 10-25%.

On PPC, this speeds up fourinarow by 18%, and probably other things as well.

llvm-svn: 31504
 2006-11-07 18:04:58 +00:00
Andrew Lenharth 0e9c437f0b Got a little ahead of myself here 
llvm-svn: 31502
 2006-11-07 16:40:19 +00:00
Andrew Lenharth 3b18505d9f Fix this test 
llvm-svn: 31501
 2006-11-07 16:32:55 +00:00
Jim Laskey 6a9342166b Use correct value for float HUGH_VAL. 
llvm-svn: 31500
 2006-11-07 12:25:45 +00:00
Evan Cheng b391e3509a Added target hook for post-indexed memory ops transformation. 
llvm-svn: 31499
 2006-11-07 09:04:16 +00:00
Evan Cheng 60c6846d21 Add post-indexed load / store transformations. 
llvm-svn: 31498
 2006-11-07 09:03:05 +00:00
Reid Spencer eb5bbd1ac8 Make some corrections to this test case. 
llvm-svn: 31497
 2006-11-07 07:58:02 +00:00
Reid Spencer 17ea4ec127 Ignore the Output dir. 
llvm-svn: 31496
 2006-11-07 07:34:56 +00:00
Reid Spencer 3b1d1b5262 Add a test case for making sure gdb can get a stack trace with our debug 
information.

llvm-svn: 31495
 2006-11-07 07:31:37 +00:00
Chris Lattner 01d039cc44 Add a new llcbeta option. This speeds up viterbi from 12.34 to 8.76s on 
X86.  If happy, I'll enable this by default.

llvm-svn: 31493
 2006-11-07 07:18:40 +00:00