RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-26 14:33:02 +02:00
Code Issues Projects Releases Wiki Activity
30,944 Commits 4 Branches 5 Tags 704 MiB
1087888176
Commit Graph

5041 Commits

Author SHA1 Message Date
Bill Wendling
f9c5de3792 Add support for integer comparison builtins. 
llvm-svn: 35384
 2007-03-27 20:21:31 +00:00
Evan Cheng
201771637e Change findRegisterUseOperand() to return operand index instead. 
llvm-svn: 35363
 2007-03-26 22:37:45 +00:00
Dale Johannesen
3ad8ab7b61 Fix reversed logic in getRegsUsed. Rename RegStates to RegsAvailable to 
hopefully forestall similar errors.

llvm-svn: 35362
 2007-03-26 22:23:54 +00:00
Bill Wendling
a42484728c Add support for the v1i64 type. This makes better code for this: 
#include <mmintrin.h>

extern __m64 C;

void baz(__v2si *A, __v2si *B)
{
  *A = C;
  _mm_empty();
}

We get this:

_baz:
        call "L1$pb"
"L1$pb":
        popl %eax
        movl L_C$non_lazy_ptr-"L1$pb"(%eax), %eax
        movq (%eax), %mm0
        movl 4(%esp), %eax
        movq %mm0, (%eax)
        emms
        ret

GCC gives us this:

_baz:
        pushl   %ebx
        call    L3
"L00000000001$pb":
L3:
        popl    %ebx
        subl    $8, %esp
        movl    L_C$non_lazy_ptr-"L00000000001$pb"(%ebx), %eax
        movl    (%eax), %edx
        movl    4(%eax), %ecx
        movl    16(%esp), %eax
        movl    %edx, (%eax)
        movl    %ecx, 4(%eax)
        emms
        addl    $8, %esp
        popl    %ebx
        ret

llvm-svn: 35351
 2007-03-26 07:53:08 +00:00
Chris Lattner
cc91b2a91d Start value symbol tables out small (space for 16 elts), not huge (space for 512). 
This is particularly useful for the JIT, which lazily deserializes functions.

llvm-svn: 35346
 2007-03-26 01:45:35 +00:00
Reid Spencer
4984ab6c01 Compute getLowBitsSet correctly. Using the complement of a 64-bit value 
and shifting down without regard for the bitwidth of the APInt can lead
to incorrect initialization values. Instead, check for the word size case
(to avoid undef results from shift) and then do (1 << loBitsSet) - 1

llvm-svn: 35344
 2007-03-25 21:58:42 +00:00
Anton Korobeynikov
6f78c59650 First step of switch lowering refactoring: perform worklist-driven 
strategy, emit JT's where possible.

llvm-svn: 35338
 2007-03-25 15:07:15 +00:00
Chris Lattner
b19069959d switch TargetLowering::getConstraintType to take the entire constraint, 
not just the first letter.  No functionality change.

llvm-svn: 35322
 2007-03-25 02:14:49 +00:00
Reid Spencer
b5b291924a Fix a typo in a comment. 
llvm-svn: 35316
 2007-03-25 01:13:46 +00:00
Reid Spencer
16d08887ef Actually, for getHighBitsSet and getLowBitsSet, don't make a 0 bit size 
illegal. Instead do the 0 valued construction for the user. This is because
the caller may not know (or care to check) that the number of bits set is
zero.

llvm-svn: 35315
 2007-03-25 00:01:47 +00:00
Reid Spencer
d24dc1f9a6 Make it illegal to set 0 bits in getHighBitsSet and getLowBitsSet. For that 
they should have used the uint64_t constructor. This avoids causing
undefined results via shifts by the word size when the bit width is an
exact multiple of the word size.

llvm-svn: 35313
 2007-03-24 23:47:58 +00:00
Reid Spencer
fb188ea881 In the getBitsSet function, don't optimize for a common case that is 
already covered by getLowBitsSet (i.e. when loBits==0). Consequently, remove
the default value for loBits and reorder the arguments to the more natural
loBits, hiBits order. This makes it more clear that this function is for bit
groups in the middle of the bit width and not towards one end or the other.

llvm-svn: 35312
 2007-03-24 23:42:47 +00:00
Reid Spencer
091d6cf3b7 Don't invoke undefined behavior in shifts in the functions getHighBitsSet 
and getLowBitsSet.

llvm-svn: 35311
 2007-03-24 23:35:54 +00:00
Reid Spencer
0e466b183f Implement the getBitsSet function. 
llvm-svn: 35310
 2007-03-24 23:27:48 +00:00
Reid Spencer
1aa3f64ef4 Implement the getHighBitsSet and getLowBitsSet functions. 
llvm-svn: 35308
 2007-03-24 23:05:35 +00:00
Reid Spencer
30351837ca Get the signs in the right place! 
llvm-svn: 35307
 2007-03-24 22:50:43 +00:00
Reid Spencer
0ceb493ee2 Undo the last change and make this really implement remainder and not 
modulus. The previous change was a result of incorrect documentation in
the LangRef.html.

llvm-svn: 35305
 2007-03-24 22:37:23 +00:00
Reid Spencer
c1dfceab8b Correct the implementation of srem to be remainder, not modulus. The sign of 
the result must follow the sign of the divisor.

llvm-svn: 35302
 2007-03-24 21:56:22 +00:00
Reid Spencer
52ea41bb97 Clean up this interface: 
1. Group similar methods into doxygen groups
2. Reorganize the groups into a consist flow.
3. Significantly improve the quality of the documentation on several methods
4. Rewrite srem and sdiv to eliminate a copy and improve readability.
5. Eliminate unneeded forward references.

llvm-svn: 35300
 2007-03-24 18:09:18 +00:00
Chris Lattner
3a16450eaf add a valuetype for v1i64, which is needed by mmx. 
llvm-svn: 35298
 2007-03-24 17:36:26 +00:00
Bill Wendling
ad01036d5c This is dead. DEAD I tells you!! 
llvm-svn: 35291
 2007-03-23 22:42:04 +00:00
Bill Wendling
124f2c8706 PR1260: 
Add final support to get the QT example to compile.

llvm-svn: 35290
 2007-03-23 22:35:46 +00:00
Dan Gohman
42a3f4e16e Add the 'explicit' keyword to several constructors that accept one 
argument that don't appear intended as implicit-conversion operators.

llvm-svn: 35280
 2007-03-23 18:44:11 +00:00
Bill Wendling
e6a9c6dfe6 We generate a shufflevector instruction, so we don't need the builtin 
intrinsic.

llvm-svn: 35269
 2007-03-22 20:29:26 +00:00
Jeff Cohen
2f87ca6cc7 Be more explicit concerning argument sizes. 
Use VC++ byteswap intrinsics.

llvm-svn: 35267
 2007-03-22 19:11:57 +00:00
Bill Wendling
1bcad4c1cd Support added for shifts and unpacking MMX instructions. 
llvm-svn: 35266
 2007-03-22 18:42:45 +00:00
Reid Spencer
17f3eeaf48 For PR1145: 
Add the NoUnwindAttribute to allow explicit marking of functions that do
not unwind the stack via the unwind instruction. This is akin to gcc's
NoThrow attribute and used for optimizing exception handling.

llvm-svn: 35250
 2007-03-22 00:32:02 +00:00
Reid Spencer
76c1fef30a Fix a comment. 
llvm-svn: 35246
 2007-03-21 22:22:19 +00:00
Dale Johannesen
44c0a5d545 repair x86 performance, dejagnu problems from previous change 
llvm-svn: 35245
 2007-03-21 21:51:52 +00:00
Dale Johannesen
3e422e3b49 do not share old induction variables when this would result in invalid 
instructions (that would have to be split later)

llvm-svn: 35227
 2007-03-20 21:54:54 +00:00
Dale Johannesen
50a22f6fe6 maintain LiveIn when splitting blocks (register scavenging needs it) 
llvm-svn: 35226
 2007-03-20 21:35:06 +00:00
Devang Patel
19dae84536 LoopSimplify::FindPHIToPartitionLoops() 
Use ETForest instead of DominatorSet.

llvm-svn: 35222
 2007-03-20 20:19:48 +00:00
Evan Cheng
0176b9859d First cut trivial re-materialization support. 
llvm-svn: 35208
 2007-03-20 08:13:50 +00:00
Evan Cheng
41f4f032ee Added MRegisterInfo hook to re-materialize an instruction. 
llvm-svn: 35205
 2007-03-20 08:09:38 +00:00
Zhou Sheng
a9c0018850 Correct the name: isStrictPositive --> isStrictlyPositive. 
llvm-svn: 35201
 2007-03-20 02:18:16 +00:00
Chris Lattner
305c3525af Add a dtor to fix leaks from all clients of BitVector. 
llvm-svn: 35200
 2007-03-20 02:10:56 +00:00
Reid Spencer
7d20e42a97 Allow ConstantInt::get(Ty, uint64_t) to interpret the 64-bit values as a 
negative number. This is needed to fix test/Assembler/2007-03-19-NegValue.ll

llvm-svn: 35181
 2007-03-19 20:39:08 +00:00
Reid Spencer
c3422e56d6 Add an indication of signedness to the uint64_t constructor so sign bits 
can be extended. This helps fix test/Assembler/2007-03-19-NegValue.ll

llvm-svn: 35179
 2007-03-19 20:36:48 +00:00
Duncan Sands
90c42a15d5 Fix obvious typo in comment. 
llvm-svn: 35164
 2007-03-19 14:25:02 +00:00
Evan Cheng
2081cf96ed Add a TargetInstrDescriptor flag to mark an instruction as "re-materializable". 
It means the instruction can be easily re-materialized at any point. e.g.
constant generation, load from constantpool.

llvm-svn: 35158
 2007-03-19 06:19:16 +00:00
Zhou Sheng
6ecf1045a6 Add isStrictPositive() to APInt to determine if this APInt Value > 0. 
llvm-svn: 35156
 2007-03-19 05:22:18 +00:00
Evan Cheng
1d943badac Fix comment. 
llvm-svn: 35139
 2007-03-18 03:26:04 +00:00
Evan Cheng
5be8544e8a Track the BB's where each virtual register is used. 
llvm-svn: 35135
 2007-03-17 09:29:54 +00:00
Evan Cheng
ce8b779c6c Added isLegalAddressExpression hook to test if the given expression can be 
folded into target addressing mode for the given type.

llvm-svn: 35121
 2007-03-16 08:42:32 +00:00
Evan Cheng
0408dc0d37 These forward declarations are not needed. 
llvm-svn: 35120
 2007-03-16 08:41:06 +00:00
Bill Wendling
feaff80149 Multiplication support for MMX. 
llvm-svn: 35118
 2007-03-15 21:24:36 +00:00
Jeff Cohen
b19ec59b5c Fix for VS 2005 problem supplied by Morten Ofstad. 
llvm-svn: 35098
 2007-03-14 15:25:21 +00:00
Zhou Sheng
dfcf9376bc Add zextOrCopy() into APInt for convenience. 
llvm-svn: 35079
 2007-03-13 06:16:26 +00:00
Evan Cheng
f132fa5bba More flexible TargetLowering LSR hooks for testing whether an immediate is 
a legal target address immediate or scale.

llvm-svn: 35071
 2007-03-12 23:26:27 +00:00
Zhou Sheng
9643d0a422 Add getSignBit() and operator<<= into APInt for convenience. 
llvm-svn: 35059
 2007-03-11 07:16:10 +00:00