RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 12:12:47 +01:00
Code Issues Projects Releases Wiki Activity
68,428 Commits 4 Branches 5 Tags 704 MiB
3bc741a0d2
Commit Graph

68428 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Chris Lattner
3bc741a0d2 recognize an unsigned add with overflow idiom into uadd. 
This resolves a README entry and technically resolves PR4916,
but we still get poor code for the testcase in that PR because
GVN isn't CSE'ing uadd with add, filed as PR8817.

Previously we got:

_test7:                                 ## @test7
	addq	%rsi, %rdi
	cmpq	%rdi, %rsi
	movl	$42, %eax
	cmovaq	%rsi, %rax
	ret

Now we get:

_test7:                                 ## @test7
	addq	%rsi, %rdi
	movl	$42, %eax
	cmovbq	%rsi, %rax
	ret

llvm-svn: 122182
 2010-12-19 19:37:52 +00:00
Chris Lattner
faef9b6bfb optimize uadd(x, cst) into a comparison when the normal 
result is dead.  This is required for my next patch to not
regress the testsuite.

llvm-svn: 122181
 2010-12-19 19:35:32 +00:00
Chris Lattner
4ce6f60782 add a version of IRBuilder::SetInsertPoint that takes an instruction. 
llvm-svn: 122180
 2010-12-19 19:16:22 +00:00
Chris Lattner
c5074803db use IC.ReplaceInstUsesWith instead of a raw RAUW so that uses of 
the old thing end up on the instcombine worklist.  Not doing this
can cause an extra top-level iteration of instcombine, burning
compile time.

llvm-svn: 122179
 2010-12-19 18:38:44 +00:00
Chris Lattner
d1f114d8f2 generalize the sadd creation code to not require that the 
sadd formed is half the size of the original type. We can
now compile this into a sadd.i8:

unsigned char X(char a, char b) {
  int res = a+b;
  if ((unsigned )(res+128) > 255U)
    abort();
  return res;
}

llvm-svn: 122178
 2010-12-19 18:35:09 +00:00
Chris Lattner
bb0d067691 fix another miscompile in the llvm.sadd formation logic: it wasn't 
checking to see if the high bits of the original add result were dead.
Inserting a smaller add and zexting back to that size is not good enough.

This is likely to be the fix for 8816.

llvm-svn: 122177
 2010-12-19 18:22:06 +00:00
Cameron Zwarich
163792fb1f Remove some checks for StrongPHIElim. These checks make it impossible to use an 
alternative register allocator that does not require LiveIntervals by specifying
it on the command-line for a target that has StrongPHIElimination enabled by
default.

These checks are pretty meaningless anyways, since StrongPHIElimination and
PHIElimination are never used at the same time.

llvm-svn: 122176
 2010-12-19 18:03:27 +00:00
Chris Lattner
c7876edb16 fix a bug (possibly 8816) in the sadd forming xform: it isn't 
profitable (or safe) to promote code when the add-with-constant
has other uses.

llvm-svn: 122175
 2010-12-19 17:59:02 +00:00
Chris Lattner
cc804862d5 rework the code added in r122072 to pull it out to its own 
helper function, clean up comments, and reduce indentation.
No functionality change.

llvm-svn: 122174
 2010-12-19 17:52:50 +00:00
Chris Lattner
bb93cd80d6 Enhance LICM to promote alias sets whose pointers themselves are stored, 
which doesn't affect the memory address being promoted.

llvm-svn: 122172
 2010-12-19 05:57:25 +00:00
Chris Lattner
71fcecf597 fix PR8602, a bug in an assertion: a volatile store *of* a pointer 
does not make the alias set for that pointer volatile, just stores
*to* the pointer.

llvm-svn: 122171
 2010-12-19 05:51:54 +00:00
Chris Lattner
ac82ea26da fix PR8642: if a critical edge has a PHI value that can trap, 
isel is *required* to split the edge.  PHI values get evaluated
on the edge, not in their predecessor block.

llvm-svn: 122170
 2010-12-19 04:58:57 +00:00
Chris Lattner
70cfded393 remove dead header 
llvm-svn: 122169
 2010-12-19 04:49:11 +00:00
Chris Lattner
0965f3f76d revert r122164, I'm going to go with a different approach. 
llvm-svn: 122168
 2010-12-19 04:23:03 +00:00
Rafael Espindola
894b34478f Fixed version of 122160 (the previous one would fold undefined symbols). 
llvm-svn: 122167
 2010-12-19 04:18:56 +00:00
Rafael Espindola
702b7a3b37 Revert 122160 while I debug it. 
llvm-svn: 122165
 2010-12-19 03:22:05 +00:00
Chris Lattner
14a3e26146 first step to fixing PR8642: don't fold away empty basic blocks 
which have trapping constant exprs in them due to PHI nodes.
Eliminating them can cause the constant expr to be evalutated
on new paths if the input edges are critical.

llvm-svn: 122164
 2010-12-19 03:02:34 +00:00
Chris Lattner
1cc35d2472 move this test into the ARM test so that it is only run when the arm backend 
is enabled.

llvm-svn: 122163
 2010-12-19 02:58:14 +00:00
Rafael Espindola
50fd4aa80f Move all folding to AttemptToFoldSymbolOffsetDifference. 
llvm-svn: 122160
 2010-12-19 02:15:04 +00:00
Michael J. Spencer
0e0c7c269f Support/PathV1: Clarify deprecation warning. 
llvm-svn: 122159
 2010-12-18 22:23:24 +00:00
Michael J. Spencer
6269cd9426 Fix whitespace. 
llvm-svn: 122158
 2010-12-18 22:23:15 +00:00
Michael J. Spencer
4701f704be Support/PathV1: Deprecate get{Basename,Dirname,Suffix}. 
llvm-svn: 122157
 2010-12-18 22:23:07 +00:00
Chris Lattner
25d76d1e53 simplify this a bit. 
llvm-svn: 122156
 2010-12-18 20:22:49 +00:00
Anton Korobeynikov
f49c9c02d6 Restore the behavior of frame lowering before my refactoring. 
It turns out that ppc backend has really weird interdependencies
over different hooks and all stuff is fragile wrt small changes.
This should fix PR8749

llvm-svn: 122155
 2010-12-18 19:53:14 +00:00
Benjamin Kramer
84d3e6cfd0 Just rename the functions, relying on matching a instruction that has the same name as a symbol is way too fragile. 
llvm-svn: 122154
 2010-12-18 14:23:57 +00:00
Benjamin Kramer
4d591385d1 Test more than just label names and make test work on non-x86 hosts. 
llvm-svn: 122153
 2010-12-18 14:07:28 +00:00
Eric Christopher
4e2b6c0e28 Header warning patrol. 
llvm-svn: 122152
 2010-12-18 10:54:29 +00:00
Roman Divacky
ed5bb14415 Add support for lexing single quotes like 'c'. 
This fixed 8615.

llvm-svn: 122150
 2010-12-18 08:56:37 +00:00
Owen Anderson
7103acd18e Revert r122143 through r122140, which collectively broke the LLVMC tests on 
the buildbots.

llvm-svn: 122149
 2010-12-18 07:37:18 +00:00
Rafael Espindola
df98fb74f6 Merge isAbsolute into IsSymbolRefDifferenceFullyResolved. 
llvm-svn: 122148
 2010-12-18 06:27:54 +00:00
Rafael Espindola
7f9be9e112 Remove the MCObjectFormat class. 
llvm-svn: 122147
 2010-12-18 05:37:28 +00:00
Rafael Espindola
4c1b83e53f Add a test that shows that we produce no fixups when computing the difference 
of two symbols in the same fragment.

llvm-svn: 122145
 2010-12-18 05:07:45 +00:00
Rafael Espindola
0b15443dd3 Add a FIXME and explain a hack. 
llvm-svn: 122144
 2010-12-18 04:19:20 +00:00
Michael J. Spencer
b70c703b01 Support/PathV1: Clarify deprecation warning. 
llvm-svn: 122143
 2010-12-18 04:13:54 +00:00
Michael J. Spencer
1bbb8752a8 Fix whitespace. 
llvm-svn: 122142
 2010-12-18 04:13:46 +00:00
Michael J. Spencer
6d64a36ced Support/PathV1: Deprecate get{Basename,Dirname,Suffix}. 
llvm-svn: 122141
 2010-12-18 04:13:36 +00:00
Rafael Espindola
b8369170e9 Fix the note. 
llvm-svn: 122139
 2010-12-18 04:01:45 +00:00
Rafael Espindola
f3aeaaa3f4 Revert 122011, 122012, 122013, 122023 adding back an important optimization. 
I added a note, but suggestions on how to add a test are really welcome.

llvm-svn: 122138
 2010-12-18 03:57:21 +00:00
Jakob Stoklund Olesen
bdf06d6c7b Apparently, operandices is not a word. 
llvm-svn: 122135
 2010-12-18 03:28:32 +00:00
Rafael Espindola
9a4bfd7755 Move some data to the TargetWriter. 
llvm-svn: 122134
 2010-12-18 03:27:34 +00:00
Jakob Stoklund Olesen
e06ded7533 Teach the inline spiller to attempt folding a load instruction into its single 
use before rematerializing the load.

This allows us to produce:

    addps	LCPI0_1(%rip), %xmm2

Instead of:

    movaps	LCPI0_1(%rip), %xmm3
    addps	%xmm3, %xmm2

Saving a register and an instruction. The standard spiller already knows how to
do this.

llvm-svn: 122133
 2010-12-18 03:04:14 +00:00
Jakob Stoklund Olesen
485b7965b3 Tweak debug spew. 
llvm-svn: 122132
 2010-12-18 03:04:11 +00:00
Bill Wendling
9b2ef1c9be r120333 changed the opcode for the Thumb1 stuff from ARM::tMOVr to 
ARM::tMOVgpr2gpr. But this check didn't change. As a result, we were getting
misaligned references to the jump table from an ADR instruction.

There is a test case, but unfortunately it's sensitive to random code changes.

<rdar://problem/8782223>

llvm-svn: 122131
 2010-12-18 02:13:59 +00:00
Bill Wendling
2dde4e38c9 RemoveUnusedCPEntries can change things. Track it. 
llvm-svn: 122129
 2010-12-18 01:53:06 +00:00
Rafael Espindola
dff6c18a5e Test for push being relaxed. 
llvm-svn: 122124
 2010-12-18 01:16:59 +00:00
Jakob Stoklund Olesen
a2f2eab8d4 Check that the register is live-in to the loop header before inserting copies in 
the loop predecessors.

The register can be live-out from a predecessor without being live-in to the
loop header if there is a critical edge from the predecessor.

llvm-svn: 122123
 2010-12-18 01:06:19 +00:00
Nick Lewycky
30eef45106 Fix GCC warning: 
lib/CodeGen/RegAllocGreedy.cpp:311: error: unused variable 'PhysReg' [-Wunused-variable]

llvm-svn: 122122
 2010-12-18 01:05:55 +00:00
Rafael Espindola
ba05bec626 Relax push instructions. 
llvm-svn: 122121
 2010-12-18 01:01:34 +00:00
Nick Lewycky
44a08d5812 Make LazyValueInfo non-recursive. 
llvm-svn: 122120
 2010-12-18 01:00:40 +00:00
Bob Wilson
5f9e78fe20 Rearrange some Neon multiclasses. No functional changes. 
llvm-svn: 122119
 2010-12-18 00:42:58 +00:00