1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-31 16:02:52 +01:00
Commit Graph

94126 Commits

Author SHA1 Message Date
NAKAMURA Takumi
ade122d371 LLVMSymbolize.cpp: Fix build. Triple::ArchType is not a namespace.
llvm-svn: 186494
2013-07-17 06:53:51 +00:00
Alexey Samsonov
5256cb6f86 llvm-symbolizer: be more careful with colons in file names
llvm-svn: 186493
2013-07-17 06:45:36 +00:00
Craig Topper
f16a718df3 Make x86 fast-isel correctly choose between aligned and unaligned operations for vector stores. Fixes PR16640.
llvm-svn: 186491
2013-07-17 05:57:45 +00:00
JF Bastien
05ee680a75 Fix ARMFastISel::ARMEmitIntExt shift emission
My patch 'r183551 - ARM FastISel integer sext/zext improvements' was incorrect when emitting ARM register-immediate ASR, LSL, LSR instructions: they are pseudo-instructions in ARMInstrInfo.td and I should have used MOVsi instead.

This is not an issue when code is generated through a .s file, but is an issue when generated straight to a .o (-filetype=obj).

llvm-svn: 186489
2013-07-17 05:46:46 +00:00
Hal Finkel
149f358122 PPC: Add CTR-register clobber to builtin setjmp
Because the builtin longjmp implementation uses a CTR-based indirect jump, when
the control flow arrives at the builtin setjmp call, the CTR register has
necessarily been clobbered. Correspondingly, this adds CTR to the list of
implicit definitions of the builtin setjmp pseudo instruction.

We don't need to add CTR to the implicit definitions of builtin longjmp
because, even though it does clobber the CTR register, the control flow cannot
return to inside the loop unless there is also a builtin setjmp call.

llvm-svn: 186488
2013-07-17 05:35:44 +00:00
Rafael Espindola
52e19a1bf7 Add simpler version of is_directory. It will be used in clang.
llvm-svn: 186486
2013-07-17 04:20:49 +00:00
Craig Topper
f263d3bfb8 Mark a method 'const' and another 'static'.
llvm-svn: 186485
2013-07-17 03:54:53 +00:00
Craig Topper
2bf5d5a2f0 Make a few more static string pointers constant.
llvm-svn: 186484
2013-07-17 03:43:10 +00:00
Rafael Espindola
2451973efd Don't fallback to copy + delete in rename.
Rename's documentation says "Files are renamed as if by POSIX rename()". and it
is used for atomically updating output files from a temporary. Having rename
fallback to a non atomic copy has the potential to hide bugs, like using
a temporary file in /tmp instead of a unique name next to the final destination.

llvm-svn: 186483
2013-07-17 03:33:41 +00:00
Craig Topper
c6d133943e Make constant string pointer into an array to remove a pointer lookup for every access.
llvm-svn: 186482
2013-07-17 03:11:32 +00:00
NAKAMURA Takumi
26b99c6974 raw_ostream.cpp: Introduce <fcntl.h> to let O_BINARY provided. Or, llvm::outs() would be set to O_TEXT by default.
llvm/test/Object/check_binary_output.ll is expected to pass on win32.

llvm-svn: 186480
2013-07-17 02:21:10 +00:00
Nadav Rotem
ae8b6de415 SLPVectorizer: Accelerate the isConsecutive check by replacing the subtraction of the two values with a simple SCEV expression that adds the offset to one of the pointers that we compare.
llvm-svn: 186479
2013-07-17 00:48:31 +00:00
Hal Finkel
e625744d86 PPC: Implement base pointer and stack realignment
This builds on some frame-lowering code that has existed since 2005 (r24224)
but was disabled in 2008 (r48188) because it needed base pointer support to
function correctly. This implementation follows the strategy suggested by Dale
Johannesen in r48188 where the following comment was added:

  This does not currently work, because the delta between old and new stack
  pointers is added to offsets that reference incoming parameters after the
  prolog is generated, and the code that does that doesn't handle a variable
  delta.  You don't want to do that anyway; a better approach is to reserve
  another register that retains to the incoming stack pointer, and reference
  parameters relative to that.

And now we do exactly that. If we don't need a frame pointer, then we use r31
as a base pointer. If we do need a frame pointer, then we use r30 as a base
pointer. The base pointer retains the value of the stack pointer before it was
decremented in the prologue. We then use the base pointer to resolve all
negative frame indicies. The basic scheme follows that for base pointers in the
X86 backend.

We use a base pointer when we need to dynamically realign the incoming stack
pointer. This currently applies only to static objects (dynamic allocas with
large alignments, and base-pointer support in SjLj lowering will come in future
commits).

llvm-svn: 186478
2013-07-17 00:45:52 +00:00
NAKAMURA Takumi
7b93767d62 llvm/test/CodeGen/X86/vec_setcc.ll: Add explicit -mtriple=x86_64-unknown-unknown to satisfy win32-targeted configuration.
llvm-svn: 186477
2013-07-17 00:42:37 +00:00
Craig Topper
0910c52f8e Move string pointer from being a static class member to just a static global in the one file its needed in.
llvm-svn: 186476
2013-07-17 00:31:35 +00:00
Manman Ren
929ebf85f2 Add getModuleFlag(StringRef Key) to query a module flag given Key.
No functionality change.

llvm-svn: 186470
2013-07-16 23:21:16 +00:00
NAKAMURA Takumi
67f8f68d48 llvm/test/Object/ar-create.test: Relax a CHECK line to satisfy localized message catalogue.
For example, 'No such file or directory' cannot be seen on Japanese version of msvcrt.

llvm-svn: 186469
2013-07-16 23:17:22 +00:00
NAKAMURA Takumi
3c613c0b6c llvm/test/Object/check_binary_output.ll: Mark it as XFAIL on Windows. Investigating.
llvm-svn: 186468
2013-07-16 23:16:57 +00:00
Nadav Rotem
adfe58a7ad flip the scev minus direction to simplify the code.
llvm-svn: 186466
2013-07-16 22:57:06 +00:00
Nadav Rotem
633bd23118 SLPVectorizer: Improve the compile time of isConsecutive by adding a simple constant-gep check before using SCEV.
This check does not always work because not all of the GEPs use a constant offset, but it happens often enough to reduce the number of times we use SCEV.

llvm-svn: 186465
2013-07-16 22:51:07 +00:00
Lang Hames
42e80f638a Related to r181161 - Indirect branches may not be the last branch in a basic
block. Blocks that have an indirect branch terminator, even if it's not the
last terminator, should still be treated as unanalyzable.

<rdar://problem/14437274>

Reducing a useful regression test case is proving difficult - I hope to have
one soon.

llvm-svn: 186461
2013-07-16 22:01:40 +00:00
Tilmann Scheller
5c5d0d2141 ARM: Add support for the Thumb2 PLI alternate literal form.
This adds an instruction alias to make the assembler recognize the alternate literal form: pli [PC, #+/-<imm>]

See A8.8.129 in the ARM ARM (DDI 0406C.b).

Fixes <rdar://problem/14403733>.

llvm-svn: 186459
2013-07-16 21:52:34 +00:00
Rafael Espindola
8abfbec9d8 Update the examples for an API change.
llvm-svn: 186453
2013-07-16 20:22:35 +00:00
Rafael Espindola
2a9326a78f Add a wrapper for open.
This centralizes the handling of O_BINARY and opens the way for hiding more
differences (like how open behaves with directories).

llvm-svn: 186447
2013-07-16 19:44:17 +00:00
Benjamin Kramer
6e6528e46d Finally, force the target for this test. Should unbreak non-x86 buildbots.
llvm-svn: 186445
2013-07-16 19:22:07 +00:00
Rafael Espindola
0e87cce1d6 XFAIL this test on mingw.
llvm-svn: 186444
2013-07-16 19:20:29 +00:00
Benjamin Kramer
876b63a443 Label names also differ between platforms. Use a relaxed regex.
llvm-svn: 186442
2013-07-16 18:54:21 +00:00
Benjamin Kramer
1459dae6ee Fix test not to fail when the target doesn't use leading underscores on symbols.
llvm-svn: 186439
2013-07-16 18:42:01 +00:00
Manman Ren
c67f77c5d6 Cleanup testing case by using a shorter name for types.
llvm-svn: 186436
2013-07-16 18:26:48 +00:00
Jakob Stoklund Olesen
e0d15d47e1 Remove floats from live range splitting costs.
These floats all represented block frequencies anyway, so just use the
BlockFrequency class directly.

Some floating point computations remain in tryLocalSplit(). They are
estimating spill weights which are still floats.

llvm-svn: 186435
2013-07-16 18:26:18 +00:00
Jakob Stoklund Olesen
0d3ae14227 Reapply r185393.
Original commit message:

Remove floating point computations from SpillPlacement.cpp.

Patch by Benjamin Kramer!

Use the BlockFrequency class instead of floats in the Hopfield network
computations. This rescales the node Bias field from a [-2;2] float
range to two block frequencies BiasN and BiasP pulling in opposite
directions. This construct has a more predictable behavior when block
frequencies saturate.

The per-node scaling factors are no longer necessary, assuming the block
frequencies around a bundle are consistent.

This patch can cause the register allocator to make different spilling
decisions. The differences should be small.

llvm-svn: 186434
2013-07-16 18:26:15 +00:00
Juergen Ributzka
e612fc1230 [X86] Use min/max to optimze unsigend vector comparison on X86
Use PMIN/PMAX for UGE/ULE vector comparions to reduce the number of required
instructions. This trick also works for UGT/ULT, but there is no advantage in
doing so. It wouldn't reduce the number of instructions and it would actually
reduce performance.

Reviewer: Ben

radar:5972691

llvm-svn: 186432
2013-07-16 18:20:45 +00:00
Peter Collingbourne
82d932d6c2 Make SpecialCaseList match full strings, as documented, using anchors.
Differential Revision: http://llvm-reviews.chandlerc.com/D1149

llvm-svn: 186431
2013-07-16 17:56:07 +00:00
Juergen Ributzka
9f40835ba8 Test commit to verify write access.
llvm-svn: 186429
2013-07-16 17:44:23 +00:00
Reid Kleckner
362d3541d3 [Support] Add a Unicode conversion wrapper from UTF16 to UTF8
This is to support parsing UTF16 response files in LLVM/lib/Option for
lld and clang.

Reviewers: hans

Differential Revision: http://llvm-reviews.chandlerc.com/D1138

llvm-svn: 186426
2013-07-16 17:14:33 +00:00
Hal Finkel
35292d605d When the inliner merges allocas, it must keep the larger alignment
For safety, the inliner cannot decrease the allignment on an alloca when
merging it with another.

I've included two variants of the test case for this: one with DataLayout
available, and one without. When DataLayout is not available, if only one of
the allocas uses the default alignment (getAlignment() == 0), then they cannot
be safely merged.

llvm-svn: 186425
2013-07-16 17:10:55 +00:00
Rafael Espindola
77c12b390e On error, close the temporary file descriptor.
With this change llvm-ar can remove the temporary file on windows too.

llvm-svn: 186423
2013-07-16 16:00:32 +00:00
Nadav Rotem
ebe95f88ed SLPVectorizer: Reduce the compile time of the consecutive store lookup.
Process groups of stores in chunks of 16.

llvm-svn: 186420
2013-07-16 15:25:17 +00:00
Rafael Espindola
3a0a77998b Create files with mode 666. This matches the behavior of other unix tools.
llvm-svn: 186414
2013-07-16 14:10:07 +00:00
Reid Kleckner
a037b922d1 [Support] Fix some warnings when self-hosting clang on Windows
llvm-svn: 186413
2013-07-16 14:04:08 +00:00
Ulrich Weigand
c1b627a527 [APFloat] PR16573: Avoid losing mantissa bits in ppc_fp128 to double truncation
When truncating to a format with fewer mantissa bits, APFloat::convert
will perform a right shift of the mantissa by the difference of the
precision of the two formats.  Usually, this will result in just the
mantissa bits needed for the target format.

One special situation is if the input number is denormal.  In this case,
the right shift may discard significant bits.  This is usually not a
problem, since truncating a denormal usually results in zero (underflow)
after normalization anyway, since the result format's exponent range is
usually smaller than the target format's.

However, there is one case where the latter property does not hold:
when truncating from ppc_fp128 to double.  In particular, truncating
a ppc_fp128 whose first double of the pair is denormal should result
in just that first double, not zero.  The current code however
performs an excessive right shift, resulting in lost result bits.
This is then caught in the APFloat::normalize call performed by
APFloat::convert and causes an assertion failure.

This patch checks for the scenario of truncating a denormal, and
attempts to (possibly partially) replace the initial mantissa
right shift by decrementing the exponent, if doing so will still
result in a valid *target format* exponent.


Index: test/CodeGen/PowerPC/pr16573.ll
===================================================================
--- test/CodeGen/PowerPC/pr16573.ll	(revision 0)
+++ test/CodeGen/PowerPC/pr16573.ll	(revision 0)
@@ -0,0 +1,11 @@
+; RUN: llc < %s | FileCheck %s
+
+target triple = "powerpc64-unknown-linux-gnu"
+
+define double @test() {
+  %1 = fptrunc ppc_fp128 0xM818F2887B9295809800000000032D000 to double
+  ret double %1
+}
+
+; CHECK: .quad -9111018957755033591
+
Index: lib/Support/APFloat.cpp
===================================================================
--- lib/Support/APFloat.cpp	(revision 185817)
+++ lib/Support/APFloat.cpp	(working copy)
@@ -1956,6 +1956,23 @@
     X86SpecialNan = true;
   }
 
+  // If this is a truncation of a denormal number, and the target semantics
+  // has larger exponent range than the source semantics (this can happen
+  // when truncating from PowerPC double-double to double format), the
+  // right shift could lose result mantissa bits.  Adjust exponent instead
+  // of performing excessive shift.
+  if (shift < 0 && isFiniteNonZero()) {
+    int exponentChange = significandMSB() + 1 - fromSemantics.precision;
+    if (exponent + exponentChange < toSemantics.minExponent)
+      exponentChange = toSemantics.minExponent - exponent;
+    if (exponentChange < shift)
+      exponentChange = shift;
+    if (exponentChange < 0) {
+      shift -= exponentChange;
+      exponent += exponentChange;
+    }
+  }
+
   // If this is a truncation, perform the shift before we narrow the storage.
   if (shift < 0 && (isFiniteNonZero() || category==fcNaN))
     lostFraction = shiftRight(significandParts(), oldPartCount, -shift);

llvm-svn: 186409
2013-07-16 13:03:25 +00:00
Richard Osborne
e37374c506 [XCore] Fix printing of inline asm operands.
Previously an asm operand with no operand modifier would give the error
"invalid operand in inline asm".

llvm-svn: 186407
2013-07-16 12:48:34 +00:00
Tim Northover
be10a35a43 ARM: allow printing of ARM atomic DAG nodes.
We'd forgotten to provide string representations for the special ARMISD atomic
nodes; this adds them in. No effect on CodeGen, just makes the output of
"-view-whatever-dags" slightly more readable.

llvm-svn: 186406
2013-07-16 12:15:36 +00:00
Richard Sandiford
ab0fb439a9 [SystemZ] Use ROSBG and non-zero form of RISBG for OR nodes
llvm-svn: 186405
2013-07-16 11:55:57 +00:00
Vladimir Medic
a067a86f5e Fixing a buildbot failure:unused function.
llvm-svn: 186403
2013-07-16 11:43:20 +00:00
Richard Sandiford
fde8a92af9 [SystemZ] Add MC support for R[NOX]SBG
CodeGen support will come later.

llvm-svn: 186401
2013-07-16 11:28:08 +00:00
Richard Sandiford
c99f769478 [SystemZ] Use RISBG for (shift (and ...))
Another patch in the series to make more use of R.SBG.  This one extends
r186072 and r186073 to handle cases where the AND is inside the shift.

llvm-svn: 186399
2013-07-16 11:02:24 +00:00
Vladimir Medic
e3aeddca47 This patch represents Mips utilization of r186388 code that alows asm matcher to emit mnemonics contain '.' characters. This makes asm parser code simpler and more efficient.
llvm-svn: 186397
2013-07-16 10:07:14 +00:00
NAKAMURA Takumi
c21b0543b6 PPCJITInfo.cpp: Tweak r186252 with s/__ppc/__powerpc/ to work on powerpc-linux Fedora 12.
g++ (GCC) 4.4.4 20100630 (Red Hat 4.4.4-10)

llvm-svn: 186396
2013-07-16 09:59:51 +00:00
Tim Northover
69d676cd12 ARM: implement ldrex, strex and clrex intrinsics
Intrinsics already existed for the 64-bit variants, so these support operations
of size at most 32-bits.

llvm-svn: 186392
2013-07-16 09:46:55 +00:00