Commit Graph

91876 Commits

Author SHA1 Message Date
Arnold Schwaighofer
87defd0924 LoopVectorize: Add support for floating point min/max reductions
Add support for min/max reductions when "no-nans-float-math" is enabled. This
allows us to assume we have ordered floating point math and treat ordered and
unordered predicates equally.

radar://13723044

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181144 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-05 01:54:48 +00:00
Arnold Schwaighofer
e79d92c592 PatternMatch: Matcher for (un)ordered floating point min/max
Add support for matching 'ordered' and 'unordered' floating point min/max
constructs.

In LLVM we can express min/max functions as a combination of compare and select.
We have support for matching such constructs for integers but not for floating
point. In floating point math there is no total order because of the presence of
'NaN'. Therefore, we have to be careful to preserve the original fcmp semantics
when interpreting floating point compare select combinations as a minimum or
maximum function. The resulting 'ordered/unordered' floating point maximum
function has to select the same value as the select/fcmp combination it is based
on.

 ordered_max(x,y)   = max(x,y) iff x and y are not NaN, y otherwise
 unordered_max(x,y) = max(x,y) iff x and y are not NaN, x otherwise
 ordered_min(x,y)   = min(x,y) iff x and y are not NaN, y otherwise
 unordered_min(x,y) = min(x,y) iff x and y are not NaN, x otherwise

This matches the behavior of the underlying select(fcmp(olt/ult/.., L, R), L, R)
construct.

Any code using this predicate has to preserve this semantics.

A follow-up patch will use this to implement floating point min/max reductions
in the vectorizer.

radar://13723044

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181143 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-05 01:54:46 +00:00
Arnold Schwaighofer
f852472823 LoopVectorizer: Cleanup of miminimum/maximum pattern match code
No need for setting the operands. The pointers are going to be bound by the
matcher.

radar://13723044

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181142 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-05 01:54:44 +00:00
Arnold Schwaighofer
c1738fdadd LoopVectorize: We don't need an identity element for min/max reductions
We can just use the initial element that feeds the reduction.

  max(max(x, y), z) == max(max(x,y), max(x,z))

radar://13723044

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181141 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-05 01:54:42 +00:00
Dmitri Gribenko
5c332dbd30 Add ArrayRef constructor from None, and do the cleanups that this constructor enables
Patch by Robert Wilhelm.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181138 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-05 00:40:33 +00:00
Nadav Rotem
76be9bf67e whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181137 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 23:27:32 +00:00
Nadav Rotem
0fb6523eae Fix an odd comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181136 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 23:24:56 +00:00
Tim Northover
7428539a1c AArch64: enable MCJIT and tests now that everything passes.
This removes dire warnings about AArch64 being unsupported and enables
the tests when appropriate on this platform.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181135 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 20:14:22 +00:00
Tim Northover
d52eaae157 AArch64: implement 64-bit absolute relocation in MCJIT
This is about the simplest relocation, but surprisingly rare in actual
code.

It occurs in (for example) the MCJIT test test-ptr-reloc.ll.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181134 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 20:14:14 +00:00
Tim Northover
4a9b6b798d AArch64: add stubs to support long function calls on MCJIT
As with global accesses, external functions could exist anywhere in
memory. Therefore the stub must create a complete 64-bit address. This
patch implements the fragment as (roughly):
    movz x16, #:abs_g3:somefunc
    movk x16, #:abs_g2_nc:somefunc
    movk x16, #:abs_g1_nc:somefunc
    movk x16, #:abs_g0_nc:somefunc
    br x16

In principle we could save 4 bytes by using a literal-load instead,
but it is unclear that would be more efficient and can only be tested
when real hardware is readily available.

This allows (for example) the MCJIT test 2003-05-07-ArgumentTest to
pass on AArch64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181133 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 20:14:09 +00:00
Tim Northover
654c2d6b7a AArch64: implement relocations for global access
The large memory model (default and main viable for JIT) emits
addresses in need of relocation as
    movz x0, #:abs_g3:somewhere
    movk x0, #:abs_g2_nc:somewhere
    movk x0, #:abs_g1_nc:somewhere
    movk x0, #:abs_g0_nc:somewhere

To support this we must implement those four relocations in the
dynamic loader.

This allows (for example) the test-global.ll MCJIT test to pass on
AArch64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181132 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 20:14:04 +00:00
Tim Northover
85829bb98a AArch64: implement first relocation required for MCJIT
R_AARCH64_PCREL32 is present in even trivial .eh_frame sections and so
is required to compile any function without the "nounwind" attribute.

This change implements very basic infrastructure in the RuntimeDyldELF
file and allows (for example) the test-shift.ll MCJIT test to pass
on AArch64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181131 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 20:13:59 +00:00
Tim Northover
bd121f5b3a Build system changes to enable MCJIT on AArch64
These changes just allow AArch64 to take part in the MCJIT world when
built correctly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181130 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 20:13:52 +00:00
Tim Northover
9a29cf281e AArch64: use __clear_cache under GCCish environments
AArch64 is going to need some kind of cache-invalidation in order to
successfully JIT since it has a weak memory-model. This is provided by
a __clear_cache builtin in libgcc, which acts very much like the
32-bit ARM equivalent (on platforms where it exists).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181129 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 18:52:44 +00:00
Richard Osborne
7153a30610 Fix buildbot failure on 64 bit linux due to std::max() having different
operand types.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181128 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 17:41:01 +00:00
Richard Osborne
43c7abee14 [XCore] Remove unused operand type.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181127 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 17:30:05 +00:00
Richard Osborne
6ffbf6ea8f [XCore] Make use of the target independent global address offset folding.
This let us to remove some custom code that matched constant offsets
from globals at instruction selection time as a special addressing mode.
No intended functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181126 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 17:24:33 +00:00
Richard Osborne
40827bc716 [XCore] Simplify code that checks for an aligned base plus a constant.
The code now makes use of ComputeMaskedBits,
SelectionDAG::isBaseWithConstantOffset and TargetLowering::isGAPlusOffset
where appropriate reducing the amount of logic needed in XCoreISelLowering.
No intended functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181125 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 17:17:10 +00:00
Richard Osborne
597432fbe5 [XCore] Move lowering of thread local storage to a separate pass.
Thread local storage is not supported by the XMOS linker so we handle
thread local variables by lowering the variable to an array of n elements
(where n is the number of hardware threads per core, currently 8
for all XMOS devices) indexed by the the current thread ID.

Previously this lowering was spread across the XCoreISelLowering and the
XCoreAsmPrinter classes. Moving this to a separate pass should be much
cleaner.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181124 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 17:01:55 +00:00
Tim Northover
effc16bb49 AArch64: assert code model is small for TLS accesses
Supporting TLS in the large memory model is rather difficult at the
moment, so make sure no-one gets into difficulties by mistake.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181121 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 16:54:11 +00:00
Tim Northover
79c1c092df AArch64: support literal pool access in large memory model.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181120 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 16:54:07 +00:00
Tim Northover
cd1b09b25b AArch64: support large code model for jump-tables
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181119 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 16:54:00 +00:00
Tim Northover
b2efdde06c AArch64: implement support for blockaddress in large code model
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181118 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 16:53:53 +00:00
Tim Northover
45db92038b AArch64: implement large code model access to global variables.
The MOVZ/MOVK instruction sequence may not be the most efficient (a
literal-pool load could be better) but adding that would require
reinstating the ConstantIslands pass.

For now the sequence is correct, and that's enough. Beware, as of
commit GNU ld does not appear to support the relocations needed for
this. Its primary purpose (for now) will be to support JITed code,
since in that case there is no guarantee of where your code will end
up in memory relative to external symbols it references.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181117 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 16:53:46 +00:00
Richard Osborne
0c66f30325 [XCore] Use static relocation model by default.
This allows us to get get rid of a hack in XCoreTargetObjectFile where the
the DataRel* sections were overridden.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181116 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 16:40:58 +00:00
Tim Northover
d66ad6c571 Allow host triple to be correctly overridden in CMake builds
The intended semantics mirror autoconf, where the user is able to
specify a host triple, but if it's left to the build system then
"config.guess" is invoked for the default.

This also renames the LLVM_HOSTTRIPLE define to LLVM_HOST_TRIPLE to
fit in with the style of the surrounding defines.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181112 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 07:36:23 +00:00
Rafael Espindola
c1685b3a50 Optimize llvm-link too.
This takes the linking of almost all modules in a clang build from 6:32
to 0:19.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181105 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 05:30:49 +00:00
Rafael Espindola
cfb320f5f9 Fix a performance bug in the Linker.
Now that we hava a convinient place to keep it, remeber the set of
identified structs as we merge modules.

This speeds up the linking of all the bitcode files in clang with the
gold plugin and -plugin-opt=emit-llvm (i.e., link only, no codegen) from
5:25 minutes to 13.6 seconds!

Patch by Xiaofei Wan!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181104 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 05:05:18 +00:00
Rafael Espindola
2e013028f2 Implement Linker::LinkModules with Linker::linkInModule.
Flipping which one is the implementation will let us optimize linkInModule.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181102 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 04:08:02 +00:00
Rafael Espindola
c7c35a9b86 Now that Linker.cpp is almost empty, merge it into LinkModules.cpp.
Also remove unused includes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181100 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 03:48:37 +00:00
Rafael Espindola
fca8863165 Last batch of cleanups to Linker.h.
Update comments, fix * placement, fix method names that are not
used in clang, add a linkInModule that takes a Mode and put it
in Linker.cpp.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181099 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 03:06:50 +00:00
Rafael Espindola
105193772d Don't construct or delete a module on the Linker.
The linker is now responsible only for actually linking the modules, it
is up to the clients to create and destroy them.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181098 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 02:43:00 +00:00
Rafael Espindola
bcc6ac93ba Don't store the context in the Linker.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181097 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 02:34:41 +00:00
Rafael Espindola
ae8f1f3fde Remove unused members and constructor arguments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181096 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 02:28:57 +00:00
Rafael Espindola
b00b4bed57 Add missing header.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181095 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 02:21:46 +00:00
Rafael Espindola
e6315cd32b Delete dead code from the linker.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181094 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 02:13:18 +00:00
Krzysztof Parzyszek
cc420a0911 Use consistent function names.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181090 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 01:30:49 +00:00
Nick Lewycky
cd4e5e9b34 Tabs to spaces. No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181082 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-04 01:08:15 +00:00
Amara Emerson
1aef163a68 Revert r181009.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181079 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 23:57:17 +00:00
Reed Kotler
2bb955a693 Remove some uneeded pseudos in the presence of the naked function attribute.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181072 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 23:17:24 +00:00
Amara Emerson
3170ad7dd4 Delete test instead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181066 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 22:39:03 +00:00
Amara Emerson
224e114e71 Temporarily disable failing test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181062 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 22:27:48 +00:00
Bill Wendling
b41528507c Remove this hack. We can support this better with function attributes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181059 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 21:53:50 +00:00
Ulrich Weigand
fcdfd5a7ff [PowerPC] Avoid using '$' in generated assembler code
PowerPC assemblers are supposed to support a stand-alone '$' symbol
as an alternative of '.' to refer to the current PC.  This does not
work in the LLVM assembler parser yet.

To avoid bootstrap failures when using the LLVM assembler as system
assembler, this patch modifies the assembler source code generated
by LLVM to avoid using '$' (and simply use '.' instead).



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181054 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 19:53:04 +00:00
Ulrich Weigand
a7e5e6b959 [PowerPC] Parse platform-specifc variant kinds in AsmParser
This patch adds support for PowerPC platform-specific variant
kinds in MCSymbolRefExpr::getVariantKindForName, and also
adds a test case to verify they are translated to the appropriate
fixup type.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181053 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 19:52:35 +00:00
Ulrich Weigand
8e4ba8f7b1 [PowerPC] Add some Book II instructions to AsmParser
This patch adds a couple of Book II instructions (isync, icbi) to the
PowerPC assembler parser.  These are needed when bootstrapping clang
with the integrated assembler forced on, because they are used in
inline asm statements in the code base.

The test case adds the full list of Book II storage control instructions,
including associated extended mnemonics.  Again, those that are not yet
supported as marked as FIXME.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181052 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 19:51:09 +00:00
Ulrich Weigand
16adfdb2e6 [PowerPC] Support extended mnemonics in AsmParser
This patch adds infrastructure to support extended mnemonics in the
PowerPC assembler parser.  It adds support specifically for those
extended mnemonics that LLVM will itself generate.

The test case lists *all* extended mnemonics according to the
PowerPC ISA v2.06 Book I, but marks those not yet supported
as FIXME.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181051 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 19:50:27 +00:00
Ulrich Weigand
5e220753ff [PowerPC] Add assembler parser
This adds assembler parser support to the PowerPC back end.

The parser will run for any powerpc-*-* and powerpc64-*-* triples,
but was tested only on 64-bit Linux.  The supported syntax is
intended to be compatible with the GNU assembler.

The parser does not yet support all PowerPC instructions, but
it does support anything that is generated by LLVM itself.
There is no support for testing restricted instruction sets yet,
i.e. the parser will always accept any instructions it knows,
no matter what feature flags are given.

Instruction operands will be checked for validity and errors
generated.  (Error handling in general could still be improved.)

The patch adds a number of test cases to verify instruction
and operand encodings.  The tests currently cover all instructions
from the following PowerPC ISA v2.06 Book I facilities:
Branch, Fixed-point, Floating-Point, and Vector. 
Note that a number of these instructions are not yet supported
by the back end; they are marked with FIXME.

A number of follow-on check-ins will add extra features.  When
they are all included, LLVM passes all tests (including bootstrap)
when using clang -cc1as as the system assembler.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181050 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 19:49:39 +00:00
Shuxin Yang
968d689ec3 Decompose GVN::processNonLocalLoad() (about 400 LOC) into smaller helper functions. No function change.
This function consists of following steps:
   1. Collect dependent memory accesses.
   2. Analyze availability.
   3. Perform fully redundancy elimination, or 
   4. Perform PRE, depending on the availability

 Step 2, 3 and 4 are now moved to three helper routines.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181047 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 19:17:26 +00:00
Akira Hatanaka
a2b2200ff8 [mips] Split the DSP control register and define one register for each field of
its fields.

This removes false dependencies between DSP instructions which access different
fields of the the control register. Implicit register operands are added to
instructions RDDSP and WRDSP after instruction selection, depending on the
value of the mask operand.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@181041 91177308-0d34-0410-b5e6-96231b3b80d8
2013-05-03 18:37:49 +00:00