llvm-svn: 148031

Restore the (obviously wrong) behavior from before r147938 without relying on
undefined behavior. Add a fat FIXME note.

This should fix nightly tester failures.

llvm-svn: 148030

Fix a bug in the AVX 256-bit shuffle code in cases where the splat element is on the boundary of two 128-bit vectors.
The attached testcase was stuck in an endless loop.

llvm-svn: 148027

X86: Generalize the x << (y & const) optimization to also catch masks with more set bits set than 31 or 63.

llvm-svn: 148024

In att style asm syntax memory operand size is derived from suffix attached with mnemonic.  In intel style asm syntax it is part of memory operand hence predicate method check is required to select appropriate instruction.

llvm-svn: 148006

need to make a deep copy of each of the std::maps. Use a std::map of the
std::map instead. This improves the compile time of sqlite3 by ~2%.

llvm-svn: 148003

This is a work in progress.

llvm-svn: 148002

same pattern. We already had this pattern is a few places, but others
tried to make a rough approximation of an actual DAG structure. As not
everywhere went to this trouble, nothing could rely on this being done.
In fact, I've checked all references to these node Ids, and the ones
that are using the topo-sort properties are actually satisfied with
a strict-weak-ordering. The requirement appears to be that Use >= Def.

I've added a big blurb of comments to this bit of the transform to
clarify why the order is so important for the next reader of the code.

I'm starting with this change as it is very small, and trivially
reverted if something breaks or the >= above really does need to be >.
If that proves the case, we can hide the problem by reverting this
patch, but the problem exists elsewhere as well, and so a more
comprehensive solution will be needed.

llvm-svn: 148001

llvm-svn: 147980

llvm-svn: 147979

This appears to improve sqlite3's compile time by ~2%.

llvm-svn: 147978

This helper method is too simplistic for RAGreedy.

llvm-svn: 147976

llvm-svn: 147975

No functional change.

llvm-svn: 147972

Re-fix the issue Bill fixed in r147899 in a slightly different way, which doesn't abuse the semantics of linker_private.  We don't really want to merge any string constant with a weak_odr global.

llvm-svn: 147971

llvm-svn: 147966

llvm-svn: 147965

When we load the v12i32 type, the GenWidenVectorLoads method generates two loads: v8i32 and v4i32 
and attempts to use CONCAT_VECTORS to join them. In this fix I concat undef values to widen 
the smaller value. The test "widen_load-2.ll" also exposes this bug on AVX.

llvm-svn: 147964

This uses TLS slot 90, which actually belongs to JavaScriptCore. We only support
frames with static size
Patch by Brian Anderson.

llvm-svn: 147960

Patch by Brian Anderson.

llvm-svn: 147958

hoped this would revive one of the llvm-gcc selfhost build bots, but it
didn't so it doesn't appear that my transform is the culprit.

If anyone else is seeing failures, please let me know!

llvm-svn: 147957

This is a comparison of two addresses, and GCC does the comparison unsigned.

Patch by Brian Anderson.

llvm-svn: 147954

[asan] extend the workaround for http://llvm.org/bugs/show_bug.cgi?id=11395: don't instrument the function at all on x86_32 if it has a large asm blob

llvm-svn: 147953

Patch by Brian Anderson.

llvm-svn: 147952

directives was in the wrong place and getting triggered incorectly with a
cpp .file directive.  This change fixes that and adds a test case.

llvm-svn: 147951

llvm-svn: 147949

Fix a bug in the lowering of BUILD_VECTOR for AVX. SCALAR_TO_VECTOR does not zero untouched elements. Use INSERT_VECTOR_ELT instead.

llvm-svn: 147948

are invalid).  Fixes a crash on array1.C from the GCC testsuite when
compiled with dragonegg.

llvm-svn: 147946

strange build bot failures that look like a miscompile into an infloop.
I'll investigate this tomorrow, but I'd both like to know whether my
patch is the culprit, and get the bots back to green.

llvm-svn: 147945

lots of lines of code. No functionality changed.

llvm-svn: 147942

SRL-rooted code.

llvm-svn: 147941

factor the differences that were hiding in one of them into its other
caller, the SRL handling code. No change in behavior.

llvm-svn: 147940

mask+shift pairs at the beginning of the ISD::AND case block, and then
hoist the final pattern into a helper function, simplifying and
reflowing it appropriately. This should have no observable behavior
change, but several simplifications fell out of this such as directly
computing the new mask constant, etc.

llvm-svn: 147939

I don't think the compact encoding code is right, but at least is has
defined behavior now.

llvm-svn: 147938

extracts and scaled addressing modes into its own helper function. No
functionality changed here, just hoisting and layout fixes falling out
of that hoisting.

llvm-svn: 147937

detect a pattern which can be implemented with a small 'shl' embedded in
the addressing mode scale. This happens in real code as follows:

  unsigned x = my_accelerator_table[input >> 11];

Here we have some lookup table that we look into using the high bits of
'input'. Each entity in the table is 4-bytes, which means this
implicitly gets turned into (once lowered out of a GEP):

  *(unsigned*)((char*)my_accelerator_table + ((input >> 11) << 2));

The shift right followed by a shift left is canonicalized to a smaller
shift right and masking off the low bits. That hides the shift right
which x86 has an addressing mode designed to support. We now detect
masks of this form, and produce the longer shift right followed by the
proper addressing mode. In addition to saving a (rather large)
instruction, this also reduces stalls in Intel chips on benchmarks I've
measured.

In order for all of this to work, one part of the DAG needs to be
canonicalized *still further* than it currently is. This involves
removing pointless 'trunc' nodes between a zextload and a zext. Without
that, we end up generating spurious masks and hiding the pattern.

llvm-svn: 147936

1. Size heuristics changed. Now we calculate number of unswitching
branches only once per loop.
2. Some checks was moved from UnswitchIfProfitable to
processCurrentLoop, since it is not changed during processCurrentLoop
iteration. It allows decide to skip some loops at an early stage.
Extended statistics:
- Added total number of instructions analyzed.

llvm-svn: 147935

This interface is misleading and dangerous, but it is actually what we need for unrolling.

llvm-svn: 147926

llvm-svn: 147924

llvm-svn: 147923