Commit Graph

16149 Commits

Author SHA1 Message Date
David Blaikie
cec7ce78d7 DebugInfo: Remove some unneeded conditionals now that DIBuilder no longer emits zero-length arrays as {i32 0}
A bunch of test cases needed to be cleaned up for this, many my fault -
when implementid imported modules I updated test cases by simply
duplicating the prior metadata field - which wasn't always the empty
metadata entry.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200731 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-04 01:23:52 +00:00
Hal Finkel
e6c04bff3c Expand vector bswap in LegalizeVectorOps
ISD::BSWAP was missing from the list of node types that should be expanded
element-wise.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200705 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-03 17:27:25 +00:00
Eli Bendersky
a7bc25e34c Remove some unused #includes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200611 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-01 13:12:54 +00:00
Josh Magee
cde5c26c46 [stackprotector] Implement the sspstrong rules for stack layout.
This changes the PrologueEpilogInserter and LocalStackSlotAllocation passes to
follow the extended stack layout rules for sspstrong and sspreq.

The sspstrong layout rules are:
 1. Large arrays and structures containing large arrays (>= ssp-buffer-size)
are closest to the stack protector.
 2. Small arrays and structures containing small arrays (< ssp-buffer-size) are
2nd closest to the protector.
 3. Variables that have had their address taken are 3rd closest to the
protector.


Differential Revision: http://llvm-reviews.chandlerc.com/D2546


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200601 91177308-0d34-0410-b5e6-96231b3b80d8
2014-02-01 01:36:16 +00:00
Reid Kleckner
8a24e83550 Implement inalloca codegen for x86 with the new inalloca design
Calls with inalloca are lowered by skipping all stores for arguments
passed in memory and the initial stack adjustment to allocate argument
memory.

Now the frontend is responsible for the memory layout, and the backend
doesn't have to do any work.  As a result these changes are pretty
minimal.

Reviewers: echristo

Differential Revision: http://llvm-reviews.chandlerc.com/D2637

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200596 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 23:50:57 +00:00
Reid Kleckner
f10743d765 Don't put non-static allocas in the static alloca map
Allocas marked inalloca are never static, but we were trying to put them
into the static alloca map if they were in the entry block.  Also add an
assertion in x86 fastisel.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200593 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 23:45:12 +00:00
Rafael Espindola
6f4efee6cb Remove a redundant call to hasRawTextSupport.
The code path it was guarding was already using emitRawComment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200591 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 23:14:01 +00:00
Paul Robinson
c2c98e7884 If we're not producing DWARF accel tables, don't waste memory
keeping track of those entries.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200572 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 20:39:19 +00:00
Eric Christopher
5ff4b203e6 Add support for DW_FORM_flag and DW_FORM_flag_present to the DIE hashing
algorithm. Sink the 'A' + Attribute hash into each form so we don't
have to check valid forms before deciding whether or not we're going
to hash which will let the default be to return without doing anything.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200571 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 20:02:58 +00:00
David Blaikie
0451736fc9 DebugInfo: Flag type unit references as declarations
This ensures DWARF consumers don't confuse these references for
definitions. I'd argue it might be nice to improve debuggers so we don't
need this, but it's just one field in an abbreviation anyway - so it
doesn't seem worth the fight.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200569 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 19:52:26 +00:00
Manman Ren
05324ab015 This patch teaches the DAGCombiner how to fold insert_subvector nodes
when the input is a concat_vectors and the insert replaces one of the
concat halves:

Lower half: fold (insert_subvector (concat_vectors X, Y), Z) ->
(concat_vectors Z, Y)
Upper half: fold (insert_subvector (concat_vectors X, Y), Z) ->
(concat_vectors X, Z)

This can be seen with the following IR:

define <8 x float> @lower_half(<4 x float> %v1, <4 x float> %v2, <4 x
float> %v3) {
  %1 = shufflevector <4 x float> %v1, <4 x float> %v2, <8 x i32> <i32
0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  %2 = tail call <8 x float> @llvm.x86.avx.vinsertf128.ps.256(<8 x
float> %1, <4 x float> %v3, i8 0)

The vinsertf128 intrinsic is converted into an insert_subvector node
in SelectionDAGBuilder.cpp.

Using AVX, without the patch this generates two vinsertf128 instructions:

vinsertf128 $1, %xmm1, %ymm0, %ymm0
vinsertf128 $0, %xmm2, %ymm0, %ymm0

With the patch this is optimized into:

vinsertf128 $1, %xmm1, %ymm2, %ymm0

Patch by Robert Lougher.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200506 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 01:10:35 +00:00
Owen Anderson
45b07e157c DAGCombine should not produce ISD::OR nodes after operation legalization if they're not legal.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200503 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 00:51:43 +00:00
Manman Ren
21f09088d3 PGO branch weight: update edge weights in SelectionDAGBuilder.
When converting from "or + br" to two branches, or converting from
"and + br" to two branches, we correctly update the edge weights of
the two branches.

The previous attempt at r200431 was reverted at r200434 because of
two testing case failures. I modified my patch a little, but forgot
to re-run "make check-all".

Testing case CodeGen/ARM/lsr-unfolded-offset.ll is updated because of
the patch's impact on branch probability which causes changes in
spill placement.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200502 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-31 00:42:44 +00:00
Juergen Ributzka
014fdcdaf0 [Stackmaps] Record the stack size of each function that contains a stackmap/patchpoint intrinsic.
Re-applying the patch, but this time without using AsmPrinter methods.

Reviewed by Andy

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200481 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-30 18:58:27 +00:00
Juergen Ributzka
d26c0e731c Revert "[Stackmaps] Record the stack size of each function that contains a stackmap/patchpoint intrinsic."
This reverts commit r200444 to unbreak buildbots.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200445 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-30 03:34:02 +00:00
Juergen Ributzka
2baaf25bf5 [Stackmaps] Record the stack size of each function that contains a stackmap/patchpoint intrinsic.
Reviewed by Andy

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200444 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-30 03:06:14 +00:00
Timur Iskhodzhanov
ab42578bdd Reland r200340 - 'Add line table debug info to COFF files when using a win32 triple'
This incorporates a couple of fixes reviewed at http://llvm-reviews.chandlerc.com/D2651

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200440 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-30 01:39:17 +00:00
Manman Ren
2227f98e32 Revert r200431 due to bot failures.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200434 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-30 00:53:27 +00:00
Manman Ren
0bdaca5058 PGO branch weight: update edge weights in SelectionDAGBuilder.
When converting from "or + br" to two branches, or converting from
"and + br" to two branches, we correctly update the edge weights of
the two branches.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200431 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-30 00:24:37 +00:00
Manman Ren
ea120b6bb5 PGO branch weight: update edge weights in IfConverter.
This commit only handles IfConvertTriangle. To update edge weights
of a successor, one interface is added to MachineBasicBlock:
/// Set successor weight of a given iterator.
setSuccWeight(succ_iterator I, uint32_t weight)

An existing testing case test/CodeGen/Thumb2/v8_IT_5.ll is updated,
since we now correctly update the edge weights, the cold block
is placed at the end of the function and we jump to the cold block.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200428 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-29 23:18:47 +00:00
Eric Christopher
65a12969f9 Move range handling for a function to endFunction rather than
when we create the subprogram DIE.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200426 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-29 23:05:43 +00:00
Eric Christopher
c55cb498c9 If we use DW_AT_ranges we need to specify a base address that ranges
are relative to in the compile unit. Currently let's just use 0...

Thanks to Greg Clayton for the catch!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200425 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-29 22:22:56 +00:00
Eric Christopher
c879504978 Turn on CU ranges if we've got multiple compile units in the same
module since there's no range guarantee that we could make given
output order. This also fixes up the testcases that have multiple
CUs to have the correct range offset.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200422 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-29 22:06:27 +00:00
Eric Christopher
4ec1a52c5f Make the compile unit map a MapVector so that we can assume a stable
output ordering.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200421 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-29 22:06:23 +00:00
Eric Christopher
8e290f5a5d Fix formatting of comment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200420 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-29 22:06:21 +00:00
Renato Golin
3fca788967 Enable EHABI by default
After all hard work to implement the EHABI and with the test-suite
passing, it's time to turn it on by default and allow users to
disable it as a work-around while we fix the eventual bugs that show
up.

This commit also remove the -arm-enable-ehabi-descriptors, since we
want the tables to be printed every time the EHABI is turned on
for non-Darwin ARM targets.

Although MCJIT EHABI is not working yet (needs linking with the right
libraries), this commit also fixes some relocations on MCJIT regarding
the EH tables/lib calls, and update some tests to avoid using EH tables
when none are needed.

The EH tests in the test-suite that were previously disabled on ARM
now pass with these changes, so a follow-up commit on the test-suite
will re-enable them.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200388 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-29 11:50:56 +00:00
NAKAMURA Takumi
52de969206 Revert r200340, "Add line table debug info to COFF files when using a win32 triple."
It was incompatible with --target=i686-win32.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200375 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-29 06:05:38 +00:00
David Woodhouse
4396f5d9d2 Change MCStreamer EmitInstruction interface to take subtarget info
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200345 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-28 23:12:42 +00:00
Timur Iskhodzhanov
38afdbea2c Add line table debug info to COFF files when using a win32 triple.
Reviewed at http://llvm-reviews.chandlerc.com/D2232

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200340 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-28 21:33:27 +00:00
Adrian Prantl
bb6f236729 typo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200323 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-28 18:13:47 +00:00
Andrea Di Biagio
5144469bb4 [DAGCombiner] Avoid introducing an illegal build_vector when folding a sign_extend.
Make sure that we don't introduce illegal build_vector dag nodes
when trying to fold a sign_extend of a build_vector.

This fixes a regression introduced by r200234.
Added test CodeGen/X86/fold-vector-sext-crash.ll
to verify that llc no longer crashes with an assertion failure
due to an illegal build_vector of type MVT::v4i64.

Thanks to Ilia Filippov for spotting this regression and for
providing a reproducible test case.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200313 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-28 12:53:56 +00:00
Juergen Ributzka
efbb39740c [TLI] Add a new hook to TargetLowering to query the target if a load of a constant should be converted to simply the constant itself.
Before this patch we used getIntImmCost from TargetTransformInfo to determine if
a load of a constant should be converted to just a constant, but the threshold
for this was set to an arbitrary value. This value works well for the two
targets (X86 and ARM) that implement this target-hook, but it isn't
target-independent at all.

Now targets have the possibility to decide directly if this optimization should
be performed. The default value is set to false to preserve the current
behavior. The target hook has been moved to TargetLowering, which removed the
last use and need of TargetTransformInfo in SelectionDAG.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200271 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-28 01:20:14 +00:00
Eric Christopher
1c2827cd6a Revert r199871 and replace it with a simple check in the debug info
code to see if we're emitting a function into a non-default
text section. This is still a less-than-ideal solution, but more
contained than r199871 to determine whether or not we're emitting
code into an array of comdat sections.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200269 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-28 00:49:26 +00:00
Eric Christopher
2ddf01b64d Reformat slightly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200264 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-27 23:50:03 +00:00
Matt Arsenault
410f38e015 Fix sext(setcc) -> select_cc using wrong type for setcc.
Also update the comment, since it actually produces a
select (setcc) instead of select_cc.

It was checking and using the setcc result type for the
type of the sext, instead of the type of the compared items.

In my problem case, the sext was to i32 and was used as the setcc type,
but the expected type was i64.

No test since I haven't been able to hit the problem with
this on any in-tree targets.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200249 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-27 21:41:54 +00:00
Andrea Di Biagio
e9c0b5aba6 [DAGCombiner] Teach how to fold sext/aext/zext of constant build vectors.
This patch teaches the DAGCombiner how to fold a sext/aext/zext dag node when
the operand in input is a build vector of constants (or UNDEFs).

The inability to fold a sext/zext of a constant build_vector was the root
cause of some pcg bugs affecting vselect expansion on x86-64 with AVX support.

Before this change, the DAGCombiner only knew how to fold a sext/zext/aext of a
ConstantSDNode.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200234 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-27 18:45:30 +00:00
David Majnemer
0fd23cd6c2 MC: Add support for .cfi_startproc simple
This commit allows LLVM MC to process .cfi_startproc directives when
they are followed by an additional `simple' identifier. This signals to
elide the emission of target specific CFI instructions that would
normally occur initially.

This fixes PR16587.

Differential Revision: http://llvm-reviews.chandlerc.com/D2624


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200227 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-27 17:20:25 +00:00
Stepan Dyatkovskiy
c628af024e Fix for PR18102.
Issue outcomes from DAGCombiner::MergeConsequtiveStores, more precisely from
mem-ops sequence sorting.

Consider, how MergeConsequtiveStores works for next example:

store i8 1, a[0]
store i8 2, a[1]
store i8 3, a[1]   ; a[1] again.
return   ; DAG starts here

1. Method will collect all the 3 stores.
2. It sorts them by distance from the base pointer (farthest with highest
index).
3. It takes first consecutive non-overlapping stores and (if possible) replaces
them with a single store instruction.

The point is, we can't determine here which 'store' instruction
would be the second after sorting ('store 2' or 'store 3').
It happens that 'store 3' would be the second, and 'store 2' would be the third.

So after merging we have the next result:

store i16 (1 | 3 << 8), base   ; is a[0] but bit-casted to i16
store i8 2, a[1]

So actually we swapped 'store 3' and 'store 2' and got wrong contents in a[1].

Fix: In sort routine just also take into account mem-op sequence number. 



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200201 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-27 09:18:31 +00:00
Rafael Espindola
550d791907 Pass a MCSubtargetInfo down to the TargetStreamer creation.
With this the target streamers will be able to know the target features that
are in use.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200135 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-26 06:38:58 +00:00
Kevin Qin
40bb464283 [AArch64 NEON] Fix pattern match failed on FP_ROUND from v1f128 to v1f64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200109 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-26 02:19:35 +00:00
Hal Finkel
397cd32bc2 Disable the use of TBAA when using AA in CodeGen
There are currently two issues, of which I currently know, that prevent TBAA
from being correctly usable in CodeGen:

  1. Stack coloring does not update TBAA when merging allocas. This is easy
     enough to fix, but is not the largest problem.

  2. CGP inserts ptrtoint/inttoptr pairs when sinking address computations.
     Because BasicAA does not handle inttoptr, we'll often miss basic type punning
     idioms that we need to catch so we don't miscompile real-world code (like LLVM).

I don't yet have a small test case for this, but this fixes self hosting a
non-asserts build of LLVM on PPC64 when using -enable-aa-sched-mi and -misched=shuffle.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200093 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-25 19:24:54 +00:00
Hal Finkel
7a1fc7ad50 Add combiner-aa-only-func (debug only)
This option (which is !NDEBUG only) allows restricting the use of alias
analysis in DAGCombiner to a specific function. This has proved extremely
valuable to isolating bugs related to this feature, and mirrors the
misched-only-func option provided by the new instruction scheduler.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200088 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-25 17:32:39 +00:00
Hal Finkel
f62acf3c54 Improve descriptions of combiner-alias-analysis and combiner-global-alias-analysis
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200087 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-25 17:32:37 +00:00
Juergen Ributzka
943ce55f39 Revert "Revert "Add Constant Hoisting Pass" (r200034)"
This reverts commit r200058 and adds the using directive for
ARMTargetTransformInfo to silence two g++ overload warnings.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200062 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-25 02:02:55 +00:00
Hans Wennborg
503793e834 Revert "Add Constant Hoisting Pass" (r200034)
This commit caused -Woverloaded-virtual warnings. The two new
TargetTransformInfo::getIntImmCost functions were only added to the superclass,
and to the X86 subclass. The other targets were not updated, and the
warning highlighted this by pointing out that e.g. ARMTTI::getIntImmCost was
hiding the two new getIntImmCost variants.

We could pacify the warning by adding "using TargetTransformInfo::getIntImmCost"
to the various subclasses, or turning it off, but I suspect that it's wrong to
leave the functions unimplemnted in those targets. The default implementations
return TCC_Free, which I don't think is right e.g. for ARM.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200058 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-25 01:18:18 +00:00
Juergen Ributzka
96172cb4a4 Add Constant Hoisting Pass
Retry commit r200022 with a fix for the build bot errors. Constant expressions
have (unlike instructions) module scope use lists and therefore may have users
in different functions. The fix is to simply ignore these out-of-function uses.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200034 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-24 20:18:00 +00:00
Hal Finkel
86720f7c65 Fix DAGCombiner::GatherAllAliases to account for non-chain dependencies
DAGCombiner::GatherAllAliases, which is only used when AA used is enabled
during DAGCombine, had a fundamentally incorrect assumption for which this
change compensates. GatherAllAliases, which is used to find aliasing
predecessor chain nodes (so that a better chain can be selected for a load or
store to enable subsequent optimizations) assumed that walking up the chain
would always catch all possibly-aliasing loads and stores. This is not true: To
really find all aliases, we also need to search for aliases through the value
operand of a store, etc.  Consider the following situation:

  Token1 = ...
  L1 = load Token1, %52
  S1 = store Token1, L1, %51
  L2 = load Token1, %52+8
  S2 = store Token1, L2, %51+8
  Token2 = Token(S1, S2)
  L3 = load Token2, %53
  S3 = store Token2, L3, %52
  L4 = load Token2, %53+8
  S4 = store Token2, L4, %52+8

If we search for aliases of S3 (which loads address %52), and we look only
through the chain, then we'll miss the trivial dependence on L1 (which loads
from %52). We then might change all loads and stores to use Token1 as their
chain operand, which could result in copying %53 into %52 before copying
%52 into %51 (which should happen first).

The problem is, however, that searching for such data dependencies can become
expensive, and the cost is not directly related to the chain depth. Instead,
we'll rule out such configurations by insisting that we've visited all chain
users (except for users of the original chain, which is not necessary).  When
doing this, we need to look through nodes we don't care about (otherwise,
things like register copies will interfere with trivial use cases).

Unfortunately, I don't have a small test case for this problem. Creating the
underlying situation is not hard (a pair of memcpys will do it), but arranging
for the default instruction schedule to be incorrect is very fragile.

This unbreaks self hosting on PPC64 when using
-mllvm -combiner-global-alias-analysis -mllvm -combiner-alias-analysis.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200033 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-24 20:12:02 +00:00
Juergen Ributzka
dc6f9b9a4f Revert "Add Constant Hoisting Pass"
This reverts commit r200022 to unbreak the build bots.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200024 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-24 18:40:30 +00:00
Hal Finkel
08aa5bf1d5 Restrict FindBetterChain DAG combines to unindexed nodes
These transformations obviously won't work for indexed (pre/post-inc) loads and
stores. In practice, I'm not sure there is any benefit to enabling them for
indexed nodes because other transformations that these might enable likely also
won't handle indexed nodes.

I don't have an in-tree test case that hits this problem, but an upcoming bug
fix will make it much more likely.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200023 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-24 18:25:26 +00:00
Juergen Ributzka
fb282c68b7 Add Constant Hoisting Pass
This pass identifies expensive constants to hoist and coalesces them to
better prepare it for SelectionDAG-based code generation. This works around the
limitations of the basic-block-at-a-time approach.

First it scans all instructions for integer constants and calculates its
cost. If the constant can be folded into the instruction (the cost is
TCC_Free) or the cost is just a simple operation (TCC_BASIC), then we don't
consider it expensive and leave it alone. This is the default behavior and
the default implementation of getIntImmCost will always return TCC_Free.

If the cost is more than TCC_BASIC, then the integer constant can't be folded
into the instruction and it might be beneficial to hoist the constant.
Similar constants are coalesced to reduce register pressure and
materialization code.

When a constant is hoisted, it is also hidden behind a bitcast to force it to
be live-out of the basic block. Otherwise the constant would be just
duplicated and each basic block would have its own copy in the SelectionDAG.
The SelectionDAG recognizes such constants as opaque and doesn't perform
certain transformations on them, which would create a new expensive constant.

This optimization is only applied to integer constants in instructions and
simple (this means not nested) constant cast experessions. For example:
%0 = load i64* inttoptr (i64 big_constant to i64*)

Reviewed by Eric

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200022 91177308-0d34-0410-b5e6-96231b3b80d8
2014-01-24 18:23:08 +00:00