Commit Graph

113570 Commits

Author SHA1 Message Date
Roman Tereshin 13229aff54 [GlobalISel] NFCI, Getting GlobalISel ~5% faster
by replacing DenseMap with IndexedMap for LLTs within MRI, as
benchmarked by cross-compiling sqlite3 amalgamation for AArch64
on x86 machine.

Reviewed By: qcolombet

Differential Revision: https://reviews.llvm.org/D46809

llvm-svn: 333125
2018-05-23 21:12:02 +00:00
Lei Huang 8b0da65bfb [Power9]Legalize and emit code for W vector extract and convert to QP
Implemente patterns to extract [Un]signed Word vector element and convert to
quad-precision.

Differential Revision: https://reviews.llvm.org/D46536

llvm-svn: 333115
2018-05-23 19:31:54 +00:00
Lei Huang 8990168a45 [Power9]Legalize and emit code for DW vector extract and convert to QP
Implemente patterns to extract [Un]signed DWord vector element and convert to
quad-precision.

Differential Revision: https://reviews.llvm.org/D46333

llvm-svn: 333112
2018-05-23 18:36:51 +00:00
Changpeng Fang 5f9154618e StructurizeCFG: Adjust the loop depth for a subregion to order the nodes correctly
Summary:
  StructurizeCFG::orderNodes basically uses a reverse post-order (RPO) traversal of the region list to get the order.
The only problem with it is that sometimes backedges for outer loops will be visited before backedges for inner loops.
To solve this problem, a loop depth based approach has been used to make sure all blocks in this loop has been visited
before moving on to outer loop.

However, we found a problem for a SubRegion which is a loop itself:

--> BB1 --> BB2 --> BB3 -->

In this case, BB2 is a SubRegion (loop), and thus its loopdepth is different than that of BB1 and BB3. This fact will lead
BB2 to be placed in the wrong order.

In this work, we treat the SubRegion as a special case and use its exit block to determine the loop and its depth
to guard the sorting.

Reviewers:
  arsenm, jlebar

Differential Revision:
  https://reviews.llvm.org/D46912

llvm-svn: 333111
2018-05-23 18:34:48 +00:00
Chad Rosier 3f66363139 [CodeGen][AArch64] Use RegUnits to track register aliases. (NFC)
Use RegUnits to track register aliases in AArch64RedundantCopyElimination.

Differential Revision: https://reviews.llvm.org/D47269

llvm-svn: 333107
2018-05-23 17:49:38 +00:00
Roman Lebedev 6b6c553bb8 [InstCombine] Fold unfolded masked merge pattern with variable mask!
Summary:
Finally fixes [[ https://bugs.llvm.org/show_bug.cgi?id=6773 | PR6773 ]].

Now that the backend is all done, we can finally fold it!

The canonical unfolded masked merge pattern is
```(x &  m) | (y & ~m)```
There is a second, equivalent variant:
```(x | ~m) & (y |  m)```
Only one of them (the or-of-and's i think) is canonical.
And if the mask is not a constant, we should fold it to:
```((x ^ y) & M) ^ y```

https://rise4fun.com/Alive/ndQw

Reviewers: spatel, craig.topper

Reviewed By: spatel

Subscribers: nicholas, RKSimon, llvm-commits

Differential Revision: https://reviews.llvm.org/D46814

llvm-svn: 333106
2018-05-23 17:47:52 +00:00
Jakub Kuderski ef33edd9b5 [Dominators] Add PDT constructor from Function
Summary: This patch adds a PDT constructor from Function and lets codes previously using a local class to do this use PostDominatorTree class directly.

Reviewers: davide, kuhar, grosser, dberlin

Reviewed By: kuhar

Author: NutshellySima

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D46709

llvm-svn: 333102
2018-05-23 17:29:21 +00:00
Craig Topper 3b768e8602 [InstCombine] Negate ABS/NABS patterns by swapping the select operands to remove the negation
Differential Revision: https://reviews.llvm.org/D47236

llvm-svn: 333101
2018-05-23 17:29:03 +00:00
Petar Jovanovic 7d37bb42a1 Silence warnings introduced with r333093
r333093 introduced several warnings (-Wlogical-not-parentheses,
-Wbool-compare).
Adding parentheses in MipsSEInstrInfo::isCopyInstr() to silence it.

llvm-svn: 333097
2018-05-23 16:27:51 +00:00
Petar Jovanovic c051000b83 [X86][MIPS][ARM] New machine instruction property 'isMoveReg'
This property is needed in order to follow values movement between
registers. This property is used in TII to implement method that
returns true if simple copy like instruction is recognized, along
with source and destination machine operands.

Patch by Nikola Prica.

Differential Revision: https://reviews.llvm.org/D45204

llvm-svn: 333093
2018-05-23 15:28:28 +00:00
Nicola Zaghen 03d0b91f43 Remove DEBUG macro.
Now that the LLVM_DEBUG() macro landed on the various sub-projects
the DEBUG macro can be removed.
Also change the new uses of DEBUG to LLVM_DEBUG.

Differential Revision: https://reviews.llvm.org/D46952

llvm-svn: 333091
2018-05-23 15:09:29 +00:00
Alex Bradbury 257d5b5639 [RISCV] Add symbol diff relocation support for RISC-V
For RISC-V it is desirable to have relaxation happen in the linker once 
addresses are known, and as such the size between two instructions/byte 
sequences in a section could change.

For most assembler expressions, this is fine, as the absolute address results 
in the expression being converted to a fixup, and finally relocations. 
However, for expressions such as .quad .L2-.L1, the assembler folds this down 
to a constant once fragments are laid out, under the assumption that the 
difference can no longer change, although in the case of linker relaxation the 
differences can change at link time, so the constant is incorrect. One place 
where this commonly appears is in debug information, where the size of a 
function expression is in a form similar to the above.

This patch extends the assembler to allow an AsmBackend to declare that it 
does not want the assembler to fold down this expression, and instead generate 
a pair of relocations that allow the linker to carry out the calculation. In 
this case, the expression is not folded, but when it comes to emitting a 
fixup, the generic FK_Data_* fixups are converted into a pair, one for the 
addition half, one for the subtraction, and this is passed to the relocation 
generating methods as usual. I have named these FK_Data_Add_* and 
FK_Data_Sub_* to indicate which half these are for.

For RISC-V, which supports this via e.g. the R_RISCV_ADD64, R_RISCV_SUB64 pair 
of relocations, these are also set to always emit relocations relative to 
local symbols rather than section offsets. This is to deal with the fact that 
if relocations were calculated on e.g. .text+8 and .text+4, the result 12 
would be stored rather than 4 as both addends are added in the linker.

Differential Revision: https://reviews.llvm.org/D45181
Patch by Simon Cook.

llvm-svn: 333079
2018-05-23 12:36:18 +00:00
Alex Bradbury 3fa69dd055 [Sparc] Use addAliasForDirective to support data directives
The Sparc asm parser currently has custom parsing logic for .half, .word, 
.nword and .xword. Rather than use this custom logic, we can just use 
addAliasForDirective to enable the reuse of AsmParser::parseDirectiveValue.

https://reviews.llvm.org/D47003

llvm-svn: 333078
2018-05-23 11:20:28 +00:00
Alex Bradbury 0a59f18951 [AArch64] Use addAliasForDirective to support data directives
The AArch64 asm parser currently has custom parsing logic for .hword, .word, 
and .xword. Rather than use this custom logic, we can just use 
addAliasForDirective to enable the reuse of AsmParser::parseDirectiveValue.

Differential Revision: https://reviews.llvm.org/D47000

llvm-svn: 333077
2018-05-23 11:17:20 +00:00
Alex Bradbury 1c010d0fa4 [RISCV] Correctly report sizes for builtin fixups
This is a different approach to fixing the problem described in D46746. 
RISCVAsmBackend currently depends on the getSize helper function returning the 
number of bytes a fixup may change (note: some other backends have a similar 
helper named getFixupNumKindBytes). As noted in that review, this doesn't 
return the correct size for FK_Data_1, FK_Data_2, or FK_Data_8 meaning that 
too few bytes will be written in the case of FK_Data_8, and there's the 
potential of writing outside the Data array for the smaller fixups.

D46746 extends getSize to recognise some of the builtin fixup types. Rather 
than having a function that needs to be kept up to date as new builtin or 
target-specific fixups are added, We can calculate an appropriate bound on the 
number of bytes that might be touched using Info.TargetSize and 
Info.TargetOffset.

Differential Revision: https://reviews.llvm.org/D46965

llvm-svn: 333076
2018-05-23 10:53:56 +00:00
Max Kazantsev d99f3bacb4 [LoopUnswitch] Fix SCEV invalidation in unswitching
Loop unswitching makes substantial changes to a loop that can also affect cached
SCEV info in its outer loops as well, but it only cares to invalidate SCEV cache for the
innermost loop in case of full unswitching and does not invalidate anything at all in
case of trivial unswitching. As result, we may end up with incorrect data in cache.

Differential Revision: https://reviews.llvm.org/D46045
Reviewed By: mzolotukhin

llvm-svn: 333072
2018-05-23 10:09:53 +00:00
Piotr Padlewski d6f7346a4b Fix aliasing of launder.invariant.group
Summary:
Patch for capture tracking broke
bootstrap of clang with -fstict-vtable-pointers
which resulted in debbugging nightmare. It was fixed
https://reviews.llvm.org/D46900 but as it turned
out, there were other parts like inliner (computing of
noalias metadata) that I found after bootstraping with enabled
assertions.

Reviewers: hfinkel, rsmith, chandlerc, amharc, kuhar

Subscribers: JDevlieghere, eraman, llvm-commits, hiraditya

Differential Revision: https://reviews.llvm.org/D47088

llvm-svn: 333070
2018-05-23 09:16:44 +00:00
Daniel Cederman 6356571ec0 [Sparc] Add mnemonic aliases for flush, stb, stba, sth, and stha
Reviewers: jyknight

Reviewed By: jyknight

Subscribers: fedor.sergeev, jrtc27, llvm-commits

Differential Revision: https://reviews.llvm.org/D47140

llvm-svn: 333068
2018-05-23 08:26:49 +00:00
Serguei Katkov 46ef8fffdf SafepointIRVerifier is made unreachable block tolerant
SafepointIRVerifier crashed while traversing blocks without a DomTreeNode.
This could happen with a custom pipeline or when some optional passes were skipped by OptBisect.

SafepointIRVerifier is fixed to traverse basic blocks that are reachable from entry. Test are added.

Patch Author: Yevgeny Rouban!
Reviewers: anna, reames, dneilson, DaniilSuchkov, skatkov
Reviewed By: reames
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D47011

llvm-svn: 333063
2018-05-23 05:54:55 +00:00
Roman Tereshin e79d656c33 [GlobalISel][ARM] Adding HPR and QPR regclasses to FPRB regbank
Also bringing ARMRegisterBankInfo::getRegBankFromRegClass
implementation up to speed with the *.td-definition.

Reviewed By: qcolombet

Differential Revision: https://reviews.llvm.org/D43982

llvm-svn: 333056
2018-05-23 02:59:31 +00:00
Heejin Ahn 1e4d35044f [WebAssembly] Add functions for EHScopes
Summary:
There are functions using the term 'funclet' to refer to both
1. an EH scopes, the structure of BBs that starts with
catchpad/cleanuppad and ends with catchret/cleanupret, and
2. a small function that gets outlined in AsmPrinter, which is the
original meaning of 'funclet'.

So far the two have been the same thing; EH scopes are always outlined
in AsmPrinter as funclets at the end of the compilation pipeline. But
now wasm also uses scope-based EH but does not outline those, so we now
need to correctly distinguish those two use cases in functions.

This patch splits `MachineBasicBlock::isFuncletEntry` into
`isFuncletEntry` and `isEHScopeEntry`, and
`MachineFunction::hasFunclets` into `hasFunclets` and `hasEHScopes`, in
order to distinguish the two different use cases. And this also changes
some uses of the term 'funclet' to 'scope' in `getFuncletMembership` and
change the function name to `getEHScopeMembership` because this function
is not about outlined funclets but about EH scope memberships.

This change is in the same vein as D45559.

Reviewers: majnemer, dschuff

Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits

Differential Revision: https://reviews.llvm.org/D47005

llvm-svn: 333045
2018-05-23 00:32:46 +00:00
Sanjay Patel 4b96935bd7 [InstCombine] use nsw negation for abs libcalls
Also, produce the canonical IR abs (s<0) to be more efficient. 

This is the libcall equivalent of the clang builtin change from:
rL333038

Pasting from that commit message:
The stdlib functions are defined in section 7.20.6.1 of the C standard with:
"If the result cannot be represented, the behavior is undefined."

That lets us mark the negation with 'nsw' because "sub i32 0, INT_MIN" would
be UB/poison.

llvm-svn: 333042
2018-05-22 23:29:40 +00:00
David Bolvansky 1f343fa0e0 [InstCombine] Remove calloc transformations
Summary: Previous patch does not care if a value is changed between calloc and strlen. This needs to be removed from InstCombine and maybe moved to DSE later after some rework.

Reviewers: efriedma

Reviewed By: efriedma

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D47218

llvm-svn: 333022
2018-05-22 20:27:36 +00:00
Matt Arsenault 606bc315d6 AMDGPU: Fix v2f16 fneg/fabs pattern
The integer operation convertion for some reason only happens
if the source is a bitcast from an integer, which happens to
always be the situation when the result is loaded. Add
an additional pattern for when the source operation is really
an FP operation.

llvm-svn: 333019
2018-05-22 20:13:34 +00:00
Eli Friedman 785acce51d Delete unused variable from r333015.
(The assertion suppressed the unused variable warning on
Release+Asserts builds, so I didn't notice.)

llvm-svn: 333018
2018-05-22 19:38:07 +00:00
Tom Stellard b12f4dec08 AMDGPU: Move AMDGPUTargetLowering::isFPExtFoldable() into SITargetLowering
Summary: This is always false for R600.

Reviewers: arsenm, nhaehnle

Reviewed By: arsenm

Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits

Differential Revision: https://reviews.llvm.org/D47180

llvm-svn: 333016
2018-05-22 19:37:55 +00:00
Eli Friedman 042dc9e092 [MachineOutliner] Add "thunk" outlining for AArch64.
When we're outlining a sequence that ends in a call, we can save up to
three instructions in the outlined function by turning the call into
a tail-call. I refer to this as thunk outlining because the resulting
outlined function looks like a thunk; suggestions welcome for a better
name.

In addition to making the outlined function shorter, thunk outlining
allows outlining calls which would otherwise be illegal to outline:
we don't need to save/restore LR, so we don't need to prove anything
about the stack access patterns of the callee.

To make this work effectively, I also added
MachineOutlinerInstrType::LegalTerminator to the generic MachineOutliner
code; this allows treating an arbitrary instruction as a terminator in
the suffix tree.

Differential Revision: https://reviews.llvm.org/D47173

llvm-svn: 333015
2018-05-22 19:11:06 +00:00
Krzysztof Parzyszek 840b02bccf [Hexagon] Add patterns for accumulating HVX compares
llvm-svn: 333009
2018-05-22 18:27:02 +00:00
Florian Hahn a6e63f176c [NewGVN] Fix handling of assumes
This patch fixes two bugs:

* test1: Previously assume(a >= 5) concluded that a == 5. That's only
         valid for assume(a == 5)...
* test2: If operands were swapped, additional users were added to the
         wrong cmp operand. This resulted in an "unsettled iteration"
         assertion failure.

Patch by Nikita Popov

Differential Revision: https://reviews.llvm.org/D46974

llvm-svn: 333007
2018-05-22 17:38:22 +00:00
Jonas Devlieghere 63eca15e95 [DebugInfo] Invert DIE order for range errors.
When printing an error for an invalid address range in a DIE, we used to
print the child above the parent, which is counter intuitive. This patch
reverses the order and indents the child to mimic the way we print the
debug info section.

llvm-svn: 333006
2018-05-22 17:38:03 +00:00
Jonas Devlieghere 7e0b023302 [DebugInfo] Fix location list check in the verifier
We weren't properly verifying location lists because we tried obtaining
the offset as a constant.

llvm-svn: 333005
2018-05-22 17:37:27 +00:00
Paul Robinson 543c0e1d50 [DWARFv5] Put the DWO ID in its place.
In DWARF v5, the DWO ID is in the (split/skeleton) CU header, not an
attribute on the CU DIE.

This changes the size of those headers, so use the parsed size whenever
we have one, for simplicitly.

Differential Revision: https://reviews.llvm.org/D47158

llvm-svn: 333004
2018-05-22 17:27:31 +00:00
Lang Hames 5261aa9f91 [ORC] Move symbol-scanning and discard from BasicIRLayerMaterializationUnit in
to a base class (IRMaterializationUnit).

The new class, IRMaterializationUnit, provides a convenient base for any client
that wants to write a materializer for LLVM IR.

llvm-svn: 332993
2018-05-22 16:15:38 +00:00
David Bolvansky 41f4b64ee1 [InstCombine] Calloc-ed strings optimizations
Summary:
Example cases:
strlen(calloc(...)) -> 0

Reviewers: efriedma, bkramer

Reviewed By: bkramer

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D47059

llvm-svn: 332990
2018-05-22 15:41:23 +00:00
Aleksandar Beserminji a5f755186a [mips] Merge MipsLongBranch and MipsHazardSchedule passes
MipsLongBranchPass and MipsHazardSchedule passes are joined to one pass
because of mutual conflict. When MipsHazardSchedule inserts 'nop's, it
potentially breaks some jumps, so they have to be expanded to long
branches. When some branch is expanded to long branch, it potentially
creates a hazard situation, which should be fixed by adding nops.
New pass is called MipsBranchExpansion, it combines these two passes,
and runs them alternately until one of them reports no changes were made.

Differential Revision: https://reviews.llvm.org/D46641

llvm-svn: 332977
2018-05-22 13:24:38 +00:00
Simon Dardis 437153bb80 [mips] Correct the predicates of the cache and pref instructions
Reviewers: atanasyan, abeserminji, smaksimovic

Differential Revision: https://reviews.llvm.org/D46949

llvm-svn: 332970
2018-05-22 10:55:05 +00:00
Simon Pilgrim 4162d77744 [TTI] Add uniform/non-uniform constant Pow2 detection to TargetTransformInfo::getInstructionThroughput
This enables us to detect more fast path sdiv cases under cost analysis.

This patch also enables us to handle non-uniform-constant pow2 cases for X86 SDIV costs.

Found while working on D46276

Future patches can then extend the vectorizers to more fully support non-uniform pow2 cases.

Differential Revision: https://reviews.llvm.org/D46637

llvm-svn: 332969
2018-05-22 10:40:09 +00:00
Karl-Johan Karlsson 11d68a619e [LowerSwitch] Fixed faulty PHI node update
Summary:
When lowerswitch merge several cases into a new default block it's not
updating the PHI nodes accordingly. The code that update the PHI nodes
for the default edge only update the first entry and do not remove the
remaining ones, to make sure the number of entries match the number of
predecessors.

This is easily fixed by replacing the code that update the PHI node with
the already existing utility function for updating PHI nodes.

Reviewers: hans, reames, arsenm

Reviewed By: arsenm

Subscribers: wdng, llvm-commits

Differential Revision: https://reviews.llvm.org/D47055

llvm-svn: 332960
2018-05-22 08:46:48 +00:00
Bjorn Pettersson fecef6be9e [LoopVersioning] Don't modify the list that we iterate over in addPHINodes
Summary:
In LoopVersioning::addPHINodes we need to iterate over all
users for a value "Inst", and if the user is outside of the
VersionedLoop we should replace the use of "Inst" by using
the value "PN" instead.

Replacing the use of "Inst" for a user of "Inst" also means
that Inst->users() is modified. So it is not safe to do the
replace while iterating over Inst->users() as we used to do.
This patch splits the task into two steps. First we iterate
over Inst->users() to find all users that should be updated.
Those users are saved into a local data structure on the stack.
And then, in the second step, we do the actual updates. This
time iterating over the local data structure.

Reviewers: mzolotukhin, anemet

Reviewed By: mzolotukhin

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D47134

llvm-svn: 332958
2018-05-22 08:33:02 +00:00
Stanislav Mekhanoshin 0e132dca53 [AMDGPU] Optimze old value of v_mov_b32_dpp
We can eliminate old value if bound_ctrl = 1 and row_mask = bank_mask = 0xf.
This is alternative implementation working with the intrinsic in InstCombine.
Original review for past-ISel optimization: D46570.

Differential Revision: https://reviews.llvm.org/D46596

llvm-svn: 332956
2018-05-22 08:04:33 +00:00
Matt Arsenault 1349a04ef5 AMDGPU: Make v2i16/v2f16 legal on VI
This usually results in better code. Fixes using
inline asm with short2, and also fixes having a different
ABI for function parameters between VI and gfx9.

Partially cleans up the mess used for lowering of the d16
operations. Making v4f16 legal will help clean this up more,
but this requires additional work.

llvm-svn: 332953
2018-05-22 06:32:10 +00:00
Dan Gohman b81848272d [WebAssembly] Fix fast-isel lowering illegal argument and return types.
For both argument and return types, promote illegal types like i24 to i32,
and if a type can't be easily promoted, clear out the signature before
bailing out, so avoid leaving it in a partially complete state.

Fixes PR37546.

llvm-svn: 332947
2018-05-22 04:58:36 +00:00
Tom Stellard 44b30b4537 AMDGPU: Remove #include "MCTargetDesc/AMDGPUMCTargetDesc.h" from common headers
Summary:
MCTargetDesc/AMDGPUMCTargetDesc.h contains enums for all the instuction
and register defintions, which are huge so we only want to include
them where needed.

This will also make it easier if we want to split the R600 and GCN
definitions into separate tablegenerated files.

I was unable to remove AMDGPUMCTargetDesc.h from SIMachineFunctionInfo.h
because it uses some enums from the header to initialize default values
for the SIMachineFunction class, so I ended up having to remove includes of
SIMachineFunctionInfo.h from headers too.

Reviewers: arsenm, nhaehnle

Reviewed By: nhaehnle

Subscribers: MatzeB, kzhuravl, wdng, yaxunl, dstuttard, tpr, t-tye, javed.absar, llvm-commits

Differential Revision: https://reviews.llvm.org/D46272

llvm-svn: 332930
2018-05-22 02:03:23 +00:00
Peter Collingbourne 2d3eaeb2d4 MC: Remove dead code. NFCI.
This code appears to have been copied from the mach-o streamer. It has
no effect in ELF because indirect symbols are specific to mach-o.

llvm-svn: 332926
2018-05-22 01:20:46 +00:00
Sanjay Patel 17a870f07c [DAG] fold FP binops with undef operands to NaN
This is the FP sibling of D43141 with the corresponding IR change in rL327212.

We can't propagate undef here because if a variable operand is a NaN, these 
binops must propagate NaN. Neither global nor node-level fast-math makes a 
difference. If we have 'nnan', I think later folds can turn the NaN into undef.

The tests in X86/fp-undef.ll are meant to be the definitive verification for 
these folds - everything reduces identically now.

The other test changes are collateral damage. They may need to be altered to
preserve their intent.

Differential Revision: https://reviews.llvm.org/D47026

llvm-svn: 332920
2018-05-21 23:54:19 +00:00
Lang Hames 373f4628a5 [LKH] Add a replacement RTDyldLayer.
llvm-svn: 332918
2018-05-21 23:45:40 +00:00
Craig Topper 358b094971 [X86] Remove 128/256-bit cvtdq2ps, cvtudq2ps, cvtqq2pd, cvtuqq2pd intrinsics.
These can all be implemented with sitofp/uitofp instructions.

llvm-svn: 332916
2018-05-21 23:15:00 +00:00
Diego Caballero 1bd5f2261d Fix warning from r332654 with LLVM_ATTRIBUTE_USED
r332654 tried to fix an unused function warning with
a void cast. This approach worked for clang and gcc 
but not for MSVC. This commit replaces the void cast
with the LLVM_ATTRIBUTE_USED approach.

llvm-svn: 332910
2018-05-21 22:12:38 +00:00
Roman Lebedev 9f65d16d5d [DAGCombiner] isAllOnesConstantOrAllOnesSplatConstant(): look through bitcasts
Summary:
As pointed out in D46528, we errneously transform cases like `xor X, -1`,
even though we use said function.
It's because the `-1` is actually a bitcast there.
So i think we can just look through it in the function.

Differential Revision: https://reviews.llvm.org/D47156

llvm-svn: 332905
2018-05-21 21:41:10 +00:00
Roman Lebedev 7772de25d0 [DAGCombine][X86][AArch64] Masked merge unfolding: vector edition.
Summary:
This **appears** to be the last missing piece for the masked merge pattern handling in the backend.

This is [[ https://bugs.llvm.org/show_bug.cgi?id=37104 | PR37104 ]].

[[ https://bugs.llvm.org/show_bug.cgi?id=6773 | PR6773 ]] will introduce an IR canonicalization that is likely bad for the end assembly.
Previously, `andps`+`andnps` / `bsl` would be generated. (see `@out`)
Now, they would no longer be generated  (see `@in`), and we need to make sure that they are generated.

Differential Revision: https://reviews.llvm.org/D46528

llvm-svn: 332904
2018-05-21 21:41:02 +00:00
Lang Hames 502f81e37e [ORC] Preserve Materializing symbol flag during resolution.
llvm-svn: 332899
2018-05-21 21:11:22 +00:00
Lang Hames 0b0b41fcce [ORC] Lookup now returns an error if any symbols are not found.
Also tightens the behavior of ExecutionSession::failQuery. Queries can usually
only be failed by marking a symbol as failed-to-materialize, but
ExecutionSession::failQuery provides a second route, and both routes may be
executed from different threads. In the case that a query has already been
failed due to a materialization error, ExecutionSession::failQuery will
direct the error to ExecutionSession::reportError instead.

llvm-svn: 332898
2018-05-21 21:11:21 +00:00
Lang Hames add9b6805c [ORC] Remove the optional MaterializationResponsibility argument from lookup.
The lookup function provides blocking symbol resolution for JIT clients (not
layers themselves) so it does not need to track symbol dependencies via a
MaterializationResponsibility.

llvm-svn: 332897
2018-05-21 21:11:21 +00:00
Lang Hames 1cf9987f6e [ORC] Add IRLayer and ObjectLayer interfaces and related MaterializationUnits.
llvm-svn: 332896
2018-05-21 21:11:13 +00:00
Craig Topper 25444c852a [DAGCombiner] Use computeKnownBits to match rotate patterns that have had their amount masking modified by simplifyDemandedBits
SimplifyDemandedBits can remove bits from the masks for the shift amounts we need to see to detect rotates.

This patch uses zeroes from computeKnownBits to fill in some of these mask bits to make the match work.

As currently written this calls computeKnownBits even when the mask hasn't been simplified because it made the code simpler. If we're worried about compile time performance we can improve this.

I know we're talking about making a rotate intrinsic, but hopefully we can go ahead and do this change and just make sure the rotate intrinsic also handles it.

Differential Revision: https://reviews.llvm.org/D47116

llvm-svn: 332895
2018-05-21 21:09:18 +00:00
Reid Kleckner 537917d13c [X86] Simplify some X86 address mode folding code, NFCI
This code should really do exactly the same thing for 32-bit x86 and
64-bit small code models, with the exception that RIP-relative
addressing can't use base and index registers.

llvm-svn: 332893
2018-05-21 21:03:19 +00:00
Craig Topper aad3aefaeb [X86] Remove masking from vpternlog intrinsics. Use a select in IR instead.
This removes 6 intrinsics since we no longer need separate mask and maskz intrinsics.

Differential Revision: https://reviews.llvm.org/D47124

llvm-svn: 332890
2018-05-21 20:58:09 +00:00
Peter Collingbourne 274c4f7ab4 Fix a make_unique ambiguity.
llvm-svn: 332889
2018-05-21 20:56:28 +00:00
Sanjay Patel b8346e3f07 [InstCombine] remove fptrunc (select) code; NFCI
This pattern is handled within commonCastTransforms(),
so the code here is dead AFAICT.

llvm-svn: 332887
2018-05-21 20:39:35 +00:00
Peter Collingbourne c5a9765cea LTO: Replace split dwarf implementation that uses objcopy with one that uses direct emission.
Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47091

llvm-svn: 332884
2018-05-21 20:26:49 +00:00
Peter Collingbourne 9a45114b3c CodeGen: Add a dwo output file argument to addPassesToEmitFile and hook it up to dwo output.
Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47089

llvm-svn: 332881
2018-05-21 20:16:41 +00:00
Peter Collingbourne 63062d9d0f MC: Introduce an ELF dwo object writer and teach llvm-mc about it.
Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47051

llvm-svn: 332875
2018-05-21 19:44:54 +00:00
Jonas Devlieghere c111382aa8 [DebugInfo] Use absolute addresses in location lists
Rather than relying on the user to do the address calculating in
DW_AT_location we should just dump the absolute address.

rdar://problem/38513870

Differential revision: https://reviews.llvm.org/D47152

llvm-svn: 332873
2018-05-21 19:36:54 +00:00
Peter Collingbourne f0226e62a8 MC: Extract a derived class from ELFObjectWriter. NFCI.
This class will be used to create regular, non-split ELF files.

Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47049

llvm-svn: 332870
2018-05-21 19:30:59 +00:00
Peter Collingbourne dcd7d6c331 MC: Separate creating a generic object writer from creating a target object writer. NFCI.
With this we gain a little flexibility in how the generic object
writer is created.

Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47045

llvm-svn: 332868
2018-05-21 19:20:29 +00:00
Peter Collingbourne a29fe579f4 MC: Extract ELFObjectWriter's ELF writing functionality into an ELFWriter class. NFCI.
The idea is that we will be able to use this class to create multiple
files.

Differential Revision: https://reviews.llvm.org/D47048

llvm-svn: 332867
2018-05-21 19:18:28 +00:00
Peter Collingbourne 2602a0d40c Fix ubsan bounds check failure.
llvm-svn: 332866
2018-05-21 19:09:47 +00:00
Craig Topper f14e62c9a5 [EarlyCSE] Improve EarlyCSE of some absolute value cases.
Change matchSelectPattern to return X and -X for ABS/NABS in a well defined order. Adjust EarlyCSE to account for this. Ensure the SPF result is some kind of min/max and not abs/nabs in one place in InstCombine that made me nervous.

Prevously we returned the two operands of the compare part of the abs pattern. The RHS is always going to be a 0i, 1 or -1 constant. This isn't a very meaningful thing to return for any one. There's also some freedom in the abs pattern as to what happens when the value is equal to 0. This freedom led to early cse failing to match when different constants were used in otherwise equivalent operations. By returning the input and its negation in a defined order we can ensure an exact match. This also makes sure both patterns use the exact same subtract instruction for the negation. I believe CSE should evebntually make this happen and properly merge the nsw/nuw flags. But I'm not familiar with CSE and what order it does things in so it seemed like it might be good to really enforce that they were the same.

Differential Revision: https://reviews.llvm.org/D47037

llvm-svn: 332865
2018-05-21 18:42:42 +00:00
Peter Collingbourne 59a6fc469f MC: Remove stream and output functions from MCObjectWriter. NFCI.
Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47043

llvm-svn: 332864
2018-05-21 18:28:57 +00:00
Peter Collingbourne 438390fae1 MC: Have the object writers return the number of bytes written. NFCI.
This removes the last external use of the stream.

Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47042

llvm-svn: 332863
2018-05-21 18:23:50 +00:00
Stanislav Mekhanoshin 9badad2051 [AMDGPU] Add divergence analysis as a dependency for ISel
AMDGPUDAGToDAGISel adds DivergenceAnalysis in getAnalysisUsage
but does not list it in pass dependencies which may lead to
crash.

Differential Revision: https://reviews.llvm.org/D47151

llvm-svn: 332862
2018-05-21 18:18:52 +00:00
Peter Collingbourne f17b149d8c MC: Change object writers to use endian::Writer. NFCI.
Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47040

llvm-svn: 332861
2018-05-21 18:17:42 +00:00
Diego Caballero 168d04d544 [VPlan] Reland r332654 and silence unused func warning
r332654 was reverted due to an unused function warning in
release build. This commit includes the same code with the
warning silenced.

Differential Revision: https://reviews.llvm.org/D44338

llvm-svn: 332860
2018-05-21 18:14:23 +00:00
Peter Collingbourne 147db3e628 MC: Change MCAssembler::writeSectionData and writeFragmentPadding to take a raw_ostream. NFCI.
Also clean up a couple of hacks where we were writing the section
contents to another stream by setting the object writer's stream,
writing and setting it back.

Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47038

llvm-svn: 332858
2018-05-21 18:11:35 +00:00
Peter Collingbourne 571a3301ae MC: Change MCAsmBackend::writeNopData() to take a raw_ostream instead of an MCObjectWriter. NFCI.
To make this work I needed to add an endianness field to MCAsmBackend
so that writeNopData() implementations know which endianness to use.

Part of PR37466.

Differential Revision: https://reviews.llvm.org/D47035

llvm-svn: 332857
2018-05-21 17:57:19 +00:00
Tom Stellard a91ce17b5f AMDGPU/GlobalISel: Address post-commit review comments for r332379
MCRegisterInfo::getPhysRegSize() will be deprecated.

llvm-svn: 332856
2018-05-21 17:49:31 +00:00
Alexey Bataev 7c9ad0db3d [InstCombine] Fix PR37526: MinMax patterns produce an infinite loop.
Summary:
This patch fixes PR37526 by simplifying the newly generated LoadInst
instructions. If the pointer address is a bitcast from the pointer to
the NewType, we can just remove this extra bitcast instead of creating
the new one. This fixes the PR37526 + may speed up the whole compilation
process.

Reviewers: spatel, RKSimon, hfinkel

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D47144

llvm-svn: 332855
2018-05-21 17:46:34 +00:00
Andrea Di Biagio b5757abefb [X86][BtVer2] Add a 'J' prefix to the PRF/RCU defs. NFC
This is to keep the Jaguar model's naming convention. Processor resources all
have a 'J' prefix in the BtVer2 scheduling model.

llvm-svn: 332851
2018-05-21 16:30:26 +00:00
Robert Widmann 38fa750b7a [LLVM-C] Add DIBuilder Bindings For ObjC Classes
Summary: Add LLVMDIBuilderCreateObjCIVar, LLVMDIBuilderCreateObjCProperty, and LLVMDIBuilderCreateInheritance to allow declaring metadata for Objective-C class hierarchies and their associated properties and instance variables.

Reviewers: whitequark, deadalnix

Reviewed By: whitequark

Subscribers: harlanhaskins, llvm-commits

Differential Revision: https://reviews.llvm.org/D47123

llvm-svn: 332850
2018-05-21 16:27:35 +00:00
Lama Saba 9417f7ff2e [X86] - Avoid SFB pass - fix bug in updating the offsets for newly created copies
Change-Id: I169ab6fe7e187727c0298c2a1e2868a683f3e688
llvm-svn: 332849
2018-05-21 16:23:16 +00:00
James Henderson 004b729ed1 [DWARF] Refactor callback usage for .debug_line error handling
Change the "recoverable" error callback to take an Error instaed of a
string.

Reviewed by: JDevlieghere

Differential Revision: https://reviews.llvm.org/D46831

llvm-svn: 332845
2018-05-21 15:30:54 +00:00
Simon Pilgrim a8869e68a9 [X86][SSE] Add an assert to ensure that rotation amount is converted to a scale
Missed in rL332832 where we added SSE v4i32 rotations for PR37426.

llvm-svn: 332844
2018-05-21 15:17:23 +00:00
Tim Northover 4e3eec39fa ARM: be conservative when asked load/store alignment of weird type.
Chances are we'll be asked again after type legalization, but before that point
it's better to claim misaligned accesses aren't allowed than to assert.

llvm-svn: 332840
2018-05-21 12:43:54 +00:00
Nico Weber e4a12cfa2f revert r332610, it breaks cfi, see D46326
llvm-svn: 332838
2018-05-21 11:44:39 +00:00
Aleksandar Beserminji 4977705727 [mips] Revert Merge MipsLongBranch and MipsHazardSchedule passes
Revert this patch due buildbot failure.

Differential Revision: https://reviews.llvm.org/D46641

llvm-svn: 332837
2018-05-21 11:38:52 +00:00
David Green 8ceab61c75 [CVP] Require DomTree for new Pass Manager
We were previously using a DT in CVP through SimplifyQuery, but not requiring it in
the new pass manager. Hence it would crash if DT was not already available. This now
gets DT directly and plumbs it through to where it is used (instead of using it
through SQ).

llvm-svn: 332836
2018-05-21 11:06:28 +00:00
Eric Christopher 563d0b9cb9 Fix up a few grammar issues.
llvm-svn: 332835
2018-05-21 10:27:36 +00:00
Aleksandar Beserminji de7be5e46f [mips] Merge MipsLongBranch and MipsHazardSchedule passes
MipsLongBranchPass and MipsHazardSchedule passes are joined to one pass
because of mutual conflict. When MipsHazardSchedule inserts 'nop's, it
potentially breaks some jumps, so they have to be expanded to long
branches. When some branch is expanded to long branch, it potentially
creates a hazard situation, which should be fixed by adding nops.
New pass is called MipsBranchExpansion, it combines these two passes,
and runs them alternately until one of them reports no changes were made.

Differential Revision: https://reviews.llvm.org/D46641

llvm-svn: 332834
2018-05-21 10:20:02 +00:00
Simon Pilgrim 5aa7cdfd70 [X86][SSE] Support v4i32 rotations (PR37426)
As suggested by Fabian on PR37426, we can use PMULUDQ to perform v4i32 vector rotations as the upper 32bits of the multiply will contain the 'wrapped' bits of the rotation.

v8i16/v16i8 rotations would be straightforward to add to lowerRotate in the future - ideally we'd mostly share code with the vector shifts lowering.

Differential Revision: https://reviews.llvm.org/D46954

llvm-svn: 332832
2018-05-21 09:45:59 +00:00
Robert Widmann 360d6e35e6 [LLVM-C] Improve Bindings For Aliases
Summary: Add wrappers for a module's alias iterators and a getter and setter for the aliasee value.

Reviewers: whitequark, deadalnix

Reviewed By: whitequark

Subscribers: llvm-commits, harlanhaskins

Differential Revision: https://reviews.llvm.org/D46808

llvm-svn: 332826
2018-05-20 23:49:08 +00:00
Craig Topper e4c045b7df [X86] Remove mask arguments from permvar builtins/intrinsics. Use a select in IR instead.
Someday maybe we'll use selects for all intrinsics.

llvm-svn: 332824
2018-05-20 23:34:04 +00:00
Nico Weber 41597b92b1 Revert 332750, llvm part (see comment on D46910).
llvm-svn: 332823
2018-05-20 23:03:17 +00:00
Simon Dardis 777afc7fbd [mips] Add microMIPSR6 ll/sc instructions.
Previously the compiler was using the microMIPSR3 variants, incorrectly.

Reviewers: atanasyan, abeserminji, smaksimovic

Differential Revision: https://reviews.llvm.org/D46948

llvm-svn: 332820
2018-05-20 17:21:00 +00:00
Sanjay Patel a003c728a5 [InstCombine] choose 1 form of abs and nabs as canonical
We already do this for min/max (see the blob above the diff), 
so we should do the same for abs/nabs.
A sign-bit check (<s 0) is used as a predicate for other IR 
transforms and it's likely the best for codegen.

This might solve the motivating cases for D47037 and D47041, 
but I think those patches still make sense. We can't guarantee 
this canonicalization if the icmp has more than one use.

Differential Revision: https://reviews.llvm.org/D47076

llvm-svn: 332819
2018-05-20 14:23:23 +00:00
Haicheng Wu 69ba0613f2 [GlobalMerge] Exit early if only one global is to be merged
To save some compilation time and prevent some unnecessary changes.

Differential Revision: https://reviews.llvm.org/D46640

llvm-svn: 332813
2018-05-19 18:00:02 +00:00
Brian Gesiak 9968e0dd49 Re-revert "[Option] Fix PR37006 prefix choice in findNearest"
Summary:
Reverting due to a test failure in an llvm-mt test on some buildbots, namely
http://green.lab.llvm.org/green/job/clang-stage2-configure-Rlto/26020/.

llvm-svn: 332812
2018-05-19 16:21:01 +00:00
Robert Widmann 025c78f5d7 [LLVM-C] Use Length-Providing Value Name Getters and Setters
Summary:
- Provide LLVMGetValueName2 and LLVMSetValueName2 that return and take the length of the provided C string respectively
- Deprecate LLVMGetValueName and LLVMSetValueName

Reviewers: whitequark, deadalnix

Reviewed By: whitequark

Subscribers: llvm-commits, harlanhaskins

Differential Revision: https://reviews.llvm.org/D46890

llvm-svn: 332810
2018-05-19 15:08:36 +00:00
Max Kazantsev c0b268f90c [IRCE] Fix miscompile with range checks against negative values
In the patch rL329547, we have lifted the over-restrictive limitation on collected range
checks, allowing to work with range checks with the end of their range not being
provably non-negative. However it appeared that the non-negativity of this value was
assumed in the utility function `ClampedSubtract`. In particular, its reasoning is based
on the fact that `0 <= SINT_MAX - X`, which is not true if `X` is negative.

The function `ClampedSubtract` is only called twice, once with `X = 0` (which is OK)
and the second time with `X = IRC.getEnd()`, where we may now see the problem if
the end is actually a negative value. In this case, we may sometimes miscompile.

This patch is the conservative fix of the miscompile problem. Rather than rejecting
non-provably non-negative `getEnd()` values, we will check it for non-negativity in
runtime. For this, we use function `smax(smin(X, 0), -1) + 1` that is equal to `1` if `X`
is non-negative and is equal to 0 if `X` is negative. If we multiply `Begin, End` of safe
iteration space by this function calculated for `X = IRC.getEnd()`, we will get the original
`[Begin, End)` if `IRC.getEnd()` was non-negative (and, thus, `ClampedSubtract` worked
correctly) and the empty range `[0, 0)` in case if ` IRC.getEnd()` was negative.

So we in fact prohibit execution of the main loop if at least one of range checks was
made against a negative value (and we figured it out in runtime). It is still better than
what we have before (non-negativity had to be proved in compile time) and prevents
us from miscompile, however it is sometiles too restrictive for unsigned range checks
against a negative value (which in fact can be eliminated).

Once we re-implement `ClampedSubtract` in a way that it handles negative `X` correctly,
this limitation can be lifted, too.

Differential Revision: https://reviews.llvm.org/D46860
Reviewed By: samparker

llvm-svn: 332809
2018-05-19 13:06:37 +00:00
Benjamin Kramer a76b64ff80 [MergeICmps] Don't crash when memcmp is not available
Fixes clang crashing with -fno-builtin, PR37527.

llvm-svn: 332808
2018-05-19 12:51:59 +00:00
Simon Pilgrim ede0e4073e Fix MSVC unused variable warning. NFCI.
AMDGPURegisterInfo::getSubRegFromChannel is a static method - we don't need to get the AMDGPURegisterInfo instance.

llvm-svn: 332807
2018-05-19 12:46:02 +00:00