Craig Topper [Tue, 6 Jun 2017 07:13:15 +0000 (07:13 +0000)]
[ValueTracking] Remove scalar only restriction from isKnownNonEqual. The computeKnownBits and isKnownNonZero calls this code relies on should work fine for vectors.
This will be used by another commit to remove some code from InstSimplify that is redundant for scalars, but was needed for vectors due to this issue.
Xin Tong [Tue, 6 Jun 2017 02:34:41 +0000 (02:34 +0000)]
Add a dominanance check interface that uses caching for instructions within same basic block.
Summary:
This problem stems from the fact that instructions are allocated using new
in LLVM, i.e. there is no relationship that can be derived by just looking
at the pointer value.
This interface dispatches to appropriate dominance check given 2 instructions,
i.e. in case the instructions are in the same basic block, ordered basicblock
(with instruction numbering and caching) are used. Otherwise, dominator tree
is used.
This is a preparation patch for https://reviews.llvm.org/D32720
[x86] Add the test for folding stack spills into pextrw.
This is a negative test as pextrw doesn't write to all 32-bits of the
spilled GPR. This fold ended up happening when D32684 was landed and
covers the regression that motivated reverting it in r304762.
[x86] Revert the X86FoldTablesEmitter due to more miscompiles.
In testing, we've found yet another miscompile caused by the new tables.
And this one is even less clear how to fix (we could teach it to fold
a 16-bit load instead of the 32-bit load it wants, or block folding
entirely).
Also, the approach to excluding instructions seems increasingly to not
scale well.
I have left a more detailed analysis on the review log for the original
patch (https://reviews.llvm.org/D32684) along with suggested path
forward. I will land an additional test case that I wrote which covers
the code that was miscompiling (folding into the output of `pextrw`) in
a subsequent commit to keep this a pure revert.
For each commit reverted here, I've restricted the revert to the
non-test code touching the x86 fold table emission until the last commit
where I did revert the test updates. This means the *new* test cases
added for `insertps` and `xchg` remain untouched (and continue to pass).
Reverted commits:
r304540: [X86] Don't fold into memory operands into insertps in the ...
r304347: [TableGen] Adapt more places to getValueAsString now ...
r304163: [X86] Don't fold away the memory operand of an xchg.
r304123: Don't capture a temporary std::string in a StringRef.
r304122: Resubmit "[X86] Adding new LLVM TableGen backend that ..."
Original commit was in r304088, and after a string of fixes was reverted
previously in r304121 to fix build bots, and then re-landed in r304122.
[cmake] Enable reverse iteration by default through build macro
Summary:
Reverse iteration can be turned on, by default, by setting -DLLVM_REVERSE_ITERATION:BOOL=ON during cmake.
With this enabled, we can uncover lots of cases of non-determinism in codegen by simply running our tests (without any other change).
We can then setup a buildbot which will have this turned on by default. Initially, a lot of unit tests will fail in this configuration.
Once we start fixing non-determinism issues, we can gradually make this a blocker for patches.
- Move ISel (and pre-isel) pass construction into TargetPassConfig
- Extract AsmPrinter construction into a helper function
Putting the ISel code into TargetPassConfig seems a lot more natural and
both changes together make make it easier to build custom pipelines
involving .mir in an upcoming commit. This moves MachineModuleInfo to an
earlier place in the pass pipeline which shouldn't have any effect.
Sanjay Patel [Tue, 6 Jun 2017 00:18:11 +0000 (00:18 +0000)]
[x86] fix over-specific triple; NFC
There's nothing darwin-specific in these tests, and using that
setting causes extra phantom diffs when the auto-generated check
lines are regenerated today.
Althought it is not wrong to spill undef values, it is useless and harms
both code size and runtime. Before spilling a value, check that its
content actually matters.
LSR: Calculate instruction cost only if InsnsCost is set to true (NFC)
Summary:
The patch guard all instruction cost calculations with InsnCosts (-lsr-insns-cost) option.
Currently even if the option set to false we calculate and print (in debug mode) instruction costs.
This is because in `SelectionDAGIsel` we split critical edges without
updating the corresponding dominator for the function (and we claim
in `MachineFunctionPass::getAnalysisUsage()` that the domtree is preserved).
We could either stop preserving the domtree in `getAnalysisUsage`
or tell `splitCriticalEdge()` to update it.
As the second option is easy to implement, that's the one I chose.
Zachary Turner [Mon, 5 Jun 2017 21:40:33 +0000 (21:40 +0000)]
[CodeView] Handle Cross Module Imports and Exports.
While it's not entirely clear why a compiler or linker might
put this information into an object or PDB file, one has been
spotted in the wild which was causing llvm-pdbdump to crash.
This patch adds support for reading-writing these sections.
Since I don't know how to get one of the native tools to
generate this kind of debug info, the only test here is one
in which we feed YAML into the tool to produce a PDB and
then spit out YAML from the resulting PDB and make sure that
it matches.
This ensures that we can emit the ObjC Image Info structure on COFF and
ELF as well. The frontend already would attempt to emit this
information but would get dropped when generating assembly or an object
file.
Craig Topper [Mon, 5 Jun 2017 20:48:05 +0000 (20:48 +0000)]
[ConstantRange] Remove costly udivrem from ConstantRange::truncate
Truncate currently uses a udivrem call which is going to be slow particularly for larger than 64-bit widths.
As far as I can tell all we were trying to do was modulo LowerDiv by (MaxValue+1) and make sure whatever value was effectively subtracted from LowerDiv was also subtracted from UpperDiv.
This patch recognizes that MaxValue+1 is a power of 2 so we can just use a bitwise AND to accomplish a modulo operation or isolate the upper bits.
Sanjay Patel [Mon, 5 Jun 2017 17:01:10 +0000 (17:01 +0000)]
[DAGCombine] Fix unchecked calls to DAGCombiner::*ExtPromoteOperand
Other calls to DAGCombiner::*PromoteOperand check the result, but here it could cause an assertion in getNode.
Falling back to any extend in this case instead of failing outright seems correct to me.
No test case because:
The failure was triggered by an out of tree backend. In order to trigger it, a backend would need to overload
TargetLowering::IsDesirableToPromoteOp to return true for a type for which ISD::SIGN_EXTEND_INREG is marked
illegal. In tree, only X86 overloads and sometimes returns true for MVT::i16 yet it marks
setOperationAction(ISD::SIGN_EXTEND_INREG, MVT::i16 , Legal);.
Adam Nemet [Mon, 5 Jun 2017 16:27:09 +0000 (16:27 +0000)]
Handle non-unique edges in edge-dominance
This removes a quadratic behavior in assert-enabled builds.
GVN propagates the equivalence from a condition into the blocks guarded by the
condition. E.g. for 'if (a == 7) { ... }', 'a' will be replaced in the block
with 7. It does this by replacing all the uses of 'a' that are dominated by
the true edge.
For a switch with N cases and U uses of the value, this will mean N * U calls
to 'dominates'. Asserting isSingleEdge in 'dominates' make this N^2 * U
because this function checks for the uniqueness of the edge. I.e. traverses
each edge between the SwitchInst's block and the cases.
The change removes the assert and makes 'dominates' works correctly in the
presence of non-unique edges.
This brings build time down by an order of magnitude for an input that has
~10k cases in a switch statement.
Dmitry Mikulin [Mon, 5 Jun 2017 16:24:25 +0000 (16:24 +0000)]
Symbols re-defined with -wrap and -defsym need to be excluded from inter-
procedural optimizations to prevent dropping symbols and allow the linker
to process re-directs.
PR33145: --wrap doesn't work with lto.
Differential Revision: https://reviews.llvm.org/D33621
Dimitry Andric [Mon, 5 Jun 2017 13:52:31 +0000 (13:52 +0000)]
Remove duplicate LLVM_NATIVE_ from llvm's config.h
Summary:
Since LLVM_NATIVE_ARCH, LLVM_NATIVE_ASMPARSER, LLVM_NATIVE_ASMPRINTER,
LLVM_NATIVE_DISASSEMBLER, LLVM_NATIVE_TARGET, LLVM_NATIVE_TARGETINFO and
LLVM_NATIVE_TARGETMC are already defined in llvm-config.h, there seems
to be no reason to also define them in config.h. Also, I can only find
usage of these macros in files that include llvm-config.h.
So let's remove the duplicated macros from config.h.
Diana Picus [Mon, 5 Jun 2017 12:54:53 +0000 (12:54 +0000)]
[ARM] GlobalISel: Constrain callee register on indirect calls
When lowering calls, we generate instructions with machine opcodes
rather than generic ones. Therefore, we need to constrain the register
classes of the operands.
Also enable the machine verifier on the arm-irtranslator.ll test, since
that would've caught this issue.
Dimitry Andric [Mon, 5 Jun 2017 11:22:18 +0000 (11:22 +0000)]
Fix building DynamicLibrary.cpp with musl libc
Summary:
The workaround added in rL301240 for stderr/out/in symbols being both
macros and globals is only necessary for glibc, and it does not compile
with musl libc. Alpine Linux has had the following fix for it:
Javed Absar [Mon, 5 Jun 2017 10:53:26 +0000 (10:53 +0000)]
Move ARM specific test to ELF/ARM dir
Moving ARM specific test clang-section.s
from MC/ELF to MC/ELF/ARM
Buildbots reported failures on
commit https://reviews.llvm.org/rL304705
Full details are available at:
http://lab.llvm.org:8011/builders/llvm-clang-lld-x86_64-scei-ps4-windows10pro-fast/builds/10333
Javed Absar [Mon, 5 Jun 2017 10:09:13 +0000 (10:09 +0000)]
Add support for #pragma clang section
This patch provides a means to specify section-names for global variables,
functions and static variables, using #pragma directives.
This feature is only defined to work sensibly for ELF targets.
One can specify section names as:
#pragma clang section bss="myBSS" data="myData" rodata="myRodata" text="myText"
One can "unspecify" a section name with empty string e.g.
#pragma clang section bss="" data="" text="" rodata=""
Reviewers: Roger Ferrer, Jonathan Roelofs, Reid Kleckner
Differential Revision: https://reviews.llvm.org/D33413
Peter Smith [Mon, 5 Jun 2017 09:37:12 +0000 (09:37 +0000)]
[ARM] Support fixup for Thumb2 modified immediate
This change adds a new fixup fixup_t2_so_imm for the t2_so_imm_asmoperand
"T2SOImm". The fixup permits code such as:
.L1:
sub r3, r3, #.L2 - .L1
.L2:
to assemble in Thumb2 as well as in ARM state.
The operand predicate isT2SOImm() explicitly doesn't match expressions
containing :upper16: and :lower16: as expressions with these operators
must match the movt and movw instructions.
The test mov r0, foo2 in thumb2-diagnostics is moved to a new file as the
fixup delays the error message till after the assembler has quit due to
the other errors.
As the mov instruction shares the t2_so_imm_asmoperand mov instructions
with a non constant expression now match t2MOVi rather than t2MOVi16 so the
error message is slightly different.
This fixes a bug that can cause extractelements with operands that
haven't been defined yet to be inserted at a wrong point when
optimising insertelements.
The issue is because we are not placing sequential vector elements together early enough, we fail to recognise many combinable patterns - consecutive scalar loads, extractions etc.
Instead, this patch unpacks progressively larger sequential vector elements together:
This does mean that we are creating UNPCKL shuffle of different value types, but the relevant combines that benefit from this are quite capable of handling the additional BITCASTs that are now included in the shuffle tree.
Ayal Zaks [Sun, 4 Jun 2017 13:29:51 +0000 (13:29 +0000)]
[LV] Make scalarizeInstruction() non-virtual. NFC.
Following the request made in https://reviews.llvm.org/D32871,
scalarizeInstruction() which is no longer overridden by InnerLoopUnroller is
hereby made non-virtual in InnerLoopVectorizer.
Craig Topper [Sun, 4 Jun 2017 08:21:51 +0000 (08:21 +0000)]
[ConstantFolding][X86] Replace an LLVM_FALLTHROUGH with a break because it really shouldn't fallthrough.
This is actually NFC because the next case starts with the same if statement as this case did. So the result will be the same and it will fallthrough to the end of the switch. But there's no reason to rely on that so we should just break.
David Blaikie [Sun, 4 Jun 2017 03:27:12 +0000 (03:27 +0000)]
Make the Twine pretty-printer work with GDB 7.11
Apparently ::NodeKind is sometimes part of the name in GDB.
Without this patch I get the following error message from GDB:
`Unhandled NodeKind llvm::Twine::NodeKind::EmptyKind`.
[AMDGPU] Untangle SDWA pass from SIShrinkInstructions
Remove dependency of SDWA pass on SIShrinkInstructions.
The goal is to move SDWA even higher in the stack to avoid second run
of MachineLICM, MachineCSE and SIFoldOperands.
Also added handling to preserve original src modifiers.
Zachary Turner [Sat, 3 Jun 2017 00:33:35 +0000 (00:33 +0000)]
[PDB] Fix use after free.
Previously MappedBlockStream owned its own BumpPtrAllocator that
it would allocate from when a read crossed a block boundary. This
way it could still return the user a contiguous buffer of the
requested size. However, It's not uncommon to open a stream, read
some stuff, close it, and then save the information for later.
After all, since the entire file is mapped into memory, the data
should always be available as long as the file is open.
Of course, the exception to this is when the data isn't *in* the
file, but rather in some buffer that we temporarily allocated to
present this contiguous view. And this buffer would get destroyed
as soon as the strema was closed.
The fix here is to force the user to specify the allocator, this
way it can provide an allocator that has whatever lifetime it
chooses.
Matthias Braun [Sat, 3 Jun 2017 00:26:35 +0000 (00:26 +0000)]
LiveRegUnits: Port recent LivePhysRegs bugfixes
Adjust code to look more like the code in LivePhysRegs and port over the
fix for LivePhysRegs from r304001 and adapt to the new CSR management in
MachineRegisterInfo.
Sanjay Patel [Fri, 2 Jun 2017 23:40:46 +0000 (23:40 +0000)]
[x86] fix over-specific triple; NFC
There's nothing darwin-specific in these tests, and using
that setting causes extra phantom diffs when the auto-generated
check lines are regenerated today.
projects / llvm / log