Tony Jiang [Wed, 5 Jul 2017 16:55:00 +0000 (16:55 +0000)]
[Power9] Exploit vector extract with variable index.
This patch adds the exploitation for new power 9 instructions which extract
variable elements from vectors:
VEXTUBLX
VEXTUBRX
VEXTUHLX
VEXTUHRX
VEXTUWLX
VEXTUWRX
Differential Revision: https://reviews.llvm.org/D34032
Commit on behalf of Zaara Syeda (syzaara@ca.ibm.com)
Tony Jiang [Wed, 5 Jul 2017 16:00:38 +0000 (16:00 +0000)]
[Power9] Exploit vector integer extend instructions when indices aren't correct.
This patch adds on to the exploitation added by https://reviews.llvm.org/D33510.
This now catches build vector nodes where the inputs are coming from sign
extended vector extract elements where the indices used by the vector extract
are not correct. We can still use the new hardware instructions by adding a
shuffle to move the elements to the correct indices. I introduced a new PPCISD
node here because adding a vector_shuffle and changing the elements of the
vector_extracts was getting undone by another DAG combine.
Commit on behalf of Zaara Syeda (syzaara@ca.ibm.com)
Differential Revision: https://reviews.llvm.org/D34009
David Blaikie [Wed, 5 Jul 2017 15:23:56 +0000 (15:23 +0000)]
DebugInfo: Generalize LoadedObjectInfoHelper from RuntimeDyld
Make it usable by any class derived (even indirectly) from
LoadedObjectInfo by allowing a custom base class to be specified and
perfect forwarding to the ctor.
Daniel Sanders [Wed, 5 Jul 2017 14:50:18 +0000 (14:50 +0000)]
[globalisel][tablegen] Finish fixing compile-time regressions by merging the matcher and emitter state machines.
Summary:
Also, made a few minor tweaks to shave off a little more cumulative memory consumption:
* All rules share a single NewMIs instead of constructing their own. Only one
will end up using it.
* Use MIs.resize(1) instead of MIs.clear();MIs.push_back(I) and prevent
GIM_RecordInsn from changing MIs[0].
David Green [Wed, 5 Jul 2017 13:25:58 +0000 (13:25 +0000)]
[IndVarSimplify] Add AShr exact flags using induction variables ranges.
This adds exact flags to AShr/LShr flags where we can statically
prove it is valid using the range of induction variables. This
allows further optimisations to remove extra loads.
[SystemZ] Simplify handling of 128-bit multiply/divide instruction
Several integer multiply/divide instructions require use of a
register pair as input and output. This patch moves setting
up the input register pair from C++ code to TableGen, simplifying
the whole process and making it more easily extensible.
Fixes a couple of whitespace errors, re-sorts the vector floating-point
instructions to make them more easily extensible, and adds a missing
pseudo instruction.
Tom Stellard [Wed, 5 Jul 2017 12:57:30 +0000 (12:57 +0000)]
CMake: Add LLVM_UTILS_INSTALL_DIR option
Summary:
This is like the LLVM_TOOLS_INSTALL_DIR option, but for the utils
that are installed when the LLVM_INSTALL_UTILS. This option
defaults to 'bin' to remain consistent with the current behavior, but
distros may want to install these to libexec/llvm.
[GlobalISel] Refactor Legalizer helpers for libcalls
We used to have a helper that replaced an instruction with a libcall.
That turns out to be too aggressive, since sometimes we need to replace
the instruction with at least two libcalls. Therefore, change our
existing helper to only create the libcall and leave the instruction
removal as a separate step. Also rename the helper accordingly.
This implements suggesting other mnemonics when an invalid one is specified,
for example:
$ echo "adXd r1,r2,#3" | llvm-mc -triple arm
<stdin>:1:1: error: invalid instruction, did you mean: add, qadd?
adXd r1,r2,#3
^
The implementation is target agnostic, but as a first step I have added it only
to the ARM backend; so the ARM backend is a good example if someone wants to
enable this too for another target.
Daniel Sanders [Wed, 5 Jul 2017 09:39:33 +0000 (09:39 +0000)]
[globalisel][tablegen] Added instruction emission to the state-machine-based matcher.
Summary:
This further improves the compile-time regressions that will be caused by a
re-commit of r303259.
Also added included preliminary work in preparation for the multi-insn emitter
since I needed to change the relevant part of the API for this patch anyway.
Max Kazantsev [Wed, 5 Jul 2017 06:38:49 +0000 (06:38 +0000)]
[IndVars] Canonicalize comparisons between non-negative values and indvars
-If there is a IndVar which is known to be non-negative, and there is a value which is also non-negative,
then signed and unsigned comparisons between them produce the same result. Both of those can be
seen in the same loop. To allow other optimizations to simplify them, we turn all instructions like
Nirav Dave [Wed, 5 Jul 2017 01:21:23 +0000 (01:21 +0000)]
Rewrite areNonVolatileConsecutiveLoads to use BaseIndexOffset
Relanding after rewriting undef.ll test to avoid host-dependant
endianness.
As discussed in D34087, rewrite areNonVolatileConsecutiveLoads using
generic checks. Also, propagate missing local handling from there to
BaseIndexOffset checks.
Tests of note:
* test/CodeGen/X86/build-vector* - Improved.
* test/CodeGen/BPF/undef.ll - Improved store alignment allows an
additional store merge
* test/CodeGen/X86/clear_upper_vector_element_bits.ll - This is a
case we already do not handle well. Here, the DAG is improved, but
scheduling causes a code size degradation.
[profiledata] Avoid creating a temporary vector in getNumValueData
getValueSitesForKind returns ArrayRef which has a cast operator
to std::vector, as a result a temporary vector is created
if the type of the variable is const std::vector&
that is suboptimal in this case.
Anna Thomas [Wed, 5 Jul 2017 01:16:29 +0000 (01:16 +0000)]
[SafepointIRVerifier] Add verifier pass for finding GC relocation bugs
Original Patch and summary by Philip Reames.
RewriteStatepointsForGC tries to rewrite a function in a manner where
the optimizer can't end up using a pointer value after it might have
been relocated by a safepoint. This pass checks the invariant that
RSForGC is supposed to establish and that (if we constructed semantics
correctly) later passes must preserve.
This has been a really useful diagnostic tool when initially developing
the rewriting scheme and has found numerous bugs.
Gadi Haber [Tue, 4 Jul 2017 21:51:05 +0000 (21:51 +0000)]
NFC.
Made some updates to the half.ll test under CodeGen to make it friendly to the update_llc_test_checks .py tool as follows:
1.Removing the llc flag -asm-verbose=false
2.Grouping the multiple check-prefix directives
3.Apply update_llc_test_checks.py tool on the test
This change is needed to easily update scheduling changes in an upcoming patch.
Recommit r307064, "[InstCombine] Add test cases demonstrating creation of extra bswap instrinsic calls when when optimizing bswap and bitwise ops when the bswaps have additional uses. NFC"
Anna Thomas [Tue, 4 Jul 2017 15:24:08 +0000 (15:24 +0000)]
[FastISel] Move gc intrinsic test to X86 directory
Move from generic to X86 directory since gc intrinsics only supposed in
X86 64 bit.
Add target triple as well.
Fixes build failure in i686-linux-RA caused by rL307084.
Anna Thomas [Tue, 4 Jul 2017 15:09:09 +0000 (15:09 +0000)]
[FastISel][SelectionDAG]Teach fastISel about GC intrinsics
Summary:
We are crashing in LLC at O0 when gc intrinsics are present in the block.
The reason being FastISel performs basic block ISel by modifying GC.relocates
to be the first instruction in the block. This can cause us to visit the GC
relocate before it's corresponding GC.statepoint is visited, which is incorrect.
When we lower the statepoint, we record the base and derived pointers, along
with the gc.relocates. After this we can visit the gc.relocate.
This patch avoids fastISel from incorrectly creating the block with gc.relocate
as the first instruction.
Daniel Sanders [Tue, 4 Jul 2017 14:35:06 +0000 (14:35 +0000)]
[globalisel][tablegen] Partially fix compile-time regressions by converting matcher to state-machine(s)
Summary:
Replace the matcher if-statements for each rule with a state-machine. This
significantly reduces compile time, memory allocations, and cumulative memory
allocation when compiling AArch64InstructionSelector.cpp.o after r303259 is
recommitted.
The following patches will expand on this further to fully fix the regressions.
Revert r307064, "[InstCombine] Add test cases demonstrating creation of extra bswap instrinsic calls when when optimizing bswap and bitwise ops when the bswaps have additional uses. NFC"
Seems confused between %tmpN and unnamed %N to give same name.
[InstCombine] Add test cases demonstrating creation of extra bswap instrinsic calls when when optimizing bswap and bitwise ops when the bswaps have additional uses. NFC
I assume bswap intrinsics are somewhat costly so we should be making sure we are getting rid of them not creating more.
[tablegen] Avoid creating a temporary vector in getInstructionCase
Record::getValues returns ArrayRef which has a cast operator
to std::vector, as a result a temporary vector is created
if the type of the variable is const std::vector&
that is suboptimal in this case.
[X86] Add comment string for broadcast loads from the constant pool.
Summary:
When broadcasting from the constant pool its useful to print out the final vector similar to what we do for normal moves from the constant pool.
I changed only a couple tests that were broadcast focused. One of them had been previously hand tweaked after running the script so that it could check the constant pool declaration. But I think this patch makes that unnecessary now since we can check the comment instead.
[tablegen] Avoid creating a temporary vector in getInstructionCase
Record::getValues returns ArrayRef which has a cast operator
to std::vector, as a result a temporary vector is created
if the type of the variable is const std::vector&
that was suboptimal in this case.
Lang Hames [Tue, 4 Jul 2017 04:42:30 +0000 (04:42 +0000)]
[Orc] Remove the memory manager argument to addModule, and de-templatize the
symbol resolver argument.
De-templatizing the symbol resolver is part of the ongoing simplification of
ORC layer API.
Removing the memory management argument (and delegating construction of memory
managers for RTDyldObjectLinkingLayer to a functor passed in to the constructor)
allows us to build JITs whose base object layers need not be compatible with
RTDyldObjectLinkingLayer's memory mangement scheme. For example, a 'remote
object layer' that sends fully relocatable objects directly to the remote does
not need a memory management scheme at all (that will be handled by the remote).
Anton Yartsev [Tue, 4 Jul 2017 01:08:55 +0000 (01:08 +0000)]
[legalize-types] Clean up softening machinery.
The patch makes SoftenFloatResult/Operand logic just the same as all other legalization routines have: SoftenFloatResult() now fills the SoftenFloats map and SoftenFloatOperand() perform all needed replacements. This prevents softening mashinery from leaving stale entries in SoftenFloats map (that resulted in errors during the legalize type checking) and clarifies softening. The patch replaces https://reviews.llvm.org/D29265.
MathExtras UnitTest: Assert that isPowerOf2(0) is false. NFC.
Summary:
This is a follow-up on D34077. Elena observed that the
correctness of the code relies on isPowerOf2(0) returning false.
Adding a test to cover this corner-case.
Summary:
Add a combine for creating a truncate to replace a build_vector composed of extracts with
indices that form a stride-2^N series.
Example:
v8i32 V = ...
v4i32 build_vector((extract_elt V, 0), (extract_elt V, 2), (extract_elt V, 4), (extract_elt V, 6))
-->
v4i32 truncate (bitcast V to v4i64)
Related discussion in llvm-dev about canonicalizing shuffles to
truncates in LLVM IR:
http://lists.llvm.org/pipermail/llvm-dev/2017-January/108936.html.
[x86] auto-generate complete checks for tests; NFC
These all used 'CHECK-NOT' which isn't necessary if we have complete checks.
There were also several over-specifications in the RUN params such as CPU model or OS requirement
projects / llvm / log