Andrea Di Biagio [Wed, 14 Oct 2015 10:03:13 +0000 (10:03 +0000)]
[x86][FastISel] Teach how to select nontemporal stores.
This patch teaches x86 fast-isel how to select nontemporal stores.
On x86, we can use MOVNTI for nontemporal stores of doublewords/quadwords.
Instructions (V)MOVNTPS/PD/DQ can be used for SSE2/AVX aligned nontemporal
vector stores.
Before this patch, fast-isel always selected 'movd/movq' instead of 'movnti'
for doubleword/quadword nontemporal stores. In the case of nontemporal stores
of aligned vectors, fast-isel always selected movaps/movapd/movdqa instead of
movntps/movntpd/movntdq.
With this patch, if we use SSE2/AVX intrinsics for nontemporal stores we now
always get the expected (V)MOVNT instructions.
The lack of fast-isel support for nontemporal stores was spotted when analyzing
the -O0 codegen for nontemporal stores.
Chris Bieneman [Wed, 14 Oct 2015 07:37:00 +0000 (07:37 +0000)]
[CMake] Set Policy CMP0048 to NEW
CMake 3.0 introduced the VERSION option for the project() command. If you don't specify the VERSION in the function it will clear out variables matching ${PROJECT_NAME}_VERSION_${MAJOR|MINOR|PATCH|TWEAK}.
This makes overriding LLVM_VERSION_* not work properly with newer versions of CMake. To make this work properly we need to:
(1) Optionally set the policy to NEW
(2) Move default versions and setting PACKAGE_VERSION to before the call to project()
(3) If the policy is set, pass the VERSION and LANGUAGES options in the new format
This change should have no behavioral change for CMake versions before 3.0, and it makes the behavior of later versions match the earlier versions.
Craig Topper [Wed, 14 Oct 2015 05:37:42 +0000 (05:37 +0000)]
[X86] Update CPU detection to only enable XSAVE features if the OS has enabled them and the saving of YMM state. This seems to be consistent with gcc behavior.
Richard Smith [Wed, 14 Oct 2015 00:04:19 +0000 (00:04 +0000)]
Rename one of our two llvm::GCOVOptions classes to llvm::GCOV::Options. We used
to get away with this because llvm/Support/GCOV.h was an implementation detail
of the llvm-gcov tool, but it's now being used by FDO.
Diego Novillo [Tue, 13 Oct 2015 22:48:46 +0000 (22:48 +0000)]
Sample profiles - Add a name table to the binary encoding.
Binary encoded profiles used to encode all function names inline at
every reference. This is clearly suboptimal in terms of space. This
patch fixes this by adding a name table to the header of the file.
Cong Hou [Tue, 13 Oct 2015 22:27:41 +0000 (22:27 +0000)]
Update MachineBranchProbabilityInfo::normalizeEdgeWeights to make sure there is no zero weight in the output, and also add a missing test for JumpThreading.
The test is for the patch in http://reviews.llvm.org/D10979 but was missing when committing that patch.
We forgot to append the terminatepad's arguments which resulted in us
treating the old terminatepad as an argument to the new terminatepad
causing us to crash immediately. Instead, add the old terminatepad's
arguments to the new terminatepad.
Kevin Enderby [Tue, 13 Oct 2015 20:48:04 +0000 (20:48 +0000)]
Tweak to r250117 and change to use ErrorOr and drop isSizeValid for
ArchiveMemberHeader, suggestion by Rafael Espíndola.
Also The clang-x86-win2008-selfhost bot still does not like the
malformed-machos 00000031.a test, so removing it for now. All
the other bots are fine with it however.
Joseph Tremoulet [Tue, 13 Oct 2015 20:18:27 +0000 (20:18 +0000)]
[WinEH] Add CoreCLR EH table emission
Summary:
Emit the handler and clause locations immediately after the standard
xdata.
Clauses are emitted in the same order and format used to communiate them
to the CLR Execution Engine.
Add a lit test to verify correct table generation on a small but
interesting example function.
One of the changes in lib/Target/AMDGPU/AMDGPUMCInstLower.cpp was a new
one. Previously, bundle iterators and single-instruction iterators
could be compared to each other (comparing on underlying pointers).
I changed a comparison from using `MBB->end()` to using
`MBB->instr_end()`, since both end iterators should point at the some
place anyway.
I don't think the implicit conversion between the two iterator types is
a good idea since it's fairly easy to accidentally compare to the wrong
thing (they aren't always end iterators). Otherwise I would have just
added the conversion.
Even with that, no there should be functionality change here.
Remove remaining `ilist_iterator` implicit conversions from
LLVMScalarOpts.
This change exposed some scary behaviour in
lib/Transforms/Scalar/SCCP.cpp around line 1770. This patch changes a
call from `Function::begin()` to `&Function::front()`, since the return
was immediately being passed into another function that takes a
`Function*`. `Function::front()` started to assert, since the function
was empty. Note that `Function::end()` does not point at a legal
`Function*` -- it points at an `ilist_half_node` -- so the other
function was getting garbage before. (I added the missing check for
`Function::isDeclaration()`.)
Cong Hou [Tue, 13 Oct 2015 18:43:10 +0000 (18:43 +0000)]
Update the branch weight metadata in JumpThreading pass.
Currently in JumpThreading pass, the branch weight metadata is not updated after CFG modification. Consider the jump threading on PredBB, BB, and SuccBB. After jump threading, the weight on BB->SuccBB should be adjusted as some of it is contributed by the edge PredBB->BB, which doesn't exist anymore. This patch tries to update the edge weight in metadata on BB->SuccBB by scaling it by 1 - Freq(PredBB->BB) / Freq(BB->SuccBB).
[PGO]: Eliminate calls to __llvm_profile_register_function for Linux.
On Linux, the profile runtime can use __start_SECTNAME and __stop_SECTNAME
symbols defined by the linker to locate the start and end location of
a named section (with C name). This eliminates the need for instrumented
binary to call __llvm_profile_register_function during start-up time.
Kevin Enderby [Tue, 13 Oct 2015 17:06:34 +0000 (17:06 +0000)]
The issue with the malformed-machos 00000031.a test is that it needed ‘-arch x86_64’
flag as it was a Mach-O universal file.
The default as to which architecture slice that is dumped without an -arch flag
depends on the host architecture and the contents of the universal file. The
malformed archive 00000031.a file has both an x86_64 and i386 slice. So for
for x86_64 hosts only that slice is dumped, for non-x86_64 hosts, which is many
of the bots both slices are dumped.
The test is intended to only check that the malformation of the x86_64 which
has a non-decimal characters in the size field of the archive header so it no
longer crashes.
The problem turned out that the i388 slice of the malformed archive had a
different malformation which was causing the non-x86_64 bots to get this error:
llvm-objdump -macho -disassemble -arch i386 00000031.a
Archive : .00000031.a 00000031.a(c_start.o):
LLVM ERROR: Symbol name entry points before beginning or past end of file.
and causing the test as it was written to fail. So by adding ‘-arch x86_64’ it
should correct the test and the malformation on the i388 slice will not be
dumped.
Also the removal of the malformed-machos mem-crup-0261.macho was not causing
the issue so that is put back in.
Joseph Tremoulet [Tue, 13 Oct 2015 16:44:30 +0000 (16:44 +0000)]
[WinEH] Iterate state changes instead of invokes
Summary:
Add an iterator that can walk across blocks and which visits the state
transitions rather than state ranges, with explicit transitions to -1
indicating the presence of top-level calls that may throw and cause the
current function to unwind to caller. This will simplify code that needs
to identify nested try regions.
Refactor SEH and C++EH table generation to use the new
InvokeStateChangeIterator, and remove the InvokeLabelIterator they were
using.
James Molloy [Tue, 13 Oct 2015 10:43:33 +0000 (10:43 +0000)]
[GlobalsAA] Don't assume anything about functions that may be overridden
Weak linkage and friends allow a symbol to be overriden outside the
code generator's model, so GlobalsAA shouldn't assume that anything it
can compute about such a symbol is valid.
BitcodeWriter: Stop using implicit ilist iterator conversion, NFC
Now LLVMBitWriter compiles without implicit ilist iterator conversions.
In these cases, the cleanest thing was to switch to range-based for
loops. Since there wasn't much noise I converted sub-loops and parent
loops as a drive-by.
Sanjoy Das [Tue, 13 Oct 2015 02:53:27 +0000 (02:53 +0000)]
[SCEV] Put some utilites in the ScalarEvolution class
In a later commit, `SplitBinaryAdd` will be used outside `IsConstDiff`,
so lift that out. And lift out `IsConstDiff` as
`computeConstantDifference` to keep things clean and to avoid playing
C++ access specifier games.
Continuing the work from last week to remove implicit ilist iterator
conversions. First related commit was probably r249767, with some more
motivation in r249925. This edition gets LLVMTransformUtils compiling
without the implicit conversions.
Matt Arsenault [Tue, 13 Oct 2015 00:49:00 +0000 (00:49 +0000)]
DAGCombiner: Don't stop finding better chain on 2 aliases
The comment says this was stopped because it was unlikely to be
profitable. This is not true if you want to combine vector loads
with multiple components.
t4 = store t0:1, t0:1
t5 = store t4, t1:0
t6 = store t5, t2:0
t7 = store t6, t3:0
We want to get all of these stores onto a chain
that is a TokenFactor of these N loads. This mostly
solves the AMDGPU merge-stores.ll regressions
with -combiner-alias-analysis for merging vector
stores of vector loads.
JF Bastien [Tue, 13 Oct 2015 00:28:47 +0000 (00:28 +0000)]
x86: preserve flags when folding atomic operations
Summary:
D4796 taught LLVM to fold some atomic integer operations into a single
instruction. The pattern was unaware that the instructions clobbered
flags.
This patch adds the missing EFLAGS definition.
Floating point operations don't set flags, the subsequent fadd
optimization is therefore correct. The same applies for surrounding
load/store optimizations.
Kevin Enderby [Tue, 13 Oct 2015 00:05:17 +0000 (00:05 +0000)]
Remove the correct unstable malformed-machos test mem-crup-0261.macho and
restore the malformed-machos 00000031.a test. Hopefully this will get all the
build bots happy again. I’ll again keep an eye on them.
Matt Arsenault [Mon, 12 Oct 2015 23:59:50 +0000 (23:59 +0000)]
DAGCombiner: Combine extract_vector_elt from build_vector
This basic combine was surprisingly missing.
AMDGPU legalizes many operations in terms of 32-bit vector components,
so not doing this results in many extra copies and subregister extracts
that need to be cleaned up later.
InstCombine already does this for the hasOneUse case. The target hook
is to fix a handful of tests which break (e.g. ARM/vmov.ll) which turn
from a vector materialize repeated immediate instruction to a constant
vector load with more scalar copies from it.
Cong Hou [Mon, 12 Oct 2015 23:02:58 +0000 (23:02 +0000)]
Assign correct edge weights to unwind destinations when lowering invoke statement.
When lowering invoke statement, all unwind destinations are directly added as successors of call site block, and the weight of those new edges are not assigned properly. Actually, default weight 16 are used for those edges. This patch calculates the proper edge weights for those edges when collecting all unwind destinations.
Simon Pilgrim [Mon, 12 Oct 2015 23:00:11 +0000 (23:00 +0000)]
[SelectionDAG] Add common vector constant folding helper function
We have a number of functions that implement constant folding of vectors (unary and binary ops) in near identical manners (and the differences don't appear to be critical).
This patch introduces a common implementation (SelectionDAG::FoldConstantVectorArithmetic) and calls this in both the unary and binary op cases.
After this initial patch I intend to begin enabling vector constant folding for a wider number of opcodes in SelectionDAG::getNode().
Kevin Enderby [Mon, 12 Oct 2015 22:04:54 +0000 (22:04 +0000)]
Fixed bugs in llvm-obdump while parsing Mach-O files from malformed archives
that caused aborts. This was because of the characters of the ‘Size’ field in
the archive header did not contain decimal characters.
Cong Hou [Mon, 12 Oct 2015 19:44:08 +0000 (19:44 +0000)]
Update the branch weight metadata in JumpThreading pass.
In JumpThreading pass, the branch weight metadata is not updated after CFG modification. Consider the jump threading on PredBB, BB, and SuccBB. After jump threading, the weight on BB->SuccBB should be adjusted as some of it is contributed by the edge PredBB->BB, which doesn't exist anymore. This patch tries to update the edge weight in metadata on BB->SuccBB by scaling it by 1 - Freq(PredBB->BB) / Freq(BB->SuccBB).
Reid Kleckner [Mon, 12 Oct 2015 19:43:34 +0000 (19:43 +0000)]
Make Win64 localescape offsets FP relative instead of SP relative
We made them SP relative back in March (r233137) because that's the
value the runtime passes to EH functions. With the new cleanuppad IR,
funclets adjust their frame argument from SP to FP, so our offsets
should now be FP-relative.
Andrea Di Biagio [Mon, 12 Oct 2015 19:22:30 +0000 (19:22 +0000)]
[x86] Fix wrong lowering of vsetcc nodes (PR25080).
Function LowerVSETCC (in X86ISelLowering.cpp) worked under the wrong
assumption that for non-AVX512 targets, the source type and destination type
of a type-legalized setcc node were always the same type.
This assumption was unfortunately incorrect; the type legalizer is not always
able to promote the return type of a setcc to the same type as the first
operand of a setcc.
In the case of a vsetcc node, the legalizer firstly checks if the first input
operand has a legal type. If so, then it promotes the return type of the vsetcc
to that same type. Otherwise, the return type is promoted to the 'next legal
type', which, for vectors of MVT::i1 is always a 128-bit integer vector type.
Example (-mattr=+avx):
%0 = trunc <8 x i32> %a to <8 x i23>
%1 = icmp eq <8 x i23> %0, zeroinitializer
The type legalizer would firstly check if 't5' has a legal type. If so, then it
would reuse that same type to promote the return type of the setcc node.
Unfortunately 't5' is of illegal type v8i23, and therefore it cannot be used to
promote the return type of the setcc node. Consequently, the setcc return type
is promoted to v8i16. Later on, 't5' is promoted to v8i32 thus leading to the
following dag node:
v8i16 = setcc t32, t25, seteq:ch
where t32 and t25 are now values of type v8i32.
Before this patch, function LowerVSETCC would have wrongly expanded the setcc
to a single X86ISD::PCMPEQ. Surprisingly, ISel was still able to match an
instruction. In our case, ISel would have matched a VPCMPEQWrr:
t37: v8i16 = X86ISD::VPCMPEQWrr t36, t25
However, t36 and t25 are both VR256, while the result type is instead of class
VR128. This inconsistency ended up causing the insertion of COPY instructions
like this:
%vreg7<def> = COPY %vreg3; VR128:%vreg7 VR256:%vreg3
Which is an invalid full copy (not a sub register copy).
Eventually, the backend would have hit an UNREACHABLE "Cannot emit physreg copy
instruction" in the attempt to expand the malformed pseudo COPY instructions.
This patch fixes the problem adding the missing logic in LowerVSETCC to handle
the corner case of a setcc with 128-bit return type and 256-bit operand type.
This problem was originally reported by Dimitry as PR25080. It has been latent
for a very long time. I have added the minimal reproducible from that bugzilla
as test setcc-lowering.ll.
Reid Kleckner [Mon, 12 Oct 2015 17:42:32 +0000 (17:42 +0000)]
Don't call PrepareEHLandingPad on non EH pads
This was a minor bug in r249492. Calling PrepareEHLandingPad on a
non-landingpad was a no-op, but it attempted to get the generic pointer
register class, which apparently doesn't exist for some targets.
Oliver Stannard [Mon, 12 Oct 2015 15:52:36 +0000 (15:52 +0000)]
[Debug] Look through bitcasts to find argument registers
On targets where f32 is not legal, we have to look through a BITCAST SDNode to
find the register that an argument is stored in when emitting debug info, or we
will not be able to emit a DW_AT_location for it.
Greg Bedwell [Mon, 12 Oct 2015 15:11:47 +0000 (15:11 +0000)]
Fix rename() sometimes failing if another process uses openFileForRead()
On Windows, fs::rename() could fail is another process was reading the
file at the same time using fs::openFileForRead(). In most cases the user
wouldn't notice as fs::rename() will continue to retry for 2000ms. Typically
this is enough for the read to complete and a retry to succeed, but if the
disk is being it too hard then the response time might be longer than the
retry time and the rename would fail with a permission error.
Add FILE_SHARE_DELETE to the sharing flags for CreateFileW() in
fs::openFileForRead() and try ReplaceFileW() prior to MoveFileExW()
in fs::rename().
projects / llvm / log