Craig Topper [Fri, 31 May 2019 05:20:27 +0000 (05:20 +0000)]
[X86] Correct the ins operand order for MASKPAIR16STORE to match other store instructions.
This makes the 5 address operands come first. And the data operand comes last.
This matches the operand order the instruction is created with. It's also the
expected order in X86MCInstLower. So everything appeared to work, but the
operands didn't match their declared type.
Fixes a -verify-machineinstrs failure.
Also remove the isel patterns from these instructions since they should only
be used for stack spills and reloads. I'm not even sure what types the patterns
were looking for to match.
Petr Hosek [Fri, 31 May 2019 01:34:51 +0000 (01:34 +0000)]
[CMake] Provide an option to use relative paths in debug info
CMake always uses absolute file paths in the generated compiler
invocation which results in absolute file paths being embedded in debug
info. This is undesirable when building a toolchain e.g. on bots as the
debug info may embed the bot source checkout path which is meaningless
anywhere else.
This change introduces the LLVM_USE_RELATIVE_PATHS_IN_DEBUG_INFO which uses
-fdebug-prefix-map (where supported) options to rewrite paths embedded
into debug info with relative ones. Additionally, LLVM_SOURCE_PREFIX can
be used to override the path to source directory with a different one.
Craig Topper [Thu, 30 May 2019 22:29:06 +0000 (22:29 +0000)]
[X86] Remove code that unnecessarily sets EXTLOAD with src type of v2f32/v4f32/v8f32 as Legal for SSE2/AVX/AVX512 respectively. NFC
The LoadExt table defaults to all combinations being Legal. For
vector types, only src VTs with an i1 element type were ever changed.
So we don't need to mark them legal manually.
Chris Bieneman [Thu, 30 May 2019 22:25:48 +0000 (22:25 +0000)]
Support codesigning bundles and forcing
Summary:
Clangd's framework is assembled by copying binaries from the lib and bin directories into a bundle shape. This results in an invalid bundle code signature because the signature only applies to the binaries not the resources.
This patch adds two new options to `llvm_codesign` to enable re-signing the library and XPC service as bundles.
The `BUNDLE_PATH` option allow specifying an explicit path to codesign, which enables signing bundles which aren't generated using CMake's `FRAMEWORK` or `BUNDLE` target properties.
The `FORCE` option allows re-signing binaries that have already been signed. This is required for how clangd exposes the clangd library and tools as both XPC and non-XPC services using the same binary.
[Remarks][NFC] Move the serialization to lib/Remarks
Separate the remark serialization to YAML from the LLVM Diagnostics.
This adds a new serialization abstraction: remarks::Serializer. It's
completely independent from lib/IR and it provides an easy way to
replace YAML by providing a new remarks::Serializer.
I don't have a test case for these, but there is a test case for D62266
where, even after all the constant-folding patches, we still end up
with endless combine loop. Which makes sense, since we don't constant
fold for opaque constants.
Martin Storsjo [Thu, 30 May 2019 20:53:21 +0000 (20:53 +0000)]
[InstCombine] Avoid use after free in DenseMap, when built with GCC
Previously, this used a statement like this:
Map[A] = Map[B];
This is equivalent to the following:
const auto &Src = Map[B];
auto &Dest = Map[A];
Dest = Src;
The second statement, "auto &Dest = Map[A];" can insert a new
element into the DenseMap, which can potentially grow and reallocate
the DenseMap's internal storage, which will invalidate the existing
reference to the source. When doing the actual assignment,
the Src reference is dereferenced, accessing memory that was
freed when the DenseMap grew.
This issue hasn't shown up when LLVM was built with Clang, because
the right hand side ended up dereferenced before evaulating the
left hand side. (If the value type is a larger data type, Clang doesn't
do this but behaves like GCC.)
With GCC, a cast to Value* isn't enough to make it dereference the
right hand side reference before invoking operator[] (while that is
enough to make Clang/LLVM do the right thing for larger types), but
storing it in an intermediate variable in a separate statement works.
Roman Lebedev [Thu, 30 May 2019 20:37:39 +0000 (20:37 +0000)]
[DAGCombine] (x - C) - y -> (x - y) - C fold. Try 3
Summary:
Again only vectors affected. Frustrating. Let me take a look into that..
https://rise4fun.com/Alive/AAq
This is a recommit, originally committed in rL361852, but reverted
to investigate test-suite compile-time hangs, and then reverted in
rL362109 to fix missing constant folds that were causing
endless combine loops.
Roman Lebedev [Thu, 30 May 2019 20:37:29 +0000 (20:37 +0000)]
[DAGCombine][X86][AArch64][AMDGPU] (x - y) + -1 -> add (xor y, -1), x fold. Try 3
Summary:
This prevents regressions in next patch,
and somewhat recovers from the regression to AMDGPU test in D62223.
It is indeed not great that we leave vector decrement,
don't transform it into vector add all-ones..
https://rise4fun.com/Alive/ZRl
This is a recommit, originally committed in rL361852, but reverted
to investigate test-suite compile-time hangs, and then reverted in
rL362109 to fix missing constant folds that were causing
endless combine loops.
Roman Lebedev [Thu, 30 May 2019 20:37:18 +0000 (20:37 +0000)]
[DAGCombiner][X86][AArch64][SPARC][SystemZ] y - (x + C) -> (y - x) - C fold. Try 3
Summary:
Direct sibling of D62223 patch.
While i don't have a direct motivational pattern for this,
it would seem to make sense to handle both patterns (or none),
for symmetry?
The aarch64 changes look neutral;
sparc and systemz look like improvement (one less instruction each);
x86 changes - 32bit case improves, 64bit case shows that LEA no longer
gets constructed, which may be because that whole test is `-mattr=+slow-lea,+slow-3ops-lea`
https://rise4fun.com/Alive/ffh
This is a recommit, originally committed in rL361852, but reverted
to investigate test-suite compile-time hangs, and then reverted in
rL362109 to fix missing constant folds that were causing
endless combine loops.
Roman Lebedev [Thu, 30 May 2019 20:36:54 +0000 (20:36 +0000)]
[DAGCombiner][X86][AArch64][AMDGPU] (x + C) - y -> (x - y) + C fold. Try 3
Summary:
The main motivation is shown by all these `neg` instructions that are now created.
In particular, the `@reg32_lshr_by_negated_unfolded_sub_b` test.
AArch64 test changes all look good (`neg` created), or neutral.
X86 changes look neutral (vectors), or good (`neg` / `xor eax, eax` created).
I'm not sure about `X86/ragreedy-hoist-spill.ll`, it looks like the spill
is now hoisted into preheader (which should still be good?),
2 4-byte reloads become 1 8-byte reload, and are elsewhere,
but i'm not sure how that affects that loop.
I'm unable to interpret AMDGPU change, looks neutral-ish?
This is hopefully a step towards solving [[ https://bugs.llvm.org/show_bug.cgi?id=41952 | PR41952 ]].
https://rise4fun.com/Alive/pkdq (we are missing more patterns, i'll submit them later)
This is a recommit, originally committed in rL361852, but reverted
to investigate test-suite compile-time hangs, and then reverted in
rL362109 to fix missing constant folds that were causing
endless combine loops.
Michael Trent [Thu, 30 May 2019 20:09:09 +0000 (20:09 +0000)]
Write new tests for r362121
Summary:
The tests for r362121 ran dsymutil against a test binary every time.
This caused problems on lld-x86_64-ubuntu-fast as dsymutil required
a lipo tool be available to process those binaries.
This change rewrites the new test cases in macho-disassemble-g-dsym
to use bespoke test binaries (exe and dwarf) simplifying the test's
runtime dependencies.
The changes to tools/llvm-objdump/MachODump.cpp are unchanged from
r362121
Lang Hames [Thu, 30 May 2019 19:59:20 +0000 (19:59 +0000)]
[RuntimeDyld] Apply padding and alignment bumps to all sections with stubs, and
increase the MachO/x86-64 stub alignment to 8.
Stub alignment should be guaranteed for any section containing RuntimeDyld
stubs/GOT-entries. To do this we should pad and align all sections containing
stubs, not just code sections.
This commit also bumps the MachO/x86-64 stub alignment to 8, so that GOT entries
will be aligned.
Tim Northover [Thu, 30 May 2019 18:48:23 +0000 (18:48 +0000)]
Reapply: IR: add optional type to 'byval' function parameters
When we switch to opaque pointer types we will need some way to describe
how many bytes a 'byval' parameter should occupy on the stack. This adds
a (for now) optional extra type parameter.
If present, the type must match the pointee type of the argument.
The original commit did not remap byval types when linking modules, which broke
LTO. This version fixes that.
Note to front-end maintainers: if this causes test failures, it's probably
because the "byval" attribute is printed after attributes without any parameter
after this change.
With LLPC, previous investigation has suggested that si-scheduler
interacts badly with SiFormMemoryClauses on an XNACK target in some
games.
That needs further investigation in the future. In the meantime, this
commit adds a target-specific attribute to allow us to disable
SIFormMemoryClauses by setting it to 1 on a per-function basis for LLPC
to use.
Florian Hahn [Thu, 30 May 2019 18:46:13 +0000 (18:46 +0000)]
[LV] Remove the redundant using LoopVectorizationPlanner:VPlanPtr
VPlan.h already contains the declaration of VPlanPtr type alias:
using VPlanPtr = std::unique_ptr<VPlan>;
The LoopVectorizationPlanner class also contains the same declaration
of VPlanPtr and therefore LoopVectorize requires a long wording when
its methods return VPlanPtr:
Puyan Lotfi [Thu, 30 May 2019 18:06:28 +0000 (18:06 +0000)]
[MIR-Canon] Add support for rewriting VRegs that are typed but don't have an RC.
There were crashes (addrspace-memoperands.mir was only one of them) in MIR that
had operands that came from before register classes were set. With these
operands, creating a replacement vreg (for MIR-Canon's renaming) needs to use
the vreg type rather than the RegisterClass which is not present.
Michael Trent [Thu, 30 May 2019 17:56:05 +0000 (17:56 +0000)]
Support Universal dSYM files in llvm-objdump
Summary:
Commonly programmers use llvm-objdump to disassemble Mach-O target
binaries with Mach-O dSYMS. While llvm-objdump allows programmers to
disassemble Universal binaries, it previously did not recognize
Universal dSYM files. This change updates llvm-objdump to support
passing in Universal files via the -dsym option. Now, when
disassembling a Mach-O file either as a stand alone file or as an entry
in a Universal binariy, llvm-objdump will search through a Universal
dSYM for a Mach-O matching the architecture flag of the file being
disassembled.
J. Ryan Stinnett [Thu, 30 May 2019 16:46:22 +0000 (16:46 +0000)]
[Docs] Modernize references to macOS
Summary:
This updates all places in documentation that refer to "Mac OS X", "OS X", etc.
to instead use the modern name "macOS" when no specific version number is
mentioned.
If a specific version is mentioned, this attempts to use the OS name at the time
of that version:
* Mac OS X for 10.0 - 10.7
* OS X for 10.8 - 10.11
* macOS for 10.12 - present
Kevin P. Neal [Thu, 30 May 2019 16:44:47 +0000 (16:44 +0000)]
[FPEnv] Added a special UnrollVectorOp method to deal with the chain on StrictFP opcodes
This change creates UnrollVectorOp_StrictFP. The purpose of this is to address a failure that consistently occurs when calling StrictFP functions on vectors whose number of elements is 3 + 2n on most platforms, such as PowerPC or SystemZ. The old UnrollVectorOp method does not expect that the vector that it will unroll will have a chain, so it has an assert that prevents it from running if this is the case. This new StrictFP version of the method deals with the chain while unrolling the vector. With this new function in place during vector widending, llc can run vector-constrained-fp-intrinsics.ll for SystemZ successfully.
Submitted by: Drew Wock <drew.wock@sas.com>
Reviewed by: Cameron McInally, Kevin P. Neal
Approved by: Cameron McInally
Differential Revision: http://reviews.llvm.org/D62546
Roman Lebedev [Thu, 30 May 2019 16:07:11 +0000 (16:07 +0000)]
[DAGCombine] Revert of recommit of "binop-with-const hoisting" patches
I was looking into an endless combine loop the uncommitted follow-up patch
was causing, and it appears even these patches can exibit such an
endless loop. The root cause is that we try to hoist one binop (add/sub) with
constant operand, and if we get two such binops both of which are
eligible for this hoisting, we get stuck.
Sjoerd Meijer [Thu, 30 May 2019 14:34:29 +0000 (14:34 +0000)]
[ARM] Change the MC names for VMAXNM/VMINNM
Now the NEON ones have a prefix "NEON_", and the VFP ones have a
prefix "VFP_". This is so that the regex in ARMScheduleA57.td can be
made to match both of _those_ classes of VMAXNM without also matching
the MVE ones that are going to be introduced soon. NFCI.
Sjoerd Meijer [Thu, 30 May 2019 12:57:04 +0000 (12:57 +0000)]
[ARM] add target arch definitions for 8.1-M and MVE
This adds:
- LLVM subtarget features to make all the new instructions conditional on,
- CPU and FPU names for use on clang's command line, with default FPUs set
so that "armv8.1-m.main+fp" and "armv8.1-m.main+fp.dp" will select the right
FPU features,
- architecture extension names "mve" and "mve.fp",
- ABI build attribute support for v8.1-M (a new value for Tag_CPU_arch) and MVE
(a new actual tag).
George Rimar [Thu, 30 May 2019 12:39:05 +0000 (12:39 +0000)]
[llvm-readobj] - Rewrite reloc-types.test to use YAML. NFCI.
This change rewrites and splits reloc-types.test
to use yaml2obj instead of precompiled binaries.
That allowed to remove 7 precompiled objects from the inputs.
I took the existent objects, used obj2yaml on them, simplified the result and
used yaml2obj in the test case with the result.
Notes:
* I converted, but did not remove relocs.obj.elf-i386, relocs.obj.elf-x86_64 or relocs.obj.elf-mips objects
because found they are used in other tests.
* I was unable to convert relocs.obj.elf-ppc64, because obj2yaml hangs on this file for me.
* I was unable to convert relocs.obj.macho-arm, relocs.obj.macho-i386 and relocs.obj.macho-x86_64
because the output produced by obj2yaml does not seem to be correct.
* Because of the above I did not remove the script for creating all
of those objects: test\tools\llvm-readobj\Inputs\relocs.py
Sjoerd Meijer [Thu, 30 May 2019 12:37:05 +0000 (12:37 +0000)]
[ARM] Introduce separate features for FP registers
The MVE extension in Arm v8.1-M permits the use of some move, load and
store isntructions which access the FP registers, even if there's no
actual FP support in the processor (in particular, if you have the
integer-only version of MVE).
Therefore, we need separate subtarget features to condition those
instructions on, which are implied by both FP and MVE but are not part
of either.
Error was:
/home/buildslave/buildslave/clang-armv7-linux-build-cache/llvm/tools/llvm-readobj/ELFDumper.cpp:3540:7:
error: non-constant-expression cannot be narrowed from type 'llvm::support::detail::packed_endian_specific_integral<unsigned long long,
llvm::support::endianness::little, 1>::value_type' (aka 'unsigned long long') to 'size_t' (aka 'unsigned int') in initializer list [-Wc++11-narrowing]
StrTabSec->sh_size};
George Rimar [Thu, 30 May 2019 10:36:52 +0000 (10:36 +0000)]
[llvm-readobj/llvm-readelf] - Implement GNU style dumper of the SHT_GNU_verdef section.
It was not implemented yet, we had only LLVM style dumper implemented.
Section description is here: https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/symversion.html
Simon Pilgrim [Thu, 30 May 2019 10:25:20 +0000 (10:25 +0000)]
[X86][SSE] Improve bool vector extload (PR26091)
We already have good codegen for (vXiY *ext(vXi1 bitcast(iX))) cases, this patch uses it for loads of vXi1 types as well - changing the load into a iX integer load, and bitcasting so that combineToExtendBoolVectorInReg can then use it.
George Rimar [Thu, 30 May 2019 10:14:41 +0000 (10:14 +0000)]
[llvm-readobj/llvm-readelf] - Implement GNU style dumper of the SHT_GNU_verneed section.
It was not implemented yet, we had only LLVM style dumper implemented.
Section description is here: https://refspecs.linuxfoundation.org/LSB_2.0.1/LSB-Core/LSB-Core/symverrqmts.html
Sjoerd Meijer [Thu, 30 May 2019 08:07:06 +0000 (08:07 +0000)]
[ARM] Add an MVE execution domain
MVE architecturally specifies a 'beat' system in which a vector
instruction executed now will complete its actual operation over the
next four cycles, so it can overlap with the execution of the previous
and next MVE instruction.
This makes it generally an advantage to avoid moving values back and
forth between MVE registers and anywhere else, if there's any sensible
way to do the same processing in whatever register type the values
already occupied.
That's just what the 'execution domain' system is supposed to achieve.
So here we add a new execution domain which will contain all the MVE
vector instructions when they are added.
If an assembly instruction has to mention an input operand name twice,
for example the MVE VMOV instruction that accesses two lanes of the
same vector by writing 'vmov r1, r2, q0[3], q0[1]', then the obvious
way to write its AsmString is to include the same operand (here $Qd)
twice. But this causes the AsmMatcher generator to omit that
instruction completely from the match table, on the basis that the
generator isn't clever enough to deal with the duplication.
But you need to have _some_ way of dealing with an instruction like
this - and in this case, where the mnemonic is shared with many other
instructions that the AsmMatcher does handle, it would be very painful
to take it out of the AsmMatcher system completely.
A nicer way is to add a custom AsmMatchConverter routine, and let that
deal with the problem if the autogenerated converter can't. But that
doesn't work, because TableGen leaves the instruction out of its table
_even_ if you provide a custom converter.
Solution: this change, which makes TableGen relax the restriction on
duplicated operands in the case where there's a custom converter.
Sjoerd Meijer [Thu, 30 May 2019 07:30:37 +0000 (07:30 +0000)]
[TableGen] New default operand "undef_tied_input"
This is a new special identifier which you can use as a default in
OperandWithDefaultOps. The idea is that you use it for an input
operand of an instruction that's tied to an output operand, and its
semantics are that (in the default case) the input operand's value is
not used at all.
The detailed effect is that when instruction selection emits the
instruction in the form of a pre-regalloc MachineInstr, it creates an
IMPLICIT_DEF node to use as that input.
If you're creating an MCInst with explicit register names, then the
right handling would be to set the input operand to the same register
as the output one (honouring the tie) and to add the 'undef' flag
indicating that that register is deemed to acquire a new don't-care
definition just before we read it. But I haven't done that in this
commit, because there was no need to - no Tablegen backend seems to
autogenerate default fields in an MCInst.
Pengfei Wang [Thu, 30 May 2019 03:59:16 +0000 (03:59 +0000)]
[X86] Add ENQCMD instructions
For more details about these instructions, please refer to the latest
ISE document:
https://software.intel.com/en-us/download/intel-architecture-instruction-set-extensions-programming-reference.
Petr Hosek [Thu, 30 May 2019 01:24:31 +0000 (01:24 +0000)]
[CMake] Set LLVM_PATH in the runtimes build
This avoids using llvm-config for inferring various paths within the
runtimes build. We also set LLVM_INCLUDE_DIR variable that's used by
these builds and move assignment of LLVM_BINARY_DIR and LLVM_LIBRARY_DIR
to the same location for consistency.
Roman Lebedev [Wed, 29 May 2019 20:03:00 +0000 (20:03 +0000)]
UpdateTestChecks: Lanai triple support
Summary:
The assembly structure most resembles the SPARC pattern:
```
.globl f6 ! -- Begin function f6
.p2align 2
.type f6,@function
f6: ! @f6
.cfi_startproc
! %bb.0:
st %fp, [--%sp]
<...>
ld -8[%fp], %fp
.Lfunc_end0:
.size f6, .Lfunc_end0-f6
.cfi_endproc
! -- End function
```
Test being affected by upcoming patch, so regenerate it.
Tim Northover [Wed, 29 May 2019 19:12:48 +0000 (19:12 +0000)]
IR: add optional type to 'byval' function parameters
When we switch to opaque pointer types we will need some way to describe
how many bytes a 'byval' parameter should occupy on the stack. This adds
a (for now) optional extra type parameter.
If present, the type must match the pointee type of the argument.
Note to front-end maintainers: if this causes test failures, it's probably
because the "byval" attribute is printed after attributes without any parameter
after this change.
Chris Bieneman [Wed, 29 May 2019 18:37:49 +0000 (18:37 +0000)]
[CMake] [Runtimes] Set *_STANDALONE_BUILD
Summary:
The runtimes use `*_STANDALONE_BUILD=OFF` to signify that clang is an in-tree target. This is not the case with the runtime builds, so we really need this set to `ON`.
In order to resolve the issues phosek was having with checks, we should use checks that don't link. We can use compiler-rt's `try_compile_only` as a basis for that.
This patch is *required* to be able to run the runtime libraries check-* targets.
Stella Stamenova [Wed, 29 May 2019 18:07:39 +0000 (18:07 +0000)]
lit: modernize the lit configuration for the lit tests
Summary: This also normalizes the config feature that represents the windows platform to "system-windows" as opposed to having both "windows" and "system-windows"
Teresa Johnson [Wed, 29 May 2019 16:50:46 +0000 (16:50 +0000)]
[ThinLTO] Use original alias visibility when importing
Summary:
When we import an alias, we do so by making a clone of the aliasee. Just
as this clone uses the original alias name and linkage, it should also
use the same visibility (not the aliasee's visibility). Otherwise,
linker behavior is affected (e.g. if the aliasee was hidden, but the
alias is not, the resulting imported clone should not be hidden,
otherwise the linker will make the final symbol hidden which is
incorrect).
projects / llvm / log