Turn LLVM_ENABLE_ABI_BREAKING_CHECKS into a 0/1 definition like
LLVM_ENABLE_THREADS. Include llvm-config.h explicitly in headers to make
sure that the definition is available.
Gor Nishanov [Fri, 30 Sep 2016 19:24:19 +0000 (19:24 +0000)]
[Coroutines] Part15c: Fix coro-split to correctly handle definitions between coro.save and coro.suspend
Summary:
In the case below, %Result.i19 is defined between coro.save and coro.suspend and used after coro.suspend. We need to correctly place such a value into the coroutine frame.
Gor Nishanov [Fri, 30 Sep 2016 19:05:06 +0000 (19:05 +0000)]
[Coroutines] Part15b: Fix dbg information handling in coro-split.
Summary:
Without the fix, if there was a function inlined into the coroutine with debug information, CloneFunctionInto(NewF, &F, VMap, /*ModuleLevelChanges=*/true, Returns); would duplicate all of the debug information including the DICompileUnit.
We know use VMap to indicate that debug metadata for a File, Unit and FunctionType should not be duplicated when we creating clones that will become f.resume, f.destroy and f.cleanup.
Gor Nishanov [Fri, 30 Sep 2016 18:41:35 +0000 (18:41 +0000)]
[Coroutines] Part 15a: Lower coro.subfn.addr in CoroCleanup
Summary: Not all coro.subfn.addr intrinsics can be eliminated in CoroElide through devirtualization. Those that remain need to be lowered in CoroCleanup.
Michal Gorny [Fri, 30 Sep 2016 18:34:23 +0000 (18:34 +0000)]
cmake: Install the OCaml libraries into a more correct path
Add a OCAML_INSTALL_PATH variable that can be used to control
the install path for OCaml libraries. The new variable defaults to
${OCAML_STDLIB_PATH}, i.e. the OCaml library path obtained from
the OCaml compiler. Install libraries into "llvm" subdirectory.
This fixes two issues:
1. OCaml library directories differ between systems, and 'lib/ocaml' is
incorrect e.g. on amd64 Gentoo where OCaml is installed
in 'lib64/ocaml'. Therefore, obtain the library path from the OCaml
compiler using 'ocamlc -where' (which is already used to set
OCAML_STDLIB_PATH), which is the method used commonly in OCaml packages.
2. The top-level directory is reserved for the standard library, and has
precedence over local directory in search path. As a result, OCaml
preferred the files installed along with previous LLVM version over the
source tree when building a new version, resulting in two versions being
mixed during the build. The new layout is used commonly by other OCaml
packages, and findlib is able to find the LLVM libraries successfully.
[WebAssembly] Make register stackification more conservative
Register stackification currently checks VNInfo for changes. Make that
more accurate by testing each intervening instruction for any other defs
to the same virtual register.
Matthew Simpson [Fri, 30 Sep 2016 15:13:52 +0000 (15:13 +0000)]
[LV] Build all scalar steps for non-uniform induction variables
When building the steps for scalar induction variables, we previously attempted
to determine if all the scalar users of the induction variable were uniform. If
they were, we would only emit the step corresponding to vector lane zero. This
optimization was too aggressive. We generally don't know the entire set of
induction variable users that will be scalar. We have
isScalarAfterVectorization, but this is only a conservative estimate of the
instructions that will be scalarized. Thus, an induction variable may have
scalar users that aren't already known to be scalar. To avoid emitting unused
steps, we can only check that the induction variable is uniform. This should
fix PR30542.
Summary:
This change adds the AVR assembly instruction printer.
No tests are included in this patch. I have left them downstream so we can
add them once `llc` successfully runs (there's very few components left
to upstream until this).
[RegAllocGreedy] Attempt to split unspillable live intervals
Summary:
Previously, when allocating unspillable live ranges, we would never
attempt to split. We would always bail out and try last ditch graph
recoloring.
This patch changes this by attempting to split all live intervals before
performing recoloring.
This fixes LLVM bug PR14879.
I can't add test cases for any backends other than AVR because none of
them have small enough register classes to trigger the bug.
Petr Hosek [Fri, 30 Sep 2016 06:29:28 +0000 (06:29 +0000)]
[CMake] Support symlinks even with LLVM_INSTALL_TOOLCHAIN_ONLY
When LLVM_INSTALL_TOOLCHAIN_ONLY is used and LLVM_TOOLCHAIN_TOOLS
contains a tool which is a symlink, it would be ignored. This already
worked before but got broken in r282510.
[X86] Add AVX-512 VTs to findRepresentativeClass as well as v16i16 which was also missing. Change register class to include the extra 16 AVX512 registers.
I'm not completely sure what this method does or why all the 256-bit VTs returned VR128RegClass when the comments on the method definiton say it should return the largest super register class. I just figured AVX-512 should be similar.
[AVX-512] Always use the full 32 register vector classes for addRegisterClass regardless of whether AVX512/VLX is enabled or not.
If AVX512 is disabled, the registers should already be marked reserved. Pattern predicates and register classes on instructions should take care of most of the rest. Loads/stores and physical register copies for XMM16-31 and YMM16-31 without VLX have already been taken care of.
I'm a little unclear why this changed the register allocation of the SSE2 run of the sad.ll test, but the registers selected appear to be valid after this change.
Matt Arsenault [Fri, 30 Sep 2016 01:50:20 +0000 (01:50 +0000)]
AMDGPU: Use unsigned compare for eq/ne
For some reason there are both of these available, except
for scalar 64-bit compares which only has u64. I'm not sure
why there are both (I'm guessing it's for the one bit inputs we
don't use), but for consistency always using the
unsigned one.
[AArch64][RegisterBankInfo] Use static mapping for 3-operands instrs.
This uses a TableGen'ed like structure for all 3-operands instrs.
The output of the RegBankSelect pass should be identical but the
RegisterBankInfo will do less dynamic allocations.
enumerate allows you to iterate over a range by pairing the
iterator's value with its index in the enumeration. This gives
you most of the benefits of using a for loop while still allowing
the range syntax.
Also, make foldSelectExtConst() a member of InstCombiner, remove
unnecessary parameters from its interface, and group visitSelectInst
helpers together in the header file.
Kevin Enderby [Thu, 29 Sep 2016 21:07:29 +0000 (21:07 +0000)]
Next set of additional error checks for invalid Mach-O files for the
load command that uses the MachO::entry_point_command type
but not used in llvm libObject code but used in llvm tool code.
[RegisterBankInfo] Change the default mapping for Copy and PHI.
Instead of producing a mapping for all the operands, we only generate a
mapping for the definition. Indeed, the other operands are not
constrained by the instruction and thus, we should leave the choice to
the actual definition to do the right thing.
In pratice this is almost NFC, but with one advantage. We will have only
one instance of OperandsMapping for each copy and phi that map to one
register bank instead of one different instance for each different
number of operands for each copy and phi.
[codeview] Use character types for all byte-sized integer types
The VS debugger doesn't appear to understand the 0x68 or 0x69 type
indices, which were probably intended for use on a platform where a C
'int' is 8 bits. So, use the character types instead. Clang was already
using the character types because '[u]int8_t' is usually defined in
terms of 'char'.
See the Rust issue for screenshots of what VS does:
https://github.com/rust-lang/rust/issues/36646
Kevin Enderby [Thu, 29 Sep 2016 17:45:23 +0000 (17:45 +0000)]
Next set of additional error checks for invalid Mach-O files for the
load command that uses the Mach::source_version_command type
but not used in llvm libObject code but used in llvm tool code.
This includes just the LC_SOURCE_VERSION load command.
[sanitizer-coverage/libFuzzer] make the guards for trace-pc 32-bit; create one array of guards per function, instead of one guard per BB. reorganize the code so that trace-pc-guard does not create unneeded globals
Adam Nemet [Thu, 29 Sep 2016 16:49:42 +0000 (16:49 +0000)]
[LV] Convert all but one opt remark in Legality to new streaming interface
The last one remaining after which emitAnalysis can be removed is when
we convert the LAA's report to a vectorization report. This requires
converting LAA to the new interface first.
projects / llvm / log