This avoids creating multiple thunks for symbols with aliases or which
belong to ICF'd sections. This patch reduces the size of Chromium for
Android by 260KB (0.8% of .text).
Differential Revision: https://reviews.llvm.org/D44284
llvm-svn: 327154
Summary: Starting from GCN 2nd generation, ISA supports ds_read_b128 on top of ds_read_b64.
This patch supports ds_read_b128 instruction pattern and generation of this instruction.
In the vectorizer, this patch also widen the vector length so that vectorizer generates
128 bit loads for local address-space which gets translated to ds_read_b128.
Since the performance benefit is not clear; compiler generates ds_read_b128 under -amdgpu-ds128.
Author: FarhanaAleen
Reviewed By: rampitec, arsenm
Subscribers: llvm-commits, AMDGPU
Differential Revision: https://reviews.llvm.org/D44210
llvm-svn: 327153
Added helpers to build G_FCONSTANT, along with matching ConstantFP and
unit tests for the same.
Sample usage.
auto MIB = Builder.buildFConstant(s32, 0.5); // Build IEEESingle
For Matching the above
const ConstantFP* Tmp;
mi_match(DstReg, MRI, m_GFCst(Tmp));
https://reviews.llvm.org/D44128
reviewed by: volkan
llvm-svn: 327152
This error case is described in Linking.md. The operand for call requires
generation of a synthetic stub.
Differential Revision: https://reviews.llvm.org/D44028
llvm-svn: 327151
In r263618, JumpThreading learned to look trough simple cast instructions, but
only if the source of those cast instructions was a phi/cmp i1 (in an effort to
limit compile time effects). I think this condition is too restrictive. For
switches with limited value range, InstCombine will readily introduce an extra
trunc instruction to a smaller integer type (e.g. from i8 to i2), leaving us in
the somewhat perverse situation that jump-threading would work before running
instcombine, but not after. Since instcombine produces this pattern, I think we
need to consider it canonical and support it in JumpThreading. In general,
for limiting recursion, I think the existing restriction to phi and cmp nodes
should be sufficient to avoid looking through unprofitable chains of
instructions.
Patch by Keno Fischer!
Differential Revision: https://reviews.llvm.org/D42262
llvm-svn: 327150
Previously we created __wasm_call_ctors with null InputFunction, and
added the InputFunction later. Now we create the SyntheticFunction with
null body, and set the body later.
Differential Revision: https://reviews.llvm.org/D44206
llvm-svn: 327149
Summary:
Both `SizeClassInfo` structures for the 32-bit primary & `RegionInfo`
structures for the 64-bit primary can be used by different threads, and as such
they should be aligned & padded to the cacheline size to avoid false sharing.
The former was padded but the array was not aligned, the latter was not padded
but we lucked up as the size of the structure was 192 bytes, and aligned by
the properties of `mmap`.
I plan on adding a couple of fields to the `RegionInfo`, and some highly
threaded tests pointed out that without proper padding & alignment, performance
was getting a hit - and it is going away with proper padding.
This patch makes sure that we are properly padded & aligned for both. I used
a template to avoid padding if the size is already a multiple of the cacheline
size. There might be a better way to do this, I am open to suggestions.
Reviewers: alekseyshl, dvyukov
Reviewed By: alekseyshl
Subscribers: kubamracek, delcypher, #sanitizers, llvm-commits
Differential Revision: https://reviews.llvm.org/D44261
llvm-svn: 327145
Also, fix the undef vs. UB example to use 'sdiv' because that can trigger div-by-zero UB.
The existing text for the constrained intrinsics says:
"By default, LLVM optimization passes assume that the rounding mode is round-to-nearest
and that floating point exceptions will not be monitored. Constrained FP intrinsics are
used to support non-default rounding modes and accurately preserve exception behavior
without compromising LLVM’s ability to optimize FP code when the default behavior is
used."
...so the additional text with the normal FP opcodes should make the different modes
clear.
Differential Revision: https://reviews.llvm.org/D44216
llvm-svn: 327138
We improved the handling of errors and warnings in dwarfdump's verifier
in rL314498. This patch does the same thing for dsymutil.
Differential revision: https://reviews.llvm.org/D44052
llvm-svn: 327137
If initialization of the task reductions requires pointer to original
variable, which is stored in the threadprivate storage, we used the
address of this pointer instead.
llvm-svn: 327136
lib/WindowsManifest/CMakeLists.txt adds it to LLVM_SYSTEM_LIBS on that
target, but it was never getting picked up in
tools/llvm-config/CMakeLists.txt.
Differential Revision: https://reviews.llvm.org/D44302
llvm-svn: 327135
Summary:
The relevant failing assertion message is:
../tools/clang/lib/Sema/SemaInit.cpp:8411: PerformCopyInitialization(): Assertion `InitE && "No initialization expression?"' failed.
See the added test case for a repro.
Reviewers: bkramer, sammccall, ioeric, hokein
Reviewed By: sammccall
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D44300
llvm-svn: 327134
The test I added in r327110 is failing on windows because of "import
pexpect". However, this import is no longer necessary as these tests
don't use pexpect anymore.
In fact, it seems that all TestCompletion tests are passing on windows
after this, so I enable all of them.
llvm-svn: 327133
The code to match and produce more x86 vector blends was enabled for all
architectures even though the transform may pessimize the code for other
architectures that do not provide a vector blend instruction.
Added an aarch64 testcase to check that a VZIP instruction is generated instead
of byte movs.
Differential Revision: https://reviews.llvm.org/D44118
llvm-svn: 327132
Summary: This also matches the range in symbol index.
Reviewers: sammccall
Subscribers: klimek, ilya-biryukov, jkorous-apple, ioeric, cfe-commits
Differential Revision: https://reviews.llvm.org/D44247
llvm-svn: 327129
Summary:
These have different USRs than the underlying entity, but are not typically
interesting in their own right and can be numerous (e.g. generated traits).
Reviewers: ioeric
Subscribers: klimek, ilya-biryukov, jkorous-apple, cfe-commits
Differential Revision: https://reviews.llvm.org/D44298
llvm-svn: 327127
In what appears to be a copy-and-paste error, lld currently only
installs libraries if the lld tools are configured to build.
Instead, lld should allow the libraries to be installed even if the lld
tools are not being built. Additionally, if users want to only install
the tools and not the libraries, the LLVM way of doing that is by
checking for LLVM_INSTALL_TOOLCHAIN_ONLY.
This fixes PR35960.
llvm-svn: 327126
Summary:
Besides being superfluous, this double merging was actually wrong and
causing some sections to be added twice. The reason for that was that
the code assumes section IDs are unique in the section list, but this is
only true if all sections in the list come from the same object file.
Reviewers: fjricci, jankratochvil
Subscribers: emaste, lldb-commits, arichardson
Differential Revision: https://reviews.llvm.org/D44157
llvm-svn: 327123
Allows capturing a list of concrete instantiated defs.
This can be combined with foreach to create parallel sets of def
instantiations with less repetition in the source. This purpose is
largely also served by multiclasses, but in some cases multiclasses
can't be used.
The motivating example for this change is having a large set of
intrinsics, which are generated from the IntrinsicsBackend.td file
included by Intrinsics.td, and a corresponding set of instruction
selection patterns, which are generated via the backend's .td files.
Multiclasses cannot be used to eliminate the redundancy in this case,
because a multiclass cannot span both LLVM's common .td files and
the backend .td files at the same time.
Change-Id: I879e35042dceea542a5e6776fad23c5e0e69e76b
Differential revision: https://reviews.llvm.org/D44109
llvm-svn: 327121
The changes to FieldInit are required to make field references (Def.field)
work inside a ForeachDeclaration: previously, Def.field wasn't resolved
immediately when Def was already a fully resolved DefInit.
Change-Id: I9875baec2fc5aac8c2b249e45b9cf18c65ae699b
llvm-svn: 327120
Use the default ParseValueMode instead of ParseForeachMode when
parsing the rule
ForeachDeclaration ::= ID '=' '[' ValueList ']'
because the only difference between the two is how an open brace '{'
is handled at the end. In the context of foreach, the 'in' keyword
will appear after the ForeachDeclaration, so this special handling
of '{' is not required.
Change-Id: I4d86bb73bab9ec26752e1273e5213df77cf28d1d
llvm-svn: 327119
Summary:
Even though the getDIEOffset offset function was common for the two
accelerator table implementations, it was doing two different things:
for the Apple tables, it was returning the die offset relative to the
start of the section, whereas for DWARF v5 tables, it was relative to
the start of the CU.
I resolve this by renaming the function to getDIESectionOffset to make
it obvious what the function returns, and change the DWARF
implementation to return the section offset. I also keep the CU-relative
accessor, but only in the DWARF implementation (there is no way to get
this information for the Apple tables). This was not caught by existing
tests because the hand-written inputs also erroneously used section
offsets instead of CU-relative ones.
While looking at this, I noticed that the Apple implementation was not
fully correct either -- the header contains a DIEOffsetBase field, which
should be added to offsets encoded with the DW_FORM_ref*** family, but
this was not being used. This went unnoticed because all current writers
set this field to zero anyway. I fix this as well and add a hand-written
test which demonstrates the issue.
Reviewers: JDevlieghere, dblaikie
Subscribers: aprantl, llvm-commits
Differential Revision: https://reviews.llvm.org/D44202
llvm-svn: 327116
Summary:
The args class is used in plenty of places (a lot of them in the lower lldb
layers) for representing a list of arguments, and most of these places don't
care about option parsing. Moving the option parsing out of the class removes
the largest external dependency (there are a couple more, but these are in
static functions), and brings us closer to being able to move it to the
Utility module).
The new home for these functions is the Options class, which was already used
as an argument to the parse calls, so this just inverts the dependency between
the two.
The functions are themselves are mainly just copied -- the biggest functional
change I've made to them is to avoid modifying the input Args argument (getopt
likes to permute the argument vector), as it was weird to have another class
reorder the entries in Args class. So now the functions don't modify the input
arguments, and (for those where it makes sense) return a new Args vector
instead. I've also made the addition of a "fake arg0" (required for getopt
compatibility) an implementation detail rather than a part of interface.
While doing that I noticed that ParseForCompletion function was recording the
option indexes in the shuffled vector, but then the consumer was looking up the
entries in the unshuffled one. This manifested itself as us not being able to
complete "watchpoint set variable foo --" (because getopt would move "foo" to
the end). Surprisingly all other completions (e.g. "watchpoint set variable foo
--w") were not affected by this. However, I couldn't find a comprehensive test
for command argument completion, so I consolidated the existing tests and added
a bunch of new ones.
Reviewers: davide, jingham, zturner
Subscribers: lldb-commits
Differential Revision: https://reviews.llvm.org/D43837
llvm-svn: 327110
Fixes PR36311.
See more detailed analysis in
https://bugs.llvm.org/show_bug.cgi?id=36311.
isUniform() information is recomputed after LV started transforming the
underlying IR and that triggered an assert in SCEV.
From vectorizer's architectural perspective, such information, while
still useful in vector code gen, should not be recomputed after the
start of transforming the LLVM IR. Instead, we should collect and cache
such information during the analysis phase of LV and use the cached info
during code gen.
From the symptom perspective, this assert as it stands right now is not
very useful. Legality already rejected loops that would trigger the
assert. As such, commenting out the assert is NFC from vectorizer's
functionality perspective. On top of that, just above the assertion, we
check for unit-strided load/store or
gather scatter. Addresses can't be uniform below that check.
From vectorization theory point of view, we don't have to reject all
cases of stores to uniform addresses. Eventually, we should support
safe/profitable cases.
This patch resolves the issue by removing the useless assertion that is
invoking LAA's isUniform() that requires up-to-date DomTree ---- once
vector code gen starts modifying CFG, we don't have an up-to-date
DomTree.
Patch by Hideki Saito <hideki.saito@intel.com>.
llvm-svn: 327109
Move the DWARF syntax highlighting into support. This has several
advantages, most notably that this makes the WithColor RAII wrapper
available outside libDebugInfo. Furthermore, several projects all have
their own code for handling colored output. This provides a place to
centralize it.
Differential revision: https://reviews.llvm.org/D44215
llvm-svn: 327108
Simon Pilgrim