Remove non-consecutive stores from store merge candidate search as
they cannot be merged and will prevent us from finding subsequent
mergeable store cases.
Reviewers: jyknight, bogner, javed.absar, spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D32086
llvm-svn: 300561
This patch is part of D28975's breakdown.
Add caching for block masks similar to the cache already used for edge masks,
replacing generation per user with reusing the first generated value which
dominates all uses.
Differential Revision: https://reviews.llvm.org/D32054
llvm-svn: 300557
This is a recommit of r300539 that was reverted in r300543 due to test failures.
The original commit message is displayed below:
The new '#pragma clang attribute' directive can be used to apply attributes to
multiple declarations. An attribute must satisfy the following conditions to
be supported by the pragma:
- It must have a subject list that's defined in the TableGen file.
- It must be documented.
- It must not be late parsed.
- It must have a GNU/C++11 spelling.
Differential Revision: https://reviews.llvm.org/D30009
llvm-svn: 300556
Summary:
* Add checks for store. That is needed because GlobalsAA is called
twice in the current pipeline with different sets of Function passes
following it. However, the loads are eliminated using instcombine
which happens everywhere. On the other hand, DeadStoreElimination is
performed only once so by checking for store we'll be able to catch
more cases when GlobalsAA is invalidated unintentionally.
* Add empty function above/below the test so that we don't depend on
the relative order of instcombine/dead-store-elimination and the
pass that invalidates the analysis (inside the same
FunctionPassManager).
Reviewers: kristof.beyls
Reviewed By: kristof.beyls
Subscribers: llvm-commits, n.bozhenov
Differential Revision: https://reviews.llvm.org/D32015
Patch by Andrei Elovikov <andrei.elovikov@intel.com>
llvm-svn: 300553
This passes an option to the ARM assembly parser to emit build
attributes for the hardware selected by command line options, when
assembling an assembly file.
This is not enabled for C/C++, as this would result in duplicate build
attribute directives being emitted in each inline assembly block, when
emitting assembly.
This also adds an option to allow disabling this behaviour for assembly
files, for users who were relying on the old behaviour.
Differential revision: https://reviews.llvm.org/D31813
llvm-svn: 300550
In the assembler, we should emit build attributes based on the target
selected with command-line options. This matches the GNU assembler's
behaviour. We only do this for build attributes which describe the
hardware that is expected to be available, not the ones that describe
ABI compatibility.
This is done by moving some of the attribute emission code to
ARMTargetStreamer, so that it can be shared between the assembly and
code-generation code paths. Since the assembler only creates a
MCSubtargetInfo, not an ARMSubtarget, the code had to be changed to
check raw features, and not use the convenience functions in
ARMSubtarget.
If different attributes are later specified using the .eabi_attribute
directive, then they will take precedence, as happens when the same
.eabi_attribute is specified twice.
This must be enabled by an option, because we don't want to do this when
parsing inline assembly. The attributes would match the ones emitted at
the start of the file, so wouldn't actually change the emitted object
file, but the extra directives would be added to every inline assembly
block when emitting assembly, which we'd like to avoid.
The majority of the changes in the build-attributes.ll test are just
re-ordering the directives, because the hardware attributes are now
emitted before the ABI ones. However, I did fix one bug which I spotted:
Tag_CPU_arch_profile was not being emitted for v6M.
Differential revision: https://reviews.llvm.org/D31812
llvm-svn: 300547
Before this patch, we always called method 'findCalleeFunctionSamples()' on
intrinsic calls. However, intrinsic calls like llvm.dbg.value() are not viable
candidates for obvious reasons.
No functional change intended.
Differential Revision: https://reviews.llvm.org/D32008
llvm-svn: 300541
The new '#pragma clang attribute' directive can be used to apply attributes to
multiple declarations. An attribute must satisfy the following conditions to
be supported by the pragma:
- It must have a subject list that's defined in the TableGen file.
- It must be documented.
- It must not be late parsed.
- It must have a GNU/C++11 spelling.
Differential Revision: https://reviews.llvm.org/D30009
llvm-svn: 300539
This reverts r300535 and r300537.
The newly added tests in test/CodeGen/AArch64/GlobalISel/arm64-fallback.ll
produces slightly different code between LLVM versions being built with different compilers.
E.g., dependent on the compiler LLVM is built with, either one of the following
can be produced:
remark: <unknown>:0:0: unable to legalize instruction: %vreg0<def>(p0) = G_EXTRACT_VECTOR_ELT %vreg1, %vreg2; (in function: vector_of_pointers_extractelement)
remark: <unknown>:0:0: unable to legalize instruction: %vreg2<def>(p0) = G_EXTRACT_VECTOR_ELT %vreg1, %vreg0; (in function: vector_of_pointers_extractelement)
Non-determinism like this is clearly a bad thing, so reverting this until
I can find and fix the root cause of the non-determinism.
llvm-svn: 300538
For subtargets that use the custom lowering for divmod, e.g. gnueabi,
we used to check if the subtarget has hardware divide and then lower to
a div-mul-sub sequence if true, or to a libcall if false.
However, judging by the usage of hasDivide vs hasDivideInARMMode, it
seems that hasDivide only refers to Thumb. For instance, in the
ARMTargetLowering constructor, the code that specifies whether to use
libcalls for (S|U)DIV looks like this:
bool hasDivide = Subtarget->isThumb() ? Subtarget->hasDivide()
: Subtarget->hasDivideInARMMode();
In the case of divmod for arm-gnueabi, using only hasDivide() to
determine what to do means that instead of lowering to __aeabi_idivmod
to get the remainder, we lower to div-mul-sub and then further lower the
div to __aeabi_idiv. Even worse, if we have hardware divide in ARM but
not in Thumb, we generate a libcall instead of using it (this is not an
issue in practice since AFAICT none of the cores that we support have
hardware divide in ARM but not Thumb).
This patch fixes the code dealing with custom lowering to take into
account the mode (Thumb or ARM) when deciding whether or not hardware
division is available.
Differential Revision: https://reviews.llvm.org/D32005
llvm-svn: 300536
This fixes PR32471.
As comment 10 on that bug report highlights
(https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a
few different defendable design tradeoffs that could be made, including
not representing pointers at all in LLT.
I decided to go for representing vector-of-pointer as a concept in LLT,
while keeping the size of the LLT type 64 bits (this is an increase from
48 bits before). My rationale for keeping pointers explicit is that on
some targets probably it's very handy to have the distinction between
pointer and non-pointer (e.g. 68K has a different register bank for
pointers IIRC). If we keep a scalar pointer, it probably is easiest to
also have a vector-of-pointers to keep LLT relatively conceptually clean
and orthogonal, while we don't have a very strong reason to break that
orthogonality. Once we gain more experience on the use of LLT, we can
of course reconsider this direction.
Rejecting vector-of-pointer types in the IRTranslator is also an option
to avoid the crash reported in PR32471, but that is only a very
short-term solution; also needs quite a bit of code tweaks in places,
and is probably fragile. Therefore I didn't consider this the best
option.
llvm-svn: 300535
Summary:
The "performance-inefficient-vector-operation" check finds vector oprations in
for-loop statements which may cause multiple memory reallocations.
This is the first version, only detects typical for-loop:
```
std::vector<int> v;
for (int i = 0; i < n; ++i) {
v.push_back(i);
}
// or
for (int i = 0; i < v2.size(); ++i) {
v.push_back(v2[i]);
}
```
We can extend it to handle more cases like for-range loop in the future.
Reviewers: alexfh, aaron.ballman
Reviewed By: aaron.ballman
Subscribers: zaks.anna, Eugene.Zelenko, mgorny, cfe-commits, djasper
Differential Revision: https://reviews.llvm.org/D31757
llvm-svn: 300534
Turned out that adding defined(_arm_) in sanitizer_stoptheworld_linux_libcdep.cc breaks android arm with some toolchains.
.../llvm/projects/compiler-rt/lib/sanitizer_common/sanitizer_stoptheworld_linux_libcdep.cc:36:11: fatal error:
'linux/user.h' file not found
# include <linux/user.h> // for pt_regs
^
1 error generated.
Context:
#if SANITIZER_ANDROID && defined(__arm__)
# include <linux/user.h> // for pt_regs
#else
This patch removes corresponding #if SANITIZER_ANDROID && defined(__arm__) and a bit rearranges adjacent сode.
Differential Revision: https://reviews.llvm.org/D32128
llvm-svn: 300531
A problem arises if a machine supports the rdtscp instruction, but the processor
frequency cannot be determined by the function getTSCFrequency(). In this case,
we want to use the emulated TSC instead. This patch implements that by adding a
call to getTSCFrequency() from probeRequiredCPUFeatures(), and the function only
returns true if both the processor supports rdtscp and the CPU frequency can be
determined.
This should fix PR32620.
Reviewers: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D32067
llvm-svn: 300525
Summary:
This patch implements [dcl.fct.def.coroutine]p8:
> The unqualified-id get_return_object_on_allocation_failure is looked up in the scope of
> class P by class member access lookup (3.4.5). If a declaration is found, ..., and if a
> global allocation function is selected, the ::operator new(size_t, nothrow_t) form shall be used.
> [...]
> The allocation function used in this case must have a non-throwing noexcept-specification.
Reviewers: GorNishanov, rsmith, majnemer, aaron.ballman
Reviewed By: GorNishanov
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D31562
llvm-svn: 300524
LLVM has changed the semantics of dbg.declare for describing function
arguments. After this patch a dbg.declare always takes the *address*
of a variable as the first argument, even if the argument is not an
alloca.
https://bugs.llvm.org/show_bug.cgi?id=32382
rdar://problem/31205000
llvm-svn: 300523
The DWARF specification knows 3 kinds of non-empty simple location
descriptions:
1. Register location descriptions
- describe a variable in a register
- consist of only a DW_OP_reg
2. Memory location descriptions
- describe the address of a variable
3. Implicit location descriptions
- describe the value of a variable
- end with DW_OP_stack_value & friends
The existing DwarfExpression code is pretty much ignorant of these
restrictions. This used to not matter because we only emitted very
short expressions that we happened to get right by accident. This
patch makes DwarfExpression aware of the rules defined by the DWARF
standard and now chooses the right kind of location description for
each expression being emitted.
This would have been an NFC commit (for the existing testsuite) if not
for the way that clang describes captured block variables. Based on
how the previous code in LLVM emitted locations, DW_OP_deref
operations that should have come at the end of the expression are put
at its beginning. Fixing this means changing the semantics of
DIExpression, so this patch bumps the version number of DIExpression
and implements a bitcode upgrade.
There are two major changes in this patch:
I had to fix the semantics of dbg.declare for describing function
arguments. After this patch a dbg.declare always takes the *address*
of a variable as the first argument, even if the argument is not an
alloca.
When lowering a DBG_VALUE, the decision of whether to emit a register
location description or a memory location description depends on the
MachineLocation — register machine locations may get promoted to
memory locations based on their DIExpression. (Future) optimization
passes that want to salvage implicit debug location for variables may
do so by appending a DW_OP_stack_value. For example:
DBG_VALUE, [RBP-8] --> DW_OP_fbreg -8
DBG_VALUE, RAX --> DW_OP_reg0 +0
DBG_VALUE, RAX, DIExpression(DW_OP_deref) --> DW_OP_reg0 +0
All testcases that were modified were regenerated from clang. I also
added source-based testcases for each of these to the debuginfo-tests
repository over the last week to make sure that no synchronized bugs
slip in. The debuginfo-tests compile from source and run the debugger.
https://bugs.llvm.org/show_bug.cgi?id=32382
<rdar://problem/31205000>
Differential Revision: https://reviews.llvm.org/D31439
llvm-svn: 300522
These tests were unconditionally asserting that optional and unique_ptr declare throwing hashes, but MSVC++ implements conditional noexcept forwarding that of the underlying hash function. As a result we were failing these tests but there's nothing forbidding strengthening noexcept in that way.
Changed the ASSERT_NOT_NOEXCEPT asserts to use types which themselves have non-noexcept hash functions.
llvm-svn: 300516
Previously, if an escaped newline was followed by a newline or a nul, we'd lex
the escaped newline as a bogus space character. This led to a bunch of
different broken corner cases:
For the pattern "\\\n\0#", we would then have a (horizontal) space whose
spelling ends in a newline, and would decide that the '#' is at the start of a
line, and incorrectly start preprocessing a directive in the middle of a
logical source line. If we were already in the middle of a directive, this
would result in our attempting to process multiple directives at the same time!
This resulted in crashes, asserts, and hangs on invalid input, as discovered by
fuzz-testing.
For the pattern "\\\n" at EOF (with an implicit following nul byte), we would
produce a bogus trailing space character with spelling "\\\n". This was mostly
harmless, but would lead to clang-format getting confused and misformatting in
rare cases. We now produce a trailing EOF token with spelling "\\\n",
consistent with our handling for other similar cases -- an escaped newline is
always part of the token containing the next character, if any.
For the pattern "\\\n\n", this was somewhat more benign, but would produce an
extraneous whitespace token to clients who care about preserving whitespace.
However, it turns out that our lexing for line comments was relying on this bug
due to an off-by-one error in its computation of the end of the comment, on the
slow path where the comment might contain escaped newlines.
llvm-svn: 300515
Instead of storing an UncommonIndex on the Symbol, use a flag bit to store
whether the Symbol has an Uncommon. This shrinks Chromium's .bc files (after
D32061) by about 1%.
Differential Revision: https://reviews.llvm.org/D32070
llvm-svn: 300514
Haojian Wu