add an intrinsic to expose the 'VSX Scalar Convert Half-Precision to
Single-Precision' instruction.

Differential review: https://reviews.llvm.org/D26536

llvm-svn: 286862

llvm-svn: 286861

Summary:
  Extend image intrinsics to support data types of V1F32 and V2F32.

  TODO: we should define a mapping table to change the opcode for data type of V2F32 but just one channel is active,
  even though such case should be very rare.

Reviewers:
  tstellarAMD

Differential Revision:
  http://reviews.llvm.org/D26472

llvm-svn: 286860

llvm-svn: 286855

Differential Revision: https://reviews.llvm.org/D26552

llvm-svn: 286853

https://reviews.llvm.org/D26516

llvm-svn: 286852

Darwin's backtrace() function does not work with sigaltstack (which was
enabled when available with r270395) — it does a sanity check to make
sure that the current frame pointer is within the expected stack area
(which it is not when using an alternate stack) and gives up otherwise.
The alternative of _Unwind_Backtrace seems to work fine on macOS, so use
that when backtrace() fails. Note that we then use backtrace_symbols_fd()
with the addresses from _Unwind_Backtrace, but I’ve tested that and it
also seems to work fine. rdar://problem/28646552

llvm-svn: 286851

llvm-svn: 286845

This restores the rest of r286297 (part was restored in r286475).
Specifically, it restores the part requiring adding a dependency from
the Analysis to Object library (downstream use changed to correctly
model split BitReader vs BitWriter libraries).

Original description of this part of patch follows:

Module level asm may also contain defs of values. We need to prevent
export of any refs to local values defined in module level asm (e.g. a
ref in normal IR), since that also requires renaming/promotion of the
local. To do that, the summary index builder looks at all values in the
module level asm string that are not marked Weak or Global, which is
exactly the set of locals that are defined. A summary is created for
each of these local defs and flagged as NoRename.

This required adding handling to the BitcodeWriter to look at GV
declarations to see if they have a summary (rather than skipping them
all).

Finally, added an assert to IRObjectFile::CollectAsmUndefinedRefs to
ensure that an MCAsmParser is available, otherwise the module asm parse
would silently fail. Initialized the asm parser in the opt tool for use
in testing this fix.

Fixes PR30610.

llvm-svn: 286844

The Stack slot coloring pass removes a store that is followed by a load
that deal with the same stack slot. The function isLoadFromStackSlot
is supposed to consider the loads that have no side-effects. This
patch fixed the issue by removing the unsafe loads from this function
Eg:
%vreg0<def> = L2_loadruh_io <fi#15>, 0
S2_storeri_io <fi#15>, 0, %vreg0

In this case, we load an unsigned extended half word and store this in to
the same stack slot. The Stack slot coloring pass considers safe to remove
the store. This patch marked all the non-vector byte and half word loads as
unsafe.

llvm-svn: 286843

The existing logic was to discard any symbols representing function template
instantiations, as the definitions were assumed to be inline. But there are
three explicit specializations of clang::Type::getAs that are only defined in
Clang's lib/AST/Type.cpp, and at least the plugin used by the LibreOffice build
(https://wiki.documentfoundation.org/Development/Clang_plugins) uses those
functions.

Differential Revision: https://reviews.llvm.org/D26455

llvm-svn: 286841

Summary:
The change in r285513 to prevent exporting of locals used in
inline asm added all locals in the llvm.used set to the reference
set of functions containing inline asm. Since these locals were marked
NoRename, this automatically prevented importing of the function.

Unfortunately, this caused an explosion in the summary reference lists
in some cases. In my particular example, it happened for a large protocol
buffer generated C++ file, where many of the generated functions
contained an inline asm call. It was exacerbated when doing a ThinLTO
PGO instrumentation build, where the PGO instrumentation included
thousands of private __profd_* values that were added to llvm.used.

We really only need to include a single llvm.used local (NoRename) value
in the reference list of a function containing inline asm to block it
being imported. However, it seems cleaner to add a flag to the summary
that explicitly describes this situation, which is what this patch does.

Reviewers: mehdi_amini

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D26402

llvm-svn: 286840

More realistic v16i8/v32i8/v64i8 MUL costs - we have to extend to vXi16, use PMULLW and then truncate the result

llvm-svn: 286838

Add explicit v16i16/v32i8 ADD/SUB costs, matching the costs of v4i64/v8i32 - they were missing for some reason.

This has side effects on the LV max bandwidth tests (AVX1 now prefers 128-bit vectors vs AVX2 which still prefers 256-bit)

llvm-svn: 286832

  Differential Revision: https://reviews.llvm.org/D26272

llvm-svn: 286829

Reverting r285406, which was a temporary workaround to get one of the documentation bots upgraded to something newer than GCC 4.7. This restores the check for GCC 4.8.

llvm-svn: 286822

Also,

Revert "test: remove the archive before modifying it"
Revert "test: explicitly use gnu format"

This reverts commits r286778, r286729 and r286767, as they are randomly failing
on many bots (AArch64, x86_64).

llvm-svn: 286820

We were already testing is the op was not a leaf, so need to then test if it was a leaf (added it to the assert instead).

llvm-svn: 286817

When calculating the cost of a call instruction we were applying a heuristic penalty as well as the cost of the instruction itself.

However, when calculating the benefit from inlining we weren't discounting the equivalent penalty for the call instruction that would be removed! This caused skew in the calculation and meant we wouldn't inline in the following, trivial case:

  int g() {
    h();
  }
  int f() {
    g();
  }

llvm-svn: 286814

'A || (!A && B)' is equivalent to 'A || B':

(LoopCycle > DefCycle) || (LoopCycle <= DefCycle && LoopStage <= DefStage)
-->
(LoopCycle > DefCycle) || (LoopStage <= DefStage)

llvm-svn: 286811

llvm-svn: 286808

Summary:
Unfolding selects was previously done with the help of a vector
of pointers that was then sorted to be able to remove duplicates.
As this sorting depends on the memory addresses, it was
non-deterministic. A SetVector is used now so that duplicates are
removed without the need of sorting first.

Reviewers: mgrang, efriedma

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D26450

llvm-svn: 286807

llvm-svn: 286806

Summary:
This patch adds explicit `(void)` casts to discarded `release()` calls to suppress -Wunused-result.

This patch fixes *all* warnings are generated as a result of [applying `[[nodiscard]]`  within libc++](https://reviews.llvm.org/D26596).
Similar fixes were applied to Clang in r286796.

Reviewers: chandlerc, dberris

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D26598

llvm-svn: 286797

Only attempt to demangle symbols which have the itanium C++ prefix of `_Z`.
This ensures that we do not treat any symbol name as a managled named.  We would
previously treat a C function `f` as a mangled name and decode that to `float`
incorrectly.

While it is easy to add tests for this, Mehdi recommended against introducing
tests for the demangler as libc++abi should cover the testing.

llvm-svn: 286795

[AVX-512] Add suffixless aliases for EVEX encoded vcvtsi2ss/vcvtsi2sd/vcvtusi2ss/vcvtusi2sd. This matches the VEX behavior.

Fixes another problem from PR28850.

llvm-svn: 286790

[X86] Cleanup 'x' and 'y' mnemonic suffixes for vcvtpd2dq/vcvttpd2dq/vcvtpd2ps and similar instructions.

-Don't print the 'x' suffix for the 128-bit reg/mem VEX encoded instructions in Intel syntax. This is consistent with the EVEX versions.
-Don't print the 'y' suffix for the 256-bit reg/reg VEX encoded instructions in Intel or AT&T syntax. This is consistent with the EVEX versions.
-Allow the 'x' and 'y' suffixes to be used for the reg/mem forms when we're assembling using Intel syntax.
-Allow the 'x' and 'y' suffixes on the reg/reg EVEX encoded instructions in Intel or AT&T syntax. This is consistent with what VEX was already allowing.

This should fix at least some of PR28850.

llvm-svn: 286787

[AVX-512] Remove and autoupgrade masked dword/qword variable shift intrinsics to the new unmasked versions and selects.

llvm-svn: 286786

`shl nsw i8 1, i8 8` is poison, but `mul i8 1, i8 128` is not.

This was discussed previously here:
http://lists.llvm.org/pipermail/llvm-dev/2015-April/084195.html.  From
the discussion, it was not clear which semantics we want for `shl`, but
for now at least make the language reference more accurate.

llvm-svn: 286785

The archive may already exist when not doing a clean test run.  The dirty state
can cause a test failure.  Remove the archive first.

llvm-svn: 286778

`c++filt` when given no arguments runs as a REPL, decoding each line as a
decorated name.  Unify the test structure to be more uniform, with the tests for
llvm-cxxfilt living under test/tools/llvm-cxxfilt.

llvm-svn: 286777

Similar to:
https://reviews.llvm.org/rL285499
https://reviews.llvm.org/rL286318

We can't minimally expose this in IR tests because we don't have min/max intrinsics,
but the difference is visible in codegen because SelectionDAGBuilder::visitSelect() 
uses matchSelectPattern().

We're not canonicalizing these patterns in IR (yet), so I don't expect there to be any
regressions as noted here:
http://lists.llvm.org/pipermail/llvm-dev/2016-November/106868.html

llvm-svn: 286776

[AVX-512] Fix a disassembler failure for AVX-512 vcmpss/vcmpsd with an immediate larger than 32. Fix the same bug with VLX vcmpps/vcmppd.

Fixes PR24941.

llvm-svn: 286775

llvm-strings was added to the test dependencies without updating the lit
substitutions.  Synchronise the list.

llvm-svn: 286773

llvm-svn: 286772

[X86][IR] Reduce the number of full string comparisons in the code that autoupgrades masked shift intrinsics.

llvm-svn: 286768

This adds support for the `-f` or `--print-file-name` option for strings.

llvm-svn: 286767

nThis avoids the nasty problems caused by using
memory instructions that read the exec mask while
spilling / restoring registers used for control flow
masking, but only for VI when these were added.

This always uses the scalar stores when enabled currently,
but it may be better to still try to spill to a VGPR
and use this on the fallback memory path.

The cache also needs to be flushed before wave termination
if a scalar store is used.

llvm-svn: 286766

llvm-svn: 286765

llvm-svn: 286764