4 years ago[X86] Cleanup ADCX/ADOX instruction definitions.
Craig Topper [Sun, 1 Apr 2018 23:58:50 +0000 (23:58 +0000)]
[X86] Cleanup ADCX/ADOX instruction definitions.

Give them both the same itineraries. Add hasSideEffects = 0 to ADOX since they don't have patterns. Rename source operands to $src1 and $src2 instead of $src0 and $src. Add ReadAfterLd to the memory form SchedRW.

llvm-svn: 328952

4 years ago[Coroutines] Schedule coro-split before asan
Brian Gesiak [Sun, 1 Apr 2018 23:55:21 +0000 (23:55 +0000)]
[Coroutines] Schedule coro-split before asan

The docs for the LLVM coroutines intrinsic `@llvm.coro.id` state that
"The second argument, if not null, designates a particular alloca instruction
to be a coroutine promise."

However, if the address sanitizer pass is run before the `@llvm.coro.id`
intrinsic is lowered, the `alloca` instruction passed to the intrinsic as its
second argument is converted, as per the
https://github.com/google/sanitizers/wiki/AddressSanitizerAlgorithm docs, to
an `inttoptr` instruction that accesses the address of the promise.

On optimization levels `-O1` and above, the `-asan` pass is run after
`-coro-early`, `-coro-split`, and `-coro-elide`, and before
`-coro-cleanup`, and so there is no issue. At `-O0`, however, `-asan`
is run in between `-coro-early` and `-coro-split`, which causes an
assertion to be hit when the `inttoptr` instruction is forcibly cast to
an `alloca`.

Rearrange the passes such that the coroutine passes are registered
before the sanitizer passes.

Test Plan:
Compile a simple C++ program that uses coroutines in `-O0` with
`-fsanitize-address`, and confirm no assertion is hit:
`clang++ coro-example.cpp -fcoroutines-ts -g -fsanitize=address -fno-omit-frame-pointer`.

Reviewers: GorNishanov, lewissbaker, EricWF

Reviewed By: GorNishanov

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D43927

llvm-svn: 328951

4 years ago[AArch64] Reserve x18 register on Fuchsia
Petr Hosek [Sun, 1 Apr 2018 23:44:04 +0000 (23:44 +0000)]
[AArch64] Reserve x18 register on Fuchsia

This register is reserved as a platform register on Fuchsia.

Differential Revision: https://reviews.llvm.org/D45105

llvm-svn: 328950

4 years ago[Coroutines] Find custom allocators in class scope
Brian Gesiak [Sun, 1 Apr 2018 22:59:22 +0000 (22:59 +0000)]
[Coroutines] Find custom allocators in class scope

https://reviews.llvm.org/rL325291 implemented Coroutines TS N4723
section [dcl.fct.def.coroutine]/7, but it performed lookup of allocator
functions within both the global and class scope, whereas the specified
behavior is to perform lookup for custom allocators within just the
class scope.

To fix, add parameters to the `Sema::FindAllocationFunctions` function
such that it can be used to lookup allocators in global scope,
class scope, or both (instead of just being able to look up in just global
scope or in both global and class scope). Then, use those parameters
from within the coroutine Sema.

This incorrect behavior had the unfortunate side-effect of causing the
bug https://bugs.llvm.org/show_bug.cgi?id=36578 (or at least the reports
of that bug in C++ programs). That bug would occur for any C++ user with
a coroutine frame that took a single pointer argument, since it would
then find the global placement form `operator new`, described in the
C++ standard This patch prevents Clang from generating code
that triggers the LLVM assert described in that bug report.

Test Plan: `check-clang`

Reviewers: GorNishanov, eric_niebler, lewissbaker

Reviewed By: GorNishanov

Subscribers: EricWF, cfe-commits

Differential Revision: https://reviews.llvm.org/D44552

llvm-svn: 328949

4 years ago[DebugCounter] Make -debug-counter cl::Hidden.
Craig Topper [Sun, 1 Apr 2018 22:16:52 +0000 (22:16 +0000)]
[DebugCounter] Make -debug-counter cl::Hidden.

llvm-svn: 328948

4 years ago[LegacyPassManager] Make 'print-module-scope' cl::Hidden like the rest of the printin...
Craig Topper [Sun, 1 Apr 2018 21:54:26 +0000 (21:54 +0000)]
[LegacyPassManager] Make 'print-module-scope' cl::Hidden like the rest of the printing options.

llvm-svn: 328947

4 years ago[X86] Give ADC8/16/32/64mi the same scheduling information as ADC8/16/32/64mr and...
Craig Topper [Sun, 1 Apr 2018 21:54:24 +0000 (21:54 +0000)]
[X86] Give ADC8/16/32/64mi the same scheduling information as ADC8/16/32/64mr and SBB8/16/32/64mi.

It doesn't make a lot of sense that it would be different.

llvm-svn: 328946

4 years ago[x86] Correct the operand structure of the ADOX instruction.
Chandler Carruth [Sun, 1 Apr 2018 21:53:18 +0000 (21:53 +0000)]
[x86] Correct the operand structure of the ADOX instruction.

This also moves to define it in the same way as ADCX which seems to use
constraints a bit better.

This is pulled out of the review for reducing the use of popf for
restoring EFLAGS, but is independent. There are still more problems with
our definitions for these instructions that Craig is going to look at
but this is at least less broken and he can start from this to improve
them more fully.

Thanks to Craig for the review here.

llvm-svn: 328945

4 years ago[x86] Expose more of the condition conversion routines in the public API
Chandler Carruth [Sun, 1 Apr 2018 21:47:55 +0000 (21:47 +0000)]
[x86] Expose more of the condition conversion routines in the public API
for X86's instruction information. I've now got a second patch under
review that needs these same APIs. This bit is nicely orthogonal and
obvious, so landing it. NFC.

llvm-svn: 328944

4 years ago[tools] Change std::sort to llvm::sort in response to r327219
Mandeep Singh Grang [Sun, 1 Apr 2018 21:24:53 +0000 (21:24 +0000)]
[tools] Change std::sort to llvm::sort in response to r327219

r327219 added wrappers to std::sort which randomly shuffle the container before sorting.
This will help in uncovering non-determinism caused due to undefined sorting
order of objects having the same key.

To make use of that infrastructure we need to invoke llvm::sort instead of std::sort.

Note: This patch is one of a series of patches to replace *all* std::sort to llvm::sort.
Refer the comments section in D44363 for a list of all the required patches.

Reviewers: JDevlieghere, zturner, echristo, dberris, friss

Reviewed By: echristo

Subscribers: gbedwell, llvm-commits

Differential Revision: https://reviews.llvm.org/D45141

llvm-svn: 328943

4 years agoFix a major swiftcall ABI bug with trivial C++ class types.
John McCall [Sun, 1 Apr 2018 21:04:30 +0000 (21:04 +0000)]
Fix a major swiftcall ABI bug with trivial C++ class types.

The problem with the previous logic was that there might not be any
explicit copy/move constructor declarations, e.g. if the type is
trivial and we've never type-checked a copy of it.  Relying on Sema's
computation seems much more reliable.

Also, I believe Richard's recommendation is exactly the rule we use
now on the Itanium ABI, modulo the trivial_abi attribute (which this
change of course fixes our handling of in Swift).

This does mean that we have a less portable rule for deciding
indirectness for swiftcall.  I would prefer it if we just applied the
Itanium rule universally under swiftcall, but in the meantime, I need
to fix this bug.

This only arises when defining functions with class-type arguments
in C++, as we do in the Swift runtime.  It doesn't affect normal Swift
operation because we don't import code as C++.

llvm-svn: 328942

4 years ago[include] Change std::sort to llvm::sort in response to r327219
Mandeep Singh Grang [Sun, 1 Apr 2018 18:39:50 +0000 (18:39 +0000)]
[include] Change std::sort to llvm::sort in response to r327219

r327219 added wrappers to std::sort which randomly shuffle the container before sorting.
This will help in uncovering non-determinism caused due to undefined sorting
order of objects having the same key.

To make use of that infrastructure we need to invoke llvm::sort instead of std::sort.

Note: This patch is one of a series of patches to replace *all* std::sort to llvm::sort.
Refer the comments section in D44363 for a list of all the required patches.

Reviewers: echristo, zturner, mzolotukhin, lhames

Reviewed By: echristo

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D45135

llvm-svn: 328940

4 years agoAMDGPU: Make isIntrinsicSourceOfDivergence table-driven
Nicolai Haehnle [Sun, 1 Apr 2018 17:09:14 +0000 (17:09 +0000)]
AMDGPU: Make isIntrinsicSourceOfDivergence table-driven

This is in preparation for the new dimension-aware image intrinsics,
which I'd rather not have to list here by hand.

Change-Id: Iaa16e3a635a11283918ce0d9e1e618591b0bf6fa

Reviewers: arsenm, rampitec, b-sumner

Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, llvm-commits, t-tye

Differential Revision: https://reviews.llvm.org/D44938

llvm-svn: 328939

4 years agoAMDGPU: Make getTgtMemIntrinsic table-driven for resource-based intrinsics
Nicolai Haehnle [Sun, 1 Apr 2018 17:09:07 +0000 (17:09 +0000)]
AMDGPU: Make getTgtMemIntrinsic table-driven for resource-based intrinsics

Avoids having to list all intrinsics manually.

This is in preparation for the new dimension-aware image intrinsics,
which I'd rather not have to list here by hand.

Change-Id: If7ced04998397ef68c4cb8f7de66b5050fb767e5

Reviewers: arsenm, rampitec, b-sumner

Subscribers: kzhuravl, wdng, mgorny, yaxunl, dstuttard, tpr, llvm-commits, t-tye

Differential Revision: https://reviews.llvm.org/D44937

llvm-svn: 328938

4 years agoTableGen: Support Intrinsic values in SearchableTable
Nicolai Haehnle [Sun, 1 Apr 2018 17:08:58 +0000 (17:08 +0000)]
TableGen: Support Intrinsic values in SearchableTable

We will use this in the AMDGPU backend in a subsequent patch
in the stack to lookup target-specific per-intrinsic information.

The generic CodeGenIntrinsic machinery is used to ensure that,
even though we don't calculate actual enum values here, we do
get the intrinsics in the right order for the binary search

Change-Id: If61cd5587963a4c5a1cc53df1e59c5e4dec1f9dc

Reviewers: arsenm, rampitec, b-sumner

Subscribers: wdng, tpr, llvm-commits

Differential Revision: https://reviews.llvm.org/D44935

llvm-svn: 328937

4 years agoTableGen: More helpful error messages
Nicolai Haehnle [Sun, 1 Apr 2018 17:08:49 +0000 (17:08 +0000)]
TableGen: More helpful error messages

Summary: Change-Id: I3c23f6f6597912423762780cd8c5315870412bbe

Reviewers: arsenm, rampitec, b-sumner

Subscribers: wdng, llvm-commits

Differential Revision: https://reviews.llvm.org/D44936

Change-Id: Ie62614a3e2d7774f46e4034478b28f57100a2c92
llvm-svn: 328936

4 years ago[DebugInfo] Change std::sort to llvm::sort in response to r327219
Mandeep Singh Grang [Sun, 1 Apr 2018 16:18:49 +0000 (16:18 +0000)]
[DebugInfo] Change std::sort to llvm::sort in response to r327219

r327219 added wrappers to std::sort which randomly shuffle the container before sorting.
This will help in uncovering non-determinism caused due to undefined sorting
order of objects having the same key.

To make use of that infrastructure we need to invoke llvm::sort instead of std::sort.

Note: This patch is one of a series of patches to replace *all* std::sort to llvm::sort.
Refer the comments section in D44363 for a list of all the required patches.

Reviewers: echristo, zturner, samsonov

Reviewed By: echristo

Subscribers: JDevlieghere, llvm-commits

Differential Revision: https://reviews.llvm.org/D45134

llvm-svn: 328935

4 years ago[ThinLTO] Add an import cutoff for debugging/triaging
Teresa Johnson [Sun, 1 Apr 2018 15:54:40 +0000 (15:54 +0000)]
[ThinLTO] Add an import cutoff for debugging/triaging

Adds -import-cutoff=N which will stop importing during the thin link
after N imports. Default is -1 (no  limit).

Reviewers: wmi

Subscribers: inglorion, llvm-commits

Differential Revision: https://reviews.llvm.org/D45127

llvm-svn: 328934

4 years ago[LoopRotate] Rotate loops with loop exiting latches
David Green [Sun, 1 Apr 2018 12:48:24 +0000 (12:48 +0000)]
[LoopRotate] Rotate loops with loop exiting latches

If a loop has a loop exiting latch, it can be profitable
to rotate the loop if it leads to the simplification of
a phi node. Perform rotation in these cases even if loop
rotate itself didnt simplify the loop to get there.

Differential Revision: https://reviews.llvm.org/D44199

llvm-svn: 328933

4 years ago[clang-tidy] Define __clang_analyzer__ macro for clang-tidy for compatibility with...
Zinovy Nis [Sun, 1 Apr 2018 11:51:57 +0000 (11:51 +0000)]
[clang-tidy] Define __clang_analyzer__ macro for clang-tidy for compatibility with clang static analyzer

This macro is widely used in many well-known projects, ex. Chromium.
But it's not set for clang-tidy, so for ex. DCHECK in Chromium is not considered as [[no-return]], and a lot of false-positive warnings about nullptr dereferenced are emitted.
This patch fixes the issue by explicitly added macro definition.

Differential Revision: https://reviews.llvm.org/D44906

llvm-svn: 328932

4 years ago[X86] Don't check for folding into a store when deciding if we can promote an i16...
Craig Topper [Sun, 1 Apr 2018 06:29:32 +0000 (06:29 +0000)]
[X86] Don't check for folding into a store when deciding if we can promote an i16 mul.

There's no RMW mul operation.

llvm-svn: 328931

4 years ago[X86] Check if the load and store are to the same pointer before preventing i16 RMW...
Craig Topper [Sun, 1 Apr 2018 06:29:28 +0000 (06:29 +0000)]
[X86] Check if the load and store are to the same pointer before preventing i16 RMW shifts and subtracts from being promoted.

llvm-svn: 328930

4 years ago[X86] Add test case to show failure to promote i16 subtract when the LHS is a load...
Craig Topper [Sun, 1 Apr 2018 06:29:27 +0000 (06:29 +0000)]
[X86] Add test case to show failure to promote i16 subtract when the LHS is a load and the result is stored to a different address.

We mistakenly believe we might be able to fold this as a RMW operation, but that doesn't end up happening.

llvm-svn: 328929

4 years ago[X86] Allow i16 subtracts to be promoted if the load is on the LHS and its not being...
Craig Topper [Sun, 1 Apr 2018 06:29:25 +0000 (06:29 +0000)]
[X86] Allow i16 subtracts to be promoted if the load is on the LHS and its not being stored.

llvm-svn: 328928

4 years ago[X86] Add test case to show failure to promote i16 subtract because we mistakenly...
Craig Topper [Sun, 1 Apr 2018 06:29:23 +0000 (06:29 +0000)]
[X86] Add test case to show failure to promote i16 subtract because we mistakenly believe the load can be folded. NFC

The left hand side of the subtract is a load, but we cna't fold those unless we also have a store.

llvm-svn: 328927

4 years ago[X86] Remove unneeded temporary variable. NFC
Craig Topper [Sun, 1 Apr 2018 06:29:21 +0000 (06:29 +0000)]
[X86] Remove unneeded temporary variable. NFC

This Promote flag was alwasys set to true except in the default case. But in the default case we don't need to set PVT and can just return false.

llvm-svn: 328926

4 years ago[Analysis] Change std::sort to llvm::sort in response to r327219
Mandeep Singh Grang [Sun, 1 Apr 2018 01:46:51 +0000 (01:46 +0000)]
[Analysis] Change std::sort to llvm::sort in response to r327219

r327219 added wrappers to std::sort which randomly shuffle the container before sorting.
This will help in uncovering non-determinism caused due to undefined sorting
order of objects having the same key.

To make use of that infrastructure we need to invoke llvm::sort instead of std::sort.

Note: This patch is one of a series of patches to replace *all* std::sort to llvm::sort.
Refer D44363 for a list of all the required patches.

Reviewers: sanjoy, dexonsmith, hfinkel, RKSimon

Reviewed By: dexonsmith

Subscribers: david2050, llvm-commits

Differential Revision: https://reviews.llvm.org/D44944

llvm-svn: 328925

4 years agoAdd missing include to ContinuousRangeMap.h
Eric Fiselier [Sun, 1 Apr 2018 00:33:51 +0000 (00:33 +0000)]
Add missing include to ContinuousRangeMap.h

llvm-svn: 328924

4 years agoAdd missing include to Visibility.h
Eric Fiselier [Sun, 1 Apr 2018 00:31:14 +0000 (00:31 +0000)]
Add missing include to Visibility.h

llvm-svn: 328923

4 years agoRevert r328845, it caused crbug.com/827810.
Nico Weber [Sat, 31 Mar 2018 18:26:25 +0000 (18:26 +0000)]
Revert r328845, it caused crbug.com/827810.

llvm-svn: 328922

4 years ago[DAGCombine] (float)((int) f) --> ftrunc (PR36617)
Sanjay Patel [Sat, 31 Mar 2018 17:55:44 +0000 (17:55 +0000)]
[DAGCombine] (float)((int) f) --> ftrunc (PR36617)

fptosi / fptoui round towards zero, and that's the same behavior as ISD::FTRUNC,
so replace a pair of casts with the equivalent node. We don't have to account for
special cases (NaN, INF) because out-of-range casts are undefined.

Differential Revision: https://reviews.llvm.org/D44909

llvm-svn: 328921

4 years ago[llvm-rtdyld] Fix the InputFileList cl::opt description: it accepts multiple
Lang Hames [Sat, 31 Mar 2018 16:01:01 +0000 (16:01 +0000)]
[llvm-rtdyld] Fix the InputFileList cl::opt description: it accepts multiple
input files.

llvm-svn: 328920

4 years ago[analyzer] Unroll the loop when it has a unsigned counter.
Henry Wong [Sat, 31 Mar 2018 12:46:46 +0000 (12:46 +0000)]
[analyzer] Unroll the loop when it has a unsigned counter.

The original implementation in the `LoopUnrolling.cpp` didn't consider the case where the counter is unsigned. This case is only handled in `simpleCondition()`, but this is not enough, we also need to deal with the unsinged counter with the counter initialization.

Since `IntegerLiteral` is `signed`, there is a `ImplicitCastExpr<IntegralCast>` in `unsigned counter = IntergerLiteral`. This patch add the `ignoringParenImpCasts()` in the `IntegerLiteral` matcher.

Reviewers: szepet, a.sidorin, NoQ, george.karpenkov

Reviewed By: szepet, george.karpenkov

Subscribers: xazax.hun, rnkovacs, cfe-commits, MTC

Differential Revision: https://reviews.llvm.org/D45086

llvm-svn: 328919

4 years ago[X86][Btver2] Add MMX_PSHUFB to the JWritePSHUFB InstRW entries
Simon Pilgrim [Sat, 31 Mar 2018 09:15:54 +0000 (09:15 +0000)]
[X86][Btver2] Add MMX_PSHUFB to the JWritePSHUFB InstRW entries

llvm-svn: 328918

4 years agoFix trailing whitespace. NFCI.
Simon Pilgrim [Sat, 31 Mar 2018 09:14:14 +0000 (09:14 +0000)]
Fix trailing whitespace. NFCI.

llvm-svn: 328917

4 years agoUnbreak the build of the go bindings after r328839.
Benjamin Kramer [Sat, 31 Mar 2018 07:41:25 +0000 (07:41 +0000)]
Unbreak the build of the go bindings after r328839.

llvm-svn: 328916

4 years ago[MIR-Canon] Adding support for local idempotent instruction hoisting.
Puyan Lotfi [Sat, 31 Mar 2018 05:48:51 +0000 (05:48 +0000)]
[MIR-Canon] Adding support for local idempotent instruction hoisting.

llvm-svn: 328915

4 years ago[X86] Add SchedRW for PMULLD
Craig Topper [Sat, 31 Mar 2018 04:54:32 +0000 (04:54 +0000)]
[X86] Add SchedRW for PMULLD

It seems many CPUs don't implement this instruction as well as the other vector multiplies. Often using a multi uop flow. Silvermont in particular has a 7 uop flow with 11 cycle throughput. Sandy Bridge implements it as a single uop with 5 cycle latency and 1 cycle throughput. But Haswell and later use 2 uops with 10 cycle latency and 2 cycle throughput.

This patch adds a new X86SchedWritePair we can use to tag this instruction separately. I've provided correct information for Silvermont, Btver2, and Sandy Bridge. I've removed the InstRWs for SandyBridge. I've left Haswell/Broadwell/Skylake InstRWs in place because I wasn't sure how to account for the different load latency between 128 and 256 bits. I also left Znver1 InstRWs in place because the existing values don't match Agner's spreadsheet.

I also left a FIXME in the SandyBridge model because it being used for the "generic" model is too optimistic for the 256/512-bit versions since those are multiple uops on all known CPUs.

Reviewers: RKSimon, GGanesh, courbet

Reviewed By: RKSimon

Subscribers: gchatelet, gbedwell, andreadb, llvm-commits

Differential Revision: https://reviews.llvm.org/D44972

llvm-svn: 328914

4 years ago[analyzer] Hopefully fix the ARM buildbot.
George Karpenkov [Sat, 31 Mar 2018 02:17:15 +0000 (02:17 +0000)]
[analyzer] Hopefully fix the ARM buildbot.

llvm-svn: 328913

4 years ago[analyzer] Fix assertion crash in CStringChecker
George Karpenkov [Sat, 31 Mar 2018 01:20:08 +0000 (01:20 +0000)]
[analyzer] Fix assertion crash in CStringChecker

An offset might be unknown.


Differential Revision: https://reviews.llvm.org/D45115

llvm-svn: 328912

4 years ago[analyzer] Cache offset computation for MemRegion
George Karpenkov [Sat, 31 Mar 2018 01:20:07 +0000 (01:20 +0000)]
[analyzer] Cache offset computation for MemRegion

Achieves almost a 200% speedup on the example where the performance of
visitors was problematic.

Performance on sqlite3 is unaffected.


Differential Revision: https://reviews.llvm.org/D45113

llvm-svn: 328911

4 years ago[analyzer] Fix liveness calculation for C++17 structured bindings
George Karpenkov [Sat, 31 Mar 2018 01:20:06 +0000 (01:20 +0000)]
[analyzer] Fix liveness calculation for C++17 structured bindings

C++ structured bindings for non-tuple-types are defined in a peculiar
way, where the resulting declaration is not a VarDecl, but a
That means a lot of existing machinery stops working.


Differential Revision: https://reviews.llvm.org/D44956

llvm-svn: 328910

4 years ago[ThinLTO] Add an option to force summary call edges cold for debugging
Teresa Johnson [Sat, 31 Mar 2018 00:18:08 +0000 (00:18 +0000)]
[ThinLTO] Add an option to force summary call edges cold for debugging

Useful to selectively disable importing into specific modules for

Reviewers: eraman

Subscribers: inglorion, llvm-commits

Differential Revision: https://reviews.llvm.org/D45062

llvm-svn: 328909

4 years ago[ELF] Simplify read32. NFC
Fangrui Song [Fri, 30 Mar 2018 23:13:00 +0000 (23:13 +0000)]
[ELF] Simplify read32. NFC

llvm-svn: 328908

4 years agoFix a bunch of typoes. NFC
Fangrui Song [Fri, 30 Mar 2018 22:22:31 +0000 (22:22 +0000)]
Fix a bunch of typoes. NFC

llvm-svn: 328907

4 years ago[ASTImporter] Add test helper Fixture
Peter Szecsi [Fri, 30 Mar 2018 22:03:29 +0000 (22:03 +0000)]
[ASTImporter] Add test helper Fixture

Add a helper test Fixture, so we can add tests which can check internal
attributes of AST nodes like getPreviousDecl(), isVirtual(), etc.
This enables us to check if a redeclaration chain is correctly built during
import, if the virtual flag is preserved during import, etc. We cannot check
such attributes with the existing testImport.
Also, this fixture makes it possible to import from several "From" contexts.

We also added several test cases here, some of them are disabled.
We plan to pass the disabled tests in other patches.

Patch by Gabor Marton!

Differential Revision: https://reviews.llvm.org/D43967

llvm-svn: 328906

4 years agoELF: Place ordered sections in the middle of the unordered section list on targets...
Peter Collingbourne [Fri, 30 Mar 2018 21:36:54 +0000 (21:36 +0000)]
ELF: Place ordered sections in the middle of the unordered section list on targets with limited-range branches.

It generally does not matter much where we place sections ordered
by --symbol-ordering-file relative to other sections. But if the
ordered sections are hot (which is the case already for some users
of --symbol-ordering-file, and is increasingly more likely to be
the case once profile-guided section layout lands) and the target
has limited-range branches, it is beneficial to place the ordered
sections in the middle of the output section in order to decrease
the likelihood that a range extension thunk will be required to call
a hot function from a cold function or vice versa.

That is what this patch does. After D44966 it reduces the size of
Chromium for Android's .text section by 60KB.

Differential Revision: https://reviews.llvm.org/D44969

llvm-svn: 328905

4 years agoPrevent data races in concurrent ThinLTO processes.
Ekaterina Romanova [Fri, 30 Mar 2018 21:35:42 +0000 (21:35 +0000)]
Prevent data races in concurrent ThinLTO processes.

Make sure ThinLTO with caching doesn't use non-atomic writes to the cache file (to prevent data races and cache files corruption).

1. Place temp file to the same place where the caching directory is (instead of creating it the directory pointed to by TMP/TEMP variable). This will help to prevent using non-atomic rename and falling back to non-atomic "direct" write to the cache file.
2. if rename failed do not write to the cache file directly (direct write to the file is non-atomic and could cause data race conditions).
3. if cache file doesn't exist (e.g., because 'rename' failed or because some other reasons), bypass using the cache altogether.

Differential Revision:  https://reviews.llvm.org/D45076

llvm-svn: 328904

4 years ago[analyzer] Fix test triple in missing-bind-temporary.cpp.
Artem Dergachev [Fri, 30 Mar 2018 21:22:35 +0000 (21:22 +0000)]
[analyzer] Fix test triple in missing-bind-temporary.cpp.

Otherwise the default triple for x86-windows-msvc2015 auto-inserts
__attribute__((thiscall)) to some calls.

Fixes the respective buildbot.

llvm-svn: 328903

4 years agoInitialize Elf Header to zero to ensure that bytes not assigned any value later on...
Rumeet Dhindsa [Fri, 30 Mar 2018 20:49:34 +0000 (20:49 +0000)]
Initialize Elf Header to zero to ensure that bytes not assigned any value later on are initialized properly.

Differential Revision: https://reviews.llvm.org/D44986

llvm-svn: 328902

4 years ago[WebAssembly] Register wasm passes with the PassRegistry
Jacob Gravelle [Fri, 30 Mar 2018 20:36:58 +0000 (20:36 +0000)]
[WebAssembly] Register wasm passes with the PassRegistry

This exposes WebAssembly passes for use on the command line (as
arguments to -print-before and the like).

Reviewers: dschuff, sunfish

Subscribers: MatzeB, jfb, sbc100, llvm-commits, aheejin

Differential Revision: https://reviews.llvm.org/D45103

llvm-svn: 328901

4 years agoMinor cleanup in __kmp_atfork_child()
Jonathan Peyton [Fri, 30 Mar 2018 19:55:11 +0000 (19:55 +0000)]
Minor cleanup in __kmp_atfork_child()

This change removes the unnecessary lock operation on __kmp_initz_lock inside
the __kmp_atfork_child() function for Linux; the lock variable is initialized
in the same function later.

Patch by Hansang Bae

Differential Revision: https://reviews.llvm.org/D44949

llvm-svn: 328900

4 years ago[Hexagon] Fix testcase
Krzysztof Parzyszek [Fri, 30 Mar 2018 19:46:28 +0000 (19:46 +0000)]
[Hexagon] Fix testcase

llvm-svn: 328899

4 years ago[Hexagon] Reduce excessive indentation in .s output
Krzysztof Parzyszek [Fri, 30 Mar 2018 19:30:28 +0000 (19:30 +0000)]
[Hexagon] Reduce excessive indentation in .s output

llvm-svn: 328898

4 years ago[Hexagon] Avoid creating invalid offsets in packetizer
Krzysztof Parzyszek [Fri, 30 Mar 2018 19:28:37 +0000 (19:28 +0000)]
[Hexagon] Avoid creating invalid offsets in packetizer

Two memory instructions with a dependency only on the address register
between the two (the first one of them being post-incrememnt) can be
packetized together after the offset on the second was updated to the
incremement value. Make sure that the new offset is valid for the

llvm-svn: 328897

4 years ago[analyzer] Track null or undef values through pointer arithmetic.
Artem Dergachev [Fri, 30 Mar 2018 19:27:42 +0000 (19:27 +0000)]
[analyzer] Track null or undef values through pointer arithmetic.

Pointer arithmetic on null or undefined pointers results in null or undefined
pointers. This is obvious for undefined pointers; for null pointers it follows
from our incorrect-but-somehow-working approach that declares that 0 (Loc)
doesn't necessarily represent a pointer of numeric address value 0, but instead
it represents any pointer that will cause a valid "null pointer dereference"
issue when dereferenced.

For now we've been seeing through pointer arithmetic at the original dereference
expression, i.e. in bugreporter::getDerefExpr(), but not during further
investigation of the value's origins in bugreporter::trackNullOrUndefValue().
The patch fixes it.

Differential Revision: https://reviews.llvm.org/D45071

llvm-svn: 328896

4 years ago[CFG] [analyzer] Work around a disappearing CXXBindTemporaryExpr.
Artem Dergachev [Fri, 30 Mar 2018 19:25:39 +0000 (19:25 +0000)]
[CFG] [analyzer] Work around a disappearing CXXBindTemporaryExpr.

Sometimes template instantiation causes CXXBindTemporaryExpr to be missing in
its usual spot. In CFG, temporary destructors work by relying on
CXXBindTemporaryExprs, so they won't work in this case.

Avoid the crash and notify the clients that we've encountered an unsupported AST
by failing to provide the ill-formed construction context for the temporary.

Differential Revision: https://reviews.llvm.org/D44955

llvm-svn: 328895

4 years ago[lldb-dotest] Don't swallow error exit codes.
Davide Italiano [Fri, 30 Mar 2018 19:24:08 +0000 (19:24 +0000)]
[lldb-dotest] Don't swallow error exit codes.

llvm-svn: 328894

4 years ago[CFG] [analyzer] Avoid modeling C++17 constructors that aren't fully supported.
Artem Dergachev [Fri, 30 Mar 2018 19:21:18 +0000 (19:21 +0000)]
[CFG] [analyzer] Avoid modeling C++17 constructors that aren't fully supported.

Not enough work has been done so far to ensure correctness of construction
contexts in the CFG when C++17 copy elision is in effect, so for now we
should drop construction contexts in the CFG and in the analyzer when
they seem different from what we support anyway.

This includes initializations with conditional operators and return values
across multiple stack frames.

Differential Revision: https://reviews.llvm.org/D44854

llvm-svn: 328893

4 years ago[X86][BtVer2] Fixed the number of micro opcodes for AVX vector converts and
Andrea Di Biagio [Fri, 30 Mar 2018 18:53:47 +0000 (18:53 +0000)]
[X86][BtVer2] Fixed the number of micro opcodes for AVX vector converts and
VSQRT instructions.

There were still a few AVX instructions with an incorrect number of opcodes.
These should be fixed now.

llvm-svn: 328892

4 years agoRemove unused CHECK lines leftover from r306928.
Eli Friedman [Fri, 30 Mar 2018 18:39:28 +0000 (18:39 +0000)]
Remove unused CHECK lines leftover from r306928.

The RUN lines were removed, but the corresponding CHECK lines never
went away.

llvm-svn: 328891

4 years agoDataFlowSanitizer: wrappers of functions with local linkage should have the same...
Peter Collingbourne [Fri, 30 Mar 2018 18:37:55 +0000 (18:37 +0000)]
DataFlowSanitizer: wrappers of functions with local linkage should have the same linkage as the function being wrapped

This patch resolves link errors when the address of a static function is taken, and that function is uninstrumented by DFSan.

This change resolves bug 36314.

Patch by Sam Kerner!

Differential Revision: https://reviews.llvm.org/D44784

llvm-svn: 328890

4 years agoELF: Try to create last thunk section at ThunkSectionSpacing bytes before the end.
Peter Collingbourne [Fri, 30 Mar 2018 18:32:24 +0000 (18:32 +0000)]
ELF: Try to create last thunk section at ThunkSectionSpacing bytes before the end.

Now that we have the ability to create short thunks, it is beneficial
for thunk sections to be surrounded by ThunkSectionSpacing bytes
of code on both sides in order to increase the likelihood that the
distance from the thunk to the target will be sufficiently small to
allow for the creation of a short thunk. This is currently the case
for most thunks that we create, except for the last one, which could,
depending on the size of the output section, potentially appear near
the end and therefore have a relatively small amount of code after it.

This patch moves the last thunk section to ThunkSectionSpacing bytes
before the end of the output section, as long as the section is larger
than 2*ThunkSectionSpacing bytes. It reduces the size of Chromium
for Android's .text section by 32KB.

Differential Revision: https://reviews.llvm.org/D44966

llvm-svn: 328889

4 years ago[OPENMP] Added emission of offloading data sections for declare target
Alexey Bataev [Fri, 30 Mar 2018 18:31:07 +0000 (18:31 +0000)]
[OPENMP] Added emission of offloading data sections for declare target

Added emission of the offloading data sections for the variables within
declare target regions + fixes emission of the declare target variables
marked as declare target not within the declare target region.

llvm-svn: 328888

4 years ago[MIR] Adding support for Named Virtual Registers in MIR.
Puyan Lotfi [Fri, 30 Mar 2018 18:15:54 +0000 (18:15 +0000)]
[MIR] Adding support for Named Virtual Registers in MIR.

llvm-svn: 328887

4 years ago[X86][BtVer2] Fix the number of uOps for horizontal operations.
Andrea Di Biagio [Fri, 30 Mar 2018 18:15:30 +0000 (18:15 +0000)]
[X86][BtVer2] Fix the number of uOps for horizontal operations.

llvm-svn: 328886

4 years ago[NVPTX] Enable StructuredCFG for NVPTX
Tim Shen [Fri, 30 Mar 2018 17:51:03 +0000 (17:51 +0000)]
[NVPTX] Enable StructuredCFG for NVPTX

Make NVPTX require structured CFG. Added a temporary flag to
"roll back" the behavior for easy deployment.

Combined with D45008, this fixes several internal Nvidia GPU test
failures that we suspect to be ptxas miscompiles (PR27738).

Reviewers: jlebar

Subscribers: jholewinski, sanjoy, llvm-commits, hiraditya

Differential Revision: https://reviews.llvm.org/D45070

llvm-svn: 328885

4 years ago[BlockPlacement] Disable block placement tail duplciation in structured CFG.
Tim Shen [Fri, 30 Mar 2018 17:51:00 +0000 (17:51 +0000)]
[BlockPlacement] Disable block placement tail duplciation in structured CFG.

Tail duplication easily breaks the structure of CFG, e.g. duplicating on
a region entry. If the structure is intended to be preserved, then we
may want to configure tail duplication, or disable it for structured
CFG. From our benchmark results disabling it doesn't cause performance

Notice that this currently affects AMDGPU backend. In the next patch, I
also plan to turn on requiresStructuredCFG for NVPTX.

All unit tests still pass.

Reviewers: jlebar, arsenm

Subscribers: jholewinski, sanjoy, wdng, tpr, hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D45008

llvm-svn: 328884

4 years ago[LLVM-C] Finish exception instruction bindings - Round 2
Robert Widmann [Fri, 30 Mar 2018 17:49:53 +0000 (17:49 +0000)]
[LLVM-C] Finish exception instruction bindings - Round 2

Previous revision caused a leak in the echo test that got caught by the ASAN bots because of missing free of the handlers array and was reverted in r328759.  Resubmitting the patch with that correction.

Add support for cleanupret, catchret, catchpad, cleanuppad and catchswitch and their associated accessors.

Test is modified from SimplifyCFG because it contains many diverse usages of these instructions.

Reviewers: whitequark, deadalnix

Reviewed By: whitequark

Subscribers: llvm-commits, vlad.tsyrklevich

Differential Revision: https://reviews.llvm.org/D45100

llvm-svn: 328883

4 years agoFix Windows buildbots.
Rui Ueyama [Fri, 30 Mar 2018 17:49:51 +0000 (17:49 +0000)]
Fix Windows buildbots.

llvm-svn: 328882

4 years agoFix some signed / unsigned conversion problems.
Zachary Turner [Fri, 30 Mar 2018 17:28:35 +0000 (17:28 +0000)]
Fix some signed / unsigned conversion problems.

llvm-svn: 328881

4 years agoImprove error message for an unknown --plugin-opt.
Rui Ueyama [Fri, 30 Mar 2018 17:22:44 +0000 (17:22 +0000)]
Improve error message for an unknown --plugin-opt.


  $ ld.lld --plugin-opt=-foo
  ld.lld: --Unknown command line argument '-abc'


  $ ld.lld --plugin-opt=-foo
  ld.lld: --plugin-opt: ld.lld --Unknown command line argument '-abc'

Differential Revision: https://reviews.llvm.org/D45075

llvm-svn: 328880

4 years ago[lld-link] Add comment explaining that /FIXED behavior is correct despite contradicti...
Nico Weber [Fri, 30 Mar 2018 17:17:04 +0000 (17:17 +0000)]
[lld-link] Add comment explaining that /FIXED behavior is correct despite contradicting MSDN.

Also add a test for /FIXED.

llvm-svn: 328879

4 years ago[llvm-pdbutil] Dig deeper into the PDB and DBI streams when explaining.
Zachary Turner [Fri, 30 Mar 2018 17:16:50 +0000 (17:16 +0000)]
[llvm-pdbutil] Dig deeper into the PDB and DBI streams when explaining.

This will show more detail when using `llvm-pdbutil explain` on an
offset in the DBI or PDB streams.  Specifically, it will dig into
individual header fields and substreams to give a more precise
description of what the byte represents.

llvm-svn: 328878

4 years ago[lld-link] Let /PROFILE imply /OPT:REF /OPT:NOICF /INCREMENTAL:NO /FIXED:NO
Nico Weber [Fri, 30 Mar 2018 17:14:50 +0000 (17:14 +0000)]

/FIXED:NO is always the default, so that part needs no work.

Also test the interaction of /ORDER: with /INCREMENTAL.


llvm-svn: 328877

4 years ago[WebAssembly] Refactor tablegen for store instructions (NFC)
Derek Schuff [Fri, 30 Mar 2018 17:02:50 +0000 (17:02 +0000)]
[WebAssembly] Refactor tablegen for store instructions (NFC)

Summary: Add patterns similar to loads.

Differential Revision: https://reviews.llvm.org/D45064

llvm-svn: 328876

4 years agoRevert "peel loops with runtime small trip counts"
Krzysztof Parzyszek [Fri, 30 Mar 2018 16:55:44 +0000 (16:55 +0000)]
Revert "peel loops with runtime small trip counts"

This reverts commit r328854, it breaks some Hexagon tests.

llvm-svn: 328875

4 years ago[AMDGPU] Fixed some instructions latencies
Stanislav Mekhanoshin [Fri, 30 Mar 2018 16:19:13 +0000 (16:19 +0000)]
[AMDGPU] Fixed some instructions latencies

Differential Revision: https://reviews.llvm.org/D45073

llvm-svn: 328874

4 years ago[WebAssembly] Error if both --export-table and --import-table are specified.
Rui Ueyama [Fri, 30 Mar 2018 16:06:14 +0000 (16:06 +0000)]
[WebAssembly] Error if both --export-table and --import-table are specified.

Differential Revision: https://reviews.llvm.org/D45001

llvm-svn: 328873

4 years ago[SelectionDAG] Removing FABS folding from DAGCombiner
Sanjay Patel [Fri, 30 Mar 2018 15:42:52 +0000 (15:42 +0000)]
[SelectionDAG] Removing FABS folding from DAGCombiner

The code has bugs dealing with -0.0.

Since D44550 introduced FABS pattern folding in InstCombine,
this patch removes the now-redundant code that causes

Patch by Mikhail Dvoretckii!

Differential Revision: https://reviews.llvm.org/D44683

llvm-svn: 328872

4 years ago[clang-format] Ensure wrapped ObjC selectors with 1 arg obey IndentWrappedFunctionNames
Ben Hamilton [Fri, 30 Mar 2018 15:38:45 +0000 (15:38 +0000)]
[clang-format] Ensure wrapped ObjC selectors with 1 arg obey IndentWrappedFunctionNames

In D43121, @Typz introduced logic to avoid indenting 2-or-more
argument ObjC selectors too far to the right if the first component
of the selector was longer than the others.

This had a small side effect of causing wrapped ObjC selectors with
exactly 1 argument to not obey IndentWrappedFunctionNames:

- (aaaaaaaaaa)

This diff fixes the issue by ensuring we align wrapped 1-argument
ObjC selectors correctly:

- (aaaaaaaaaa)

Test Plan: New tests added. Test failed before change, passed
  after change. Ran tests with:
  % make -j12 FormatTests && ./tools/clang/unittests/Format/FormatTests

Reviewers: djasper, klimek, Typz, jolesiak

Reviewed By: djasper, jolesiak

Subscribers: cfe-commits, Typz

Differential Revision: https://reviews.llvm.org/D44994

llvm-svn: 328871

4 years ago[Hexagon] Recognize and handle :endloop01
Krzysztof Parzyszek [Fri, 30 Mar 2018 15:29:47 +0000 (15:29 +0000)]
[Hexagon] Recognize and handle :endloop01

llvm-svn: 328870

4 years ago[Hexagon] Fix printing :mem_noshuf on compiler-generated packets
Krzysztof Parzyszek [Fri, 30 Mar 2018 15:09:05 +0000 (15:09 +0000)]
[Hexagon] Fix printing :mem_noshuf on compiler-generated packets

llvm-svn: 328869

4 years ago[Hexagon] Fix flags for store-related intrinsics
Krzysztof Parzyszek [Fri, 30 Mar 2018 14:57:01 +0000 (14:57 +0000)]
[Hexagon] Fix flags for store-related intrinsics

llvm-svn: 328868

4 years ago[X86][BtVer2] Add missing ReadAfterLd to RM variants of AVX horizontal adds and
Andrea Di Biagio [Fri, 30 Mar 2018 14:48:08 +0000 (14:48 +0000)]
[X86][BtVer2] Add missing ReadAfterLd to RM variants of AVX horizontal adds and
most vector logic instructions.

Fixed a few InstRW that forgot to specify a ReadAfterLd for the register input

llvm-svn: 328867

4 years ago[Hexagon] Remove unused scheduling classes
Krzysztof Parzyszek [Fri, 30 Mar 2018 14:34:32 +0000 (14:34 +0000)]
[Hexagon] Remove unused scheduling classes

llvm-svn: 328866

4 years ago[X86][BtVer2] Add tests that show how ReadAfterLd is missing for some
Andrea Di Biagio [Fri, 30 Mar 2018 14:29:33 +0000 (14:29 +0000)]
[X86][BtVer2] Add tests that show how ReadAfterLd is missing for some

In the Btver2 model, there are a few InstRW overrides that don't specify a
ReadAfterLd for the register input operand.

As a result, a few AVX variants of horizontal operations and most vector logic
operations with a folded memory operand don't have a ReadAdvance info associated
to their input register operands.

llvm-svn: 328865

4 years ago[Hexagon] Pass pointer to SelectionDAG to dump functions
Krzysztof Parzyszek [Fri, 30 Mar 2018 14:29:15 +0000 (14:29 +0000)]
[Hexagon] Pass pointer to SelectionDAG to dump functions

llvm-svn: 328864

4 years agoSimplify test more.
Nico Weber [Fri, 30 Mar 2018 13:48:11 +0000 (13:48 +0000)]
Simplify test more.

llvm-svn: 328863

4 years agoSimplify test.
Nico Weber [Fri, 30 Mar 2018 13:44:15 +0000 (13:44 +0000)]
Simplify test.

As of rL215127, FileCheck has an -allow-empty flag,
so this could be used instead of writing a dummy line.
But it looks like the log is never empty now, so not
even that is needed.

llvm-svn: 328862

4 years ago[X86] Add llvm-mca tests for r328834.
Andrea Di Biagio [Fri, 30 Mar 2018 13:38:37 +0000 (13:38 +0000)]
[X86] Add llvm-mca tests for r328834.

Verify that the ReadAfterLd is correctly applied to FMA and 4-ops variable blend

As Craig pointed out in D44726, some Intel models still have to be fixed.

llvm-svn: 328861

4 years ago[analyzer] Remove the unused method declaration in `ValistChecker.cpp`.
Henry Wong [Fri, 30 Mar 2018 13:37:50 +0000 (13:37 +0000)]
[analyzer] Remove the unused method declaration in `ValistChecker.cpp`.

Summary: `getVariableNameFromRegion()` seems useless.

Reviewers: xazax.hun, george.karpenkov

Reviewed By: xazax.hun

Subscribers: szepet, rnkovacs, a.sidorin, cfe-commits, MTC

Differential Revision: https://reviews.llvm.org/D45081

llvm-svn: 328860

4 years ago[X86] Add tests to verify the presence of "ReadAfterLd" after r328823.
Andrea Di Biagio [Fri, 30 Mar 2018 11:44:48 +0000 (11:44 +0000)]
[X86] Add tests to verify the presence of "ReadAfterLd" after r328823.

This change adds a couple of tests to verify the change introduced by revision
328823 ([X86] Correct the placement of ReadAfterLd in BEXTR and BZHI).

llvm-svn: 328859

4 years agoRevert "[LLVM-C] Finish exception instruction bindings"
Vlad Tsyrklevich [Fri, 30 Mar 2018 06:21:28 +0000 (06:21 +0000)]
Revert "[LLVM-C] Finish exception instruction bindings"

This reverts commit r328759. It was causing LSan failures on sanitizer-x86_64-linux-bootstrap

llvm-svn: 328858

4 years ago[Modules] Improve fixit for framework private module maps
Bruno Cardoso Lopes [Fri, 30 Mar 2018 05:17:58 +0000 (05:17 +0000)]
[Modules] Improve fixit for framework private module maps

The wrong source range was being provided in some case, fix that to get
a better fixit.


llvm-svn: 328857

4 years ago[AMDGPU] Fix the SDWA Peephole phase to handle src for dst:UNUSED_PRESERVE.
Michael Bedy [Fri, 30 Mar 2018 05:03:36 +0000 (05:03 +0000)]
[AMDGPU] Fix the SDWA Peephole phase to handle src for dst:UNUSED_PRESERVE.

The phase attempts to transform operations that extract a portion of a value
into an SDWA src operand in cases where that value is used only once. It
was not prepared for this use to be the preserved portion of a value for
dst:UNUSED_PRESERVE, resulting in a crash or assert.

This change either rejects the illegal SDWA attempt, or in the case where
dst:WORD_1 and the src_sel would be WORD_0, removes the unneeded
extract instruction.

Reviewers: arsenm, #amdgpu

Reviewed By: arsenm, #amdgpu

Subscribers: arsenm, kzhuravl, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, llvm-commits

Differential Revision: https://reviews.llvm.org/D44364

llvm-svn: 328856

4 years ago[Hexagon] add missing lit config file
Ikhlas Ajbar [Fri, 30 Mar 2018 03:32:24 +0000 (03:32 +0000)]
[Hexagon] add missing lit config file

llvm-svn: 328855

4 years agopeel loops with runtime small trip counts
Ikhlas Ajbar [Fri, 30 Mar 2018 03:05:34 +0000 (03:05 +0000)]
peel loops with runtime small trip counts

For Hexagon, peeling loops with small runtime trip count is beneficial for our
benchmarks. We set PeelCount in HexagonTargetInfo.cpp and we use PeelCount set
by the target for computing the desired peel count.

Differential Revision: https://reviews.llvm.org/D44880

llvm-svn: 328854

4 years agoAdding UNSUPPORTED: system-windows at George's request until the problem can be debugged.
Douglas Yung [Fri, 30 Mar 2018 01:29:07 +0000 (01:29 +0000)]
Adding UNSUPPORTED: system-windows at George's request until the problem can be debugged.

llvm-svn: 328853

4 years agoRe-implement --just-symbols as a regular object file.
Rui Ueyama [Fri, 30 Mar 2018 01:15:36 +0000 (01:15 +0000)]
Re-implement --just-symbols as a regular object file.

I tried a few different designs to find a way to implement it without
too much hassle and settled down with this. Unlike before, object files
given as arguments for --just-symbols are handled as object files, with
an exception that their section tables are handled as if they were all

Differential Revision: https://reviews.llvm.org/D42025

llvm-svn: 328852