oota-llvm.git
8 years ago[WinEH] Update catchrets with cloned successors
Joseph Tremoulet [Sat, 2 Jan 2016 15:22:36 +0000 (15:22 +0000)]
[WinEH] Update catchrets with cloned successors

Summary:
Add a pass to update catchrets when their successors get cloned; the
existing pass doesn't catch these because it walks the funclet whose
blocks are being cloned but the catchret is in a child funclet.

Also update the test for removing incoming PHI values; when the
predecessor is a catchret, the relevant color is the catchret's parentPad,
not its block's color.

Reviewers: andrew.w.kaylor, rnk, majnemer

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D15840

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256689 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoCorrect misleading formatting of several ifs followed by two statements without braces.
Yaron Keren [Sat, 2 Jan 2016 13:40:36 +0000 (13:40 +0000)]
Correct misleading formatting of several ifs followed by two statements without braces.
While the original code would work with or without braces, it makes sense to
set HaveSemi to true only if (!HaveSemi), otherwise it's already true, so I
put the assignment inside the if block. This addresses PR25998.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256688 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[WinEH] Add additional verification
David Majnemer [Sat, 2 Jan 2016 09:26:36 +0000 (09:26 +0000)]
[WinEH] Add additional verification

Recolor the IR to make sure our computed colors are not hiding any bugs.
Also, verifyFunction if we are running some post-preparation operations;
some of these operations can hide latent bugs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256687 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Add intrinsics for reading and writing to the flags register
David Majnemer [Fri, 1 Jan 2016 06:50:01 +0000 (06:50 +0000)]
[X86] Add intrinsics for reading and writing to the flags register

LLVM's targets need to know if stack pointer adjustments occur after the
prologue.  This is needed to correctly determine if the red-zone is
appropriate to use or if a frame pointer is required.

Normally, LLVM can figure this out very precisely by reasoning about the
contents of the MachineFunction.  There is an interesting corner case:
inline assembly.

The vast majority of inline assembly which will perform a push or pop is
done so to pair up with pushf or popf as appropriate.  Unfortunately,
this inline assembly doesn't mark the stack pointer as clobbered
because, well, it isn't.  The stack pointer is decremented and then
immediately incremented.  Because of this, LLVM was changed in r256456
to conservatively assume that inline assembly contain a sequence of
stack operations.  This is unfortunate because the vast majority of
inline assembly will not end up manipulating the stack pointer in any
way at all.

Instead, let's provide a more principled solution: an intrinsic.
FWIW, other compilers (MSVC and GCC among them) also provide this
functionality as an intrinsic.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256685 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[LibCallSimplifier] propagate FMF when shrinking binary calls
Sanjay Patel [Thu, 31 Dec 2015 23:40:59 +0000 (23:40 +0000)]
[LibCallSimplifier] propagate FMF when shrinking binary calls

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256682 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Remove a return after llvm_unreachable.
Craig Topper [Thu, 31 Dec 2015 22:40:48 +0000 (22:40 +0000)]
[X86] Remove a return after llvm_unreachable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256681 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Move shuffle decoding for constant pool into the X86CodeGen library to remove...
Craig Topper [Thu, 31 Dec 2015 22:40:45 +0000 (22:40 +0000)]
[X86] Move shuffle decoding for constant pool into the X86CodeGen library to remove a layering violation in the Util library.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256680 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[LibCallSimplifier] propagate FMF when shrinking unary calls
Sanjay Patel [Thu, 31 Dec 2015 21:52:31 +0000 (21:52 +0000)]
[LibCallSimplifier] propagate FMF when shrinking unary calls

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256679 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agochange function names to avoid accidentally matching the substring
Sanjay Patel [Thu, 31 Dec 2015 21:25:25 +0000 (21:25 +0000)]
change function names to avoid accidentally matching the substring

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256678 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoadd 'fast' attribute to calls to show that the flag isn't being propagated
Sanjay Patel [Thu, 31 Dec 2015 21:12:19 +0000 (21:12 +0000)]
add 'fast' attribute to calls to show that the flag isn't being propagated

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256677 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoVariable names start with an upper case letter; NFC
Sanjay Patel [Thu, 31 Dec 2015 16:16:58 +0000 (16:16 +0000)]
Variable names start with an upper case letter; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256676 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agofix formatting; NFC
Sanjay Patel [Thu, 31 Dec 2015 16:10:49 +0000 (16:10 +0000)]
fix formatting; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256675 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoadd FMF for CreateCall variant
Sanjay Patel [Thu, 31 Dec 2015 15:39:34 +0000 (15:39 +0000)]
add FMF for CreateCall variant

The version with OpBundles was missed in:
http://reviews.llvm.org/rL255555

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256674 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AVX512] add PSRLQ and PSRLD Intrinsic
Michael Zuckerman [Thu, 31 Dec 2015 15:22:04 +0000 (15:22 +0000)]
[AVX512] add PSRLQ and PSRLD Intrinsic

Differential Revision: http://reviews.llvm.org/D15770

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256673 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Avoid folding scalar loads into unary sse intrinsics
Michael Kuperstein [Thu, 31 Dec 2015 09:45:16 +0000 (09:45 +0000)]
[X86] Avoid folding scalar loads into unary sse intrinsics

Not folding these cases tends to avoid partial register updates:
sqrtss (%eax), %xmm0
Has a partial update of %xmm0, while
movss (%eax), %xmm0
sqrtss %xmm0, %xmm0
Has a clobber of the high lanes immediately before the partial update,
avoiding a potential stall.

Given this, we only want to fold when optimizing for size.
This is consistent with the patterns we already have for some of
the fp/int converts, and in X86InstrInfo::foldMemoryOperandImpl()

Differential Revision: http://reviews.llvm.org/D15741

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256671 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86][PKU] Add {RD,WR}PKRU intrinsics
Asaf Badouh [Thu, 31 Dec 2015 08:31:13 +0000 (08:31 +0000)]
[X86][PKU] Add {RD,WR}PKRU intrinsics

Differential Revision: http://reviews.llvm.org/D15808

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256670 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Modify the AsmMatcherEmitter to only apply the table growth from r252440...
Craig Topper [Thu, 31 Dec 2015 08:18:23 +0000 (08:18 +0000)]
[TableGen] Modify the AsmMatcherEmitter to only apply the table growth from r252440 to the Hexagon target.

This restores the previous behavior of not including the mnemonic in the classes table for every target that starts instruction lines with the mnemonic. Not only did the table size increase by 1 entry, but the class enum increased in size which caused every class in the array to increase in size. It also grew the size of the function that parsers tokens into classes by a substantial amount.

This adds a new HasMnemonicFirst flag to all AsmParsers. It's set to 1 by default and Hexagon target overrides it to 0.

For the X86 target alone this recovers 324KB of size on the llvm-mc executable.

I believe the current state is still a bad design choice for the Hexagon target as it causes most of the parsing to do a linear search through the entire match table to comparing operands against every instruction until it finds one that works. At least for the other targets we do a binary search based on mnemonic over which to do the linear scan.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256669 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Use range-based for loops. NFC
Craig Topper [Thu, 31 Dec 2015 08:18:20 +0000 (08:18 +0000)]
[TableGen] Use range-based for loops. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256668 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[PGO]: Implement Func PGO name string compression
Xinliang David Li [Thu, 31 Dec 2015 07:57:16 +0000 (07:57 +0000)]
[PGO]: Implement Func PGO name string compression

This is part of the effort/prepration to reduce the size
instr-pgo (object, binary, memory footprint, and raw data).

The functionality is currently off by default and not yet
used by any clients.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256667 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Move determination of IsIsolatedToken into the tokenizer instead of trying...
Craig Topper [Thu, 31 Dec 2015 05:01:45 +0000 (05:01 +0000)]
[TableGen] Move determination of IsIsolatedToken into the tokenizer instead of trying to search characters around the token. No functional change intended. Verified for in-tree targets.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256660 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRevert "Revert "[ptr-traits] Implement the base pointer traits using the actual""
Chandler Carruth [Wed, 30 Dec 2015 23:37:25 +0000 (23:37 +0000)]
Revert "Revert "[ptr-traits] Implement the base pointer traits using the actual""

This reverts commit r256642 and restores r256620 now that Tobias has
updated Polly.

There are still some potential problems with the code in Polly that I've
sent post-commit review about, but they're unlikely to break anything in
practice, and I'd like to avoid the rest of LLVM and Clang regressing
here.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256656 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ValueTracking] fix bug computing isKnownToBeAPowerOfTwo() with arithmetic shift...
Sanjay Patel [Wed, 30 Dec 2015 22:40:52 +0000 (22:40 +0000)]
[ValueTracking] fix bug computing isKnownToBeAPowerOfTwo() with arithmetic shift right (PR25900)

This is a fix for:
https://llvm.org/bugs/show_bug.cgi?id=25900

If we think that an arithmetic right shift of a power of two is always a power of two,
an sdiv gets wrongly converted to udiv.

Differential Revision: http://reviews.llvm.org/D15827

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256655 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ThinLTO] Rename variables used in metadata linking (NFC)
Teresa Johnson [Wed, 30 Dec 2015 21:13:55 +0000 (21:13 +0000)]
[ThinLTO] Rename variables used in metadata linking (NFC)

As suggested in review for r255909, rename MDMaterialized to AllowTemps,
and identify the name of the boolean flag being set in calls to
saveMetadataList.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256653 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoEnsure MDNode used as key in metadata linking map cannot be RAUWed
Teresa Johnson [Wed, 30 Dec 2015 19:32:24 +0000 (19:32 +0000)]
Ensure MDNode used as key in metadata linking map cannot be RAUWed

As suggested in review for r255909, add a way to ensure that temporary
MD used as keys in the MetadataToID map during ThinLTO importing are not
RAUWed.

Add support for marking an MDNode as not replaceable. Clear the new
CanReplace flag when adding a temporary MD node to the MetadataToID map
and clear it when destroying the map.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256648 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ThinLTO] Check MDNode values saved for metadata linking (NFC)
Teresa Johnson [Wed, 30 Dec 2015 19:13:57 +0000 (19:13 +0000)]
[ThinLTO] Check MDNode values saved for metadata linking (NFC)

Add an assert suggested in review for r255909 to ensure that MDNodes
saved in the map used for metadata linking are either temporary or
resolved.

Also add a comment clarifying why we may need to save off non-MDNode
metadata.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256646 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agofix formatting; NFC
Sanjay Patel [Wed, 30 Dec 2015 18:31:30 +0000 (18:31 +0000)]
fix formatting; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256645 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRevert "[ptr-traits] Implement the base pointer traits using the actual"
Tobias Grosser [Wed, 30 Dec 2015 14:02:58 +0000 (14:02 +0000)]
Revert "[ptr-traits] Implement the base pointer traits using the actual"

The commit we revert is rather small, but it enables a larger piece of new
infrastructure that allows to detected misuses of pointer-traits at compile
time. Unfortunately, this change breaks with the use of incomplete types (e.g.
in Polly). As I am not aware of a simple fix on the Polly side, I temporarely
revert this commit to clean the bots and sync-up with Chandler how to best
adapt to these recent changes.

This reverts commit https://llvm.org/svn/llvm-project/llvm/trunk@256620.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256642 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Remove unnecessary conversion from StringRef to std::string when outputtin...
Craig Topper [Wed, 30 Dec 2015 06:00:24 +0000 (06:00 +0000)]
[TableGen] Remove unnecessary conversion from StringRef to std::string when outputting to a raw_ostream. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256628 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Remove raw_string_ostream by just emitting the header for the switch the...
Craig Topper [Wed, 30 Dec 2015 06:00:22 +0000 (06:00 +0000)]
[TableGen] Remove raw_string_ostream by just emitting the header for the switch the first time we emit a case. If the header was never emitted just print the default at the end. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256627 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Use range-based for loops. NFC
Craig Topper [Wed, 30 Dec 2015 06:00:20 +0000 (06:00 +0000)]
[TableGen] Use range-based for loops. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256626 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Move more things that come from variant into the AsmVariantInfo class...
Craig Topper [Wed, 30 Dec 2015 06:00:18 +0000 (06:00 +0000)]
[TableGen] Move more things that come from variant into the AsmVariantInfo class so we can reduce some parameters. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256625 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Use 'size_t' instead of 'unsigned' to better match the argument types...
Craig Topper [Wed, 30 Dec 2015 06:00:15 +0000 (06:00 +0000)]
[TableGen] Use 'size_t' instead of 'unsigned' to better match the argument types of addAsmOperand. Simplify some code by using StringRef::find instead of std::find. These were previously done in r247527 and r247528, but another commit seems to have erased them. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256624 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Implement the base pointer traits using the actual
Chandler Carruth [Wed, 30 Dec 2015 04:00:24 +0000 (04:00 +0000)]
[ptr-traits] Implement the base pointer traits using the actual
alignment of the pointee type!

This is the culmination of the ptr-traits work. Now the compiler will
catch me if I try to use a pointer to an empty struct as a key in
a dense map or inside a PointerIntPair or PointerUnion! This is much,
much better than sometimes corrupting data (and other times working
fine) due to insufficient alignment.

It also means that we will be much more diligent about rejecting other
uses of these constructs that aren't safe.

It also means that we can now be more aggressive with the constructs
when we actually have guaranteed higher alignment without specializing
stuff. I'll be going through and cleaning up all the current overrides
of these traits which are no longer necessary.

Many thanks to Richard, David, and others who helped me get all of this
together.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256620 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Refactor how PointerIntPair does its pointer manipulation
Chandler Carruth [Wed, 30 Dec 2015 03:56:19 +0000 (03:56 +0000)]
[ptr-traits] Refactor how PointerIntPair does its pointer manipulation
to isolate it in a dependent helper class.

Without doing this, we end up requiring all of the pointer traits the
moment you even define a PointerIntPair. That makes them *incredibly*
hard to use, for example you can't use them at all inside a class for
pointers to that class!

This change sinks all the logic into a helper template class that only
needs to be fully instantiated when *using* the PointerIntPair. We still
get compile-time checking, but it is deferred long enough to make
tradition out-of-line method definitions (or just the normal deferred
method body parsing) sufficient to handle cycling references.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256618 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Add one more #include necessary to do strict alignment
Chandler Carruth [Wed, 30 Dec 2015 03:56:17 +0000 (03:56 +0000)]
[ptr-traits] Add one more #include necessary to do strict alignment
checking of pointers used in PointerIntPairs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256617 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRename MDValue* to Metadata* (NFC)
Teresa Johnson [Tue, 29 Dec 2015 23:00:22 +0000 (23:00 +0000)]
Rename MDValue* to Metadata* (NFC)

Renamed MDValue* to Metadata*, and MDValueToValIDMap to MetadataToIDs,
as per review for r255909.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256593 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[RS4GC] Use DenseMap::count() instead of DenseMap::find()/DenseMap::end(). NFC.
Manuel Jacob [Tue, 29 Dec 2015 22:16:41 +0000 (22:16 +0000)]
[RS4GC] Use DenseMap::count() instead of DenseMap::find()/DenseMap::end().  NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256586 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agodon't repeat function names in comments; NFC
Sanjay Patel [Tue, 29 Dec 2015 22:11:50 +0000 (22:11 +0000)]
don't repeat function names in comments; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256584 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agouse auto with dyn_casted values; NFC
Sanjay Patel [Tue, 29 Dec 2015 22:00:37 +0000 (22:00 +0000)]
use auto with dyn_casted values; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256581 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[PlaceSafepoints] Assert that the gc.safepoint_poll function is present in the module.
Manuel Jacob [Tue, 29 Dec 2015 21:57:55 +0000 (21:57 +0000)]
[PlaceSafepoints] Assert that the gc.safepoint_poll function is present in the module.

If running the PlaceSafepoints pass on a module which doesn't have the
gc.safepoint_poll function without disabling entry and backedge safepoints,
previously the pass crashed with an obscure error because of a null pointer.
Now it fails the assert instead.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256580 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agouse auto with dyn_casted values; NFC
Sanjay Patel [Tue, 29 Dec 2015 21:49:08 +0000 (21:49 +0000)]
use auto with dyn_casted values; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256579 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agofix formatting; NFC
Sanjay Patel [Tue, 29 Dec 2015 19:34:53 +0000 (19:34 +0000)]
fix formatting; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256574 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agouse range-based for-loops; NFCI
Sanjay Patel [Tue, 29 Dec 2015 19:14:23 +0000 (19:14 +0000)]
use range-based for-loops; NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256573 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agouse range-based for-loop; NFCI
Sanjay Patel [Tue, 29 Dec 2015 18:30:09 +0000 (18:30 +0000)]
use range-based for-loop; NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256572 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAdd command line options to force function/loop alignments.
Chad Rosier [Tue, 29 Dec 2015 18:18:07 +0000 (18:18 +0000)]
Add command line options to force function/loop alignments.

These are being added for testing purposes.
http://reviews.llvm.org/D15648

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256571 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agodon't repeat function names in comments; NFC
Sanjay Patel [Tue, 29 Dec 2015 18:14:06 +0000 (18:14 +0000)]
don't repeat function names in comments; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256569 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[JumpThreading] Fix opcode bonus in getJumpThreadDuplicationCost()
Geoff Berry [Tue, 29 Dec 2015 18:10:16 +0000 (18:10 +0000)]
[JumpThreading] Fix opcode bonus in getJumpThreadDuplicationCost()

The code that was meant to adjust the duplication cost based on the
terminator opcode was not being executed in cases where the initial
threshold was hit inside the loop.

Subscribers: mcrosier, llvm-commits

Differential Revision: http://reviews.llvm.org/D15536

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256568 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agouse range-based for-loops; NFCI
Sanjay Patel [Tue, 29 Dec 2015 17:15:22 +0000 (17:15 +0000)]
use range-based for-loops; NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256566 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[MemoryBuiltins] Delete dead code [NFC]
Philip Reames [Tue, 29 Dec 2015 17:04:43 +0000 (17:04 +0000)]
[MemoryBuiltins] Delete dead code [NFC]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256565 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AVX512] add PSRLW Intrinsic
Michael Zuckerman [Tue, 29 Dec 2015 14:34:58 +0000 (14:34 +0000)]
[AVX512] add PSRLW Intrinsic

Fixing tab/space indentation.
Differential Revision: http://reviews.llvm.org/D15751

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256561 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[llvm-objdump] Mark noreturn function as such.
Davide Italiano [Tue, 29 Dec 2015 13:41:02 +0000 (13:41 +0000)]
[llvm-objdump] Mark noreturn function as such.

Match attribute in the header to make MSVC happy.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256560 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AVX512] add PSRLW Intrinsic
Michael Zuckerman [Tue, 29 Dec 2015 13:04:35 +0000 (13:04 +0000)]
[AVX512] add PSRLW Intrinsic

Differential Revision: http://reviews.llvm.org/D15751

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256558 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ADT] Teach alignment helpers to work correctly for abstract classes.
Chandler Carruth [Tue, 29 Dec 2015 09:52:41 +0000 (09:52 +0000)]
[ADT] Teach alignment helpers to work correctly for abstract classes.
This is necessary to use them as part of pointer traits and is generally
useful. I've added unit test coverage to isolate and ensure this works
correctly.

I'll watch the build bots to try to see if any compilers can't tolerate
this bit of magic (and much credit goes to Richard Smith for coming up
with this magical production!) but give a shout if you see issues.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256553 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Provide a real MCFragment address for the sentinel instead
Chandler Carruth [Tue, 29 Dec 2015 09:32:18 +0000 (09:32 +0000)]
[ptr-traits] Provide a real MCFragment address for the sentinel instead
of casting the integer '4' to such a pointer. There is no reason to
expect '4' to be a portable or reliable pointer of this form. The only
reason this ever worked is because the PointerIntPair that this actually
gets used with has an artificially *low* presumed alignment that allowed
it to work. When the alignment of PointerIntPair is derived from the
actual type's alignment, the asserts start firing on this pointer. I'm
amazed we never managed to do anything that triggered the alignment
sanitizer with it, as this is just flat out UB.

If folks dislike this approach to providing a sentinel fragment address,
there are a myriad of other alternatives, suggestions welcome. But this
one has the distinct advantage of not requiring the friend dance of
ilist's sentinel (which I'll point out is *also* in play for
MCFragment!) and seems to be using a nicely provided facility in
MCFragment to establish just such dummy nodes.

This is part of a series of patches to allow LLVM to check for complete
pointee types when computing its pointer traits. This is absolutely
necessary to get correct (or reproducible) results for things like how
many low bits are guaranteed to be zero.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256552 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Sink several in-body method definitions to be out-of-line
Chandler Carruth [Tue, 29 Dec 2015 09:24:42 +0000 (09:24 +0000)]
[ptr-traits] Sink several in-body method definitions to be out-of-line
inline definitions after the mutually recursive pair of types have been
defined. The two types mutually recurse specifically through
abstractions that require pointer traits which makes this kind of mutual
recursion especially tricky to get right in terms of ordering.

This is part of a series of patches to allow LLVM to check for complete
pointee types when computing its pointer traits. This is absolutely
necessary to get correct (or reproducible) results for things like how
many low bits are guaranteed to be zero.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256551 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Sink a constructor definition to the .cpp file and add
Chandler Carruth [Tue, 29 Dec 2015 09:24:39 +0000 (09:24 +0000)]
[ptr-traits] Sink a constructor definition to the .cpp file and add
missing includes so that the pointee types for DenseMap pointer keys and
such are complete prior to us querying the pointer traits for them.

This is part of a series of patches to allow LLVM to check for complete
pointee types when computing its pointer traits. This is absolutely
necessary to get correct (or reproducible) results for things like how
many low bits are guaranteed to be zero.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256550 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Add a bunch of includes to provide complete types that are
Chandler Carruth [Tue, 29 Dec 2015 09:06:21 +0000 (09:06 +0000)]
[ptr-traits] Add a bunch of includes to provide complete types that are
used in pointer dense map key types or in other ways that require
pointer traits.

This is part of a series of patches to allow LLVM to check for complete
pointee types when computing its pointer traits. This is absolutely
necessary to get correct (or reproducible) results for things like how
many low bits are guaranteed to be zero.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256549 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Split the MCFragment type hierarchy out of the MCAssembler
Chandler Carruth [Tue, 29 Dec 2015 09:06:16 +0000 (09:06 +0000)]
[ptr-traits] Split the MCFragment type hierarchy out of the MCAssembler
header to its own header, allowing users of fragments to have a narrower
header file, and avoid circular header dependencies when getting the
definition of MCSection prior to inspecting traits on MCSection
pointers.

This is part of a series of patches to allow LLVM to check for complete
pointee types when computing its pointer traits. This is absolutely
necessary to get correct (or reproducible) results for things like how
many low bits are guaranteed to be zero.

Note that this doesn't in any way change the design of MC, it is just
moving code around to allow the *header files* to be more fine grained.
Without this, it is impossible to get a complete type for MCSection
where it is needed.

If anyone would prefer a different slicing of the header files, I'm
happy to oblige of course. =]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256548 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[MC] Fix the comment header for this header file. It was just
Chandler Carruth [Tue, 29 Dec 2015 09:06:10 +0000 (09:06 +0000)]
[MC] Fix the comment header for this header file. It was just
copy/pasted.

Happy for anyone to suggest a more precise or refined set of boilerplate
here, but the comments on the actual code seem descriptive and accurate.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256547 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDe-virtualize mnemonicIsValid and remove from the base class. It's not called by...
Craig Topper [Tue, 29 Dec 2015 07:43:03 +0000 (07:43 +0000)]
De-virtualize mnemonicIsValid and remove from the base class. It's not called by any common code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256544 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Remove MnemonicContainsDot from AsmParser. It isn't used. NFC
Craig Topper [Tue, 29 Dec 2015 07:03:30 +0000 (07:03 +0000)]
[TableGen] Remove MnemonicContainsDot from AsmParser. It isn't used. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256542 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Remove declaration of ATTAsmParser. Its equivalent to the DefaultAsmParser...
Craig Topper [Tue, 29 Dec 2015 07:03:27 +0000 (07:03 +0000)]
[X86] Remove declaration of ATTAsmParser. Its equivalent to the DefaultAsmParser. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256541 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Add missing space to output.
Craig Topper [Tue, 29 Dec 2015 07:03:25 +0000 (07:03 +0000)]
[TableGen] Add missing space to output.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256540 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TableGen] Use range-based for loops. NFC
Craig Topper [Tue, 29 Dec 2015 07:03:23 +0000 (07:03 +0000)]
[TableGen] Use range-based for loops. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256539 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TrailingObjects] Dynamically realign under-aligned trailing objects.
James Y Knight [Tue, 29 Dec 2015 04:00:43 +0000 (04:00 +0000)]
[TrailingObjects] Dynamically realign under-aligned trailing objects.

Previously, the code enforced non-decreasing alignment of each trailing
type. However, it's easy enough to allow for realignment as needed, and
thus avoid the developer having to think about the possiblilities for
alignment requirements on all architectures.

(E.g. on Linux/x86, a struct with an int64 member is 4-byte aligned,
while on other 32-bit archs -- and even with other OSes on x86 -- it has
8-byte alignment. This sort of thing is irritating to have to manually
deal with.)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256533 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix gold test after r256465.
James Y Knight [Tue, 29 Dec 2015 03:48:37 +0000 (03:48 +0000)]
Fix gold test after r256465.

That commit added a new pass, and this test is sensitive to what the
first pass after verify is called.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256532 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ptr-traits] Merge the MetadataTracking helpers into the Metadata
Chandler Carruth [Tue, 29 Dec 2015 02:14:50 +0000 (02:14 +0000)]
[ptr-traits] Merge the MetadataTracking helpers into the Metadata
header.

This is part of a series of patches to allow LLVM to check for complete
pointee types when computing its pointer traits. This is absolutely
necessary to get correct (or reproducible) results for things like how
many low bits are guaranteed to be zero.

The MetadataTracking helpers aren't actually independent. They rely on
constructing a PointerUnion between Metadata and MetadataAsValue
pointers, which requires know the alignment of pointers to those types
which requires them to be complete.

The .cpp file even defined a method declared in Metadata.h! These really
don't seem like something that is separable, and there is no real
layering problem with just placing them together.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256531 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ADT] Run clang-format over another file before I try to enhance it so
Chandler Carruth [Tue, 29 Dec 2015 02:14:45 +0000 (02:14 +0000)]
[ADT] Run clang-format over another file before I try to enhance it so
I don't commit the bug in r256518 again.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256530 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ADT] Use a nonce type with at least 4 byte alignment.
Chandler Carruth [Tue, 29 Dec 2015 00:03:24 +0000 (00:03 +0000)]
[ADT] Use a nonce type with at least 4 byte alignment.

We didn't actually statically check this, and so it worked 25% of the
time for me. =/ Really sorry it took so long to fix, I shouldn't leave
the commit log editor window open without saving and landing the commit.
=[

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256528 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAccept dwarf version 5 for CIE versions.
Eric Christopher [Mon, 28 Dec 2015 23:02:42 +0000 (23:02 +0000)]
Accept dwarf version 5 for CIE versions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256527 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Thumb] Fix assembler error 'cannot honor width suffix pop {lr}'
Artyom Skrobov [Mon, 28 Dec 2015 21:40:45 +0000 (21:40 +0000)]
[Thumb] Fix assembler error 'cannot honor width suffix pop {lr}'

Summary:
* avoid generating POP {LR} in Thumb1 epilogues
* combine MOV LR, Rx + BX LR -> BX Rx in a peephole optimization pass
* combine POP {LR} + B + BX LR -> POP {PC} on v5T+

Test cases by Ana Pazos

Differential Revision: http://reviews.llvm.org/D15707

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256523 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[x86] lower calls to fmin and llvm.minnum.* using minss/minsd/minps/minpd (PR24475)
Sanjay Patel [Mon, 28 Dec 2015 21:16:55 +0000 (21:16 +0000)]
[x86] lower calls to fmin and llvm.minnum.* using minss/minsd/minps/minpd (PR24475)

This is a follow-on to:
http://reviews.llvm.org/rL255700
http://reviews.llvm.org/rL256454
http://reviews.llvm.org/rL256510

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256522 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRefactor inline costs analysis by removing the InlineCostAnalysis class
Easwaran Raman [Mon, 28 Dec 2015 20:28:19 +0000 (20:28 +0000)]
Refactor inline costs analysis by removing the InlineCostAnalysis class

InlineCostAnalysis is an analysis pass without any need for it to be one.
Once it stops being an analysis pass, it doesn't maintain any useful state
and the member functions inside can be made free functions. NFC.

Differential Revision: http://reviews.llvm.org/D15701

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256521 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[RS4GC] Fix rematerialization of bitcast of bitcast.
Manuel Jacob [Mon, 28 Dec 2015 20:14:05 +0000 (20:14 +0000)]
[RS4GC] Fix rematerialization of bitcast of bitcast.

Summary:
Previously, only the outer (last) bitcast was rematerialized, resulting in a
use of the unrelocated inner (first) bitcast after the statepoint.  See the
test case for an example.

Reviewers: igor-laevsky, reames

Subscribers: reames, alex, llvm-commits, sanjoy

Differential Revision: http://reviews.llvm.org/D15789

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256520 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoImplemented cost model for masked gather and scatter operations
Elena Demikhovsky [Mon, 28 Dec 2015 20:10:59 +0000 (20:10 +0000)]
Implemented cost model for masked gather and scatter operations
The cost is calculated for all X86 targets. When gather/scatter instruction
is not supported we calculate the cost of scalar sequence.

Differential revision: http://reviews.llvm.org/D15677

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256519 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ADT] Don't use a fixture just to get a nonce type for this unittest.
Chandler Carruth [Mon, 28 Dec 2015 20:03:16 +0000 (20:03 +0000)]
[ADT] Don't use a fixture just to get a nonce type for this unittest.

Instead, actually produce a nonce type in the test and use that. This
makes the test, IMO, both simpler and more clear.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256518 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAs a prelude to the rest of DWARF5 being supported go ahead and accept
Eric Christopher [Mon, 28 Dec 2015 19:58:36 +0000 (19:58 +0000)]
As a prelude to the rest of DWARF5 being supported go ahead and accept
version 5 in llvm-dwarfdump.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256515 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ADT] Run clang-format over more code, fixing numerous formatting
Chandler Carruth [Mon, 28 Dec 2015 19:54:36 +0000 (19:54 +0000)]
[ADT] Run clang-format over more code, fixing numerous formatting
oddities.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256514 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ADT] Run clang-format over PointerUnion.h and clean up its comments.
Chandler Carruth [Mon, 28 Dec 2015 19:52:46 +0000 (19:52 +0000)]
[ADT] Run clang-format over PointerUnion.h and clean up its comments.
NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256513 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[vim] Add token type to Vim syntax file.
Manuel Jacob [Mon, 28 Dec 2015 19:51:04 +0000 (19:51 +0000)]
[vim] Add token type to Vim syntax file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256512 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[x86] lower calls to fmax and llvm.maxnum.* using maxps/maxpd (PR24475)
Sanjay Patel [Mon, 28 Dec 2015 19:20:19 +0000 (19:20 +0000)]
[x86] lower calls to fmax and llvm.maxnum.* using maxps/maxpd (PR24475)

This is a follow-on to:
http://reviews.llvm.org/rL255700
http://reviews.llvm.org/rL256454

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256510 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoSpecify triple so 'make check' passes on darwin x86-64
Sanjay Patel [Mon, 28 Dec 2015 18:28:44 +0000 (18:28 +0000)]
Specify triple so 'make check' passes on darwin x86-64

The check lines were added with:
http://reviews.llvm.org/rL256458
http://reviews.llvm.org/rL256460

but on a darwin target, the output looks like:
  ## InlineAsm Start
  rorq  %rdi
  ## InlineAsm End
  ## InlineAsm Start
  rorq  %rsi
  ## InlineAsm End
  leaq  (%rsi,%rdi), %rax
  retq

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256507 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agotidy up; NFC
Sanjay Patel [Mon, 28 Dec 2015 18:18:22 +0000 (18:18 +0000)]
tidy up; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256506 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoSupport clrex instruction on ARMv6k. Patch by Andrew Turner.
Roman Divacky [Mon, 28 Dec 2015 17:47:23 +0000 (17:47 +0000)]
Support clrex instruction on ARMv6k. Patch by Andrew Turner.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256505 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRefactor: Simplify boolean conditional return statements in lib/Transforms/ObjCARC
Alexander Kornienko [Mon, 28 Dec 2015 16:19:08 +0000 (16:19 +0000)]
Refactor: Simplify boolean conditional return statements in lib/Transforms/ObjCARC

Summary: Use clang-tidy to simplify boolean conditional return statements

Reviewers: craig.topper, bkramer, chandlerc, gottesmm

Subscribers: llvm-commits

Patch by Richard Thomson!

Differential Revision: http://reviews.llvm.org/D9999

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256502 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRefactor: Simplify boolean conditional return statements in llvm/lib/Support
Alexander Kornienko [Mon, 28 Dec 2015 15:46:15 +0000 (15:46 +0000)]
Refactor: Simplify boolean conditional return statements in llvm/lib/Support

Summary: Use clang-tidy to simplify boolean conditional return statements

Reviewers: rafael, bkramer, ddunbar, Bigcheese, chandlerc, chapuni, nicholas, alexfh

Subscribers: alexfh, craig.topper, llvm-commits

Patch by Richard Thomson!

Differential Revision: http://reviews.llvm.org/D9978

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256500 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Better support for the MCU psABI (LLVM part)
Michael Kuperstein [Mon, 28 Dec 2015 14:39:21 +0000 (14:39 +0000)]
[X86] Better support for the MCU psABI (LLVM part)

This adds support for the MCU psABI in a way different from r251223 and r251224,
basically reverting most of these two patches. The problem with the approach
taken in r251223/4 is that it only handled libcalls that originated from the backend.
However, the mid-end also inserts quite a few libcalls and assumes these use the
platform's default calling convention.

The previous patch tried to insert inregs when necessary both in the FE and,
somewhat hackily, in the CG. Instead, we now define a new default calling convention
for the MCU, which doesn't use inreg marking at all, similarly to what x86-64 does.

Differential Revision: http://reviews.llvm.org/D15054

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256494 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRefactor: Simplify boolean conditional return statements in lib/Target/PowerPC
Alexander Kornienko [Mon, 28 Dec 2015 13:38:42 +0000 (13:38 +0000)]
Refactor: Simplify boolean conditional return statements in lib/Target/PowerPC

Summary: Use clang-tidy to simplify boolean conditional return statements

Reviewers: uweigand, rafael, wschmidt

Subscribers: craig.topper, llvm-commits

Patch by Richard Thomson!

Differential Revision: http://reviews.llvm.org/D9984

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256493 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86][AVX512] Lower broadcast sub vector to vector inrtrinsics
Asaf Badouh [Mon, 28 Dec 2015 08:26:26 +0000 (08:26 +0000)]
[X86][AVX512] Lower broadcast sub vector to vector inrtrinsics
lower broadcast<type>x<vector> to shuffles.
 there are two cases:
1.src is 128 bits and dest is 512 bits: in this case we will lower it to shuffle with imm = 0.
2.src is 256 bit and dest is 512 bits: in this case we will lower it to shuffle with imm = 01000100b (0x44) that way we will broadcast the 256bit source: ymm[0,1,2,3] => zmm[0,1,2,3,0,1,2,3] then it will mask it with the passthru value (in case it's mask op).

Differential Revision: http://reviews.llvm.org/D15790

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256490 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86][AVX512] add fp scalar broadcast intrinsics
Asaf Badouh [Mon, 28 Dec 2015 08:09:25 +0000 (08:09 +0000)]
[X86][AVX512] add fp scalar broadcast intrinsics

Differential Revision: http://reviews.llvm.org/D15790

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256489 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AVX512] Remove VEX_LIG from vmovd/vmovq instructions. From what I can tell from...
Craig Topper [Mon, 28 Dec 2015 06:32:47 +0000 (06:32 +0000)]
[AVX512] Remove VEX_LIG from vmovd/vmovq instructions. From what I can tell from the Intel docs these instructions require the L-bit to be 0.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256486 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AVX512] Fix some places that used FR64 instead of FR64X.
Craig Topper [Mon, 28 Dec 2015 06:11:45 +0000 (06:11 +0000)]
[AVX512] Fix some places that used FR64 instead of FR64X.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256484 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AVX512] Bring vmovq instructions names into alignment with the AVX and SSE names...
Craig Topper [Mon, 28 Dec 2015 06:11:42 +0000 (06:11 +0000)]
[AVX512] Bring vmovq instructions names into alignment with the AVX and SSE names. Add a missing encoding to disassembler and assembler.

I believe this also fixes a case where a 64-bit memory form that is documented as being unsupported in 32-bit mode was able to be selected there.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256483 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Move address for store target from outs to ins on a couple instructions.
Craig Topper [Mon, 28 Dec 2015 06:11:39 +0000 (06:11 +0000)]
[X86] Move address for store target from outs to ins on a couple instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256482 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[X86] Add proper Uses/Defs/mayLoad flags for AAA/AAD/AAM/AAS/DAA/DAS/XLAT instructions.
Craig Topper [Mon, 28 Dec 2015 06:11:37 +0000 (06:11 +0000)]
[X86] Add proper Uses/Defs/mayLoad flags for AAA/AAD/AAM/AAS/DAA/DAS/XLAT instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256481 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[lcg] Fix a few more formatting goofs found by clang-format. NFC.
Chandler Carruth [Mon, 28 Dec 2015 01:54:20 +0000 (01:54 +0000)]
[lcg] Fix a few more formatting goofs found by clang-format. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256480 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[lcg] Fix formatting errors found with clang-format, remove the now
Chandler Carruth [Mon, 28 Dec 2015 01:54:18 +0000 (01:54 +0000)]
[lcg] Fix formatting errors found with clang-format, remove the now
optional '\brief' tag and reflow some comments based on the added
horizontal space. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256479 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AVX512] Remove separate instruction and patterns for lowering ctlz_zero_undef. Chang...
Craig Topper [Sun, 27 Dec 2015 21:33:50 +0000 (21:33 +0000)]
[AVX512] Remove separate instruction and patterns for lowering ctlz_zero_undef. Change the operation for CTLZ_ZERO_UNDEF to Expand so SelectionDAG will convert them to CTLZ before lowering.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256477 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[SelectionDAG] Teach LegalizeVectorOps to not unroll CTLZ_ZERO_UNDEF and CTTZ_ZERO_UN...
Craig Topper [Sun, 27 Dec 2015 21:33:47 +0000 (21:33 +0000)]
[SelectionDAG] Teach LegalizeVectorOps to not unroll CTLZ_ZERO_UNDEF and CTTZ_ZERO_UNDEF if the non-ZERO_UNDEF form is legal or custom. Will be used to simplify X86 code in a follow on commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256476 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AVX512] Remove alternate data type versions of VALIGND, VALIGNQ, VMOVSHDUP and VMOVS...
Craig Topper [Sun, 27 Dec 2015 19:45:21 +0000 (19:45 +0000)]
[AVX512] Remove alternate data type versions of VALIGND, VALIGNQ, VMOVSHDUP and VMOVSLDUP. They don't have any tests and I don't think they can be selected. If they are truly needed they should be implemented with patterns against the normal instructions and not separate instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256475 91177308-0d34-0410-b5e6-96231b3b80d8