Matt Arsenault [Wed, 8 Mar 2017 01:06:58 +0000 (01:06 +0000)]
AMDGPU: Don't wait at end of block with a trivial successor
If there is only one successor, and that successor only
has one predecessor the wait can obviously be delayed until
uses or the end of the next block. This avoids code quality
regressions when there are trivial fallthrough blocks inserted
for structurization.
Sanjay Patel [Tue, 7 Mar 2017 23:27:14 +0000 (23:27 +0000)]
[InstCombine] shrink truncated insertelement into undef vector
This is the 2nd part of solving:
http://lists.llvm.org/pipermail/llvm-dev/2017-February/110293.html
D30123 moves the trunc ahead of the shuffle, and this moves the trunc ahead of the insertelement.
We're limiting this transform to undef rather than any constant to avoid backend problems.
Daniel Sanders [Tue, 7 Mar 2017 23:20:35 +0000 (23:20 +0000)]
Recommit: [globalisel] Change LLT constructor string into an LLT-based object that knows how to generate it.
Summary:
This will allow future patches to inspect the details of the LLT. The implementation is now split between
the Support and CodeGen libraries to allow TableGen to use this class without introducing layering concerns.
Thanks to Ahmed Bougacha for finding a reasonable way to avoid the layering issue and providing the version of this patch without that problem.
The problem with the previous commit appears to have been that TableGen was including CodeGen/LowLevelType.h instead of Support/LowLevelTypeImpl.h.
Fix one-after-the-end type metadata handling in globalsplit.
Itanium ABI may have an address point one byte after the end of a
vtable. When such vtable global is split, the !type metadata needs to
follow the right vtable.
Fix SmallPtrSet::iterator behaviour and creation ReverseIterate is true.
- Any function that creates an iterator now uses
SmallPtrSet::makeIterator, which creates an iterator that
dereferences to the given pointer.
- In reverse-iterate mode, initialze iterator::End with "CurArray"
instead of EndPointer.
- In reverse-iterate mode, the current node is iterator::Buffer[-1].
iterator::operator* and SmallPtrSet::makeIterator are the only ones
that need to know.
- Fix the assertions for reverse-iterate mode.
This fixes the tests Danny B added in r297182, and adds a couple of
others to confirm that dereferencing does the right thing, regardless of
how the iterator was found, and that iteration works correctly from each
return from find.
This was committed at r297155 and reverted at r297166 because of an
over-reaching clang test. That should be fixed with r297189.
This is one part of solving a recent bug report:
http://lists.llvm.org/pipermail/llvm-dev/2017-February/110293.html
This keeps with our general approach: changing arbitrary shuffles is off-limts,
but changing splat is ok. The transform is very similar to the existing
shrinkBitwiseLogic() canonicalization.
Chris Bieneman [Tue, 7 Mar 2017 21:34:35 +0000 (21:34 +0000)]
[ObjectYAML] Fix issue with DWARF2 AddrSize 8
In my refactoring I introduced a bug where we were using the reference size instead of the offset size for DW_FORM_strp and similar forms.
This patch resolves the error and adds a test case testing all the DWARF forms for DWARF2 AddrSize 8. There is similar coverage already in the DWARFDebugInfoTest sources that covers the parser. Once I migrate the DWARFGenerator APIs to be built on the YAML tools they will be fully covered under the same tests.
Tim Northover [Tue, 7 Mar 2017 21:24:33 +0000 (21:24 +0000)]
GlobalISel: fix legalization of G_INSERT
We were calculating incorrect extract/insert offsets by trying to be too
tricksy with min/max. It's clearer to just split the logic up into "register
starts before this segment" vs "after".
Gor Nishanov [Tue, 7 Mar 2017 21:00:54 +0000 (21:00 +0000)]
[coroutines] Add handling for unwind coro.ends
Summary:
The purpose of coro.end intrinsic is to allow frontends to mark the cleanup and
other code that is only relevant during the initial invocation of the coroutine
and should not be present in resume and destroy parts.
In landing pads coro.end is replaced with an appropriate instruction to unwind to
caller. The handling of coro.end differs depending on whether the target is
using landingpad or WinEH exception model.
For landingpad based exception model, it is expected that frontend uses the
`coro.end`_ intrinsic as follows:
```
The `CoroSpit` pass replaces `coro.end` with ``True`` in the resume functions,
thus leading to immediate unwind to the caller, whereas in start function it
is replaced with ``False``, thus allowing to proceed to the rest of the cleanup
code that is only needed during initial invocation of the coroutine.
For Windows Exception handling model, a frontend should attach a funclet bundle
referring to an enclosing cleanuppad as follows:
The `CoroSplit` pass, if the funclet bundle is present, will insert
``cleanupret from %tok unwind to caller`` before
the `coro.end`_ intrinsic and will remove the rest of the block.
Ahmed Bougacha [Tue, 7 Mar 2017 20:53:09 +0000 (20:53 +0000)]
[GlobalISel] Don't translate intrinsics with metadata parameters.
Some intrinsics take metadata parameters. These all need custom
handling of some form, and cannot possibly be lowered generically to
G_INTRINSIC calls with vreg operands.
Reject them, instead of hitting an assert later in getOrCreateVReg.
Ahmed Bougacha [Tue, 7 Mar 2017 20:53:06 +0000 (20:53 +0000)]
[GlobalISel] Avoid invalidating ValToVReg when translating no-op bitcast.
When we translate a no-op (same type) bitcast, we try to be clever and
only emit a COPY if we already assigned a vreg to the defined value.
However, when we didn't, we tried to assign to a reference into the
ValToVReg DenseMap, even though the RHS of the assignment
(getOrCreateVReg) could potentially grow that DenseMap, invalidating the
reference.
Avoid that by getting the source vreg first.
I audited the rest of the translator; this is the only tricky case.
The test is quite unwieldy, as the problem is caused by the DenseMap
growing, which happens after the 47th mapped value.
Ahmed Bougacha [Tue, 7 Mar 2017 20:53:03 +0000 (20:53 +0000)]
[GlobalISel] Relax vector G_SELECT assertion.
For vector operands, the `select` instruction supports both vector and
non-vector conditions. The MIR builder had an overly restrictive
assertion, that only accepted vector conditions for vector selects
(in effect implementing ISD::VSELECT).
Make it possible to express the full range of G_SELECTs.
Ahmed Bougacha [Tue, 7 Mar 2017 20:34:20 +0000 (20:34 +0000)]
[GlobalISel] Emit DBG_VALUE %noreg for non-int/fp constant values.
When a dbg_value has a constant operand that isn't representable in MI,
there isn't much we can do. Use %noreg (0) for those situations.
This matches the SelectionDAG behavior.
SjLjEHPrepare: Fix the pass for swifterror arguments
We cannot leave the identity copies 'select true, arg, undef' that this pass
inserts for arguments to simplify handling of values on swifterror arguments.
swifterror arguments have restrictions on their uses.
[cmake] Include openmp with add_llvm_external_project
Summary:
Include projects/openmp into the build using add_llvm_external_project
instead of add_subdirectory. This creates an option
LLVM_TOOL_OPENMP_BUILD that selects whether this project gets included
in an in-tree build.
Daniel Berlin [Tue, 7 Mar 2017 18:47:50 +0000 (18:47 +0000)]
Add PointerLikeTypeTraits for const things, as long as there is one for the non-const version. Clang and other users have a number of types they use as pointers, and this avoids having to define both const and non-const versions of PointerLikeTraits.
Daniel Berlin [Tue, 7 Mar 2017 18:47:48 +0000 (18:47 +0000)]
Make SmallPtrSet count and find able to take const PtrType's
Summary:
For our set/map types, count/find normally take const references.
This works well for non-pointer types, but can suck for pointer
types.
DenseSet<int *> foo;
const int *b = nullptr;
foo.count(b) does not work
but the equivalent reference version does work
(patch to fix DenseSet/DenseMap coming up)
For SmallPtrSet, you have no such option.
The following will not work right now:
SmallPtrSet<int *> foo;
const int *b = nullptr;
foo.count(b);
This makes const correctness hard in some cases.
Example:
SmallPtrSet<Instruction *> InstructionsToErase;
You can't make this SmallPtrSet<const Instruction *> because then you
can't erase the instruction. If I want to see if something is in the
set, I may only have a const Instruction *. Given that count and find
are non-mutating, this should just work.
The places in our code base that do this resort to const_cast :(.
This patch makes count and find able to be used with const Instruction
* in the above SmallPtrSet examples.
This is a bit annoying because of where C++ applies the const, so we
have to remove the pointer type from the passed-in-type and rebuild it
with const.
Matthew Simpson [Tue, 7 Mar 2017 18:47:30 +0000 (18:47 +0000)]
[LV] Consider users that are memory accesses in uniforms expansion step
When expanding the set of uniform instructions beyond the seed instructions
(e.g., consecutive pointers), we mark a new instruction uniform if all its
loop-varying users are uniform. We should also allow users that are consecutive
or interleaved memory accesses. This fixes cases where we have an instruction
that is used as the pointer operand of a consecutive access but also used by a
non-memory instruction that later becomes uniform as part of the expansion.
Sanjoy Das [Tue, 7 Mar 2017 18:47:22 +0000 (18:47 +0000)]
[X86] Add option to specify preferable loop alignment
Summary:
Loop alignment can cause a significant change of
the perfromance for short loops.
To be able to evaluate the impact of loop alignment this change
introduces the new option x86-experimental-pref-loop-alignment.
The alignment will be 2^Value bytes, the default value is 4.
Daniel Sanders [Tue, 7 Mar 2017 18:32:25 +0000 (18:32 +0000)]
[globalisel] Change LLT constructor string into an LLT-based object that knows how to generate it.
Summary:
This will allow future patches to inspect the details of the LLT. The implementation is now split between
the Support and CodeGen libraries to allow TableGen to use this class without introducing layering concerns.
Thanks to Ahmed Bougacha for finding a reasonable way to avoid the layering issue and providing the version of this patch without that problem.
Teresa Johnson [Tue, 7 Mar 2017 18:15:13 +0000 (18:15 +0000)]
Fix test and add missing return for llvm-lto2 error case
Summary:
This test was missing the target triple.
Once I fixed that, the case with the invalid character error stopped
returning 1 from llvm-lto2 and the test reported a failure. Fixed by
adding the missing return from llvm-lto2. Apparently we were failing
when we eventually tried to get the target.
Adrian Prantl [Tue, 7 Mar 2017 17:28:57 +0000 (17:28 +0000)]
Revert "Strip debug info when inlining into a nodebug function."
This reverts commit r296488.
As noted by David Blaikie on llvm-commits, I overlooked the case of a
debug function being inlined into a nodebug function being inlined
into a debug function.
Sanjay Patel [Tue, 7 Mar 2017 16:10:36 +0000 (16:10 +0000)]
[InstCombine] shrink truncated splat shuffle
This is one part of solving a recent bug report:
http://lists.llvm.org/pipermail/llvm-dev/2017-February/110293.html
This keeps with our general approach: changing arbitrary shuffles is off-limts,
but changing splat is ok. The transform is very similar to the existing
shrinkBitwiseLogic() canonicalization.
John Brawn [Tue, 7 Mar 2017 14:42:03 +0000 (14:42 +0000)]
[ARM] Correct handling of LSL #0 in an IT block
The check for LSL #0 in an IT block was checking if operand 4 was zero, but
operand 4 is the condition code operand so it was actually checking for LSLEQ.
Fix this by checking operand 3, which really is the immediate operand, and add
some tests.
Ranjeet Singh [Tue, 7 Mar 2017 11:17:53 +0000 (11:17 +0000)]
[ARM] Reapply r296865 "[ARM] fpscr read/write intrinsics not aware of each other""
The original patch r296865 was reverted as it broke the chromium builds for
Android https://bugs.llvm.org/show_bug.cgi?id=32134, this patch reapplies
r296865 with a fix to make sure it doesn't cause the build regression.
The problem was that intrinsic selection on int_arm_get_fpscr was failing in
ISel this was because the code to manually select this intrinsic still thought
it was the version with no side-effects (INTRINSIC_WO_CHAIN) which is wrong as
it doesn't semantically match the definition in the tablegen code which says it
does have side-effects, I've fixed this by updating the intrinsic type to
INTRINSIC_W_CHAIN (has side-effects). I've also added a test for this based on
Hans original reproducer.
Ayman Musa [Tue, 7 Mar 2017 08:11:19 +0000 (08:11 +0000)]
[X86][AVX512] Adding new LLVM TableGen backend which generates the EVEX2VEX compressing tables.
X86EvexToVex machine instruction pass compresses EVEX encoded instructions by replacing them with their identical VEX encoded instructions when possible.
It uses manually supported 2 large tables that map the EVEX instructions to their VEX ideticals.
This TableGen backend replaces the tables by automatically generating them.
Craig Topper [Tue, 7 Mar 2017 05:36:19 +0000 (05:36 +0000)]
[APInt] Add rvalue reference support to and, or, xor operations to allow their memory allocation to be reused when possible
This extends an earlier change that did similar for add and sub operations.
With this first patch we lose the fastpath for the single word case as operator&= and friends don't support it. This can be added there if we think that's important.
I had to change some functions in the APInt class since the operator overloads were moved out of the class and can't be used inside the class now. The getBitsSet change collides with another outstanding patch to implement it with setBits. But I didn't want to make this patch dependent on that series.
I've also removed the Or, And, Xor functions which were rarely or never used. I already commited two changes to remove the only uses of Or that existed.
Zachary Turner [Tue, 7 Mar 2017 03:43:17 +0000 (03:43 +0000)]
Use LLVM for all stat-related functionality.
This deletes LLDB's FileType enumeration and replaces all
users, and all calls to functions that check whether a file
exists etc with corresponding calls to LLVM.
Craig Topper [Tue, 7 Mar 2017 02:58:36 +0000 (02:58 +0000)]
[APInt] Add getBitsSetFrom and setBitsFrom to set upper bits starting at a bit
We currently have methods to set a specified number of low bits, a specified number of high bits, or a range of bits. But looking at some existing code it seems sometimes we want to set the high bits starting from a certain bit. Currently we do this with something like getHighBits(BitWidth, BitWidth - StartBit). Or once we start switching to setHighBits, setHighBits(BitWidth - StartBit) or setHighBits(getBitWidth() - StartBit).
Particularly for the latter case it would be better to have a convenience method like setBitsFrom(StartBit) so we don't need to mention the bit width that's already known to the APInt object.
I considered just making setBits have a default value of UINT_MAX for the hiBit argument and we would internally MIN it with the bit width. So if it wasn't specified it would be treated as bit width. This would require removing the assertion we currently have on the value of hiBit and may not be as readable.
Craig Topper [Tue, 7 Mar 2017 02:19:45 +0000 (02:19 +0000)]
[APInt] Implement getLowBitsSet/getHighBitsSet/getBitsSet using setLowBits/setHighBits/setBits
This patch implements getLowBitsSet/getHighBitsSet/getBitsSet in terms of the new setLowBits/setHighBits/setBits methods by making an all 0s APInt and then calling the appropriate set method.
This also adds support to setBits to allow loBits/hiBits to be in the other order to match with getBitsSet behavior.
For BitWidths larger than 64 this creates a short lived APInt with malloced storage. I think it might even call malloc twice. Its better to just provide methods that can set the necessary bits without the temporary APInt.
I'll update usages that benefit in a separate patch.
Reviewers: majnemer, MatzeB, davide, RKSimon, hans
Bob Wilson [Tue, 7 Mar 2017 00:51:07 +0000 (00:51 +0000)]
remove Cmake option for LLVM_DISABLE_ABI_BREAKING_CHECKS_ENFORCING
This is a follow-up to my change in r295090, which added support for
disabling these checks selectively based on setting the preprocessor
macro without relying on the Cmake setting. Swift has moved over to use
that approach, so we can clean up here and remove the Cmake setting.
Sanjay Patel [Mon, 6 Mar 2017 22:32:40 +0000 (22:32 +0000)]
[DAG] refactor related div/rem folds; NFCI
This is known incomplete and not called in the right order relative to
other folds, but that's the current behavior. I'm just trying to clean
this up before making actual functional changes to make the patch smaller.
The logic here should mimic the IR equivalents that are in InstSimplify's
simplifyDivRem().
Paul Robinson [Mon, 6 Mar 2017 22:20:03 +0000 (22:20 +0000)]
[DWARFv5] Update definitions to match published spec.
Some late additions to DWARF v5 were not in Dwarf.def; also one form
was redefined. Add the new cases to relevant switches in different
parts of LLVM. Replace DW_FORM_ref_sup with DW_FORM_ref_sup[4,8].
I did not add support for DW_FORM_strx3/addrx3 other that defining the
constants. We don't have any infrastructure to support these.
Fixed the asan bot failure which led to the last commit of the outliner being reverted.
The change is in lib/CodeGen/MachineOutliner.cpp in the SuffixTree's constructor. LeafVector
is no longer initialized using reserve but just a standard constructor.
Chad Rosier [Mon, 6 Mar 2017 21:20:00 +0000 (21:20 +0000)]
[AArch64][Redundant Copy Elim] Add support for CMN and shifted imm.
This patch extends the current functionality of the AArch64 redundant copy
elimination pass to handle CMN instructions as well as a shifted
immediates.
Summary: This patch refactors the DWARFYAML code for dumping compile units to use a visitor pattern. Using this design will, in the future, enable the DWARF YAML code to perform analysis and mutations of the DWARF DIEs. An example of such mutations would be calculating the length of a compile unit and updating the CU's Length field before writing the DIE. This support will make it easier to craft or modify DWARF tests by hand.
Daniel Berlin [Mon, 6 Mar 2017 20:01:31 +0000 (20:01 +0000)]
NewGVN: We were not really failing this testcase, because the instructions it was looking for are unused. GVN value numbers unused instructions, NewGVN does not. Fix the instructions to be used, so we eliminate the redundancies it's checking for, and un-XFAIL it
projects / llvm / log