Matthew Simpson [Fri, 16 Dec 2016 16:52:35 +0000 (16:52 +0000)]
[LV] Don't attempt to type-shrink scalarized instructions
After r288909, instructions feeding predicated instructions may be scalarized
if profitable. Since these instructions will remain scalar, we shouldn't
attempt to type-shrink them. We should only truncate vector types to their
minimal bit widths. This bug was exposed by enabling the vectorization of loops
containing conditional stores by default.
atomic_load_add returns the value before addition, but sets EFLAGS based on the
result of the addition. That means it's setting the flags based on effectively
subtracting C from the value at x, which is also what the outer cmp does.
This targets a pattern that occurs frequently with reference counting pointers:
David Blaikie [Fri, 16 Dec 2016 16:31:10 +0000 (16:31 +0000)]
dwarfdump: Support/process relocations on a CU's abbrev_off
Input can be produced by ld -r, for example (a normal LLVM workflow
never hits this - LLVM only ever produces a single abbrev table in an
object (shared by multiple CUs), so the reloc's always 0, and when it's
linked together the relocation's resolved so it doesn't need to be
handled)
Jun Bum Lim [Fri, 16 Dec 2016 16:03:31 +0000 (16:03 +0000)]
[CodeGenPrep] Skip merging empty case blocks
This is recommit of r287553 after fixing the invalid loop info after eliminating an empty block:
Summary: Merging an empty case block into the header block of switch could cause ISel to add COPY instructions in the header of switch, instead of the case block, if the case block is used as an incoming block of a PHI. This could potentially increase dynamic instructions, especially when the switch is in a loop. I added a test case which was reduced from the benchmark I was targetting.
Daniel Jasper [Fri, 16 Dec 2016 13:53:46 +0000 (13:53 +0000)]
Move VerifierSupport into namespace llvm.
It currently is in an unnamed namespace and then it shouldn't be used
from something in the header file. This actually triggers a warning with
GCC:
../include/llvm/IR/Verifier.h:39:7: warning: ‘llvm::TBAAVerifier’ has a field ‘llvm::TBAAVerifier::Diagnostic’ whose type uses the anonymous namespace [enabled by default]
Diana Picus [Fri, 16 Dec 2016 12:54:46 +0000 (12:54 +0000)]
[ARM] GlobalISel: Select add i32, i32
Add the minimal support necessary to select a function that returns the sum of
two i32 values.
This includes some support for argument/return lowering of i32 values through
registers, as well as the handling of copy and add instructions throughout the
GlobalISel pipeline.
Chandler Carruth [Fri, 16 Dec 2016 11:31:39 +0000 (11:31 +0000)]
Revert r289863: [LV] Enable vectorization of loops with conditional
stores by default
This uncovers a crasher in the loop vectorizer on PPC when building the
Python runtime. I'll send the testcase to the review thread for the
original commit.
Florian Hahn [Fri, 16 Dec 2016 11:10:26 +0000 (11:10 +0000)]
[codegen] Add generic functions to skip debug values.
Summary:
This commits moves skipDebugInstructionsForward and
skipDebugInstructionsBackward from lib/CodeGen/IfConversion.cpp
to include/llvm/CodeGen/MachineBasicBlock.h and updates
some codgen files to use them.
This refactoring was suggested in https://reviews.llvm.org/D27688
and I thought it's best to do the refactoring in a separate
review, but I could also put both changes in a single review
if that's preferred.
Also, the names for the functions aren't the snappiest and
I would be happy to rename them if anybody has suggestions.
Diana Picus [Fri, 16 Dec 2016 10:35:20 +0000 (10:35 +0000)]
[ARM] Expose methods to get the CCAssignFn. NFCI
Add two public methods to ARMTargetLowering: CCAssignFnForCall and
CCAssignFnForReturn, which are just calling the already existing private method
CCAssignFnForNode. These will come in handy for GlobalISel on ARM.
We also replace all calls to CCAssignFnForNode in ARMISelLowering.cpp, because
the new methods are friendlier to the reader.
Chandler Carruth [Fri, 16 Dec 2016 07:31:20 +0000 (07:31 +0000)]
Revert r289638: [PowerPC] Fix logic dealing with nop after calls (and tail-call eligibility)
This patch appears to result in trampolines in vtables being miscompiled
when they in turn tail call a method.
I've posted some preliminary details about the failure on the thread for
this commit and talked to Hal. He was comfortable going ahead and
reverting until we sort out what is wrong.
Mehdi Amini [Fri, 16 Dec 2016 06:29:14 +0000 (06:29 +0000)]
Extract a TBAAVerifier out of the verifier (NFC)
This is intended to be used (in a later patch) by the BitcodeReader
to detect invalid TBAA and drop them when loading bitcode, so that
we don't break client that have legacy bitcode with possible invalid
TBAA.
Update .debug_line section version information to match DWARF version.
One more attempt to re-commit the patch r285355, which I had to revert in r285362, because some tests were failing (the reason is because the size of the line_table varied depending on the full file name).
In the past the compiler always emitted .debug_line version 2, though some opcodes from DWARF 3 (e.g. DW_LNS_set_prologue_end, DW_LNS_set_epilogue_begin or DW_LNS_set_isa) and from DWARF 4 could be emitted by the compiler.
This patch changes version information of .debug_line to exactly match the DWARF version. For .debug_line version 4, a new field maximum_operations_per_instruction is emitted.
Adrian Prantl [Fri, 16 Dec 2016 04:25:54 +0000 (04:25 +0000)]
[IR] Remove the DIExpression field from DIGlobalVariable.
This patch implements PR31013 by introducing a
DIGlobalVariableExpression that holds a pair of DIGlobalVariable and
DIExpression.
Currently, DIGlobalVariables holds a DIExpression. This is not the
best way to model this:
(1) The DIGlobalVariable should describe the source level variable,
not how to get to its location.
(2) It makes it unsafe/hard to update the expressions when we call
replaceExpression on the DIGLobalVariable.
(3) It makes it impossible to represent a global variable that is in
more than one location (e.g., a variable with multiple
DW_OP_LLVM_fragment-s). We also moved away from attaching the
DIExpression to DILocalVariable for the same reasons.
This reapplies r289902 with additional testcase upgrades.
Teresa Johnson [Fri, 16 Dec 2016 04:11:51 +0000 (04:11 +0000)]
[ThinLTO] Thin link efficiency: More efficient export list computation
Summary:
Instead of checking whether a global referenced by a function being
imported is defined in the same module, speculatively always add the
referenced globals to the module's export list. After all imports are
computed, for each module prune any not in its defined set from its
export list.
For a huge C++ app with aggressive importing thresholds, even with
D27687 we spent a lot of time invoking modulePath() from
exportGlobalInModule (modulePath() was still the 2nd hottest routine in
profile). The reason is that with comdat/linkonce the summary lists for
each GUID can be long. For the app in question, for example, we were
invoking exportGlobalInModule almost 2 million times, and we traversed
an average of 63 entries in the summary list each time.
This patch reduced the thin link time for the app by about 10% (on top
of D27687) when using aggressive importing thresholds, and about 3.5% on
average with default importing thresholds.
Chandler Carruth [Fri, 16 Dec 2016 04:05:22 +0000 (04:05 +0000)]
Revert patch series introducing the DAG combine to match a load-by-bytes
idiom.
r289538: Match load by bytes idiom and fold it into a single load
r289540: Fix a buildbot failure introduced by r289538
r289545: Use more detailed assertion messages in the code ...
r289646: Add a couple of assertions to the load combine code ...
This DAG combine has a bad crash in it that is quite hard to trigger
sadly -- it relies on sneaking code with UB through the SDAG build and
into this particular combine. I've responded to the original commit with
a test case that reproduces it.
However, the code also has other problems that will require substantial
changes to address and so I'm going ahead and reverting it for now. This
should unblock us and perhaps others that are hitting the crash in the
wild and will let a fresh patch with updated approach come in cleanly
afterward.
Adrian Prantl [Fri, 16 Dec 2016 00:36:43 +0000 (00:36 +0000)]
[IR] Remove the DIExpression field from DIGlobalVariable.
This patch implements PR31013 by introducing a
DIGlobalVariableExpression that holds a pair of DIGlobalVariable and
DIExpression.
Currently, DIGlobalVariables holds a DIExpression. This is not the
best way to model this:
(1) The DIGlobalVariable should describe the source level variable,
not how to get to its location.
(2) It makes it unsafe/hard to update the expressions when we call
replaceExpression on the DIGLobalVariable.
(3) It makes it impossible to represent a global variable that is in
more than one location (e.g., a variable with multiple
DW_OP_LLVM_fragment-s). We also moved away from attaching the
DIExpression to DILocalVariable for the same reasons.
This pass prepares a module containing type metadata for ThinLTO by splitting
it into regular and thin LTO parts if possible, and writing both parts to
a multi-module bitcode file. Modules that do not contain type metadata are
written unmodified as a single module.
All globals with type metadata are added to the regular LTO module, and
the rest are added to the thin LTO module.
Teresa Johnson [Thu, 15 Dec 2016 23:50:06 +0000 (23:50 +0000)]
[ThinLTO] Thin link efficiency improvement: don't re-export globals (NFC)
Summary:
We were reinvoking exportGlobalInModule numerous times redundantly.
No need to re-export globals referenced by a global that was already
imported from its module. This resulted in a large speedup in the thin
link for a big application, particularly when importing aggressiveness
was cranked up.
Quentin Colombet [Thu, 15 Dec 2016 23:32:25 +0000 (23:32 +0000)]
[IRTranslator] Merge the entry and ABI lowering blocks.
The IRTranslator uses an additional block before the LLVM-IR entry block
to perform all the ABI lowering and the constant hoisting. Thus, this
block is the actual entry block and it falls through the LLVM-IR entry
block. However, with such representation, we end up with two basic
blocks that are not maximal.
Therefore, this patch adds a bit of canonicalization by merging both the
LLVM-IR entry block and the ABI lowering/constants hoisting into one
block, making the resulting block more likely to be maximal (indeed the
LLVM-IR entry block might not have been maximal).
Yichao Yu [Thu, 15 Dec 2016 22:36:53 +0000 (22:36 +0000)]
Fix R_AARCH64_MOVW_UABS_G3 relocation
Summary: The relocation is missing mask so an address that has non-zero bits in 47:43 may overwrite the register number. (Frequently shows up as target register changed to `xzr`....)
Teresa Johnson [Thu, 15 Dec 2016 21:39:42 +0000 (21:39 +0000)]
[ThinLTO] Revert part of r289843 that belonged to another patch.
The code change for D27687 accidentally got committed along with the
main change in r289843. Revert it temporarily, so that I can recommit it
along with its test as intended.
Eli Friedman [Thu, 15 Dec 2016 21:36:59 +0000 (21:36 +0000)]
Don't combine a shuffle of two BUILD_VECTORs with duplicate elements.
Targets can't handle this case well in general; we often transform
a shuffle of two cheap BUILD_VECTORs to element-by-element insertion,
which is very inefficient.
Sanjoy Das [Thu, 15 Dec 2016 21:23:44 +0000 (21:23 +0000)]
[Verifier] Allow TBAA metadata on atomicrmw and atomiccmpxchg
This used to be allowed before r289402 by default (before r289402 you
could have TBAA metadata on any instruction), and while I'm not sure
that it helps, it does sound reasonable enough to not fail the verifier
and we have out-of-tree users who use this.
Teresa Johnson [Thu, 15 Dec 2016 20:48:19 +0000 (20:48 +0000)]
[ThinLTO] Thin link efficiency: skip candidate added later with higher threshold (NFC)
Summary:
Thin link efficiency improvement. After adding an importing candidate to
the worklist we might have later added it again with a higher threshold.
Skip it when popped from the worklist if we recorded a higher threshold
than the current worklist entry, it will get processed again at the
higher threshold when that entry is popped.
This required adding the summary's GUID to the worklist, so that it can
be used to query the recorded highest threshold for it when we pop from the
worklist.
Andrea Di Biagio [Thu, 15 Dec 2016 20:01:26 +0000 (20:01 +0000)]
[SimplifyCFG] Merge debug locations when hoisting an instruction from a then/else branch. NFC.
Now that a new API to merge debug locations has been committed at r289661 (see
review D26256 for more details), we can use it to "improve" the code added by
revision r280995.
Instead of nulling the debugloc of a commoned instruction, we use the 'merged'
debug location. At the moment, this is just a no functional change since
function `DILocation::getMergedLocation()` is just a stub and would always
return a null location.
Geoff Berry [Thu, 15 Dec 2016 19:55:19 +0000 (19:55 +0000)]
[LiveRangeEdit] Change eliminateDeadDef assert to if condition.
The assert could potentially fire (though no cases have been
encountered), so just check that the instruction we're handling
specially for rematerialization only has one def to begin with.
LibDriver: Allow resource files to be archive members.
It seems pointless to add a resource to an archive because it won't have
any symbols to link against (and link.exe doesn't have an equivalent of
--whole-archive), but lib.exe allows it for some reason.
Sanjay Patel [Thu, 15 Dec 2016 19:13:37 +0000 (19:13 +0000)]
[InstCombine] add folds for icmp (smin X, Y), X
Min/max canonicalization (r287585) exposes the fact that we're missing combines for min/max patterns.
This patch won't solve the example that was attached to that thread, so something else still needs fixing.
The line between InstCombine and InstSimplify gets blurry here because sometimes the icmp instruction that
we want to fold to already exists, but sometimes it's the swapped form of what we want.
Corresponding changes for smax/umin/umax to follow.
Zachary Turner [Thu, 15 Dec 2016 19:02:43 +0000 (19:02 +0000)]
[StringRef] Add enable-if to StringLiteral.
to prevent StringLiteral from being created with a non-literal
char array, clang has a macro enable_if() that can be used
in such a way as to guarantee that the constructor is disabled
unless the length fo the string can be computed at compile time.
This only works on clang, but at least it should allow bots
to catch abuse of StringLiteral.
Ahmed Bougacha [Thu, 15 Dec 2016 18:45:30 +0000 (18:45 +0000)]
[GlobalISel] Drop workaround for Legalizer member/class sharing a name. NFC.
MachineLegalizer used to be the name of both the class and the member,
causing GCC errors. r276522 fixed that by renaming the member to just
'Legalizer'. The 'class' workaround isn't necessary anymore; drop it.
Teresa Johnson [Thu, 15 Dec 2016 18:21:01 +0000 (18:21 +0000)]
[ThinLTO] Ensure callees get hot threshold when first seen on cold path
This is split out from D27696, since it turned out to be a bug fix and
not part of the NFC efficiency change.
Keep the same adjusted (possibly decayed) threshold in both the worklist
and the ImportList. Otherwise if we encountered it first along a cold
path, the callee would be added to the worklist with a lower decayed
threshold than when it is later encountered along a hot path. But the
logic uses the threshold recorded in the ImportList entry to check if
we should re-add it, and without this patch the threshold recorded there
is the same along both paths so we don't re-add it. Using the
same possibly decayed threshold in the ImportList ensures we re-add it
later with the higher non-decayed hot path threshold.
Chris Bieneman [Thu, 15 Dec 2016 18:17:07 +0000 (18:17 +0000)]
[CMake] Minor change to symlink generation for LLDB
If OUTPUT_DIR is not specified we can assume the symlink is linking to a file in the same directory, so we can use $<TARGET_FILE_NAME:${target}> to create a relative symlink.
In the case of LLDB, when we build a framework, we are creating symlinks in a different directory than the file we're pointing to, and we don't install those links. To make this work in the build directory we can use $<TARGET_FILE:${target}> instead, which uses the full path to the target.
And this happens several times in the diffs. For chips with domain-crossing penalties,
the instruction count and size reduction should usually overcome any potential
domain-crossing penalty due to using an FP op in a sequence of int ops. For chips such
as recent Intel big cores and Atom, there is no domain-crossing penalty for shufps, so
using shufps is a pure win.
So the test case diffs all appear to be improvements except one test in
vector-shuffle-combining.ll where we miss an opportunity to use a shift to generate
zero elements and one test in combine-sra.ll where multiple uses prevent the expected
shuffle combining.
Jacques Pienaar [Thu, 15 Dec 2016 16:56:16 +0000 (16:56 +0000)]
[lanai] Simplify small section check in LowerGlobalAddress and treat ldata sections specially.
Move the check for the code model into isGlobalInSmallSectionImpl and return false (not in small section) for variables placed in sections prefixed with .ldata (workaround for a tool limitation).
Robert Lougher [Thu, 15 Dec 2016 16:17:53 +0000 (16:17 +0000)]
[SimplifyCFG] In sinkLastInstruction correctly set debugloc of "common" inst
Simplify CFG will try to sink the last instruction in a series of basic blocks,
creating a "common" instruction in the successor block (sinkLastInstruction).
When it does this, the debug location of the single instruction should be the
merged debug locations of the commoned instructions.
Specifically avoid implicit conversions from/to integral types to
avoid potential errors when changing the underlying type. For example,
a typical initialization of a "full" mask was "LaneMask = ~0u", which
would result in a value of 0x00000000FFFFFFFF if the type was extended
to uint64_t.
Ehsan Amiri [Thu, 15 Dec 2016 12:25:13 +0000 (12:25 +0000)]
[InstCombine] New opportunities for FoldAndOfICmp and FoldXorOfICmp
A number of new patterns for simplifying and/xor of icmp:
(icmp ne %x, 0) ^ (icmp ne %y, 0) => icmp ne %x, %y if the following is true:
1- (%x = and %a, %mask) and (%y = and %b, %mask)
2- %mask is a power of 2.
(icmp eq %x, 0) & (icmp ne %y, 0) => icmp ult %x, %y if the following is true:
1- (%x = and %a, %mask1) and (%y = and %b, %mask2)
2- Let %t be the smallest power of 2 where %mask1 & %t != 0. Then for any
%s that is a power of 2 and %s & %mask2 != 0, we must have %s <= %t.
For example if %mask1 = 24 and %mask2 = 16, setting %s = 16 and %t = 8
violates condition (2) above. So this optimization cannot be applied.
Nemanja Ivanovic [Thu, 15 Dec 2016 11:16:20 +0000 (11:16 +0000)]
[Power9] Allow AnyExt immediates for XXSPLTIB
In some situations, the BUILD_VECTOR node that builds a v18i8 vector by
a splat of an i8 constant will end up with signed 8-bit values and other
situations, it'll end up with unsigned ones. Handle both situations.
projects / llvm / log