Some of the AMDGPU stack addressing modes require knowing the sign
bit is zero. We used to accomplish this by custom lowering
frame indexes, and then putting an AssertZext around a
TargetFrameIndex. This required specifically looking for
the AssextZext + frame index pattern which was moderately
disgusting. The same could probably be accomplished
with a target specific node, but would still
require special handling of frame indexes.
llvm-svn: 317671
This patch enables the folding of address computation in
memory instruction in case adress is represented by Phi node.
The inputs of Phi node might be different in base register.
Differential Revision: https://reviews.llvm.org/D36073
llvm-svn: 317665
This header includes CodeGen headers, and is not, itself, included by
any Target headers, so move it into CodeGen to match the layering of its
implementation.
llvm-svn: 317647
Reland r317100 with minor fix regarding ComputeCommonTailLength function in
BranchFolding.cpp. Skipping top CFI instructions block needs to executed on
several more return points in ComputeCommonTailLength().
Original r317100 message:
"Correct dwarf unwind information in function epilogue for X86"
This patch aims to provide correct dwarf unwind information in function
epilogue for X86.
It consists of two parts. The first part inserts CFI instructions that set
appropriate cfa offset and cfa register in emitEpilogue() in
X86FrameLowering. This part is X86 specific.
The second part is platform independent and ensures that:
- CFI instructions do not affect code generation
- Unwind information remains correct when a function is modified by
different passes. This is done in a late pass by analyzing information
about cfa offset and cfa register in BBs and inserting additional CFI
directives where necessary.
Changed CFI instructions so that they:
- are duplicable
- are not counted as instructions when tail duplicating or tail merging
- can be compared as equal
Added CFIInstrInserter pass:
- analyzes each basic block to determine cfa offset and register valid at
its entry and exit
- verifies that outgoing cfa offset and register of predecessor blocks match
incoming values of their successors
- inserts additional CFI directives at basic block beginning to correct the
rule for calculating CFA
Having CFI instructions in function epilogue can cause incorrect CFA
calculation rule for some basic blocks. This can happen if, due to basic
block reordering, or the existence of multiple epilogue blocks, some of the
blocks have wrong cfa offset and register values set by the epilogue block
above them.
CFIInstrInserter is currently run only on X86, but can be used by any target
that implements support for adding CFI instructions in epilogue.
Patch by Violeta Vukobrat.
llvm-svn: 317579
This changes the interface of how targets describe how to legalize, see
the below description.
1. Interface for targets to describe how to legalize.
In GlobalISel, the API in the LegalizerInfo class is the main interface
for targets to specify which types are legal for which operations, and
what to do to turn illegal type/operation combinations into legal ones.
For each operation the type sizes that can be legalized without having
to change the size of the type are specified with a call to setAction.
This isn't different to how GlobalISel worked before. For example, for a
target that supports 32 and 64 bit adds natively:
for (auto Ty : {s32, s64})
setAction({G_ADD, 0, s32}, Legal);
or for a target that needs a library call for a 32 bit division:
setAction({G_SDIV, s32}, Libcall);
The main conceptual change to the LegalizerInfo API, is in specifying
how to legalize the type sizes for which a change of size is needed. For
example, in the above example, how to specify how all types from i1 to
i8388607 (apart from s32 and s64 which are legal) need to be legalized
and expressed in terms of operations on the available legal sizes
(again, i32 and i64 in this case). Before, the implementation only
allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0,
s128}, NarrowScalar). A worse limitation was that if you'd wanted to
specify how to legalize all the sized types as allowed by the LLVM-IR
LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times
and probably would need a lot of memory to store all of these
specifications.
Instead, the legalization actions that need to change the size of the
type are specified now using a "SizeChangeStrategy". For example:
setLegalizeScalarToDifferentSizeStrategy(
G_ADD, 0, widenToLargerAndNarrowToLargest);
This example indicates that for type sizes for which there is a larger
size that can be legalized towards, do it by Widening the size.
For example, G_ADD on s17 will be legalized by first doing WidenScalar
to make it s32, after which it's legal.
The "NarrowToLargest" indicates what to do if there is no larger size
that can be legalized towards. E.g. G_ADD on s92 will be legalized by
doing NarrowScalar to s64.
Another example, taken from the ARM backend is:
for (unsigned Op : {G_SDIV, G_UDIV}) {
setLegalizeScalarToDifferentSizeStrategy(Op, 0,
widenToLargerTypesUnsupportedOtherwise);
if (ST.hasDivideInARMMode())
setAction({Op, s32}, Legal);
else
setAction({Op, s32}, Libcall);
}
For this example, G_SDIV on s8, on a target without a divide
instruction, would be legalized by first doing action (WidenScalar,
s32), followed by (Libcall, s32).
The same principle is also followed for when the number of vector lanes
on vector data types need to be changed, e.g.:
setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal);
setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal);
setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal);
setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal);
setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal);
setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal);
setLegalizeVectorElementToDifferentSizeStrategy(
G_ADD, 0, widenToLargerTypesUnsupportedOtherwise);
As currently implemented here, vector types are legalized by first
making the vector element size legal, followed by then making the number
of lanes legal. The strategy to follow in the first step is set by a
call to setLegalizeVectorElementToDifferentSizeStrategy, see example
above. The strategy followed in the second step
"moreToWiderTypesAndLessToWidest" (see code for its definition),
indicating that vectors are widened to more elements so they map to
natively supported vector widths, or when there isn't a legal wider
vector, split the vector to map it to the widest vector supported.
Therefore, for the above specification, some example legalizations are:
* getAction({G_ADD, LLT::vector(3, 3)})
returns {WidenScalar, LLT::vector(3, 8)}
* getAction({G_ADD, LLT::vector(3, 8)})
then returns {MoreElements, LLT::vector(8, 8)}
* getAction({G_ADD, LLT::vector(20, 8)})
returns {FewerElements, LLT::vector(16, 8)}
2. Key implementation aspects.
How to legalize a specific (operation, type index, size) tuple is
represented by mapping intervals of integers representing a range of
size types to an action to take, e.g.:
setScalarAction({G_ADD, LLT:scalar(1)},
{{1, WidenScalar}, // bit sizes [ 1, 31[
{32, Legal}, // bit sizes [32, 33[
{33, WidenScalar}, // bit sizes [33, 64[
{64, Legal}, // bit sizes [64, 65[
{65, NarrowScalar} // bit sizes [65, +inf[
});
Please note that most of the code to do the actual lowering of
non-power-of-2 sized types is currently missing, this is just trying to
make it possible for targets to specify what is legal, and how non-legal
types should be legalized. Probably quite a bit of further work is
needed in the actual legalizing and the other passes in GlobalISel to
support non-power-of-2 sized types.
I hope the documentation in LegalizerInfo.h and the examples provided in the
various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well
enough how this is meant to be used.
This drops the need for LLT::{half,double}...Size().
Differential Revision: https://reviews.llvm.org/D30529
llvm-svn: 317560
This patch disables the handling of selects in optimization
extensing scope of optimizeMemoryInst.
The optimization itself is disable by default.
The idea here is just to switch optimiztion level step by step.
Specifically, first optimization will be enabled only for Phi nodes,
then select instructions will be added.
In case someone will complain about perfromance it will be easier to
detect what part of optimizations is responsible for that.
Differential Revision: https://reviews.llvm.org/D36073
llvm-svn: 317555
Summary:
Print %subreg.<subregidxname> instead of just the subregister
index when printing immediate operands corresponding to subreg
indices in INSERT_SUBREG, EXTRACT_SUBREG, SUBREG_TO_REG and
REG_SEQUENCE.
Reviewers: qcolombet, MatzeB
Reviewed By: MatzeB
Subscribers: nhaehnle, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D39696
llvm-svn: 317513
As discussed on llvm-dev:
http://lists.llvm.org/pipermail/llvm-dev/2016-November/107104.html
and again more recently:
http://lists.llvm.org/pipermail/llvm-dev/2017-October/118118.html
...this is a step in cleaning up our fast-math-flags implementation in IR to better match
the capabilities of both clang's user-visible flags and the backend's flags for SDNode.
As proposed in the above threads, we're replacing the 'UnsafeAlgebra' bit (which had the
'umbrella' meaning that all flags are set) with a new bit that only applies to algebraic
reassociation - 'AllowReassoc'.
We're also adding a bit to allow approximations for library functions called 'ApproxFunc'
(this was initially proposed as 'libm' or similar).
...and we're out of bits. 7 bits ought to be enough for anyone, right? :) FWIW, I did
look at getting this out of SubclassOptionalData via SubclassData (spacious 16-bits),
but that's apparently already used for other purposes. Also, I don't think we can just
add a field to FPMathOperator because Operator is not intended to be instantiated.
We'll defer movement of FMF to another day.
We keep the 'fast' keyword. I thought about removing that, but seeing IR like this:
%f.fast = fadd reassoc nnan ninf nsz arcp contract afn float %op1, %op2
...made me think we want to keep the shortcut synonym.
Finally, this change is binary incompatible with existing IR as seen in the
compatibility tests. This statement:
"Newer releases can ignore features from older releases, but they cannot miscompile
them. For example, if nsw is ever replaced with something else, dropping it would be
a valid way to upgrade the IR."
( http://llvm.org/docs/DeveloperPolicy.html#ir-backwards-compatibility )
...provides the flexibility we want to make this change without requiring a new IR
version. Ie, we're not loosening the FP strictness of existing IR. At worst, we will
fail to optimize some previously 'fast' code because it's no longer recognized as
'fast'. This should get fixed as we audit/squash all of the uses of 'isFast()'.
Note: an inter-dependent clang commit to use the new API name should closely follow
commit.
Differential Revision: https://reviews.llvm.org/D39304
llvm-svn: 317488
This is an implementation of PR26223.
Currently optimizeMemoryInst optimization tries to fold address computation
if all possible way to get compute the address are of the form
baseGV + base + scale * Index + offset
where scale and offset are constants and baseGV, base and Index are exactly
the same instructions if defined.
The patch extends this optimization to allow different bases. In this case
it tries to find/build a Phi node merging all possible bases and use this Phi node
as a base for sunk address computation. Also it supports Select instruction on
the way.
The main motivation for this scope extension is GCRelocateInst.
If there is a relocation of derived pointer it will be represented as relocation of base + offset.
Also there will be a Phi node merging address computation for relocated derived pointer
and derived pointer itself. If we have a Phi node merging original base and relocated base
and can fold the address computation of derived pointer then we can potentially reduce
the code size and Phi node for derived pointer. The later can have a positive impact to
register allocator.
Reviewers: efriedma, dberlin, mkazantsev, reames, john.brawn
Reviewed By: john.brawn
Subscribers: javed.absar, john.brawn, dneilson, llvm-commits
Differential Revision: https://reviews.llvm.org/D36073
llvm-svn: 317429
This header already includes a CodeGen header and is implemented in
lib/CodeGen, so move the header there to match.
This fixes a link error with modular codegeneration builds - where a
header and its implementation are circularly dependent and so need to be
in the same library, not split between two like this.
llvm-svn: 317379
This preserves the debug info for the cast operation in the original location.
rdar://problem/33460652
Reapplied r317340 with the test moved into an ARM-specific directory.
llvm-svn: 317375
DenseMaps require the definition of a type to be available when using a
pointer to that type as a key to know how many bits are available for
tombstone/etc.
llvm-svn: 317360
Make doSpillCalleeSavedRegs a member function, instead of passing most of the
members of PEI as arguments.
Differential Review: https://reviews.llvm.org/D35642
llvm-svn: 317309
mir-canon (MIRCanonicalizerPass) is a pass designed to reorder instructions and
rename operands so that two similar programs will diff more cleanly after being
run through mir-canon than they would otherwise. This project is still a work
in progress and there are ideas still being discussed for improving diff
quality.
M include/llvm/InitializePasses.h
M lib/CodeGen/CMakeLists.txt
M lib/CodeGen/CodeGen.cpp
A lib/CodeGen/MIRCanonicalizerPass.cpp
llvm-svn: 317285
Summary:
Currently the block frequency analysis is an approximation for irreducible
loops.
The new irreducible loop metadata is used to annotate the irreducible loop
headers with their header weights based on the PGO profile (currently this is
approximated to be evenly weighted) and to help improve the accuracy of the
block frequency analysis for irreducible loops.
This patch is a basic support for this.
Reviewers: davidxl
Reviewed By: davidxl
Subscribers: mehdi_amini, llvm-commits, eraman
Differential Revision: https://reviews.llvm.org/D39028
llvm-svn: 317278
undefined reference to `llvm::TargetPassConfig::ID' on
clang-ppc64le-linux-multistage
This reverts commit eea333c33fa73ad225ef28607795984829f65688.
llvm-svn: 317213
Summary:
This is mostly a noop (most of the test diffs are renamed blocks).
There are a few temporary register renames (eax<->ecx) and a few blocks are
shuffled around.
See the discussion in PR33325 for more details.
Reviewers: spatel
Subscribers: mgorny
Differential Revision: https://reviews.llvm.org/D39456
llvm-svn: 317211
When splitting a large load to smaller legally-typed loads, the last load should be padded to reach the size of the previous one so a CONCAT_VECTORS node could reunite them again.
The code currently pads the last load to reach the size of the first load (instead of the previous).
Differential Revision: https://reviews.llvm.org/D38495
Change-Id: Ib60b55ed26ce901fabf68108daf52683fbd5013f
llvm-svn: 317206
As of today we only use .cfi_offset to specify the offset of a CSR, but
we never use .cfi_restore when the CSR is restored.
If we want to perform a more advanced type of shrink-wrapping, we need
to use .cfi_restore in order to switch the CFI state between blocks.
This patch only aims at adding support for the directive.
Differential Revision: https://reviews.llvm.org/D36114
llvm-svn: 317199
This patch aims to provide correct dwarf unwind information in function
epilogue for X86.
It consists of two parts. The first part inserts CFI instructions that set
appropriate cfa offset and cfa register in emitEpilogue() in
X86FrameLowering. This part is X86 specific.
The second part is platform independent and ensures that:
- CFI instructions do not affect code generation
- Unwind information remains correct when a function is modified by
different passes. This is done in a late pass by analyzing information
about cfa offset and cfa register in BBs and inserting additional CFI
directives where necessary.
Changed CFI instructions so that they:
- are duplicable
- are not counted as instructions when tail duplicating or tail merging
- can be compared as equal
Added CFIInstrInserter pass:
- analyzes each basic block to determine cfa offset and register valid at
its entry and exit
- verifies that outgoing cfa offset and register of predecessor blocks match
incoming values of their successors
- inserts additional CFI directives at basic block beginning to correct the
rule for calculating CFA
Having CFI instructions in function epilogue can cause incorrect CFA
calculation rule for some basic blocks. This can happen if, due to basic
block reordering, or the existence of multiple epilogue blocks, some of the
blocks have wrong cfa offset and register values set by the epilogue block
above them.
CFIInstrInserter is currently run only on X86, but can be used by any target
that implements support for adding CFI instructions in epilogue.
Patch by Violeta Vukobrat.
Differential Revision: https://reviews.llvm.org/D35844
llvm-svn: 317100
Change the map key from DIFile* to the absolute path string. Computing
the absolute path isn't expensive because we already have a map that
caches the full path keyed on DIFile*.
llvm-svn: 317041
Issue found by llvm-isel-fuzzer on OSS fuzz, https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=3725
If anyone actually cares about > 64 bit arithmetic, there's a lot more to do in this area. There's a bunch of obviously wrong code in the same function. I don't have the time to fix all of them and am just using this to understand what the workflow for fixing fuzzer cases might look like.
llvm-svn: 316967
Summary:
For reference, see: http://lists.llvm.org/pipermail/llvm-dev/2017-August/116589.html
This patch fleshes out the instruction class hierarchy with respect to atomic and
non-atomic memory intrinsics. With this change, the relevant part of the class
hierarchy becomes:
IntrinsicInst
-> MemIntrinsicBase (methods-only class)
-> MemIntrinsic (non-atomic intrinsics)
-> MemSetInst
-> MemTransferInst
-> MemCpyInst
-> MemMoveInst
-> AtomicMemIntrinsic (atomic intrinsics)
-> AtomicMemSetInst
-> AtomicMemTransferInst
-> AtomicMemCpyInst
-> AtomicMemMoveInst
-> AnyMemIntrinsic (both atomicities)
-> AnyMemSetInst
-> AnyMemTransferInst
-> AnyMemCpyInst
-> AnyMemMoveInst
This involves some class renaming:
ElementUnorderedAtomicMemCpyInst -> AtomicMemCpyInst
ElementUnorderedAtomicMemMoveInst -> AtomicMemMoveInst
ElementUnorderedAtomicMemSetInst -> AtomicMemSetInst
A script for doing this renaming in downstream trees is included below.
An example of where the Any* classes should be used in LLVM is when reasoning
about the effects of an instruction (ex: aliasing).
---
Script for renaming AtomicMem* classes:
PREFIXES="[<,([:space:]]"
CLASSES="MemIntrinsic|MemTransferInst|MemSetInst|MemMoveInst|MemCpyInst"
SUFFIXES="[;)>,[:space:]]"
REGEX="(${PREFIXES})ElementUnorderedAtomic(${CLASSES})(${SUFFIXES})"
REGEX2="visitElementUnorderedAtomic(${CLASSES})"
FILES=$( grep -E "(${REGEX}|${REGEX2})" -r . | tr ':' ' ' | awk '{print $1}' | sort | uniq )
SED_SCRIPT="s~${REGEX}~\1Atomic\2\3~g"
SED_SCRIPT2="s~${REGEX2}~visitAtomic\1~g"
for f in $FILES; do
echo "Processing: $f"
sed -i ".bak" -E "${SED_SCRIPT};${SED_SCRIPT2};${EA_SED_SCRIPT};${EA_SED_SCRIPT2}" $f
done
Reviewers: sanjoy, deadalnix, apilipenko, anna, skatkov, mkazantsev
Reviewed By: sanjoy
Subscribers: hfinkel, jholewinski, arsenm, sdardis, nhaehnle, JDevlieghere, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D38419
llvm-svn: 316950
- Targets that want to support memcmp expansions now return the list of
supported load sizes.
- Expansion codegen does not assume that all power-of-two load sizes
smaller than the max load size are valid. For examples, this is not the
case for x86(32bit)+sse2.
Fixes PR34887.
llvm-svn: 316905
Introduce a isConstOrDemandedConstSplat helper function that can recognise a constant splat build vector for at least the demanded elts we care about.
llvm-svn: 316866
For cases where we know the floating point representations match the bitcasted integer equivalent, allow bitcasting to these types.
This is especially useful for the X86 floating point compare results which return all/zero bits but as a floating point type.
Differential Revision: https://reviews.llvm.org/D39289
llvm-svn: 316831
In function DAGCombiner::visitSIGN_EXTEND_INREG, sext can be combined with extload even if sextload is not supported by target, then
if sext is the only user of extload, there is no big difference, no harm no benefit.
if extload has more than one user, the combined sextload may block extload from combining with other zext, causes extra zext instructions generated. As demonstrated by the attached test case.
This patch add the constraint that when sextload is not supported by target, sext can only be combined with extload if it is the only user of extload.
Differential Revision: https://reviews.llvm.org/D39108
llvm-svn: 316802
Not having the subclass data on an MemIntrinsicSDNodes means it was possible
to try to fold 2 nodes with the same operands but differing MMO flags. This
would trip an assertion when trying to refine the alignment between the 2
MachineMemOperands.
Differential Revision: https://reviews.llvm.org/D38898
llvm-svn: 316737
Summary:
Currently we skip merging when extra moves may be added in the header of switch instead of the case block, if the case block is used as an incoming
block of a PHI. If all the incoming values of the PHIs are non-constants and the destination block is dominated by the switch block then extra moves are likely not added by ISel, so there is no need to skip merging in this case.
Reviewers: efriedma, junbuml, davidxl, hfinkel, qcolombet
Reviewed By: efriedma
Subscribers: dberlin, kuhar, mcrosier, llvm-commits
Differential Revision: https://reviews.llvm.org/D37343
llvm-svn: 316711
Summary:
This seems to be the only place in llvm we directly call qsort. We can replace
this with a call to array_pod_sort. Also minor cleanup of the sorting function.
Reviewers: bkramer, Eugene.Zelenko, rafael
Reviewed By: bkramer
Subscribers: efriedma, llvm-commits
Differential Revision: https://reviews.llvm.org/D39214
llvm-svn: 316671
Summary: Make sure shifts are legal/specified by the legalizerinfo before creating it
Reviewers: qcolombet, dsanders, rovka, t.p.northover
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39264
llvm-svn: 316602
Compute the actual decomposition only after deciding whether to expand
of not. Else, it's easy to make the compiler OOM with:
`memcpy(dst, src, 0xffffffffffffffff);`, which typically happens if
someone mistakenly passes a negative value. Add a test.
This reverts commit f8fc02fbd4ab33383c010d33675acf9763d0bd44.
llvm-svn: 316567
Duplicated code found in three places put into a new static function:
/// Given a Count of resource usage and a Latency value, return true if a
/// SchedBoundary becomes resource limited.
static bool checkResourceLimit(unsigned LFactor, unsigned Count,
unsigned Latency) {
return (int)(Count - (Latency * LFactor)) > (int)LFactor;
}
Review: Florian Hahn, Matthias Braun
https://reviews.llvm.org/D39235
llvm-svn: 316560
This code added in r297930 assumed that it could create
a select with a condition type that is just an integer
bitcast of the selected type. For AMDGPU any vselect is
going to be scalarized (although the vector types are legal),
and all select conditions must be i1 (the same as getSetCCResultType).
This logic doesn't really make sense to me, but there's
never really been a consistent policy in what the select
condition mask type is supposed to be. Try to extend
the logic for skipping the transform for condition types
that aren't setccs. It doesn't seem quite right to me though,
but checking conditions that seem more sensible (like whether the
vselect is going to be expanded) doesn't work since this
seems to depend on that also.
llvm-svn: 316554
Similar to how llvm::salvagDebugInfo hooks into InstCombine, this adds
a hook that can be invoked before an SDNode that is associated with an
SDDbgValue is erased to capture the effect of the deleted node in a
DIExpression.
The motivating example is an SDDebugValue attached to an ADD operation
that gets folded into a LOAD+OFFSET operation.
rdar://problem/32121503
llvm-svn: 316525
This reverts commit r316417, which causes internal compiles to OOM.
I don't unfortunately have a self-contained test case but will follow up
with courbet.
llvm-svn: 316497
This updates the MIRPrinter to include the regclass when printing
virtual register defs, which is already valid syntax for the
parser. That is, given 64 bit %0 and %1 in a "gpr" regbank,
%1(s64) = COPY %0(s64)
would now be written as
%1:gpr(s64) = COPY %0(s64)
While this change alone introduces a bit of redundancy with the
registers block, it allows us to update the tests to be more concise
and understandable and brings us closer to being able to remove the
registers block completely.
Note: We generally only print the class in defs, but there is one
exception. If there are uses without any defs whatsoever, we'll print
the class on all uses. I'm not completely convinced this comes up in
meaningful machine IR, but for now the MIRParser and MachineVerifier
both accept that kind of stuff, so we don't want to have a situation
where we can print something we can't parse.
llvm-svn: 316479
Refactor ExpandMemcmp:
- Stop duplicating the logic for computation of the sequence of loads to
generate (thsi was done in three different places), this is now done
only once in MemCmpExpansion::MemCmpExpansion().
- Add a FIXME to expose a bug with the computation of the number of loads
when not all sizes are loadable. For example, on X86-32 + SSE, possible
loads are {16,4,2,1} bytes. The current code considers that all loads
starting at MaxLoadSize are possible. This is not an issue right now as
vector loads are not enabled, so I'm not fixing the issue here to keep
the change as small as possible. I'm going to address this in a
subsequent revision, where I enable vector loads.
See https://bugs.llvm.org/show_bug.cgi?id=34887
Differential Revision: https://reviews.llvm.org/D38498
llvm-svn: 316417
Infrastructure designed for padding code with nop instructions in key places such that preformance improvement will be achieved.
The infrastructure is implemented such that the padding is done in the Assembler after the layout is done and all IPs and alignments are known.
This patch by itself in a NFC. Future patches will make use of this infrastructure to implement required policies for code padding.
Reviewers:
aaboud
zvi
craig.topper
gadi.haber
Differential revision: https://reviews.llvm.org/D34393
Change-Id: I92110d0c0a757080a8405636914a93ef6f8ad00e
llvm-svn: 316413
This commit adds optimisation remarks for outlining which fire when a function
is successfully outlined.
To do this, OutlinedFunctions must now contain references to their Candidates.
Since the Candidates must still be sorted and worked on separately, this is
done by working on everything in terms of shared_ptrs to Candidates. This is
good; it means that we can easily move everything to outlining in terms of
the OutlinedFunctions rather than the individual Candidates. This is far more
intuitive than what's currently there!
(Remarks are output when a function is created for some group of Candidates.
In a later commit, all of the outlining logic should be rewritten so that we
loop over OutlinedFunctions rather than over Candidates.)
llvm-svn: 316396
This fixes a bug where we'd crash given code like the test-case from
https://bugs.llvm.org/show_bug.cgi?id=30792 . Instead, we let the
offending clobber silently slide through.
This doesn't fully fix said bug, since the assembler will still complain
the moment it sees a crypto/fp/vector op, and we still don't diagnose
calls that require vector regs.
Differential Revision: https://reviews.llvm.org/D39030
llvm-svn: 316374
combineShuffleOfScalars is very conservative about shuffled BUILD_VECTORs that can be combined together.
This patch adds one additional case - if both BUILD_VECTORs represent splats of the same scalar value but with different UNDEF elements, then we should create a single splat BUILD_VECTOR, sharing only the UNDEF elements defined by the shuffle mask.
Differential Revision: https://reviews.llvm.org/D38696
llvm-svn: 316331
This fixes bugzilla 26810
https://bugs.llvm.org/show_bug.cgi?id=26810
This is intended to prevent sequences like:
movl %ebp, 8(%esp) # 4-byte Spill
movl %ecx, %ebp
movl %ebx, %ecx
movl %edi, %ebx
movl %edx, %edi
cltd
idivl %esi
movl %edi, %edx
movl %ebx, %edi
movl %ecx, %ebx
movl %ebp, %ecx
movl 16(%esp), %ebp # 4 - byte Reload
Such sequences are created in 2 scenarios:
Scenario #1:
vreg0 is evicted from physreg0 by vreg1
Evictee vreg0 is intended for region splitting with split candidate physreg0 (the reg vreg0 was evicted from)
Region splitting creates a local interval because of interference with the evictor vreg1 (normally region spliiting creates 2 interval, the "by reg" and "by stack" intervals. Local interval created when interference occurs.)
one of the split intervals ends up evicting vreg2 from physreg1
Evictee vreg2 is intended for region splitting with split candidate physreg1
one of the split intervals ends up evicting vreg3 from physreg2 etc.. until someone spills
Scenario #2
vreg0 is evicted from physreg0 by vreg1
vreg2 is evicted from physreg2 by vreg3 etc
Evictee vreg0 is intended for region splitting with split candidate physreg1
Region splitting creates a local interval because of interference with the evictor vreg1
one of the split intervals ends up evicting back original evictor vreg1 from physreg0 (the reg vreg0 was evicted from)
Another evictee vreg2 is intended for region splitting with split candidate physreg1
one of the split intervals ends up evicting vreg3 from physreg2 etc.. until someone spills
As compile time was a concern, I've added a flag to control weather we do cost calculations for local intervals we expect to be created (it's on by default for X86 target, off for the rest).
Differential Revision: https://reviews.llvm.org/D35816
Change-Id: Id9411ff7bbb845463d289ba2ae97737a1ee7cc39
llvm-svn: 316295
We don't need to do any additional recursion, we just need to analyze the APInt stored in the node. This matches what the ValueTracking versions do for IR.
llvm-svn: 316256
Summary:
We shouldn't recurse any further but it doesn't mean we shouldn't be able to give the known bits for a constant. The caller would probably like that we always return the right answer for a constant RHS. This matches what InstCombine does in this case.
I don't have a test case because this showed up while trying to revive D31724.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: arsenm, llvm-commits
Differential Revision: https://reviews.llvm.org/D38967
llvm-svn: 316255
Move the prune logic in pruneOverlaps to a new function, prune. This lets us
reuse the prune functionality. Makes the code a bit more readable. It'll also
make it easier to emit remarks/debug statements for pruned functions.
llvm-svn: 316031
This commit moves the decrement logic for outlined functions into the class,
and makes OccurrenceCount private. It can now be accessed via
getOccurrenceCount().
This makes it more difficult to accidentally introduce bugs by incorrectly
decrementing the occurrence count on OutlinedFunctions.
llvm-svn: 316020
Cleanup to Candidate that moves all end index calculations into
Candidate.endIdx(). For the sake of consistency, StartIdx and Len are now
private members, and can be accessed with length() and startIdx() respectively.
llvm-svn: 316019
Summary:
It seems that negative offset was accidentally allowed in D17967.
AFAICT small negative offset should be valid (always raise segfault) on all archs that I'm aware of (especially x86, which is the only one with this optimization enabled) and such case can be useful when loading hiden metadata from an object.
However, like the positive side, it should only be done within a certain limit.
For now, use the same limit on the positive side for the negative side.
A separate option can be added if needs appear.
Reviewers: mcrosier, skatkov
Reviewed By: skatkov
Subscribers: sanjoy, llvm-commits
Differential Revision: https://reviews.llvm.org/D38925
llvm-svn: 315991
Minor addition and follow up of r314773 and r311533: this adds more
debug messages to the type legalizer. For each node, it dumps
legalization info for results and operands nodes, rather than just the
final legalized node.
Differential Revision: https://reviews.llvm.org/D38726
llvm-svn: 315904
Summary:
iPTR is a pointer of subtarget-specific size to any address space. Therefore
type checks on this size derive the SizeInBits from a subtarget hook.
At this point, we can import the simplests G_LOAD rules and select load
instructions using them. Further patches will support for the predicates to
enable additional loads as well as the stores.
The previous commit failed on MSVC due to a failure to convert an
initializer_list to a std::vector. Hopefully, MSVC will accept this version.
Depends on D37457
Reviewers: ab, qcolombet, t.p.northover, rovka, aditya_nandakumar
Reviewed By: qcolombet
Subscribers: kristof.beyls, javed.absar, llvm-commits, igorb
Differential Revision: https://reviews.llvm.org/D37458
llvm-svn: 315887
Summary:
iPTR is a pointer of subtarget-specific size to any address space. Therefore
type checks on this size derive the SizeInBits from a subtarget hook.
At this point, we can import the simplests G_LOAD rules and select load
instructions using them. Further patches will support for the predicates to
enable additional loads as well as the stores.
Depends on D37457
Reviewers: ab, qcolombet, t.p.northover, rovka, aditya_nandakumar
Reviewed By: qcolombet
Subscribers: kristof.beyls, javed.absar, llvm-commits, igorb
Differential Revision: https://reviews.llvm.org/D37458
llvm-svn: 315885
Summary:
It's possible for a ComplexPattern to be used as an operator in a match
pattern. This is used by the load/store patterns in AArch64 to name the
suboperands returned by ComplexPattern predicate so that they can be broken
apart and referenced independently in the result pattern.
This patch adds support for this in order to enable the import of load/store
patterns.
Depends on D37445
Hopefully fixed the ambiguous constructor that a large number of bots reported.
Reviewers: ab, qcolombet, t.p.northover, rovka, aditya_nandakumar
Reviewed By: qcolombet
Subscribers: aemerson, javed.absar, igorb, llvm-commits, kristof.beyls
Differential Revision: https://reviews.llvm.org/D37456
llvm-svn: 315869
Summary:
It's possible for a ComplexPattern to be used as an operator in a match
pattern. This is used by the load/store patterns in AArch64 to name the
suboperands returned by ComplexPattern predicate so that they can be broken
apart and referenced independently in the result pattern.
This patch adds support for this in order to enable the import of load/store
patterns.
Depends on D37445
Reviewers: ab, qcolombet, t.p.northover, rovka, aditya_nandakumar
Reviewed By: qcolombet
Subscribers: aemerson, javed.absar, igorb, llvm-commits, kristof.beyls
Differential Revision: https://reviews.llvm.org/D37456
llvm-svn: 315863
TargetRegisterInfo::getMinimalPhysRegClass is actually pretty expensive
because it has to iterate over all the register classes.
Cache this information as we need and get it so that we limit its usage.
Right now, we heavily rely on it, because this is how we get the mapping
for vregs defined by copies from physreg (i.e., the one that are ABI
related).
Improve compile time by up to 10% for that pass.
NFC
llvm-svn: 315759
Prior to this patch we used to create SetVectors in temporaries that
were created and destroyed for each instruction. Now, instead we create
and destroyed them only once, but clear the content for each
instruction.
This speeds up the pass by ~25%.
NFC.
llvm-svn: 315756
This is only currently used for mad/fma transforms.
This is the only case where it should be used for AMDGPU,
so add an opcode to be sure.
llvm-svn: 315740
I don't know if we ever hit this case or not. Turning it into an assert only fired on expanding some atomic operation in a SystemZ lit test.
llvm-svn: 315648
Reverting to investigate layering effects of MCJIT not linking
libCodeGen but using TargetMachine::getNameWithPrefix() breaking the
lldb bots.
This reverts commit r315633.
llvm-svn: 315637
The clang frontend already creates a DIExpression that replicates the
logic in addBlockByrefAddress() exactly, thus making this function
effectively unreachable. To guard against human error I'm hereby
marking the function with an assertion and let it hit the bots before
eventually removing it.
rdar://problem/31629055
llvm-svn: 315636
Merge LLVMTargetMachine into TargetMachine.
- There is no in-tree target anymore that just implements TargetMachine
but not LLVMTargetMachine.
- It should still be possible to stub out all the various functions in
case a target does not want to use lib/CodeGen
- This simplifies the code and avoids methods ending up in the wrong
interface.
Differential Revision: https://reviews.llvm.org/D38489
llvm-svn: 315633
The comparator passed to std::sort must provide a strict weak ordering;
otherwise, the behavior is undefined.
Fixes an assertion failure generating debug info for globals
split by GlobalOpt. I have a testcase, but not sure how to reduce it,
so not included here. (Someone else came up with a testcase, but I
can't reproduce the crash with it, presumably because my version of LLVM
ends up sorting the array differently.)
This isn't really a complete fix (see the FIXME in the patch), but at
least it doesn't have undefined behavior.
Differential Revision: https://reviews.llvm.org/D38830
llvm-svn: 315619
Summary:
Add LLVM_FORCE_ENABLE_DUMP cmake option, and use it along with
LLVM_ENABLE_ASSERTIONS to set LLVM_ENABLE_DUMP.
Remove NDEBUG and only use LLVM_ENABLE_DUMP to enable dump methods.
Move definition of LLVM_ENABLE_DUMP from config.h to llvm-config.h so
it'll be picked up by public headers.
Differential Revision: https://reviews.llvm.org/D38406
llvm-svn: 315590
MachineInstr::isIdenticalTo has a lot of logic for dealing with register
Defs (i.e. deciding whether to take them into account or ignore them).
This logic gets things wrong in some obscure cases, for instance if an
operand is not a Def for both the current MI and the one we are
comparing to.
I'm not sure if it's possible for this to happen for regular register
operands, but it may happen in the ARM backend for special operands
which use sentinel values for the register (i.e. 0, which is neither a
physical register nor a virtual one).
This causes MachineInstrExpressionTrait::isEqual (which uses
MachineInstr::isIdenticalTo) to return true for the following
instructions, which are the same except for the fact that one sets the
flags and the other one doesn't:
%1114 = ADDrsi %1113, %216, 17, 14, _, def _
%1115 = ADDrsi %1113, %216, 17, 14, _, _
OTOH, MachineInstrExpressionTrait::getHashValue returns different values
for the 2 instructions due to the different isDef on the last operand.
In practice this means that when trying to add those instructions to a
DenseMap, they will be considered different because of their different
hash values, but when growing the map we might get an assertion while
copying from the old buckets to the new buckets because isEqual
misleadingly returns true.
This patch makes sure that isEqual and getHashValue agree, by improving
the checks in MachineInstr::isIdenticalTo when we are ignoring virtual
register definitions (which is what the Trait uses). Firstly, instead of
checking isPhysicalRegister, we use !isVirtualRegister, so that we cover
both physical registers and sentinel values. Secondly, instead of
checking MachineOperand::isReg, we use MachineOperand::isIdenticalTo,
which checks isReg, isSubReg and isDef, which are the same values that
the hash function uses to compute the hash.
Note that the function is symmetric with this change, since if the
current operand is not a Def, we check MachineOperand::isIdenticalTo,
which returns false if the operands have different isDef's.
Differential Revision: https://reviews.llvm.org/D38789
llvm-svn: 315579
While this shouldn't be necessary anymore, we have cases where we run
into the assertion below, i.e. cases with two non-fragment entries for the
same variable at different frame indices.
This should be fixed, but for now, we should revert to a version that
does not trigger asserts.
llvm-svn: 315576
This patch fixes the bug introduced in https://reviews.llvm.org/D35907; the bug is reported by http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20171002/491452.html.
Before D35907, when GetUnderlyingObjects fails to find an identifiable object, allMMOsOkay lambda in getUnderlyingObjectsForInstr returns false and Objects vector is cleared. This behavior is unintentionally changed by D35907.
This patch makes the behavior for such case same as the previous behavior.
Since D35907 introduced a wrapper function getUnderlyingObjectsForCodeGen around GetUnderlyingObjects, getUnderlyingObjectsForCodeGen is modified to return a boolean value to ask the caller to clear the Objects vector.
Differential Revision: https://reviews.llvm.org/D38735
llvm-svn: 315565
Summary:
The comments in the code said
// Remove <def,read-undef> flags. This def is now a partial redef.
but the code didn't just remove read-undef, it could introduce new ones which
could cause errors.
E.g. if we have something like
%vreg1<def> = IMPLICIT_DEF
%vreg2:subreg1<def, read-undef> = op %vreg3, %vreg4
%vreg2:subreg2<def> = op %vreg6, %vreg7
and we merge %vreg1 and %vreg2 then we should not set undef on the second subreg
def, which the old code did.
Now we solve this by actually do what the code comment says. We remove
read-undef flags rather than remove or introduce them.
Reviewers: qcolombet, MatzeB
Reviewed By: MatzeB
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D38616
llvm-svn: 315564
MCObjectStreamer owns its MCCodeEmitter -- this fixes the types to reflect that,
and allows us to remove the last instance of MCObjectStreamer's weird "holding
ownership via someone else's reference" trick.
llvm-svn: 315531
Summary:
This adds a set of new directives that describe 32-bit x86 prologues.
The directives are limited and do not expose the full complexity of
codeview FPO data. They are merely a convenience for the compiler to
generate more readable assembly so we don't need to generate tons of
labels in CodeGen. If our prologue emission changes in the future, we
can change the set of available directives to suit our needs. These are
modelled after the .seh_ directives, which use a different format that
interacts with exception handling.
The directives are:
.cv_fpo_proc _foo
.cv_fpo_pushreg ebp/ebx/etc
.cv_fpo_setframe ebp/esi/etc
.cv_fpo_stackalloc 200
.cv_fpo_endprologue
.cv_fpo_endproc
.cv_fpo_data _foo
I tried to follow the implementation of ARM EHABI CFI directives by
sinking most directives out of MCStreamer and into X86TargetStreamer.
This helps avoid polluting non-X86 code with WinCOFF specific logic.
I used cdb to confirm that this can show locals in parent CSRs in a few
cases, most importantly the one where we use ESI as a frame pointer,
i.e. the one in http://crbug.com/756153#c28
Once we have cdb integration in debuginfo-tests, we can add integration
tests there.
Reviewers: majnemer, hans
Subscribers: aemerson, mgorny, kristof.beyls, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D38776
llvm-svn: 315513
Summary:
Fixes a bogus iterator resulting from the removal of a block's first instruction at the point that incremental update is enabled.
Patch by Paul Walker.
Reviewers: fhahn, Gerolf, efriedma, MatzeB
Reviewed By: fhahn
Subscribers: aemerson, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D38734
llvm-svn: 315502
parameterized emit() calls
Summary: This is not functional change to adopt new emit() API added in r313691.
Reviewed By: anemet
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D38285
llvm-svn: 315476
The software pipeliner and the packetizer try to break dependence
between the post-increment instruction and the dependent memory
instructions by changing the base register and the offset value.
However, in some cases, the existing logic didn't work properly
and created incorrect offset value.
Patch by Jyotsna Verma.
llvm-svn: 315468
The pipeliner is generating a serial sequence that causes poor
register allocation when a post-increment instruction appears
prior to the use of the post-increment register. This occurs when
there is a circular set of dependences involved with a sequence
of instructions in the same cycle. In this case, there is no
serialization of the parallel semantics that will not cause an
additional register to be allocated.
This patch fixes the problem by changing the instructions so that
the post-increment instruction is used by the subsequent
instruction, which enables the register allocator to make a
better decision and not require another register.
Patch by Brendon Cahoon.
llvm-svn: 315466
Eg:
insert v4i32 V, (v2i16 X), 2 --> shuffle v8i16 V', X', {0,1,2,3,8,9,6,7}
This is a generalization of the IR fold in D38316 to handle insertion into a non-undef vector.
We may want to abandon that one if we can't find value in squashing the more specific pattern sooner.
We're using the existing legal shuffle target hook to avoid AVX512 horror with vXi1 shuffles.
There may be room for improvement in the shuffle lowering here, but that would be follow-up work.
Differential Revision: https://reviews.llvm.org/D38388
llvm-svn: 315460
The NumFixedArgs field of CallLoweringInfo is used by
TargetLowering::LowerCallTo to determine whether a given argument is passed
using the vararg calling convention or not (specifically, to set IsFixed for
each ISD::OutputArg).
Firstly, CallLoweringInfo::setLibCallee and CallLoweringInfo::setCallee both
incorrectly set NumFixedArgs based on the _previous_ args list. Secondly,
TargetLowering::LowerCallTo failed to increment NumFixedArgs when modifying
the argument list so a pointer is passed for the return value.
If your backend uses the IsFixed property or directly accesses NumFixedArgs,
it is _possible_ this change could result in codegen changes (although the
previous behaviour would have been incorrect). No such cases have been
identified during code review for any in-tree architecture.
Differential Revision: https://reviews.llvm.org/D37898
llvm-svn: 315457
MCObjectStreamer owns its MCAsmBackend -- this fixes the types to reflect that,
and allows us to remove another instance of MCObjectStreamer's weird "holding
ownership via someone else's reference" trick.
llvm-svn: 315410
Similarly to how Instruction has getFunction, this adds a less verbose
way to write MI->getParent()->getParent(). I'll follow up shortly with
a change that changes a bunch of the uses.
llvm-svn: 315388
Summary:
See https://llvm.org/PR33743 for more details
It seems that for non-power of 2 vector sizes, the algorithm can produce
non-matching sizes for input and result causing an assert.
This usually isn't a problem as the isAnyExtend check will weed these out, but
in some cases (most often with lots of undefined values for the mask indices) it
can pass this check for non power of 2 vectors.
Adding in an extra check that ensures that bit size will match for the result
and input (as required)
Subscribers: nhaehnle
Differential Revision: https://reviews.llvm.org/D35241
llvm-svn: 315307
Some passes might duplicate calls to llvm.dbg.declare creating
duplicate frame index expression which currently trigger an assertion
which is meant to catch erroneous, overlapping fragment declarations.
But identical frame index expressions are just redundant and don't
actually conflict with each other, so we can be more lenient and just
ignore the duplicates.
Reviewers: aprantl, rnk
Subscribers: llvm-commits, JDevlieghere
Differential Revision: https://reviews.llvm.org/D38540
llvm-svn: 315279
We end up creating COPY's that are either truncating/extending and this
should be illegal.
https://reviews.llvm.org/D37640
Patch for X86 and ARM by igorb, rovka
llvm-svn: 315240
Say you have two identical linkonceodr functions, one in M1 and one in M2.
Say that the outliner outlines A,B,C from one function, and D,E,F from another
function (where letters are instructions). Now those functions are not
identical, and cannot be deduped. Locally to M1 and M2, these outlining
choices would be good-- to the whole program, however, this might not be true!
To mitigate this, this commit makes it so that the outliner sees linkonceodr
functions as unsafe to outline from. It also adds a flag,
-enable-linkonceodr-outlining, which allows the user to specify that they
want to outline from such functions when they know what they're doing.
Changing this handles most code size regressions in the test suite caused by
competing with linker dedupe. It also doesn't have a huge impact on the code
size improvements from the outliner. There are 6 tests that regress > 5% from
outlining WITH linkonceodrs to outlining WITHOUT linkonceodrs. Overall, most
tests either improve or are not impacted.
Not outlined vs outlined without linkonceodrs:
https://hastebin.com/raw/qeguxavuda
Not outlined vs outlined with linkonceodrs:
https://hastebin.com/raw/edepoqoqic
Outlined with linkonceodrs vs outlined without linkonceodrs:
https://hastebin.com/raw/awiqifiheb
Numbers generated using compare.py with -m size.__text. Tests run for AArch64
with -Oz -mllvm -enable-machine-outliner -mno-red-zone.
llvm-svn: 315136
Summary:
After r303360, we initialize UsesCalleeSaves in runOnMachineFunction,
which runs after getRequiredProperties. UsesCalleeSaves was initialized
to 'false', so getRequiredProperties would always return an empty set.
We don't have a TargetMachine available early anymore after r303360.
Just removing the requirement of NoVRegs seems to make things work, so
let's do that.
Reviewers: thegameg, dschuff, MatzeB
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D38597
llvm-svn: 315089
The patch that this assert comes with is fixing a bug in MBP. The assert is
invalid however.
Thanks to @sergey.k.okunev for finding this
Currently this fails SPECCPU2006 LTO. I will add a test case when I do more
investigation and have one.
llvm-svn: 315032
Summary:
When reinserting debug values after register allocation, make sure to
insert debug values after each redefinition of debug value register in
the slot index range. The reason for this is that DwarfDebug will end
the range of a debug variable when the physical reg is defined. For
instructions with e.g. tied operands this result in prematurely ended
debug range.
This resolves pr34545
Patch by Karl-Johan Karlsson and Bjorn Pettersson
Reviewers: rnk, aprantl
Reviewed By: rnk
Subscribers: bjope, llvm-commits
Differential Revision: https://reviews.llvm.org/D38229
llvm-svn: 314974
Summary:
FastISel::hasTrivialKill() was the only user of the "IntPtrTy" version of
Cast::isNoopCast(). According to review comments in D37894 we could instead
use the "DataLayout" version of the method, and thus get rid of the
"IntPtrTy" versions of isNoopCast() completely.
With the above done, the remaining isNoopCast() could then be simplified
a bit more.
Reviewers: arsenm
Reviewed By: arsenm
Subscribers: wdng, llvm-commits
Differential Revision: https://reviews.llvm.org/D38497
llvm-svn: 314969
This is a follow-up to https://reviews.llvm.org/D38138.
I fixed the capitalization of some functions because we're changing those
lines anyway and that helped verify that we weren't accidentally dropping
any options by using default param values.
llvm-svn: 314930
It broke the Chromium / SQLite build; see PR34830.
> Summary:
> 1/ Operand folding during complex pattern matching for LEAs has been
> extended, such that it promotes Scale to accommodate similar operand
> appearing in the DAG.
> e.g.
> T1 = A + B
> T2 = T1 + 10
> T3 = T2 + A
> For above DAG rooted at T3, X86AddressMode will no look like
> Base = B , Index = A , Scale = 2 , Disp = 10
>
> 2/ During OptimizeLEAPass down the pipeline factorization is now performed over LEAs
> so that if there is an opportunity then complex LEAs (having 3 operands)
> could be factored out.
> e.g.
> leal 1(%rax,%rcx,1), %rdx
> leal 1(%rax,%rcx,2), %rcx
> will be factored as following
> leal 1(%rax,%rcx,1), %rdx
> leal (%rdx,%rcx) , %edx
>
> 3/ Aggressive operand folding for AM based selection for LEAs is sensitive to loops,
> thus avoiding creation of any complex LEAs within a loop.
>
> Reviewers: lsaba, RKSimon, craig.topper, qcolombet, jmolloy
>
> Reviewed By: lsaba
>
> Subscribers: jmolloy, spatel, igorb, llvm-commits
>
> Differential Revision: https://reviews.llvm.org/D35014
llvm-svn: 314919
Before the patch this was in Analysis. Moving it to IR and making it implicit
part of LLVMContext::diagnose allows the full opt-remark facility to be used
outside passes e.g. the pass manager. Jessica is planning to use this to
report function size after each pass. The same could be used for time
reports.
Tested with BUILD_SHARED_LIBS=On.
llvm-svn: 314909
Summary:
1/ Operand folding during complex pattern matching for LEAs has been
extended, such that it promotes Scale to accommodate similar operand
appearing in the DAG.
e.g.
T1 = A + B
T2 = T1 + 10
T3 = T2 + A
For above DAG rooted at T3, X86AddressMode will no look like
Base = B , Index = A , Scale = 2 , Disp = 10
2/ During OptimizeLEAPass down the pipeline factorization is now performed over LEAs
so that if there is an opportunity then complex LEAs (having 3 operands)
could be factored out.
e.g.
leal 1(%rax,%rcx,1), %rdx
leal 1(%rax,%rcx,2), %rcx
will be factored as following
leal 1(%rax,%rcx,1), %rdx
leal (%rdx,%rcx) , %edx
3/ Aggressive operand folding for AM based selection for LEAs is sensitive to loops,
thus avoiding creation of any complex LEAs within a loop.
Reviewers: lsaba, RKSimon, craig.topper, qcolombet, jmolloy
Reviewed By: lsaba
Subscribers: jmolloy, spatel, igorb, llvm-commits
Differential Revision: https://reviews.llvm.org/D35014
llvm-svn: 314886
This time invoking llc with "-march=x86-64" in the testcase, so we don't assume
the default target is x86.
Summary:
If we have
%vreg0<def> = PHI %vreg2<undef>, <BB#0>, %vreg3, <BB#2>; GR32:%vreg0,%vreg2,%vreg3
%vreg3<def,tied1> = ADD32ri8 %vreg0<kill,tied0>, 1, %EFLAGS<imp-def>; GR32:%vreg3,%vreg0
then we can't just change %vreg0 into %vreg3, since %vreg2 is actually
undef. We would have to also copy the undef flag to be able to change the
register.
Instead we deal with this case like other cases where we can't just
replace the register: we insert a COPY. The code creating the COPY already
copied all flags from the PHI input, so the undef flag will be transferred
as it should.
Reviewers: kparzysz
Reviewed By: kparzysz
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D38235
llvm-svn: 314882
Summary:
If we have
%vreg0<def> = PHI %vreg2<undef>, <BB#0>, %vreg3, <BB#2>; GR32:%vreg0,%vreg2,%vreg3
%vreg3<def,tied1> = ADD32ri8 %vreg0<kill,tied0>, 1, %EFLAGS<imp-def>; GR32:%vreg3,%vreg0
then we can't just change %vreg0 into %vreg3, since %vreg2 is actually
undef. We would have to also copy the undef flag to be able to change the
register.
Instead we deal with this case like other cases where we can't just
replace the register: we insert a COPY. The code creating the COPY already
copied all flags from the PHI input, so the undef flag will be transferred
as it should.
Reviewers: kparzysz
Reviewed By: kparzysz
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D38235
llvm-svn: 314879
This commit does two things. Firstly, it cleans up some of the benefit
calculation wrt outlined functions and candidates. Secondly, it fixes an
off-by-one bug in the cost model which was caused by the benefit value of
an OutlinedFunction and Candidate differing by 1. It updates the remarks test
to reflect this change.
llvm-svn: 314836
Summary:
This should fix a regression introduced by r313786, which switched from
MachineInstr::isIndirectDebugValue() to checking if operand 1 is an
immediate. I didn't have a test case for it until now.
A single UserValue, which approximates a user variable, may have many
DBG_VALUE instructions that disagree about whether the variable is in
memory or in a virtual register. This will become much more common once
we have llvm.dbg.addr, but you can construct such a test case manually
today with llvm.dbg.value.
Before this change, we would get two UserValues: one for direct and one
for indirect DBG_VALUE instructions describing the same variable. If we
build separate interval maps for direct and indirect locations, we will
end up accidentally coalescing identical DBG_VALUE intervals that need
to remain separate because they are broken up by intervals of the
opposite direct-ness.
Reviewers: aprantl
Subscribers: llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D37932
llvm-svn: 314819
Currently optimizeMemoryInst requires that all of the AddrModes it sees are
identical. This patch makes it capable of tracking multiple AddrModes, so long
as they differ in at most one field.
This patch does nothing by itself, but later patches will make use of it to
insert or reuse phi or select instructions for the differing fields.
Differential Revision: https://reviews.llvm.org/D38278
llvm-svn: 314795
This lets us optimize away selects that perform the same address computation in
two different ways and is also the first step towards being able to handle
selects between two different, but compatible, address computations.
Differential Revision: https://reviews.llvm.org/D38242
llvm-svn: 314794
Summary:
Take the target's endianness into account when splitting the
debug information in DAGTypeLegalizer::SetExpandedInteger.
This patch fixes so that, for big-endian targets, the fragment
expression corresponding to the high part of a split integer
value is placed at offset 0, in order to correctly represent
the memory address order.
I have attached a PPC32 reproducer where the resulting DWARF
pieces for a 64-bit integer were incorrectly reversed.
Original patch was reverted due to using -stop-after=isel in
the test case (but that is only working when AMDGPU target
is included in the llc build). The test case has now been
updated to use -stop-before=expand-isel-pseudos instead.
Patch by: dstenb
Reviewers: JDevlieghere, aprantl, dblaikie
Reviewed By: JDevlieghere, aprantl, dblaikie
Subscribers: nemanjai
Differential Revision: https://reviews.llvm.org/D38172
llvm-svn: 314781
This adds some more debug messages to the type legalizer and functions
like PromoteNode, ExpandNode, ExpandLibCall in an attempt to make
the debug messages a little bit more informative and useful.
Differential Revision: https://reviews.llvm.org/D38450
llvm-svn: 314773
Issues addressed since original review:
- Avoid bug in regalloc greedy/machine verifier when forwarding to use
in an instruction that re-defines the same virtual register.
- Fixed bug when forwarding to use in EarlyClobber instruction slot.
- Fixed incorrect forwarding to register definitions that showed up in
explicit_uses() iterator (e.g. in INLINEASM).
- Moved removal of dead instructions found by
LiveIntervals::shrinkToUses() outside of loop iterating over
instructions to avoid instructions being deleted while pointed to by
iterator.
- Fixed ARMLoadStoreOptimizer bug exposed by this change in r311907.
- The pass no longer forwards COPYs to physical register uses, since
doing so can break code that implicitly relies on the physical
register number of the use.
- The pass no longer forwards COPYs to undef uses, since doing so
can break the machine verifier by creating LiveRanges that don't
end on a use (since the undef operand is not considered a use).
[MachineCopyPropagation] Extend pass to do COPY source forwarding
This change extends MachineCopyPropagation to do COPY source forwarding.
This change also extends the MachineCopyPropagation pass to be able to
be run during register allocation, after physical registers have been
assigned, but before the virtual registers have been re-written, which
allows it to remove virtual register COPY LiveIntervals that become dead
through the forwarding of all of their uses.
llvm-svn: 314729
Summary:
Take the target's endianness into account when splitting the
debug information in DAGTypeLegalizer::SetExpandedInteger.
This patch fixes so that, for big-endian targets, the fragment
expression corresponding to the high part of a split integer
value is placed at offset 0, in order to correctly represent
the memory address order.
I have attached a PPC32 reproducer where the resulting DWARF
pieces for a 64-bit integer were incorrectly reversed.
Patch by: dstenb
Reviewers: JDevlieghere, aprantl, dblaikie
Reviewed By: JDevlieghere, aprantl, dblaikie
Subscribers: nemanjai
Differential Revision: https://reviews.llvm.org/D38172
llvm-svn: 314666
Currently expandUnalignedLoad/Store uses place holder pointer info for temporary memory operand
in stack, which does not have correct address space. This causes unaligned private double16 load/store to be
lowered to flat_load instead of buffer_load for amdgcn target.
This fixes failures of OpenCL conformance test basic/vload_private/vstore_private on target amdgcn---amdgizcl.
Differential Revision: https://reviews.llvm.org/D35361
llvm-svn: 314566
Implement shouldCoalesce() to help regalloc avoid running out of GR128
registers.
If a COPY involving a subreg of a GR128 is coalesced, the live range of the
GR128 virtual register will be extended. If this happens where there are
enough phys-reg clobbers present, regalloc will run out of registers (if
there is not a single GR128 allocatable register available).
This patch tries to allow coalescing only when it can prove that this will be
safe by checking the (local) interval in question.
Review: Ulrich Weigand, Quentin Colombet
https://reviews.llvm.org/D37899https://bugs.llvm.org/show_bug.cgi?id=34610
llvm-svn: 314516
This commit yanks out the repeated sections of code in pruneCandidates into
two lambdas: ShouldSkipCandidate and Prune. This simplifies the logic in
pruneCandidates significantly, and reduces the chance of introducing bugs by
folding all of the shared logic into one place.
llvm-svn: 314475
This patch implements the dwarfdump option --find=<name>. This option
looks for a DIE in the accelerator tables and dumps it if found. This
initial patch only adds support for .apple_names to keep the review
small, adding the other sections and pubnames support should be
trivial though.
Differential Revision: https://reviews.llvm.org/D38282
llvm-svn: 314439
Summary:
A DBG_VALUE that is referring to a physical register is
valid up until the next def of the register, or the end
of the basic block that it belongs to.
LiveDebugVariables is computing live intervals (slot index
ranges) for DBG_VALUE instructions, before regalloc, in order
to be able to re-insert DBG_VALUE instructions again after
regalloc. When the DBG_VALUE is mapping a variable to a
physical register we do not need to compute the range. We
should simply re-insert the DBG_VALUE at the start position.
The problem that was found, resulting in this patch, was a
situation when the DBG_VALUE was the last real use of the
physical register. The computeIntervals/extendDef methods
extended the range to cover the whole basic block, even though
the physical register very well could be allocated to some
virtual register inside the basic block. So the extended
range could not be trusted.
This patch is a preparation for https://reviews.llvm.org/D38229,
where the goal is to insert DBG_VALUE after each new definition
of a variable, even if the virtual registers that the variable
was connected to has been coalesced into using the same physical
register (e.g. due to two address instructions). For more info
see https://bugs.llvm.org/show_bug.cgi?id=34545
Reviewers: aprantl, rnk, echristo
Reviewed By: aprantl
Subscribers: Ka-Ka, llvm-commits
Differential Revision: https://reviews.llvm.org/D38140
llvm-svn: 314414
It's currently quite difficult to test passes like branch relaxation, which
requires branches with large displacement to be generated. The .space assembler
directive makes it easy to create arbitrarily large basic blocks, but
getInlineAsmLength is not able to parse it and so the size of the block is not
correctly estimated. Other backends (AArch64, AMDGPU) introduce options just
for testing that artificially restrict the ranges of branch instructions (e.g.
aarch64-tbz-offset-bits). Although parsing a single form of the .space
directive feels inelegant, it does allow a more direct testing approach.
This patch adapts the .space parsing code from
Mips16InstrInfo::getInlineAsmLength and removes it now the extra functionality
is provided by the base implementation. I want to move this functionality to
the generic getInlineAsmLength as 1) I need the same for RISC-V, and 2) I feel
other backends will benefit from more direct testing of large branch
displacements.
Differential Revision: https://reviews.llvm.org/D37798
llvm-svn: 314393
Summary:
If we have a non-allocated register, we allow us to try recoloring of an
already allocated and "Done" register, even if they are of the same
register class, if the non-allocated register has at least one tied def
and the allocated one has none.
It should be easier to recolor the non-tied register than the tied one, so
it might be an improvement even if they use the same regclasses.
Reviewers: qcolombet
Reviewed By: qcolombet
Subscribers: llvm-commits, MatzeB
Differential Revision: https://reviews.llvm.org/D38309
llvm-svn: 314388
Without this, we could end up trying to get the Nth (0-indexed) element
from a subvector of size N.
Differential Revision: https://reviews.llvm.org/D37880
llvm-svn: 314380
This commit allows the outliner to avoid saving and restoring the link register
on AArch64 when it is dead within an entire class of candidates.
This introduces changes to the way the outliner interfaces with the target.
For example, the target now interfaces with the outliner using a
MachineOutlinerInfo struct rather than by using getOutliningCallOverhead and
getOutliningFrameOverhead.
This also improves several comments on the outliner's cost model.
https://reviews.llvm.org/D36721
llvm-svn: 314341
Summary:
According to https://gcc.gnu.org/wiki/SplitStacks, the linker expects a zero-sized .note.GNU-split-stack section if split-stack is used (and also .note.GNU-no-split-stack section if it also contains non-split-stack functions), so it can handle the cases where a split-stack function calls non-split-stack function.
This change adds the sections if needed.
Fixes PR #34670.
Reviewers: thanm, rnk, luqmana
Reviewed By: rnk
Subscribers: llvm-commits
Patch by Cherry Zhang <cherryyz@google.com>
Differential Revision: https://reviews.llvm.org/D38051
llvm-svn: 314335
This was intended to be no-functional-change, but it's not - there's a test diff.
So I thought I should stop here and post it as-is to see if this looks like what was expected
based on the discussion in PR34603:
https://bugs.llvm.org/show_bug.cgi?id=34603
Notes:
1. The test improvement occurs because the existing 'LateSimplifyCFG' marker is not carried
through the recursive calls to 'SimplifyCFG()->SimplifyCFGOpt().run()->SimplifyCFG()'.
The parameter isn't passed down, so we pick up the default value from the function signature
after the first level. I assumed that was a bug, so I've passed 'Options' down in all of the
'SimplifyCFG' calls.
2. I split 'LateSimplifyCFG' into 2 bits: ConvertSwitchToLookupTable and KeepCanonicalLoops.
This would theoretically allow us to differentiate the transforms controlled by those params
independently.
3. We could stash the optional AssumptionCache pointer and 'LoopHeaders' pointer in the struct too.
I just stopped here to minimize the diffs.
4. Similarly, I stopped short of messing with the pass manager layer. I have another question that
could wait for the follow-up: why is the new pass manager creating the pass with LateSimplifyCFG
set to true no matter where in the pipeline it's creating SimplifyCFG passes?
// Create an early function pass manager to cleanup the output of the
// frontend.
EarlyFPM.addPass(SimplifyCFGPass());
-->
/// \brief Construct a pass with the default thresholds
/// and switch optimizations.
SimplifyCFGPass::SimplifyCFGPass()
: BonusInstThreshold(UserBonusInstThreshold),
LateSimplifyCFG(true) {} <-- switches get converted to lookup tables and loops may not be in canonical form
If this is unintended, then it's possible that the current behavior of dropping the 'LateSimplifyCFG'
setting via recursion was masking this bug.
Differential Revision: https://reviews.llvm.org/D38138
llvm-svn: 314308
This teach simplifyDemandedBits to handle constant splat vector shifts.
This required changing some uses of getZExtValue to getLimitedValue since we can't rely on legalization using getShiftAmountTy for the shift amount.
I believe there may have been a bug in the ((X << C1) >>u ShAmt) handling where we didn't check if the inner shift was too large. I've fixed that here.
I had to add new patterns to ARM because the zext/sext the patterns were trying to look for got turned into an any_extend with this patch. Happy to split that out too, but not sure how to test without this change.
Differential Revision: https://reviews.llvm.org/D37665
llvm-svn: 314139
Summary:
This code iterates the 'Orders' vector in parallel with the DbgValue
list, emitting all DBG_VALUEs that occurred between the last IR order
insertion point and the next insertion point. This assumes the
SDDbgValue list is sorted in IR order, which it usually is. However, it
is not sorted when a node with a debug value is replaced with another
one. When this happens, TransferDbgValues is called, and the new value
is added to the end of the list.
The problem can be solved by stably sorting the list by IR order.
Reviewers: aprantl, Ka-Ka
Reviewed By: aprantl
Subscribers: MatzeB, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D38197
llvm-svn: 314114
Summary:
Right now there are two functions with the same name, one does the work
and the other one returns true if expansion is needed. Rename
TargetTransformInfo::expandMemCmp to make it more consistent with other
members of TargetTransformInfo.
Remove the unused Instruction* parameter.
Differential Revision: https://reviews.llvm.org/D38165
llvm-svn: 314096
Summary: Conditional returns were not taken into consideration at all. Implement them by turning them into jumps and normal returns. This means there is a slightly higher performance penalty for conditional returns, but this is the best we can do, and it still disturbs little of the rest.
Reviewers: dberris, echristo
Subscribers: sanjoy, nemanjai, hiraditya, kbarton, llvm-commits
Differential Revision: https://reviews.llvm.org/D38102
llvm-svn: 314005
At least for the 64-bit and less case, we should be able to determine if we even have a mask without counting any bits. This also removes the need to explicitly check for 0 active bits, isMask will return false for 0.
llvm-svn: 313908
The fix is to avoid invalidating our insertion point in
replaceDbgDeclare:
Builder.insertDeclare(NewAddress, DIVar, DIExpr, Loc, InsertBefore);
+ if (DII == InsertBefore)
+ InsertBefore = &*std::next(InsertBefore->getIterator());
DII->eraseFromParent();
I had to write a unit tests for this instead of a lit test because the
use list order matters in order to trigger the bug.
The reduced C test case for this was:
void useit(int*);
static inline void inlineme() {
int x[2];
useit(x);
}
void f() {
inlineme();
inlineme();
}
llvm-svn: 313905
Summary:
SelectionDAGISel::LowerArguments is associating arguments
with frame indices (FuncInfo->setArgumentFrameIndex). That
information is later on used by EmitFuncArgumentDbgValue to
create DBG_VALUE instructions that denotes that a variable
can be found on the stack.
I discovered that for our (big endian) out-of-tree target
the association created by SelectionDAGISel::LowerArguments
sometimes is wrong. I've seen this happen when a 64-bit value
is passed on the stack. The argument will occupy two stack
slots (frame index X, and frame index X+1). The fault is
that a call to setArgumentFrameIndex is associating the
64-bit argument with frame index X+1. The effect is that the
debug information (DBG_VALUE) will point at the least significant
part of the arguement on the stack. When printing the
argument in a debugger I will get the wrong value.
I managed to create a test case for PowerPC that seems to
show the same kind of problem.
The bugfix will look at the datalayout, taking endianness into
account when examining a BUILD_PAIR node, assuming that the
least significant part is in the first operand of the BUILD_PAIR.
For big endian targets we should use the frame index from
the second operand, as the most significant part will be stored
at the lower address (using the highest frame index).
Reviewers: bogner, rnk, hfinkel, sdardis, aprantl
Reviewed By: aprantl
Subscribers: nemanjai, aprantl, llvm-commits, igorb
Differential Revision: https://reviews.llvm.org/D37740
llvm-svn: 313901
.. as well as the two subsequent changes r313826 and r313875.
This leads to segfaults in combination with ASAN. Will forward repro
instructions to the original author (rnk).
llvm-svn: 313876
This patch contains fix for reverted commit
rL312318 which was causing failure due to use
of unchecked dyn_cast to CIInit.
Patch by: Nikola Prica.
llvm-svn: 313870
Summary:
This implements the design discussed on llvm-dev for better tracking of
variables that live in memory through optimizations:
http://lists.llvm.org/pipermail/llvm-dev/2017-September/117222.html
This is tracked as PR34136
llvm.dbg.addr is intended to be produced and used in almost precisely
the same way as llvm.dbg.declare is today, with the exception that it is
control-dependent. That means that dbg.addr should always have a
position in the instruction stream, and it will allow passes that
optimize memory operations on local variables to insert llvm.dbg.value
calls to reflect deleted stores. See SourceLevelDebugging.rst for more
details.
The main drawback to generating DBG_VALUE machine instrs is that they
usually cause LLVM to emit a location list for DW_AT_location. The next
step will be to teach DwarfDebug.cpp how to recognize more DBG_VALUE
ranges as not needing a location list, and possibly start setting
DW_AT_start_offset for variables whose lifetimes begin mid-scope.
Reviewers: aprantl, dblaikie, probinson
Subscribers: eraman, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D37768
llvm-svn: 313825
Add support for passing SwiftError through a register on the Windows x64
calling convention. This allows the use of swifterror attributes on
parameters which is used by the swift front end for the `Error`
parameter. This partially enables building the swift standard library
for Windows x86_64.
llvm-svn: 313791
After r313775, it's easier to maintain a parallel BitVector of spilled
locations indexed by location number.
I wasn't able to build a good reduced test case for this iteration of
the bug, but I added a more direct assertion that spilled values must
use frame index locations. If this bug reappears, it won't only fire on
the NEON vector code that we detected it on, but on medium-sized
integer-only programs as well.
llvm-svn: 313786
Summary:
The new code should be linear in the number of DBG_VALUEs, while the old
code was quadratic. NFC intended.
This is also hopefully a more direct expression of the problem, which is
to:
1. Rewrite all virtual register operands to stack slots or physical
registers
2. Uniquely number those machine operands, assigning them location
numbers
3. Rewrite all uses of the old location numbers in the interval map to
use the new location numbers
In r313400, I attempted to track which locations were spilled in a
parallel bitvector indexed by location number. My code was broken
because these location numbers are not stable during rewriting.
Reviewers: aprantl, hans
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D38068
llvm-svn: 313775
This version of the patch fixes an off-by-one error causing PR34596. We
do not need to use std::next(BlockIter) when calling updateDepths, as
BlockIter already points to the next element.
Original commit message:
> For large basic blocks with lots of combinable instructions, the
> MachineTraceMetrics computations in MachineCombiner can dominate the compile
> time, as computing the trace information is quadratic in the number of
> instructions in a BB and it's relevant successors/predecessors.
> In most cases, knowing the instruction depth should be enough to make
> combination decisions. As we already iterate over all instructions in a basic
> block, the instruction depth can be computed incrementally. This reduces the
> cost of machine-combine drastically in cases where lots of instructions
> are combined. The major drawback is that AFAIK, computing the critical path
> length cannot be done incrementally. Therefore we only compute
> instruction depths incrementally, for basic blocks with more
> instructions than inc_threshold. The -machine-combiner-inc-threshold
> option can be used to set the threshold and allows for easier
> experimenting and checking if using incremental updates for all basic
> blocks has any impact on the performance.
>
> Reviewers: sanjoy, Gerolf, MatzeB, efriedma, fhahn
>
> Reviewed By: fhahn
>
> Subscribers: kiranchandramohan, javed.absar, efriedma, llvm-commits
>
> Differential Revision: https://reviews.llvm.org/D36619
llvm-svn: 313751
This re-applies commit r313685, this time with the proper updates to
the test cases.
Original commit message:
Unreachable blocks in the machine instr representation are these
weird empty blocks with no successors.
The MIR printer used to not print empty lists of successors. However,
the MIR parser now treats non-printed list of successors as "please
guess it for me". As a result, the parser tries to guess the list of
successors and given the block is empty, just assumes it falls through
the next block (if any).
For instance, the following test case used to fail the verifier.
The MIR printer would print
entry
/ \
true (def) false (no list of successors)
|
split.true (use)
The MIR parser would understand this:
entry
/ \
true (def) false
| / <-- invalid edge
split.true (use)
Because of the invalid edge, we get the "def does not
dominate all uses" error.
The fix consists in printing empty successor lists, so that the parser
knows what to do for unreachable blocks.
rdar://problem/34022159
llvm-svn: 313696
Unreachable blocks in the machine instr representation are these
weird empty blocks with no successors.
The MIR printer used to not print empty lists of successors. However,
the MIR parser now treats non-printed list of successors as "please
guess it for me". As a result, the parser tries to guess the list of
successors and given the block is empty, just assumes it falls through
the next block (if any).
For instance, the following test case used to fail the verifier.
The MIR printer would print
entry
/ \
true (def) false (no list of successors)
|
split.true (use)
The MIR parser would understand this:
entry
/ \
true (def) false
| / <-- invalid edge
split.true (use)
Because of the invalid edge, we get the "def does not
dominate all uses" error.
The fix consists in printing empty successor lists, so that the parser
knows what to do for unreachable blocks.
rdar://problem/34022159
llvm-svn: 313685
This reverts r313640, originally r313400, one more time for essentially
the same issue. My BitVector of spilled location numbers isn't working
because we coalesce identical DBG_VALUE locations as we rewrite them,
invalidating the location numbers used to index the BitVector.
llvm-svn: 313679
I forgot to zero out the BitVector when reusing it between UserValues.
Later uses of the same location number for a different UserValue would
falsely indicate that they were spilled. Usually this would lead to
incorrect debug info, but in some cases they would indicate something
nonsensical like a memory location based on a vector register (Q8 on
ARM).
llvm-svn: 313640
This caused asserts in Chromium. See http://crbug.com/766261
> Summary:
> This comes up in optimized debug info for C++ programs that pass and
> return objects indirectly by address. In these programs,
> llvm.dbg.declare survives optimization, which causes us to emit indirect
> DBG_VALUE instructions. The fast register allocator knows to insert
> DW_OP_deref when spilling indirect DBG_VALUE instructions, but the
> LiveDebugVariables did not until this change.
>
> This fixes part of PR34513. I need to look into why this doesn't work at
> -O0 and I'll send follow up patches to handle that.
>
> Reviewers: aprantl, dblaikie, probinson
>
> Subscribers: qcolombet, hiraditya, llvm-commits
>
> Differential Revision: https://reviews.llvm.org/D37911
llvm-svn: 313589
If we have an AssertZext of a truncated value that has already been AssertZext'ed,
we can assert on the wider source op to improve the zext-y knowledge:
assert (trunc (assert X, i8) to iN), i1 --> trunc (assert X, i1) to iN
This moves a fold from being Mips-specific to general combining, and x86 shows
improvements.
Differential Revision: https://reviews.llvm.org/D37017
llvm-svn: 313577
rL310710 allowed store merging to occur after legalization to catch stores that are created late,
but this exposes a logic hole seen in PR34217:
https://bugs.llvm.org/show_bug.cgi?id=34217
We will miss merging stores if the target lowers vector extracts into target-specific operations.
This patch allows store merging to occur both before and after legalization if the target chooses
to get maximum merging.
I don't think the potential regressions in the other tests are relevant. The tests are for
correctness of weird IR constructs rather than perf tests, and I think those are still correct.
Differential Revision: https://reviews.llvm.org/D37987
llvm-svn: 313564
r313390 taught 'allowExtraAnalysis' to check whether remarks are
enabled at all. Use that to only do the expensive instruction printing
if they are.
llvm-svn: 313552
For cases where we are BITCASTing to vectors of smaller elements, then if the entire source was a splatted sign (src's NumSignBits == SrcBitWidth) we can say that the dst's NumSignBit == DstBitWidth, as we're just splitting those sign bits across multiple elements.
We could generalize this but at the moment the only use case I have is to peek through bitcasts to vector comparison results.
Differential Revision: https://reviews.llvm.org/D37849
llvm-svn: 313543
This reverts commit 6389e7aa724ea7671d096f4770f016c3d86b0d54.
There is a bug in this implementation where the string value of the
checksum is outputted, instead of the actual hex bytes. Therefore the
checksum is incorrect, and this prevent pdbs from being loaded by visual
studio. Revert this until the checksum is emitted correctly.
llvm-svn: 313431
Summary:
This comes up in optimized debug info for C++ programs that pass and
return objects indirectly by address. In these programs,
llvm.dbg.declare survives optimization, which causes us to emit indirect
DBG_VALUE instructions. The fast register allocator knows to insert
DW_OP_deref when spilling indirect DBG_VALUE instructions, but the
LiveDebugVariables did not until this change.
This fixes part of PR34513. I need to look into why this doesn't work at
-O0 and I'll send follow up patches to handle that.
Reviewers: aprantl, dblaikie, probinson
Subscribers: qcolombet, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D37911
llvm-svn: 313400
Summary:
Fixes PR34513.
Indirect DBG_VALUEs typically come from dbg.declares of non-trivially
copyable C++ objects that must be passed by address. We were already
handling the case where the virtual register gets allocated to a
physical register and is later spilled. That's what usually happens for
normal parameters that aren't NRVO variables: they usually appear in
physical register parameters, and are spilled later in the function,
which would correctly add deref.
NRVO variables are different because the dbg.declare can come much later
after earlier instructions cause the incoming virtual register to be
spilled.
Also, clean up this code. We only need to look at the first operand of a
DBG_VALUE, which eliminates the operand loop.
Reviewers: aprantl, dblaikie, probinson
Subscribers: MatzeB, qcolombet, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D37929
llvm-svn: 313399
This means that we can honor -fdata-sections rather than
always creating a segment for each symbol.
It also allows for a followup change to add .init_array and friends.
Differential Revision: https://reviews.llvm.org/D37876
llvm-svn: 313395
Previously the 'Padding' argument was the number of padding
bytes to add. However most callers that use 'Padding' know
how many overall bytes they need to write. With the previous
code this would mean encoding the LEB once to find out how
many bytes it would occupy and then using this to calulate
the 'Padding' value.
See: https://reviews.llvm.org/D36595
Differential Revision: https://reviews.llvm.org/D37494
llvm-svn: 313393
This caused PR34629: asserts firing when building Chromium. It also broke some
buildbots building test-suite as reported on the commit thread.
> Summary:
> 1/ Operand folding during complex pattern matching for LEAs has been
> extended, such that it promotes Scale to accommodate similar operand
> appearing in the DAG.
> e.g.
> T1 = A + B
> T2 = T1 + 10
> T3 = T2 + A
> For above DAG rooted at T3, X86AddressMode will no look like
> Base = B , Index = A , Scale = 2 , Disp = 10
>
> 2/ During OptimizeLEAPass down the pipeline factorization is now performed over LEAs
> so that if there is an opportunity then complex LEAs (having 3 operands)
> could be factored out.
> e.g.
> leal 1(%rax,%rcx,1), %rdx
> leal 1(%rax,%rcx,2), %rcx
> will be factored as following
> leal 1(%rax,%rcx,1), %rdx
> leal (%rdx,%rcx) , %edx
>
> 3/ Aggressive operand folding for AM based selection for LEAs is sensitive to loops,
> thus avoiding creation of any complex LEAs within a loop.
>
> Reviewers: lsaba, RKSimon, craig.topper, qcolombet
>
> Reviewed By: lsaba
>
> Subscribers: spatel, igorb, llvm-commits
>
> Differential Revision: https://reviews.llvm.org/D35014
llvm-svn: 313376
Summary:
The checksums had already been placed in the IR, this patch allows
MCCodeView to actually write it out to an MCStreamer.
Subscribers: llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D37157
llvm-svn: 313374
removing them"
This was temporarily reverted, but now that the fix has been commited (r313197)
it should be put back in place.
https://bugs.llvm.org/show_bug.cgi?id=34502
This reverts commit 9ef93d9dc4c51568e858cf8203cd2c5ce8dca796.
llvm-svn: 313349
Summary:
1/ Operand folding during complex pattern matching for LEAs has been
extended, such that it promotes Scale to accommodate similar operand
appearing in the DAG.
e.g.
T1 = A + B
T2 = T1 + 10
T3 = T2 + A
For above DAG rooted at T3, X86AddressMode will no look like
Base = B , Index = A , Scale = 2 , Disp = 10
2/ During OptimizeLEAPass down the pipeline factorization is now performed over LEAs
so that if there is an opportunity then complex LEAs (having 3 operands)
could be factored out.
e.g.
leal 1(%rax,%rcx,1), %rdx
leal 1(%rax,%rcx,2), %rcx
will be factored as following
leal 1(%rax,%rcx,1), %rdx
leal (%rdx,%rcx) , %edx
3/ Aggressive operand folding for AM based selection for LEAs is sensitive to loops,
thus avoiding creation of any complex LEAs within a loop.
Reviewers: lsaba, RKSimon, craig.topper, qcolombet
Reviewed By: lsaba
Subscribers: spatel, igorb, llvm-commits
Differential Revision: https://reviews.llvm.org/D35014
llvm-svn: 313343
These are removed in C++17. We still have some users of
unary_function::argument_type, so just spell that typedef out. No
functionality change intended.
Note that many of the argument types are actually wrong :)
llvm-svn: 313287
This replaces TableGen's type inference to operate on parameterized
types instead of MVTs, and as a consequence, some interfaces have
changed:
- Uses of MVTs are replaced by ValueTypeByHwMode.
- EEVT::TypeSet is replaced by TypeSetByHwMode.
This affects the way that types and type sets are printed, and the
tests relying on that have been updated.
There are certain users of the inferred types outside of TableGen
itself, namely FastISel and GlobalISel. For those users, the way
that the types are accessed have changed. For typical scenarios,
these replacements can be used:
- TreePatternNode::getType(ResNo) -> getSimpleType(ResNo)
- TreePatternNode::hasTypeSet(ResNo) -> hasConcreteType(ResNo)
- TypeSet::isConcrete -> TypeSetByHwMode::isValueTypeByHwMode(false)
For more information, please refer to the review page.
Differential Revision: https://reviews.llvm.org/D31951
llvm-svn: 313271
We already have a combine for this pattern when the input to shl is add, so we just need to enable the transformation when the input is or.
Original patch by @tstellar
Differential Revision: https://reviews.llvm.org/D19325
llvm-svn: 313251
Use RotAmt.urem(VTBits) instead of AND(RotAmt, VTBits - 1)
TBH I don't expect non-power-of-2 types to be created, but it makes the logic clearer and matches what we do in other rotation combines.
llvm-svn: 313245
Summary:
XRay had been assuming that the previous section is the "text" section
of the function when lowering the instrumentation map. Unfortunately
this is not a safe assumption, because we may be coming from lowering
debug type information for the function being lowered.
This fixes an issue with combining -gsplit-dwarf, -generate-type-units,
-debug-compile and -fxray-instrument for sole member functions. When the
split dwarf section is stripped, we're left with references from the
xray_instr_map to the debug section. The change now uses the function's
symbol instead of the previous section's start symbol.
We found the bug while attempting to strip the split debug sections off
an XRay-instrumented object file, which had a peculiar edge-case for
single-function classes where the single function is being lowered.
Because XRay had assocaited the instrumentation map for a function to
the debug types section instead of the function's section, the objcopy
call will fail due to the misplaced reference from the xray_instr_map
section.
Reviewers: pcc, dblaikie, echristo
Subscribers: llvm-commits, aprantl
Differential Revision: https://reviews.llvm.org/D37791
llvm-svn: 313233
This caused PR34596.
> [MachineCombiner] Update instruction depths incrementally for large BBs.
>
> Summary:
> For large basic blocks with lots of combinable instructions, the
> MachineTraceMetrics computations in MachineCombiner can dominate the compile
> time, as computing the trace information is quadratic in the number of
> instructions in a BB and it's relevant successors/predecessors.
>
> In most cases, knowing the instruction depth should be enough to make
> combination decisions. As we already iterate over all instructions in a basic
> block, the instruction depth can be computed incrementally. This reduces the
> cost of machine-combine drastically in cases where lots of instructions
> are combined. The major drawback is that AFAIK, computing the critical path
> length cannot be done incrementally. Therefore we only compute
> instruction depths incrementally, for basic blocks with more
> instructions than inc_threshold. The -machine-combiner-inc-threshold
> option can be used to set the threshold and allows for easier
> experimenting and checking if using incremental updates for all basic
> blocks has any impact on the performance.
>
> Reviewers: sanjoy, Gerolf, MatzeB, efriedma, fhahn
>
> Reviewed By: fhahn
>
> Subscribers: kiranchandramohan, javed.absar, efriedma, llvm-commits
>
> Differential Revision: https://reviews.llvm.org/D36619
llvm-svn: 313213
MachineScheduler when clustering loads or stores checks if base
pointers point to the same memory. This check is done through
comparison of base registers of two memory instructions. This
works fine when instructions have separate offset operand. If
they require a full calculated pointer such instructions can
never be clustered according to such logic.
Changed shouldClusterMemOps to accept base registers as well and
let it decide what to do about it.
Differential Revision: https://reviews.llvm.org/D37698
llvm-svn: 313208
Previously we used a size of '1' for VLAs because we weren't sure what
MSVC did. However, MSVC does support declaring an array without a size,
for which it emits an array type with a size of zero. Clang emits the
same DI metadata for VLAs and arrays without bound, so we would describe
arrays without bound as having one element. This lead to Microsoft
debuggers only printing a single element.
Emitting a size of zero appears to cause these debuggers to search the
symbol information to find a definition of the variable with accurate
array bounds.
Fixes http://crbug.com/763580
llvm-svn: 313203
This is to fix PR34502. After rL311401, the live range of spilled vreg will be
cleared. HoistSpill need to use the live range of the original vreg before splitting
to know the moving range of the spills. The patch saves a copy of live interval for
the spilled vreg inside of HoistSpillHelper.
Differential Revision: https://reviews.llvm.org/D37578
llvm-svn: 313197
Summary:
To improve CodeView quality for static member functions, we need to make the
static explicit. In addition to a small change in LLVM's CodeViewDebug to
return the appropriate MethodKind, this requires a small change in Clang to
note the staticness in the debug info metadata.
Subscribers: aprantl, hiraditya
Differential Revision: https://reviews.llvm.org/D37715
llvm-svn: 313192
This flag is unnecessary for testing because we can get the coverage
we need by adjusting CU attributes.
Differential Revision: https://reviews.llvm.org/D37725
llvm-svn: 313079
Implementing this pass as a PowerPC specific pass. Branch coalescing utilizes
the analyzeBranch method which currently does not include any implicit operands.
This is not an issue on PPC but must be handled on other targets.
Pass is currently off by default. Enabled via -enable-ppc-branch-coalesce.
Differential Revision : https: // reviews.llvm.org/D32776
llvm-svn: 313061
Looks like these were copied from the ELF sections but
don't apply to Wasm and were not used anywhere.
Also remove unused Wasm methods in MCContext.
Differential Revision: https://reviews.llvm.org/D37633
llvm-svn: 313058
A prologue-end line record is emitted with an incorrect associated address,
which causes a debugger to show the beginning of function body to be inside
the prologue.
Patch written by Carlos Alberto Enciso.
Differential Revision: https://reviews.llvm.org/D37625
llvm-svn: 313047
Summary:
GEP merging can sometimes increase the number of live values and register
pressure across control edges and cause performance problems particularly if the
increased register pressure results in spills.
This change implements GEP unmerging around an IndirectBr in certain cases to
mitigate the issue. This is in the CodeGenPrepare pass (after all the GEP
merging has happened.)
With this patch, the Python interpreter loop runs faster by ~5%.
Reviewers: sanjoy, hfinkel
Reviewed By: hfinkel
Subscribers: eastig, junbuml, llvm-commits
Differential Revision: https://reviews.llvm.org/D36772
llvm-svn: 312930
After the split of the Scatter operation, the order of the new instructions is well defined - Lo goes before Hi. Otherwise the semantic of Scatter (from LSB to MSB) is broken.
I'm chaining 2 nodes to prevent reordering.
Differential Revision https://reviews.llvm.org/D37670
llvm-svn: 312894
- Use range based for
- Variable names should start with upper case
- Add `const`
- Change class name to match filename
- Fix doxygen comments
- Use MCPhysReg instead of unsigned
- Use references instead of pointers where things cannot be nullptr
- Misc coding style improvements
llvm-svn: 312846
rL312641 Allowed llvm.memcpy/memset/memmove to be tail calls when parent
function return the intrinsics's first argument. However on arm-none-eabi
platform, llvm.memcpy will be expanded to __aeabi_memcpy which doesn't
have return value. The fix is to check the libcall name after expansion
to match "memcpy/memset/memmove" before allowing those intrinsic to be
tail calls.
llvm-svn: 312799
Summary:
This fixes code-gen for XRay in PPC. The regression wasn't caught by
codegen tests which we add in this change.
What happened was the following:
- For tail exits, we used to unconditionally prepend the returns/exits
with a pseudo-instruction that gets lowered to the instrumentation
sled (and leave the actual return/exit instruction as-is).
- Changes to the XRay instrumentation pass caused the tail exits to
suddenly also emit the tail exit pseudo-instruction, since the check
for whether a return instruction was also a call instruction meant it
was a tail exit instruction.
- None of the tests caught the regression either due to non-existent
tests, or the tests being disabled/removed for continuous breakage.
This change re-introduces some of the basic tests and verifies that
we're back to a state that allows the back-end to generate appropriate
XRay instrumented binaries for PPC in the presence of tail exits.
Reviewers: echristo, timshen
Subscribers: nemanjai, kbarton, llvm-commits
Differential Revision: https://reviews.llvm.org/D37570
llvm-svn: 312772
r312318 - Debug info for variables whose type is shrinked to bool
r312325, r312424, r312489 - Test case for r312318
Revision 312318 introduced a null dereference bug.
Details in https://bugs.llvm.org/show_bug.cgi?id=34490
llvm-svn: 312758
Fixes some combine issues for AMDGPU where we weren't
getting the many extract_vector_elt combines expected
in a future patch.
This should really be checking isOperationLegalOrCustom on
the extract. That improves a number of x86 lit tests, but
a few get stuck in an infinite loop from one place
where a similar looking extract is created. I have a
different workaround in the backend for that which
keeps many of those improvements, but also adds a few
regressions.
llvm-svn: 312730
Summary:
For large basic blocks with lots of combinable instructions, the
MachineTraceMetrics computations in MachineCombiner can dominate the compile
time, as computing the trace information is quadratic in the number of
instructions in a BB and it's relevant successors/predecessors.
In most cases, knowing the instruction depth should be enough to make
combination decisions. As we already iterate over all instructions in a basic
block, the instruction depth can be computed incrementally. This reduces the
cost of machine-combine drastically in cases where lots of instructions
are combined. The major drawback is that AFAIK, computing the critical path
length cannot be done incrementally. Therefore we only compute
instruction depths incrementally, for basic blocks with more
instructions than inc_threshold. The -machine-combiner-inc-threshold
option can be used to set the threshold and allows for easier
experimenting and checking if using incremental updates for all basic
blocks has any impact on the performance.
Reviewers: sanjoy, Gerolf, MatzeB, efriedma, fhahn
Reviewed By: fhahn
Subscribers: kiranchandramohan, javed.absar, efriedma, llvm-commits
Differential Revision: https://reviews.llvm.org/D36619
llvm-svn: 312719
Summary:
This function is used in D36619 to update the instruction depths
incrementally.
Reviewers: efriedma, Gerolf, MatzeB, fhahn
Reviewed By: fhahn
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D36696
llvm-svn: 312714
Tail merging can convert an undef use into a normal one when creating a
common tail. Doing so can make the register live out from a block which
previously contained the undef use. To keep the liveness up-to-date,
insert IMPLICIT_DEFs in such blocks when necessary.
To enable this patch the computeLiveIns() function which used to
compute live-ins for a block and set them immediately is split into new
functions:
- computeLiveIns() just computes the live-ins in a LivePhysRegs set.
- addLiveIns() applies the live-ins to a block live-in list.
- computeAndAddLiveIns() is a convenience function combining the other
two functions and behaving like computeLiveIns() before this patch.
Based on a patch by Krzysztof Parzyszek <kparzysz@codeaurora.org>
Differential Revision: https://reviews.llvm.org/D37034
llvm-svn: 312668
When if-converting a diamond, two separate blocks will be placed back
to back to form a straight line code. To ensure correctness of the
liveness information, any registers that are live in the second block
should not be killed in the first block, even if they were in the
original code.
Additionally, when the two blocks share common instructions at the
beginning, these instructions will not be duplicated, but only placed
once, before both of the blocks. Since the function "isIdenticalTo"
(as used here) ignores kill flags, the common initial code in one
block may have a kill flag for a register that is live in the other
block.
Because the code that removes kill flags only runs for the non-common
parts of the predicated blocks, a kill flag mismatch in the common
code could still lead to a live register being killed prematurely.
llvm-svn: 312654
function return the intrinsics's first argument.
llvm.memcpy/memset/memmove return void but they will return the first
argument after they are expanded as libcalls. Now if the parent function
has any return value, llvm.memcpy cannot be turned into tail call after
expansion.
The patch is to handle that case in SelectionDAGBuilder so when caller
function return the same value as the first argument of llvm.memcpy,
tail call is allowed.
Differential Revision: https://reviews.llvm.org/D37406
llvm-svn: 312641
S_UDT records are basically the "bridge" between the debugger's
expression evaluator and the type information. If you type
(Foo*)nullptr into the watch window, the debugger looks for an
S_UDT record named Foo. If it can find one, it displays your type.
Otherwise you get an error.
We have always understood this to mean that if you have code like
this:
struct A {
int X;
};
struct B {
typedef A AT;
AT Member;
};
that you will get 3 S_UDT records. "A", "B", and "B::AT". Because
if you were to type (B::AT*)nullptr into the debugger, it would
need to find an S_UDT record named "B::AT".
But "B::AT" is actually the S_UDT record that would be generated
if B were a namespace, not a struct. So the debugger needs to be
able to distinguish this case. So what it does is:
1. Look for an S_UDT named "B::AT". If it finds one, it knows
that AT is in a namespace.
2. If it doesn't find one, split at the scope resolution operator,
and look for an S_UDT named B. If it finds one, look up the type
for B, and then look for AT as one of its members.
With this algorithm, S_UDT records for nested typedefs are not just
unnecessary, but actually wrong!
The results of implementing this in clang are dramatic. It cuts
our /DEBUG:FASTLINK PDB sizes by more than 50%, and we go from
being ~20% larger than MSVC PDBs on average, to ~40% smaller.
It also slightly speeds up link time. We get about 10% faster
links than without this patch.
Differential Revision: https://reviews.llvm.org/D37410
llvm-svn: 312583
Summary:
This intrinsic represents a label with a list of associated metadata
strings. It is modelled as reading and writing inaccessible memory so
that it won't be removed as dead code. I think the intention is that the
annotation strings should appear at most once in the debug info, so I
marked it noduplicate. We are allowed to inline code with annotations as
long as we strip the annotation, but that can be done later.
Reviewers: majnemer
Subscribers: eraman, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D36904
llvm-svn: 312569
Summary:
This is a re-roll of D36615 which uses PLT relocations in the back-end
to the call to __xray_CustomEvent() when building in -fPIC and
-fxray-instrument mode.
Reviewers: pcc, djasper, bkramer
Subscribers: sdardis, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D37373
llvm-svn: 312466
The function combineShuffleToVectorExtend in DAGCombine might generate an illegal typed node after "legalize types" phase, causing assertion on non-simple type to fail afterwards.
Adding a type check in case the combine is running after the type legalize pass.
Differential Revision: https://reviews.llvm.org/D37330
llvm-svn: 312438
If getHexUint reads in a hex 0, it will create an APInt with a value of 0.
The number of active bits on this APInt is used to calculate the bitwidth of
Result. The number of active bits is defined as an APInt's bitwidth - its
number of leading 0s. Since this APInt is 0, its bitwidth and number of leading
0s are equal.
Thus, Result is constructed with a bitwidth of 0, triggering an APInt assert.
This commit fixes that by checking if the APInt is equal to 0, and setting the
bitwidth to 32 if it is. Otherwise, it sets the bitwidth using getActiveBits.
This caused issues when compiling MIR files with successor probabilities. In
the case that a successor is tagged with a probability of 0, this assert would
fire on debug builds.
https://reviews.llvm.org/D37401
llvm-svn: 312387
A register in CodeGen can be marked as reserved: In that case we
consider the register always live and do not use (or rather ignore)
kill/dead/undef operand flags.
LiveIntervalAnalysis however tracks liveness per register unit (not per
register). We already needed adjustments for this in r292871 to deal
with super/sub registers. However I did not look at aliased register
there. Looking at ARM:
FPSCR (regunits FPSCR, FPSCR~FPSCR_NZCV) aliases with FPSCR_NZCV
(regunits FPSCR_NZCV, FPSCR~FPSCR_NZCV) hence they share a register unit
(FPSCR~FPSCR_NZCV) that represents the aliased parts of the registers.
This shared register unit was previously considered non-reserved,
however given that we uses of the reserved FPSCR potentially violate
some rules (like uses without defs) we should make FPSCR~FPSCR_NZCV
reserved too and stop tracking liveness for it.
This patch:
- Defines a register unit as reserved when: At least for one root
register, the root register and all its super registers are reserved.
- Adjust LiveIntervals::computeRegUnitRange() for new reserved
definition.
- Add MachineRegisterInfo::isReservedRegUnit() to have a canonical way
of testing.
- Stop computing LiveRanges for reserved register units in HMEditor even
with UpdateFlags enabled.
- Skip verification of uses of reserved reg units in the machine
verifier (this usually didn't happen because there would be no cached
liverange but there is no guarantee for that and I would run into this
case before the HMEditor tweak, so may as well fix the verifier too).
Note that this should only affect ARMs FPSCR/FPSCR_NZCV registers today;
aliased registers are rarely used, the only other cases are hexagons
P0-P3/P3_0 and C8/USR pairs which are not mixing reserved/non-reserved
registers in an alias.
Differential Revision: https://reviews.llvm.org/D37356
llvm-svn: 312348
Issues addressed since original review:
- Moved removal of dead instructions found by
LiveIntervals::shrinkToUses() outside of loop iterating over
instructions to avoid instructions being deleted while pointed to by
iterator.
- Fixed ARMLoadStoreOptimizer bug exposed by this change in r311907.
- The pass no longer forwards COPYs to physical register uses, since
doing so can break code that implicitly relies on the physical
register number of the use.
- The pass no longer forwards COPYs to undef uses, since doing so
can break the machine verifier by creating LiveRanges that don't
end on a use (since the undef operand is not considered a use).
[MachineCopyPropagation] Extend pass to do COPY source forwarding
This change extends MachineCopyPropagation to do COPY source forwarding.
This change also extends the MachineCopyPropagation pass to be able to
be run during register allocation, after physical registers have been
assigned, but before the virtual registers have been re-written, which
allows it to remove virtual register COPY LiveIntervals that become dead
through the forwarding of all of their uses.
llvm-svn: 312328
This patch provides such debug information for integer
variables whose type is shrinked to bool by providing
dwarf expression which returns either constant initial
value or other value.
Patch by Nikola Prica.
Differential Revision: https://reviews.llvm.org/D35994
llvm-svn: 312318
comparisons into memcmp.
Thanks to recent improvements in the LLVM codegen, the memcmp is typically
inlined as a chain of efficient hardware comparisons.
This typically benefits C++ member or nonmember operator==().
For now this is disabled by default until:
- https://bugs.llvm.org/show_bug.cgi?id=33329 is complete
- Benchmarks show that this is always useful.
Differential Revision:
https://reviews.llvm.org/D33987
llvm-svn: 312315
Matt Arsenault