This is just as fast, and it makes it possible to avoid leaking the
UsedPhysRegs BitVector implementation through
MachineRegisterInfo::addPhysRegsUsed().
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166083 91177308-0d34-0410-b5e6-96231b3b80d8
Stack is formed improperly for long structures passed as byval arguments for
EABI mode.
If we took AAPCS reference, we can found the next statements:
A: "If the argument requires double-word alignment (8-byte), the NCRN (Next
Core Register Number) is rounded up to the next even register number." (5.5
Parameter Passing, Stage C, C.3).
B: "The alignment of an aggregate shall be the alignment of its most-aligned
component." (4.3 Composite Types, 4.3.1 Aggregates).
So if we have structure with doubles (9 double fields) and 3 Core unused
registers (r1, r2, r3): caller should use r2 and r3 registers only.
Currently r1,r2,r3 set is used, but it is invalid.
Callee VA routine should also use r2 and r3 regs only. All is ok here. This
behaviour is guessed by rounding up SP address with ADD+BFC operations.
Fix:
Main fix is in ARMTargetLowering::HandleByVal. If we detected AAPCS mode and
8 byte alignment, we waste odd registers then.
P.S.:
I also improved LDRB_POST_IMM regression test. Since ldrb instruction will
not generated by current regression test after this patch.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166018 91177308-0d34-0410-b5e6-96231b3b80d8
This is a medium term workaround until we have a more robust solution
in the form of a register liveness utility for postRA passes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166001 91177308-0d34-0410-b5e6-96231b3b80d8
Using the cached bit vector in MRI avoids comstantly allocating and
recomputing the reserved register bit vector.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165983 91177308-0d34-0410-b5e6-96231b3b80d8
Also provide an MRI::getReservedRegs() function to access the frozen
register set, and isReserved() and isAllocatable() methods to test
individual registers.
The various implementations of TRI::getReservedRegs() are quite
complicated, and many passes need to look at the reserved register set.
This patch makes it possible for these passes to use the cached copy in
MRI, avoiding a lot of malloc traffic and repeated calculations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165982 91177308-0d34-0410-b5e6-96231b3b80d8
inline assembly. For the time being, these will be called directly by clang.
However, in the near future I expect these to be sunk back into the MC layer
and more basic APIs (e.g., getClobbers(), getConstraints(), etc.) will be called
by clang.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165946 91177308-0d34-0410-b5e6-96231b3b80d8
This patch replaces the EmitRawText by a EmitTCEntry class (specialized for
each Streamer) in PowerPC64 TOC entry creation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165940 91177308-0d34-0410-b5e6-96231b3b80d8
Convert the internal representation of the Attributes class into a pointer to an
opaque object that's uniqued by and stored in the LLVMContext object. The
Attributes class then becomes a thin wrapper around this opaque
object. Eventually, the internal representation will be expanded to include
attributes that represent code generation options, etc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165917 91177308-0d34-0410-b5e6-96231b3b80d8
This patch implements the new LibCallSimplifier class as outlined in [1].
In addition to providing the new base library simplification infrastructure,
all the fortified library call simplifications were moved over to the new
infrastructure. The rest of the library simplification optimizations will
be moved over with follow up patches.
NOTE: The original fortified library call simplifier located in the
SimplifyFortifiedLibCalls class was not removed because it is still
used by CodeGenPrepare. This class will eventually go away too.
[1] http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-August/052283.html
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165873 91177308-0d34-0410-b5e6-96231b3b80d8
the interface between the front-end and the MC layer when parsing inline
assembly. Unfortunately, this is too deep into the parsing stack. Specifically,
we're unable to handle target-independent assembly (i.e., assembly directives,
labels, etc.). Note the MatchAndEmitInstruction() isn't the correct
abstraction either. I'll be exposing target-independent hooks shortly, so this
is really just a cleanup.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165858 91177308-0d34-0410-b5e6-96231b3b80d8
isa<> et al. automatically infer when the cast is an upcast (including a
self-cast), so these are no longer necessary.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165767 91177308-0d34-0410-b5e6-96231b3b80d8
Additionally, all such cases are handled with no dynamic check.
All `classof()` of the form
class Foo {
[...]
static bool classof(const Bar *) { return true; }
[...]
}
where Foo is an ancestor of Bar are no longer necessary.
Don't write them!
Note: The exact test is `is_base_of<Foo, Bar>`, which is non-strict, so
that Foo is considered an ancestor of itself.
This leads to the following rule of thumb for LLVM-style RTTI:
The argument type of `classof()` should be a strict ancestor.
For more information about implementing LLVM-style RTTI, see
docs/HowToSetUpLLVMStyleRTTI.rst
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165765 91177308-0d34-0410-b5e6-96231b3b80d8
to the instruction position. The old encoding would give an absolute
ID which counts up within a function, and only resets at the next function.
I.e., Instead of having:
... = icmp eq i32 n-1, n-2
br i1 ..., label %bb1, label %bb2
it will now be roughly:
... = icmp eq i32 1, 2
br i1 1, label %bb1, label %bb2
This makes it so that ids remain relatively small and can be encoded
in fewer bits.
With this encoding, forward reference operands will be given
negative-valued IDs. Use signed VBRs for the most common case
of forward references, which is phi instructions.
To retain backward compatibility we bump the bitcode version
from 0 to 1 to distinguish between the different encodings.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165739 91177308-0d34-0410-b5e6-96231b3b80d8
Patch from Preston Briggs <preston.briggs@gmail.com>.
This is an updated version of the dependence-analysis patch, including an MIV
test based on Banerjee's inequalities.
It's a fairly complete implementation of the paper
Practical Dependence Testing
Gina Goff, Ken Kennedy, and Chau-Wen Tseng
PLDI 1991
It cannot yet propagate constraints between coupled RDIV subscripts (discussed
in Section 5.3.2 of the paper).
It's organized as a FunctionPass with a single entry point that supports testing
for dependence between two instructions in a function. If there's no dependence,
it returns null. If there's a dependence, it returns a pointer to a Dependence
which can be queried about details (what kind of dependence, is it loop
independent, direction and distance vector entries, etc). I haven't included
every imaginable feature, but there's a good selection that should be adequate
for supporting many loop transformations. Of course, it can be extended as
necessary.
Included in the patch file are many test cases, commented with C code showing
the loops and array references.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165708 91177308-0d34-0410-b5e6-96231b3b80d8
SchedulerDAGInstrs::buildSchedGraph ignores dependencies between FixedStack
objects and byval parameters. So loading byval parameters from stack may be
inserted *before* it will be stored, since these operations are treated as
independent.
Fix:
Currently ARMTargetLowering::LowerFormalArguments saves byval registers with
FixedStack MachinePointerInfo. To fix the problem we need to store byval
registers with MachinePointerInfo referenced to first the "byval" parameter.
Also commit adds two new fields to the InputArg structure: Function's argument
index and InputArg's part offset in bytes relative to the start position of
Function's argument. E.g.: If function's argument is 128 bit width and it was
splitted onto 32 bit regs, then we got 4 InputArg structs with same arg index,
but different offset values.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165616 91177308-0d34-0410-b5e6-96231b3b80d8
Allows the new machine model to be used for NumMicroOps and OutputLatency.
Allows the HazardRecognizer to be disabled along with itineraries.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165603 91177308-0d34-0410-b5e6-96231b3b80d8
This wasn't contributing anything significant to postRA heuristics except compile time (by my measurements) and will be replaced by a more general heuristic for cross-region dependencies within the scheduler itself.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165563 91177308-0d34-0410-b5e6-96231b3b80d8
The next step is to update the optimizers to allow them to optimize the different address spaces with this information.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165505 91177308-0d34-0410-b5e6-96231b3b80d8
DeadArgumentElimination pass can replace one LLVM function with another,
invalidating a pointer stored in debug info metadata entry for this function.
To fix this, we collect debug info descriptors for functions before
running a DeadArgumentElimination pass and "patch" pointers in metadata nodes
if we replace a function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165490 91177308-0d34-0410-b5e6-96231b3b80d8
We use the enums to query whether an Attributes object has that attribute. The
opaque layer is responsible for knowing where that specific attribute is stored.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165488 91177308-0d34-0410-b5e6-96231b3b80d8
This class is used by LSR and a number of places in the codegen.
This is the first step in de-coupling LSR from TLI, and creating
a new interface in between them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165455 91177308-0d34-0410-b5e6-96231b3b80d8
Start using the AttributesImpl object to hold the value of the attributes. All
queries go through the interfaces now.
This has one unfortunate consequence. I needed to move the AttributesImpl.h file
into include/llvm. But this is only temporary! Otherwise, the changes needed to
support this would be too large.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165433 91177308-0d34-0410-b5e6-96231b3b80d8
The internal representation of the Attributes class will be opaque. All of the
query methods will need to query the opaque class. Therefore, these methods need
to be out-of-line.
No functionality change intended.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165305 91177308-0d34-0410-b5e6-96231b3b80d8
Use isa<> or cast<> when semantically that is what is happening. Also
some trivial "style" cleanups at fix sites.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165292 91177308-0d34-0410-b5e6-96231b3b80d8
This is a mechanical change of dynamic_cast<> to dyn_cast<>. A number of
these uses are actually more like isa<> or cast<>, and will be changed
to the semanticaly appropriate one in a future patch.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165291 91177308-0d34-0410-b5e6-96231b3b80d8
This allows the processor-specific machine model to override selected
base opcodes without any fanciness.
e.g. InstRW<[CoreXWriteVANDP], (instregex "VANDP")>.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165180 91177308-0d34-0410-b5e6-96231b3b80d8
In order to avoid rev-lock with Clang when moving to the new API, also
preserve the current API temporarily and insert a shim to implement the
new API in terms of the old.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165165 91177308-0d34-0410-b5e6-96231b3b80d8
most of the behavior we want, but wrap the predicate in one which erases
elements from the set if they pass the predicate. Oh what I wouldn't
give for a lambda here.
Let me know if the predicate wrapping is too much magic. ;]
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165076 91177308-0d34-0410-b5e6-96231b3b80d8
preserves the values of the relocated entries, unlikely remove_if. This
allows walking them and erasing them.
Also flesh out the predicate we are using for this to support the
various constraints actually imposed on a UnaryPredicate -- without this
we can't compose it with std::not1.
Thanks to Sean Silva for the review here and noticing the issue with
std::remove_if.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165073 91177308-0d34-0410-b5e6-96231b3b80d8
scheduled for processing on the worklist eventually gets deleted while
we are processing another alloca, fixing the original test case in
PR13990.
To facilitate this, add a remove_if helper to the SetVector abstraction.
It's not easy to use the standard abstractions for this because of the
specifics of SetVectors types and implementation.
Finally, a nice small test case is included. Thanks to Benjamin for the
fantastic reduced test case here! All I had to do was delete some empty
basic blocks!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165065 91177308-0d34-0410-b5e6-96231b3b80d8
of operand is specific to MS-style inline assembly and should not be generated
when parsing normal assembly.
The purpose of the wildcard operands are to allow the AsmParser to match
multiple instructions (i.e., MCInsts) to a given ms-style asm statement. For
the time being the matcher just returns the first match. This patch only
implements wildcard matches for memory operands. Support for register
wildcards will be added in the near future.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165057 91177308-0d34-0410-b5e6-96231b3b80d8
This adds 'elf' as a recognized target triple environment value and overrides the default generated object format on Windows platforms if that value is present. This patch also enables MCJIT tests on Windows using the new environment value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165030 91177308-0d34-0410-b5e6-96231b3b80d8
map constraints and MCInst operands to inline asm operands. This replaces the
getMCInstOperandNum() function.
The logic to determine the constraints are not in place, so we still default to
a register constraint (i.e., "r"). Also, we no longer build the MCInst but
rather return just the opcode to get the MCInstrDesc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164979 91177308-0d34-0410-b5e6-96231b3b80d8
The target backend can support data-in-code load commands even when
the assembler doesn't, or vice-versa. Allow targets to opt-in for
direct-to-object.
PR13973.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164974 91177308-0d34-0410-b5e6-96231b3b80d8
- Update maximal stack alignment when stack arguments are prepared before a
call.
- Test cases are enhanced to show it's not a Win32 specific issue but a generic
one.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164946 91177308-0d34-0410-b5e6-96231b3b80d8
Reduces runtime of i386-large-relocations.s by 10x in Release builds, even more
in Debug+Asserts builds.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164945 91177308-0d34-0410-b5e6-96231b3b80d8
folders and not having it here fails to compile if you actually try to use it.
Also, CreatePointerCast was failing to do the part where it does TD-aware
constant folding. Granted there is exactly one case where that it will ever
do anything, but there's no reason to skip it. For reference, that case is a
subtraction between two constant offsets on the same global variable, eg.,
"&A[123] - &A[4].f".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164760 91177308-0d34-0410-b5e6-96231b3b80d8
This method can be called with a '0' argument which checks the return
value. However, the method it calls doesn't expect '0' as a valid value. Call the
correct method when it's 0.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164735 91177308-0d34-0410-b5e6-96231b3b80d8
The hasFnAttr method has been replaced by querying the Attributes explicitly. No
intended functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164725 91177308-0d34-0410-b5e6-96231b3b80d8
This opaque class will contain all of the attributes. All attribute queries will
go through this object. This object will also be uniqued in the LLVMContext.
Currently not used, so no implementation change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164722 91177308-0d34-0410-b5e6-96231b3b80d8
- Put statistics in alphabetical order
- Don't use getZextValue when building TableInt, just use APInts
- Introduce Create{Z,S}ExtOrTrunc in IRBuilder.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164696 91177308-0d34-0410-b5e6-96231b3b80d8
and stores. These will be used in subsequnet patches to SROA to more
systematically manage the alignment on loads and stores.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164688 91177308-0d34-0410-b5e6-96231b3b80d8
Provide interface in TargetLowering to set or get the minimum number of basic
blocks whereby jump tables are generated for switch statements rather than an
if sequence.
getMinimumJumpTableEntries() defaults to 4.
setMinimumJumpTableEntries() allows target configuration.
This patch changes the default for the Hexagon architecture to 5
as it improves performance on some benchmarks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164628 91177308-0d34-0410-b5e6-96231b3b80d8
This silences several analyzer warnings within LLVM, and provides a slightly
nicer crash experience when someone calls isa<>, cast<>, or dyn_cast<> with
a null pointer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164439 91177308-0d34-0410-b5e6-96231b3b80d8
This silences literally dozens of analyzer warnings on LLVM (since DenseMap
is such a commonly-used class).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164438 91177308-0d34-0410-b5e6-96231b3b80d8
whether or not we want to print out backtrace information. Useful
for libraries that don't need backtrace information on a crash.
rdar://11844710
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164426 91177308-0d34-0410-b5e6-96231b3b80d8
The expression based expansion too often results in IR level optimizations
splitting the intermediate values into separate basic blocks, preventing
the formation of the VBSL instruction as the code author intended. In
particular, LICM would often hoist part of the computation out of a loop.
rdar://11011471
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164340 91177308-0d34-0410-b5e6-96231b3b80d8
Make the TargetPrefix setting one big setting instead of being spread out
everywhere. No functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164265 91177308-0d34-0410-b5e6-96231b3b80d8
This fixes some obscure failure cases involving registers defined inside multiclasses or foreach constructs that would not receive a unique ID, and would end up being omitted from the AsmMatcher tables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164251 91177308-0d34-0410-b5e6-96231b3b80d8
to improve compatibility with GNU as.
Based on a patch by PaX Team.
Fixed assertion failures on non-Darwin and added additional test cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164248 91177308-0d34-0410-b5e6-96231b3b80d8
provide insertion order iteration, instead of the old option of
DenseMap order iteration over keys and insertion order iteration over
values.
This is implemented by keeping two copies of each key.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164221 91177308-0d34-0410-b5e6-96231b3b80d8
This is a generally useful utility; there's no reason to have it hidden
in CodeGenDAGPatterns.cpp.
Also, rename it to fit the other comparators in Record.h
Review by Jakob.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164189 91177308-0d34-0410-b5e6-96231b3b80d8
store this and use it to not emit long nops when the CPU is geode which
doesnt support them.
Fixes PR11212.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164132 91177308-0d34-0410-b5e6-96231b3b80d8
Now where we used to call ReInitMCSubtargetInfo, we actually recompute
the same information as InitMCSubtargetInfo instead of only setting
the feature bits.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164105 91177308-0d34-0410-b5e6-96231b3b80d8
Add LIS::pruneValue() and extendToIndices(). These two functions are
used by the register coalescer when merging two live ranges requires
more than a trivial value mapping as supported by LiveInterval::join().
The pruneValue() function can remove the part of a value number that is
going to conflict in join(). Afterwards, extendToIndices can restore the
live range, using any new dominating value numbers and updating the SSA
form.
Use this complex value mapping to support merging a register into a
vector lane that has a conflicting value, but the clobbered lane is
undef.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164074 91177308-0d34-0410-b5e6-96231b3b80d8
A value that is live in to a basic block should be returned by valueIn()
in LiveRangeQuery(getMBBStartIdx(MBB)), unless it is a PHI-def which
should be returned by valueDefined() instead.
Current code isn't using this functionality. Future code will.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163990 91177308-0d34-0410-b5e6-96231b3b80d8
If a PHI value happens to be live out from the layout predecessor of its
def block, the def slot index will be in the middle of the segment:
%vreg11 = [192r,240B:0)[352r,416B:2)[416B,496r:1) 0@192r 1@480B-phi %2@352r
A LiveRangeQuery for 480 should return NULL from valueIn() since the
PHI value is defined at the block entry, not live in to the block.
No test case, future code depends on this functionality.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163971 91177308-0d34-0410-b5e6-96231b3b80d8
new one, and add support for running the new pass in that mode and in
that slot of the pass manager. With this the new pass can completely
replace the old one within the pipeline.
The strategy for enabling or disabling the SSAUpdater logic is to do it
by making the requirement of the domtree analysis optional. By default,
it is required and we get the standard mem2reg approach. This is usually
the desired strategy when run in stand-alone situations. Within the
CGSCC pass manager, we disable requiring of the domtree analysis and
consequentially trigger fallback to the SSAUpdater promotion.
In theory this would allow the pass to re-use a domtree if one happened
to be available even when run in a mode that doesn't require it. In
practice, it lets us have a single pass rather than two which was
simpler for me to wrap my head around.
There is a hidden flag to force the use of the SSAUpdater code path for
the purpose of testing. The primary testing strategy is just to run the
existing tests through that path. One notable difference is that it has
custom code to handle lifetime markers, and one of the tests has been
enhanced to exercise that code.
This has survived a bootstrap and the test suite without serious
correctness issues, however my run of the test suite produced *very*
alarming performance numbers. I don't entirely understand or trust them
though, so more investigation is on-going.
To aid my understanding of the performance impact of the new SROA now
that it runs throughout the optimization pipeline, I'm enabling it by
default in this commit, and will disable it again once the LNT bots have
picked up one iteration with it. I want to get those bots (which are
much more stable) to evaluate the impact of the change before I jump to
any conclusions.
NOTE: Several Clang tests will fail because they run -O3 and check the
result's order of output. They'll go back to passing once I disable it
again.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163965 91177308-0d34-0410-b5e6-96231b3b80d8
- The current_pos function is supposed to return all the written bytes, not the
current position of the underlying stream.
- This caused tell() to be broken whenever the underlying stream had buffered
content.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163948 91177308-0d34-0410-b5e6-96231b3b80d8
* wrap code blocks in \code ... \endcode;
* refer to parameter names in paragraphs correctly (\arg is not what most
people want -- it starts a new paragraph);
* use \param instead of \arg to document parameters in order to be consistent
with the rest of the codebase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163902 91177308-0d34-0410-b5e6-96231b3b80d8
This is essentially a ground up re-think of the SROA pass in LLVM. It
was initially inspired by a few problems with the existing pass:
- It is subject to the bane of my existence in optimizations: arbitrary
thresholds.
- It is overly conservative about which constructs can be split and
promoted.
- The vector value replacement aspect is separated from the splitting
logic, missing many opportunities where splitting and vector value
formation can work together.
- The splitting is entirely based around the underlying type of the
alloca, despite this type often having little to do with the reality
of how that memory is used. This is especially prevelant with unions
and base classes where we tail-pack derived members.
- When splitting fails (often due to the thresholds), the vector value
replacement (again because it is separate) can kick in for
preposterous cases where we simply should have split the value. This
results in forming i1024 and i2048 integer "bit vectors" that
tremendously slow down subsequnet IR optimizations (due to large
APInts) and impede the backend's lowering.
The new design takes an approach that fundamentally is not susceptible
to many of these problems. It is the result of a discusison between
myself and Duncan Sands over IRC about how to premptively avoid these
types of problems and how to do SROA in a more principled way. Since
then, it has evolved and grown, but this remains an important aspect: it
fixes real world problems with the SROA process today.
First, the transform of SROA actually has little to do with replacement.
It has more to do with splitting. The goal is to take an aggregate
alloca and form a composition of scalar allocas which can replace it and
will be most suitable to the eventual replacement by scalar SSA values.
The actual replacement is performed by mem2reg (and in the future
SSAUpdater).
The splitting is divided into four phases. The first phase is an
analysis of the uses of the alloca. This phase recursively walks uses,
building up a dense datastructure representing the ranges of the
alloca's memory actually used and checking for uses which inhibit any
aspects of the transform such as the escape of a pointer.
Once we have a mapping of the ranges of the alloca used by individual
operations, we compute a partitioning of the used ranges. Some uses are
inherently splittable (such as memcpy and memset), while scalar uses are
not splittable. The goal is to build a partitioning that has the minimum
number of splits while placing each unsplittable use in its own
partition. Overlapping unsplittable uses belong to the same partition.
This is the target split of the aggregate alloca, and it maximizes the
number of scalar accesses which become accesses to their own alloca and
candidates for promotion.
Third, we re-walk the uses of the alloca and assign each specific memory
access to all the partitions touched so that we have dense use-lists for
each partition.
Finally, we build a new, smaller alloca for each partition and rewrite
each use of that partition to use the new alloca. During this phase the
pass will also work very hard to transform uses of an alloca into a form
suitable for promotion, including forming vector operations, speculating
loads throguh PHI nodes and selects, etc.
After splitting is complete, each newly refined alloca that is
a candidate for promotion to a scalar SSA value is run through mem2reg.
There are lots of reasonably detailed comments in the source code about
the design and algorithms, and I'm going to be trying to improve them in
subsequent commits to ensure this is well documented, as the new pass is
in many ways more complex than the old one.
Some of this is still a WIP, but the current state is reasonbly stable.
It has passed bootstrap, the nightly test suite, and Duncan has run it
successfully through the ACATS and DragonEgg test suites. That said, it
remains behind a default-off flag until the last few pieces are in
place, and full testing can be done.
Specific areas I'm looking at next:
- Improved comments and some code cleanup from reviews.
- SSAUpdater and enabling this pass inside the CGSCC pass manager.
- Some datastructure tuning and compile-time measurements.
- More aggressive FCA splitting and vector formation.
Many thanks to Duncan Sands for the thorough final review, as well as
Benjamin Kramer for lots of review during the process of writing this
pass, and Daniel Berlin for reviewing the data structures and algorithms
and general theory of the pass. Also, several other people on IRC, over
lunch tables, etc for lots of feedback and advice.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163883 91177308-0d34-0410-b5e6-96231b3b80d8
This is mostly documentation for the new machine model. It is designed
to be flexible, easy to incrementally refine for a subtarget, and
provide all the information that MachineScheduler will need.
If all goes well, I will follow up with an example of the new model in
use for ARM.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163877 91177308-0d34-0410-b5e6-96231b3b80d8
.set a, b - c + CONSTANT
d = b - c + CONSTANT
Both 'a' and 'd' should be marked as absolute symbols (N_ABS).
rdar://12219394
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163853 91177308-0d34-0410-b5e6-96231b3b80d8