SelectionDAG's 'init' has not been called when the SelectionDAGBuilder is
constructed (in SelectionDAGISel's constructor), so this was previously always
initialized with 0.
llvm-svn: 162333
Even looking at the revision history I couldn't quite piece together why this
cast was ever written in the first place, but I assume it was because of some
change in the inheritance, perhaps this function was reimplemented in a
derived type & this caller was meant to get the base version (& it wasn't
virtual)?
llvm-svn: 162301
The getSumForBlock function was quadratic in the number of successors
because getSuccWeight would perform a linear search for an already known
iterator.
This patch was originally committed as r161460, but reverted again
because of assertion failures. Now that duplicate Machine CFG edges have
been eliminated, this works properly.
llvm-svn: 162233
IR that hasn't been through SimplifyCFG can look like this:
br i1 %b, label %r, label %r
Make sure we don't create duplicate Machine CFG edges in this case.
Fix the machine code verifier to accept conditional branches with a
single CFG edge.
llvm-svn: 162230
The DAGCombiner tries to optimise a BUILD_VECTOR by checking if it
consists purely of get_vector_elts from one or two source vectors. If
so, it either makes a concat_vectors node or a shufflevector node.
However, it doesn't check the element type width of the underlying
vector, so if you have this sequence:
Node0: v4i16 = ...
Node1: i32 = extract_vector_elt Node0
Node2: i32 = extract_vector_elt Node0
Node3: v16i8 = BUILD_VECTOR Node1, Node2, ...
It will attempt to:
Node0: v4i16 = ...
NewNode1: v16i8 = concat_vectors Node0, ...
Where this is actually invalid because the element width is completely
different. This causes an assertion failure on DAG legalization stage.
Fix:
If output item type of BUILD_VECTOR differs from input item type.
Make concat_vectors based on input element type and then bitcast it to the output vector type. So the case described above will transformed to:
Node0: v4i16 = ...
NewNode1: v8i16 = concat_vectors Node0, ...
NewNode2: v16i8 = bitcast NewNode1
llvm-svn: 162195
make it more consistent with its intended semantics.
The `linker_private_weak_def_auto' linkage type was meant to automatically hide
globals which never had their addresses taken. It has nothing to do with the
`linker_private' linkage type, which outputs the symbols with a `l' (ell) prefix
among other things.
The intended semantic is more like the `linkonce_odr' linkage type.
Change the name of the linkage type to `linkonce_odr_auto_hide'. And therefore
changing the semantics so that it produces the correct output for the linker.
Note: The old linkage name `linker_private_weak_def_auto' will still parse but
is not a synonym for `linkonce_odr_auto_hide'. This should be removed in 4.0.
<rdar://problem/11754934>
llvm-svn: 162114
Increment the MBB iterator at the top of the loop to properly handle the
current (and previous) instructions getting erased.
This fixes PR13625.
llvm-svn: 162099
Select instructions pick one of two virtual registers based on a
condition, like x86 cmov. On targets like ARM that support predication,
selects can sometimes be eliminated by predicating the instruction
defining one of the operands.
Teach PeepholeOptimizer to recognize select instructions, and ask the
target to optimize them.
llvm-svn: 162059
It never does anything when running 'make check', and it get's in the
way of updating live intervals in 2-addr.
The hook was originally added to help form IT blocks in Thumb2 code
before register allocation, but the pass ordering has changed since
then, and we run if-conversion after register allocation now.
When the MI scheduler is enabled, there will be no less than two
schedulers between 2-addr and Thumb2ITBlockPass, so this hook is
unlikely to help anything.
llvm-svn: 161794
It is still possible to if-convert if the tail block has extra
predecessors, but the tail phis must be rewritten instead of being
removed.
llvm-svn: 161781
Detect when there is not enough available ILP, so if-conversion can't
speculate instructions for free.
Compute the lengthening of the critical path when inserting a select
instruction that depends on the condition as well as both sides of the
if.
Reject conversions that would stretch the critical path by more than
half a mispredict penalty.
llvm-svn: 161713
Trace::getResourceLength() computes the number of cycles required to
execute the trace when ignoring data dependencies. The number can be
compared to the critical path to estimate the trace ILP.
Trace::getPHIDepth() computes the data dependency depth of a PHI in a
trace successor that isn't necessarily part of the trace.
llvm-svn: 161711
When a trace ends with a back-edge, include PHIs in the loop header in
the height computations. This makes the critical path through a loop
more accurate by including the latencies of the last instructions in the
loop.
llvm-svn: 161688
When replacing Old with New, it can happen that New is already a
successor. Add the old and new edge weights instead of creating a
duplicate edge.
llvm-svn: 161653
This makes it possible to speed up def_iterator by stopping at the first
use. This makes def_empty() and getUniqueVRegDef() much faster when
there are many uses.
In a +Asserts build, LiveVariables is 100x faster in one case because
getVRegDef() has an assertion that would scan to the end of a
def_iterator chain.
Spill weight calculation is significantly faster (300x in one case)
because isTriviallyReMaterializable() calls MRI->isConstantPhysReg(%RIP)
which calls def_empty(%RIP).
llvm-svn: 161634
Use a more conventional doubly linked list where the Prev pointers form
a cycle. This means it is no longer necessary to adjust the Prev
pointers when reallocating the VRegInfo array.
The test changes are required because the register allocation hint is
using the use-list order to break ties.
llvm-svn: 161633
Register MachineOperands are kept in linked lists accessible via MRI's
reg_iterator interfaces. The linked list management was handled partly
by MachineOperand methods, partly by MRI methods.
Move all of the list management into MRI, delete
MO::AddRegOperandToRegInfo() and MO::RemoveRegOperandFromRegInfo().
Be more explicit about handling the cases where an MRI pointer isn't
available.
llvm-svn: 161632
We filter out MachineLoop back-edges during the trace-building PO
traversals, but it is possible to have CFG cycles that aren't natural
loops, and MachineLoopInfo doesn't include such cycles.
Use a standard visited set to detect such CFG cycles, and completely
ignore them when picking traces.
llvm-svn: 161532
We perform the following:
1> Use SUB instead of CMP for i8,i16,i32 and i64 in ISel lowering.
2> Modify MachineCSE to correctly handle implicit defs.
3> Convert SUB back to CMP if possible at peephole.
Removed pattern matching of (a>b) ? (a-b):0 and like, since they are handled
by peephole now.
rdar://11873276
llvm-svn: 161462
The getSumForBlock function was quadratic in the number of successors
because getSuccWeight would perform a linear search for an already known
iterator.
llvm-svn: 161460
This adds support for TargetIndex operands during isel. The meaning of
these (index, offset, flags) operands is entirely defined by the target.
llvm-svn: 161453
A target index operand looks a lot like a constant pool reference, but
it is completely target-defined. It contains the 8-bit TargetFlags, a
32-bit index, and a 64-bit offset. It is preserved by all code generator
passes.
TargetIndex operands can be used to carry target-specific information in
cases where immediate operands won't suffice.
llvm-svn: 161441
Compare the critical paths of the two traces through an if-conversion
candidate. If the difference is larger than the branch brediction
penalty, reject the if-conversion. If would never pay.
llvm-svn: 161433
Previously, MBP essentially aligned every branch target it could. This
bloats code quite a bit, especially non-looping code which has no real
reason to prefer aligned branch targets so heavily.
As Andy said in review, it's still a bit odd to do this without a real
cost model, but this at least has much more plausible heuristics.
Fixes PR13265.
llvm-svn: 161409
If the result of a common subexpression is used at all uses of the candidate
expression, CSE should not increase the live range of the common subexpression.
rdar://11393714 and rdar://11819721
llvm-svn: 161396
This patch is mostly just refactoring a bunch of copy-and-pasted code, but
it also adds a check that the call instructions are readnone or readonly.
That check was already present for sin, cos, sqrt, log2, and exp2 calls, but
it was missing for the rest of the builtins being handled in this code.
llvm-svn: 161282
This option runs LiveIntervals before TwoAddressInstructionPass which
will eventually learn to exploit and update the analysis.
Eventually, LiveIntervals will run before PHIElimination, and we can get
rid of LiveVariables.
llvm-svn: 161270
The previous change caused fast isel to not attempt handling any calls to
builtin functions. That included things like "printf" and caused some
noticable regressions in compile time. I wanted to avoid having fast isel
keep a separate list of functions that had to be kept in sync with what the
code in SelectionDAGBuilder.cpp was handling. I've resolved that here by
moving the list into TargetLibraryInfo. This is somewhat redundant in
SelectionDAGBuilder but it will ensure that we keep things consistent.
llvm-svn: 161263
I noticed that SelectionDAGBuilder::visitCall was missing a check for memcmp
in TargetLibraryInfo, so that it would use custom code for memcmp calls even
with -fno-builtin. I also had to add a new -disable-simplify-libcalls option
to llc so that I could write a test for this.
llvm-svn: 161262
The 'unused' state of a value number can be represented as an invalid
def SlotIndex. This also exposed code that shouldn't have been looking
at unused value VNInfos.
llvm-svn: 161258
The only real user of the flag was removeCopyByCommutingDef(), and it
has been switched to LiveIntervals::hasPHIKill().
All the code changed by this patch was only concerned with computing and
propagating the flag.
llvm-svn: 161255
The VNInfo::HAS_PHI_KILL is only half supported. We precompute it in
LiveIntervalAnalysis, but it isn't properly updated by live range
splitting and functions like shrinkToUses().
It is only used in one place: RegisterCoalescer::removeCopyByCommutingDef().
This patch changes that function to use a new LiveIntervals::hasPHIKill()
function that computes the flag for a given value number.
llvm-svn: 161254
Fast isel doesn't currently have support for translating builtin function
calls to target instructions. For embedded environments where the library
functions are not available, this is a matter of correctness and not
just optimization. Most of this patch is just arranging to make the
TargetLibraryInfo available in fast isel. <rdar://problem/12008746>
llvm-svn: 161232
Add more comments and use early returns to reduce nesting in isLoadFoldable.
Also disable folding for V_SET0 to avoid introducing a const pool entry and
a const pool load.
rdar://10554090 and rdar://11873276
llvm-svn: 161207
Whenever both instruction depths and instruction heights are known in a
block, it is possible to compute the length of the critical path as
max(depth+height) over the instructions in the block.
The stored live-in lists make it possible to accurately compute the
length of a critical path that bypasses the current (small) block.
llvm-svn: 161197
LiveRangeEdit::eliminateDeadDefs() can delete a dead instruction that
reads unreserved physregs. This would leave the corresponding regunit
live interval dangling because we don't have shrinkToUses() for physical
registers.
Fix this problem by turning the instruction into a KILL instead of
deleting it. This happens in a landing pad in
test/CodeGen/X86/2012-05-19-CoalescerCrash.ll:
%vreg27<def,dead> = COPY %EDX<kill>; GR32:%vreg27
becomes:
KILL %EDX<kill>
An upcoming fix to the machine verifier will catch problems like this by
verifying regunit live intervals.
This fixes PR13498. I am not including the test case from the PR since
we already have one exposing the problem once the verifier is fixed.
llvm-svn: 161182
Machine CSE and other optimizations can remove instructions so folding
is possible at peephole while not possible at ISel.
This patch is a rework of r160919 and was tested on clang self-host on my local
machine.
rdar://10554090 and rdar://11873276
llvm-svn: 161152
The height on an instruction is the minimum number of cycles from the
instruction is issued to the end of the trace. Heights are computed for
all instructions in and below the trace center block.
The method for computing heights is different from the depth
computation. As we visit instructions in the trace bottom-up, heights of
used instructions are pushed upwards. This way, we avoid scanning long
use lists, looking for uses in the current trace.
At each basic block boundary, a list of live-in registers and their
minimum heights is saved in the trace block info. These live-in lists
are used when restarting depth computations on a trace that
converges with an already computed trace. They will also be used to
accurately compute the critical path length.
llvm-svn: 161138
Assuming infinite issue width, compute the earliest each instruction in
the trace can issue, when considering the latency of data dependencies.
The issue cycle is record as a 'depth' from the beginning of the trace.
This is half the computation required to find the length of the critical
path through the trace. Heights are next.
llvm-svn: 161074
One motivating example is to sink an instruction from a basic block which has
two successors: one outside the loop, the other inside the loop. We should try
to sink the instruction outside the loop.
rdar://11980766
llvm-svn: 161062
We are extending live ranges, so kill flags are not accurate. They
aren't needed until they are recomputed after RA anyway.
<rdar://problem/11950722>
llvm-svn: 161023
We branch to the successor with higher edge weight first.
Convert from
je LBB4_8 --> to outer loop
jmp LBB4_14 --> to inner loop
to
jne LBB4_14
jmp LBB4_8
PR12750
rdar: 11393714
llvm-svn: 161018
This lets traces include the final iteration of a nested loop above the
center block, and the first iteration of a nested loop below the center
block.
We still don't allow traces to contain backedges, and traces are
truncated where they would leave a loop, as seen from the center block.
llvm-svn: 161003
When computing a trace, all the candidates for pred/succ must have been
visited. Filter out back-edges first, though. The PO traversal ignores
them.
Thanks to Andy for spotting this in review.
llvm-svn: 160995
This is a cleaned up version of the isFree() function in
MachineTraceMetrics.cpp.
Transient instructions are very unlikely to produce any code in the
final output. Either because they get eliminated by RegisterCoalescing,
or because they are pseudo-instructions like labels and debug values.
llvm-svn: 160977
A->isPredecessor(B) is the same as B->isSuccessor(A), but it can
tolerate a B that is null or dangling. This shouldn't happen normally,
but it it useful for verification code.
llvm-svn: 160968
Machine CSE and other optimizations can remove instructions so folding
is possible at peephole while not possible at ISel.
rdar://10554090 and rdar://11873276
llvm-svn: 160919
A value number is a PHI def if and only if it begins at a block
boundary. This can be derived from the def slot, a separate flag is not
necessary.
llvm-svn: 160893
This option replaces the existing live interval computation with one
based on LiveRangeCalc.cpp. The new algorithm does not depend on
LiveVariables, and it can be run at any time, before or after leaving
SSA form.
llvm-svn: 160892
This is still a work in progress.
Out-of-order CPUs usually execute instructions from multiple basic
blocks simultaneously, so it is necessary to look at longer traces when
estimating the performance effects of code transformations.
The MachineTraceMetrics analysis will pick a typical trace through a
given basic block and provide performance metrics for the trace. Metrics
will include:
- Instruction count through the trace.
- Issue count per functional unit.
- Critical path length, and per-instruction 'slack'.
These metrics can be used to determine the performance limiting factor
when executing the trace, and how it will be affected by a code
transformation.
Initially, this will be used by the early if-conversion pass.
llvm-svn: 160796
It is redundant; RegisterCoalescer will do the remat if it can't eliminate
the copy. Collected instruction counts before and after this. A few extra
instructions are generated due to spilling but it is normal to see these kinds
of changes with almost any small codegen change, according to Jakob.
This also fixed rdar://11830760 where xor is expected instead of movi0.
llvm-svn: 160749
When a live range splits into multiple connected components, we would
arbitrarily assign <undef> uses to component 0. This is wrong when the
use is tied to a def that gets assigned to a different component:
%vreg69<def> = ADD8ri %vreg68<undef>, 1
The use and def must get the same virtual register.
Fix this by assigning <undef> uses to the same component as the value
defined by the instruction, if any:
%vreg69<def> = ADD8ri %vreg69<undef>, 1
This fixes PR13402. The PR has a test case which I am not including
because it is unlikely to keep exposing this behavior in the future.
llvm-svn: 160739
LiveRangeEdit::foldAsLoad() can eliminate a register by folding a load
into its only use. Only do that when the load is safe to move, and it
won't extend any live ranges.
This fixes PR13414.
llvm-svn: 160575
PHIElimination splits critical edges when it predicts it can resolve
interference and eliminate copies. It doesn't split the edge if the
interference wouldn't be resolved anyway because the phi-use register is
live in the critical edge anyway.
Teach PHIElimination to split loop exiting edges with interference, even
if it wouldn't resolve the interference. This removes the necessary
copies from the loop, which is still an improvement from injecting the
copies into the loop.
The test case demonstrates the improvement. Before:
LBB0_1:
cmpb $0, (%rdx)
leaq 1(%rdx), %rdx
movl %esi, %eax
je LBB0_1
After:
LBB0_1:
cmpb $0, (%rdx)
leaq 1(%rdx), %rdx
je LBB0_1
movl %esi, %eax
llvm-svn: 160571
LiveIntervals due to the two-addr pass generating bogus MI code.
The crux of the issue was a loop nesting problem. The intent of the code
which attempts to transform instructions before converting them to
two-addr form is to defer and reprocess any transformed instructions as
the second processing is likely to have more opportunities to coalesce
copies, etc. Unfortunately, there was one section of processing that was
not deferred -- the INSERT_SUBREG rewriting. Due to quirks of how this
rewriting proceeded, not only did it occur early, it removed the bits of
information needed for the deferred processing to correctly generate the
necessary two address form (specifically inserting a copy), but didn't
trigger any immediate assertions and produced what appeared to be
already valid two-address from code. Thus, the assertion only fired much
later in the pipeline.
The fix is to hoist the transformation logic up layer to where it can
more firmly defer all further processing, and to teach the normal
processing to handle an edge case previously handled as part of the
transformation logic. This edge case (already matched tied register
operands) needs to *not* defer any steps.
As has been brought up repeatedly in the process: wow does this code
need refactoring. I *may* squeeze in some time to at least bring sanity
to this loop... but wow... =]
Thanks to Jakob for helpful hints on the way here, and the review.
llvm-svn: 160443
large immediates. Add dag combine logic to recover in case the large
immediates doesn't fit in cmp immediate operand field.
int foo(unsigned long l) {
return (l>> 47) == 1;
}
we produce
%shr.mask = and i64 %l, -140737488355328
%cmp = icmp eq i64 %shr.mask, 140737488355328
%conv = zext i1 %cmp to i32
ret i32 %conv
which codegens to
movq $0xffff800000000000,%rax
andq %rdi,%rax
movq $0x0000800000000000,%rcx
cmpq %rcx,%rax
sete %al
movzbl %al,%eax
ret
TargetLowering::SimplifySetCC would transform
(X & -256) == 256 -> (X >> 8) == 1
if the immediate fails the isLegalICmpImmediate() test. For x86,
that's immediates which are not a signed 32-bit immediate.
Based on a patch by Eli Friedman.
PR10328
rdar://9758774
llvm-svn: 160346
In the added testcase the constant 55 was behind an AssertZext of type i1, and ComputeDemandedBits
reported that some of the bits were both known to be one and known to be zero.
Together with Michael Kuperstein <michael.m.kuperstein@intel.com>
llvm-svn: 160305
Add a micro-optimization to getNode of CONCAT_VECTORS when both operands are undefs.
Can't find a testcase for this because VECTOR_SHUFFLE already handles undef operands, but Duncan suggested that we add this.
Together with Michael Kuperstein <michael.m.kuperstein@intel.com>
llvm-svn: 160229
The notable fix is to look at any dependencies attached to the kill
instruction (or other instructions between MI nad the kill) where the
dependencies are specific to the register in question.
The old code implicitly handled this by rejecting the transform if *any*
other uses were found within the block, but after the start point. The
new code directly finds the kill, and has to re-use the existing
dependency scan to check for non-kill uses.
This was caught by self-host, but I found the bug via inspection and use
of absurd assert scaffolding to compute the kills in two ways and
compare them. So I have no useful testcase for this other than
"bootstrap". I'd work harder to reduce a test case if this particular
code were likely to live for a long time.
Thanks to Benjamin Kramer for reviewing the fix itself.
llvm-svn: 160228
Catch uses of undefined physregs that haven't been added to basic block
live-in lists. Run the verifier to pinpoint the problem.
Also run the verifier when a virtual register use is not jointly
dominated by defs.
llvm-svn: 160207
removes the largest scaling problem in the test cases from PR13225 when
ASan is switched to insert basic blocks in the natural CFG order.
It may also solve some scaling problems for more normal code with large
numbers of basic blocks and variables.
llvm-svn: 160194
When dumping the DAG for a fatal 'Cannot select' back-end error, also
provide the name of the function the construct is in. Useful when dealing
with large testcases, as the next step is to llvm-extract the function
in question to get a small(er) testcase.
llvm-svn: 160152
the input vector, it can be bigger (this is helpful for powerpc where <2 x i16>
is a legal vector type but i16 isn't a legal type, IIRC). However this wasn't
being taken into account by ExpandRes_EXTRACT_VECTOR_ELT, causing PR13220.
Lightly tweaked version of a patch by Michael Liao.
llvm-svn: 160116
r1025 = s/zext r1024, 4
r1026 = extract_subreg r1025, 4
to a copy:
r1026 = copy r1024
This is correct. However it uses TII->isCoalescableExtInstr() which can return
true for instructions which essentially does a sext_in_reg so this can end up
with an illegal copy where the source and destination register classes do not
match. Add a check to avoid it. Sorry, no test case possible at this time.
rdar://11849816
llvm-svn: 160059
generalizing its implementation sufficiently to support this value
number scenario as well.
This cuts out another significant performance hit in large functions
(over 10k basic blocks, etc), especially those with "natural" CFG
structures.
llvm-svn: 160026
This ordering allows nested if-conversion without using a work list, and
it makes it possible to update the dominator tree on the fly as well.
Any erased basic blocks will always be dominated by the current
post-order position, so the domtree can be pruned without invalidating
the iterator.
llvm-svn: 160025
back of it.
I don't have anything even remotely close to a test case for this. It
only broke two build bots, both of them doing bootstrap builds, one of
them a dragonegg bootstrap. It doesn't break for me when I bootstrap
either. It doesn't reproduce every time or on many machines during the
bootstrap. Many thanks to Duncan Sands who got the exact command (and
stage of the bootstrap) which failed on the dragonegg bootstrap and
managed to get it to trigger under valgrind with debug symbols. The fix
was then found by inspection.
llvm-svn: 159993
multiple scalars and insert them into a vector. Next, we shuffle the elements
into the correct places, as before.
Also fix a small dagcombine bug in SimplifyBinOpWithSameOpcodeHands, when the
migration of bitcasts happened too late in the SelectionDAG process.
llvm-svn: 159991
quadratic behavior when performing pathological merges. Fixes the core
element of PR12652.
There is only one user of addRangeFrom left: join. I'm hoping to
refactor further in a future patch and have join use this merge
operation as well.
llvm-svn: 159982
of the trick merge routines. This adds a layer of testing that was
necessary when implementing more efficient (and complex) merge logic for
this datastructure.
No functionality changed here.
llvm-svn: 159981
subtarget CPU descriptions and support new features of
MachineScheduler.
MachineModel has three categories of data:
1) Basic properties for coarse grained instruction cost model.
2) Scheduler Read/Write resources for simple per-opcode and operand cost model (TBD).
3) Instruction itineraties for detailed per-cycle reservation tables.
These will all live side-by-side. Any subtarget can use any
combination of them. Instruction itineraries will not change in the
near term. In the long run, I expect them to only be relevant for
in-order VLIW machines that have complex contraints and require a
precise scheduling/bundling model. Once itineraries are only actively
used by VLIW-ish targets, they could be replaced by something more
appropriate for those targets.
This tablegen backend rewrite sets things up for introducing
MachineModel type #2: per opcode/operand cost model.
llvm-svn: 159891
DwarfDebug class could generate the same (inlined) DIVariable twice:
1) when trying to find abstract debug variable for a concrete inlined instance.
2) when explicitly collecting info for variables that were optimized out.
This change makes sure that this duplication won't happen and makes
Clang pass "gdb.opt/inline-locals" test from gdb testsuite.
Reviewed by Eric Christopher.
llvm-svn: 159811
hash_value overload for MachineOperands. This addresses a FIXME
sufficient for me to remove it, and cleans up the code nicely too.
The important changes to the hashing logic:
- TargetFlags are now included in all of the hashes. These were complete
missed.
- Register operands have their subregisters and whether they are a def
included in the hash.
- We now actually hash all of the operand types. Previously, many
operand types were simply *dropped on the floor*. For example:
- Floating point immediates
- Large integer immediates (>64-bit)
- External globals!
- Register masks
- Metadata operands
- It removes the offset from the block-address hash; I'm a bit
suspicious of this, but isIdenticalTo doesn't consider the offset for
black addresses.
Any patterns involving these entities could have triggered extreme
slowdowns in MachineCSE or PHIElimination. Let me know if there are PRs
you think might be closed now... I'm looking myself, but I may miss
them.
llvm-svn: 159743
broken. This patch fixes the superficial problems which lead to the
intractably slow compile times reported in PR13225.
The specific issue is that we were failing to include the *offset* of
a global variable in the hash code. Oops. This would in turn cause all
MIs which were only distinguishable due to operating on different
offsets of a global variable to produce identical hash functions. In
some of the test cases attached to the PR I saw hash table activity
where there were O(1000) probes-per-lookup *on average*. A very few
entries were responsible for most of these probes.
There is still quite a bit more to do here. The ad-hoc layering of data
in MachineOperands makes them *extremely* brittle to hash correctly.
We're missing quite a few other cases, the only ones I've fixed here are
the specific MO types which were allowed through the assert() in
getOffset().
llvm-svn: 159741
change.
Move the "Not profitable, avoid CSE!" debug message next to where we fail the
check for profitability and use a different message for avoiding CSE due to
being in different register classes.
llvm-svn: 159729
Also allow trailing register mask operands on non-variadic both
MachineSDNodes and MachineInstrs.
The extra physreg RegisterSDNode operands are added to the MI as
<imp-use> operands. This makes it possible to have non-variadic call
instructions.
Call and return instructions really are non-variadic, the argument
registers should only be used implicitly - they are not part of the
encoding.
llvm-svn: 159727
IntegersSubsetMapping
- Replaced type of Items field from std::list with std::map. In neares future I'll test it with DenseMap and do the correspond replacement
if possible.
llvm-svn: 159703
This pass performs if-conversion on SSA form machine code by
speculatively executing both sides of the branch and using a cmov
instruction to select the result. This can help lower the number of
branch mispredictions on architectures like x86 that don't have
predicable instructions.
The current implementation is very aggressive, and causes regressions on
mosts tests. It needs good heuristics that have yet to be implemented.
llvm-svn: 159694
IntegersSubsetMapping
- Replaced type of Items field from std::list with std::map. In neares future I'll test it with DenseMap and do the correspond replacement
if possible.
llvm-svn: 159659
It appears to have caught a use-after-free introduced as by r159567
and/or friends which call 'addPass' from many more places. The bug in
'addPass' doesn't appear to be new, and was spotted by inspection when
ASan shown a bright light of a stacktrace at these functions.
Hopefully this will fix the ASan failure -- I have no test case other
than running an ASan-built clang over the test suite.
llvm-svn: 159614
This is still a work in progress but I believe it is currently good enough
to fix PR13122 "Need unit test driver for codegen IR passes". For example,
you can run llc with -stop-after=loop-reduce to have it dump out the IR after
running LSR. Serializing machine-level IR is not yet supported but we have
some patches in progress for that.
The plan is to serialize the IR to a YAML file, containing separate sections
for the LLVM IR, machine-level IR, and whatever other info is needed. Chad
suggested that we stash the stop-after pass in the YAML file and use that
instead of the start-after option to figure out where to restart the
compilation. I think that's a great idea, but since it's not implemented yet
I put the -start-after option into this patch for testing purposes.
llvm-svn: 159570
This is a preliminary step toward having TargetPassConfig be able to
start and stop the compilation at specified passes for unit testing
and debugging. No functionality change.
llvm-svn: 159567
implicit_def, the other instruction can be anything, including instructions
that define multiple values. Be careful about that and don't assume what operand
0 is.
Fixes pr13249.
llvm-svn: 159509
This was always part of the VMCore library out of necessity -- it deals
entirely in the IR. The .cpp file in fact was already part of the VMCore
library. This is just a mechanical move.
I've tried to go through and re-apply the coding standard's preferred
header sort, but at 40-ish files, I may have gotten some wrong. Please
let me know if so.
I'll be committing the corresponding updates to Clang and Polly, and
Duncan has DragonEgg.
Thanks to Bill and Eric for giving the green light for this bit of cleanup.
llvm-svn: 159421
The TargetInstrInfo::getNumMicroOps API does not change, but soon it
will be used by MachineScheduler. Now each subtarget can specify the
number of micro-ops per itinerary class. For ARM, this is currently
always dynamic (-1), because it is used for load/store multiple which
depends on the number of register operands.
Zero is now a valid number of micro-ops. This can be used for
nop pseudo-instructions or instructions that the hardware can squash
during dispatch.
llvm-svn: 159406
Teach vector legalization how to honor Promote for int to float
conversions. The code checking whether to promote the operation knew
to look at the operand, but the actual promotion code didn't. This
fixes that. The operand is promoted up via [zs]ext.
rdar://11762659
llvm-svn: 159378
include/llvm/Analysis/DebugInfo.h to include/llvm/DebugInfo.h.
The reasoning is because the DebugInfo module is simply an interface to the
debug info MDNodes and has nothing to do with analysis.
llvm-svn: 159312
Such passes can be used to tweak the register assignments in a
target-dependent way, for example to avoid write-after-write
dependencies.
llvm-svn: 159209
The primary advantage is that loop optimizations will be applied in a
stable order. This helps debugging and unit test creation. It is also
a better overall implementation without pathologically bad performance
on deep functions.
On large functions (llvm-stress --size=200000 | opt -loops)
Before: 0.1263s
After: 0.0225s
On deep functions (after tweaking llvm-stress, thanks Nadav):
Before: 0.2281s
After: 0.0227s
See r158790 for more comments.
The loop tree is now consistently generated in forward order, but loop
passes are applied in reverse order over the program. If we have a
loop optimization that prefers forward order, that can easily be
achieved by adding a different type of LoopPassManager.
llvm-svn: 159183
Verify that all paths from the entry block to a virtual register read
pass through a def. Enable this check even when MRI->isSSA() is false.
Verify that the live range of a virtual register is live out of all
predecessor blocks, even for PHI-values.
This requires that PHIElimination sometimes inserts IMPLICIT_DEF
instruction in predecessor blocks.
llvm-svn: 159150
Implicitly defined virtual registers can simply have the <undef> bit set
on all uses, and copies can be turned into implicit defs recursively.
Physical registers are a bit trickier. We handle the common case where a
physreg def is used by a nearby instruction in the same basic block. For
more complicated cases, just leave the IMPLICIT_DEF instruction in.
llvm-svn: 159149
When a PHI use is <undef>, don't emit a copy in the predecessor block,
but insert an IMPLICIT_DEF instruction instead. This ensures that
virtual register uses are always jointly dominated by defs, even if some
of them are IMPLICIT_DEF.
llvm-svn: 159121
When the source register to a 2-addr instruction is undefined, there is
no need to attempt any transformations - simply replace the source
register with the destination register.
This also comes up when lowering IMPLICIT_DEF instructions - make sure
the <undef> flag is moved to the new partial register def operand:
%vreg8<def> = INSERT_SUBREG %vreg9<undef>, %vreg0<kill>, sub_16bit
rewrite undef:
%vreg8<def> = INSERT_SUBREG %vreg8<undef>, %vreg0<kill>, sub_16bit
convert to:
%vreg8:sub_16bit<def,read-undef> = COPY %vreg0<kill>
llvm-svn: 159120
It's simple: Don't treat <undef> operands as uses, and don't assume a
virtual register has a defining instruction unless a real use has been
seen.
llvm-svn: 159061
Original commit message:
Allow up to 64 functional units per processor itinerary.
This patch changes the type used to hold the FU bitset from unsigned to uint64_t.
This will be needed for some upcoming PowerPC itineraries.
llvm-svn: 159027
With regunit liveness permanently enabled, this function would always
return true.
Also remove now obsolete code for checking physreg interference.
llvm-svn: 159006
boolean flag to an enum: { Fast, Standard, Strict } (default = Standard).
This option controls the creation by optimizations of fused FP ops that store
intermediate results in higher precision than IEEE allows (E.g. FMAs). The
behavior of this option is intended to match the behaviour specified by a
soon-to-be-introduced frontend flag: '-ffuse-fp-ops'.
Fast mode - allows formation of fused FP ops whenever they're profitable.
Standard mode - allow fusion only for 'blessed' FP ops. At present the only
blessed op is the fmuladd intrinsic. In the future more blessed ops may be
added.
Strict mode - allow fusion only if/when it can be proven that the excess
precision won't effect the result.
Note: This option only controls formation of fused ops by the optimizers. Fused
operations that are explicitly requested (e.g. FMA via the llvm.fma.* intrinsic)
will always be honored, regardless of the value of this option.
Internally TargetOptions::AllowExcessFPPrecision has been replaced by
TargetOptions::AllowFPOpFusion.
llvm-svn: 158956
to be generic across architectures. It has the
following description in the gnu sources:
Negate the immediate constant
Several Architectures such as x86 have local implementations
of operand modifier 'n' which go beyond the above description
slightly. This won't affect them.
Affected files:
lib/CodeGen/AsmPrinter/AsmPrinterInlineAsm.cpp
Added 'n' to the switch cases.
test/CodeGen/Generic/asm-large-immediate.ll
Generic compiled test (x86 for me)
test/CodeGen/Mips/asm-large-immediate.ll
Mips compiled version of the generic one
Contributer: Jack Carter
llvm-svn: 158939
to be generic across architectures. It has the
following description in the gnu sources:
Substitute immediate value without immediate syntax
Several Architectures such as x86 have local implementations
of operand modifier 'c' which go beyond the above description
slightly. To make use of the generic modifiers without overriding
local implementation one can make a call to the base class method
for AsmPrinter::PrintAsmOperand() in the locally derived method's
"default" case in the switch statement. That way if it is already
defined locally the generic version will never get called.
This change is needed when test/CodeGen/generic/asm-large-immediate.ll
failed on a native Mips board. The test was assuming a generic
implementation was in place.
Affected files:
lib/Target/Mips/MipsAsmPrinter.cpp:
Changed the default case to call the base method.
lib/CodeGen/AsmPrinter/AsmPrinterInlineAsm.cpp
Added 'c' to the switch cases.
test/CodeGen/Mips/asm-large-immediate.ll
Mips compiled version of the generic one
Contributer: Jack Carter
llvm-svn: 158925
_umodsi3 libcalls if they have the same arguments. This optimization
was apparently broken if one of the node was replaced in place.
rdar://11714607
llvm-svn: 158900
Stop depending on the LiveIntervalUnions in RegAllocBase, they are about
to be removed.
The changes are mostly replacing register alias iterators with regunit
iterators, and querying LiveRegMatrix instrad of RegAllocBase.
InterferenceCache is converted to work with per-regunit
LiveIntervalUnions, and it checks fixed regunit interference separately,
using the fixed live intervals provided by LiveIntervalAnalysis.
The local splitting helper calcGapWeights() is also considering fixed
regunit interference which is kept on the side now.
llvm-svn: 158867
That is a DenseMap iterator keyed by pointers, so the iteration order is
nondeterministic.
I would like to replace the DenseMap with an IndexedMap which doesn't
allow iteration.
llvm-svn: 158856
Regunit live ranges are computed on demand, so when mi-sched calls
handleMove, some regunits may not have live ranges yet.
That makes updating them easier: Just skip the non-existing ranges. They
will be computed correctly from the rescheduled machine code when they
are needed.
llvm-svn: 158831
I'll admit I'm not entirely satisfied with this change, but it seemed
the cleanest option. Other suggestions quite welcome
The issue is that the traits specializations have static methods which
return the typedef'ed PHI_iterator type. In both the IR and MI layers
this is typedef'ed to a custom iterator class defined in an anonymous
namespace giving the types and the functions returning them internal
linkage. However, because the traits specialization is defined in the
'llvm' namespace (where it has to be, specialized template lives there),
and is in turn used in the templated implementation of the SSAUpdater.
This led to the linkage conflict that Clang now warns about.
The simplest solution to me was just to define the PHI_iterator as
a nested class inside the trait specialization. That way it still
doesn't get scoped widely, it can't be accidentally reused somewhere,
etc. This is a little gross just because nested class definitions are
a little gross, but the alternatives seem more ad-hoc.
llvm-svn: 158799
-stable-loops enables a new algorithm for generating the Loop
forest. It differs from the original algorithm in a few respects:
- Not determined by use-list order.
- Initially guarantees RPO order of block and subloops.
- Linear in the number of CFG edges.
- Nonrecursive.
I didn't want to change the LoopInfo API yet, so the block lists are
still inclusive. This seems strange to me, and it means that building
LoopInfo is not strictly linear, but it may not be a problem in
practice. At least the block lists start out in RPO order now. In the
future we may add an attribute or wrapper analysis that allows other
passes to assume RPO order.
The primary motivation of this work was not to optimize LoopInfo, but
to allow reproducing performance issues by decomposing the compilation
stages. I'm often unable to do this with the current LoopInfo, because
the loop tree order determines Loop pass order. Serializing the IR
tends to invert the order, which reverses the optimization order. This
makes it nearly impossible to debug interdependent loop optimizations
such as LSR.
I also believe this will provide more stable performance results across time.
llvm-svn: 158790
The implementation only needs inclusion from LoopInfo.cpp and
MachineLoopInfo.cpp. Clients of the interface should only include the
interface. This makes the interface readable and speeds up rebuilds
after modifying the implementation.
llvm-svn: 158787
When LiveIntervals is tracking fixed interference in regunits, make sure
to update those intervals as well. Currently guarded by -live-regunits.
llvm-svn: 158766
ensureAlignment() in MachineFunction). Also, drop setMaxAlignment() in
favor of this new function. This creates a main entry point to setting
MaxAlignment, which will be helpful for future work. No functionality
change intended.
llvm-svn: 158758
This patch adds DAG combines to form FMAs from pairs of FADD + FMUL or
FSUB + FMUL. The combines are performed when:
(a) Either
AllowExcessFPPrecision option (-enable-excess-fp-precision for llc)
OR
UnsafeFPMath option (-enable-unsafe-fp-math)
are set, and
(b) TargetLoweringInfo::isFMAFasterThanMulAndAdd(VT) is true for the type of
the FADD/FSUB, and
(c) The FMUL only has one user (the FADD/FSUB).
If your target has fast FMA instructions you can make use of these combines by
overriding TargetLoweringInfo::isFMAFasterThanMulAndAdd(VT) to return true for
types supported by your FMA instruction, and adding patterns to match ISD::FMA
to your FMA instructions.
llvm-svn: 158757
The PPC::EXTSW instruction preserves the low 32 bits of its input, just
like some of the x86 instructions. Use it to reduce register pressure
when the low 32 bits have multiple uses.
This requires a small change to PeepholeOptimizer since EXTSW takes a
64-bit input register.
This is related to PR5997.
llvm-svn: 158743
TargetLoweringObjectFileELF. Use this to support it on X86. Unlike ARM,
on X86 it is not easy to find out if .init_array should be used or not, so
the decision is made via TargetOptions and defaults to off.
Add a command line option to llc that enables it.
llvm-svn: 158692
This patch changes the type used to hold the FU bitset from unsigned to uint64_t.
This will be needed for some upcoming PowerPC itineraries.
llvm-svn: 158679
For store->load dependencies that may alias, we should always use
TrueMemOrderLatency, which may eventually become a subtarget hook. In
effect, we should guarantee at least TrueMemOrderLatency on at least
one DAG path from a store to a may-alias load.
This should fix the standard mode as well as -enable-aa-sched-mi".
llvm-svn: 158380
The LiveRegMatrix represents the live range of assigned virtual
registers in a Live interval union per register unit. This is not
fundamentally different from the interference tracking in RegAllocBase
that both RABasic and RAGreedy use.
The important differences are:
- LiveRegMatrix tracks interference per register unit instead of per
physical register. This makes interference checks cheaper and
assignments slightly more expensive. For example, the ARM D7 reigster
has 24 aliases, so we would check 24 physregs before assigning to one.
With unit-based interference, we check 2 units before assigning to 2
units.
- LiveRegMatrix caches regmask interference checks. That is currently
duplicated functionality in RABasic and RAGreedy.
- LiveRegMatrix is a pass which makes it possible to insert
target-dependent passes between register allocation and rewriting.
Such passes could tweak the register assignments with interference
checking support from LiveRegMatrix.
Eventually, RABasic and RAGreedy will be switched to LiveRegMatrix.
llvm-svn: 158255
This deduplicates some code from the optimizing register allocators, and
it means that it is now possible to change the register allocators'
solutions simply by editing the VirtRegMap between the register
allocator pass and the rewriter.
llvm-svn: 158249
OK, not really. We don't want to reintroduce the old rewriter hacks.
This patch extracts virtual register rewriting as a separate pass that
runs after the register allocator. This is possible now that
CodeGen/Passes.cpp can configure the full optimizing register allocator
pipeline.
The rewriter pass uses register assignments in VirtRegMap to rewrite
virtual registers to physical registers, and it inserts kill flags based
on live intervals.
These finalization steps are the same for the optimizing register
allocators: RABasic, RAGreedy, and PBQP.
llvm-svn: 158244
Bulk move of TargetInstrInfo implementation into
TargetInstrInfoImpl. This is dirty because the code isn't part of
TargetInstrInfoImpl class, nor should it be, because the methods are
not target hooks. However, it's the current mechanism for keeping
libTarget useful outside the backend. You'll get a not-so-nice link
error if you invoke a TargetInstrInfo method that depends on CodeGen.
The TargetInstrInfoImpl class should probably be removed since it
doesn't really solve this problem.
To really fix this, we probably need separate interfaces for the
CodeGen/nonCodeGen sides of TargetInstrInfo.
llvm-svn: 158212
The commit is intended to fix rdar://11540023.
It is implemented as part of peephole optimization. We can actually implement
this in the SelectionDAG lowering phase.
llvm-svn: 158122
Bundles should be treated as one atomic transaction when checking
liveness. That is how the register allocator (and VLIW targets) treats
bundles.
llvm-svn: 158116
LLVM is now -Wunused-private-field clean except for
- lib/MC/MCDisassembler/Disassembler.h. Not sure why it keeps all those unaccessible fields.
- gtest.
llvm-svn: 158096
There are some that I didn't remove this round because they looked like
obvious stubs. There are dead variables in gtest too, they should be
fixed upstream.
llvm-svn: 158090
Soon we'll be making LiveIntervalUnions for register units as well.
This was the only place using the RepReg member, so just remove it.
llvm-svn: 158038
Don't print out the register number and spill weight, making the TRI
argument unnecessary.
This allows callers to interpret the reg field. It can currently be a
virtual register, a physical register, a spill slot, or a register unit.
llvm-svn: 158031
Instead of computing a live interval per physreg, LiveIntervals can
compute live intervals per register unit. This makes impossible the
confusing situation where aliasing registers could have overlapping live
intervals. It should also make fixed interferernce checking cheaper
since registers have fewer register units than aliases.
Live intervals for regunits are computed on demand, using MRI use-def
chains and the new LiveRangeCalc class. Only regunits live in to ABI
blocks are precomputed during LiveIntervals::runOnMachineFunction().
The regunit liveness computations don't depend on LiveVariables.
llvm-svn: 158029
expression (a * b + c) that can be implemented as a fused multiply-add (fma)
if the target determines that this will be more efficient. This intrinsic
will be used to implement FP_CONTRACT support and an aggressive FMA formation
mode.
If your target has a fast FMA instruction you should override the
isFMAFasterThanMulAndAdd method in TargetLowering to return true.
llvm-svn: 158014
This allows a subtarget to explicitly specify the issue width and
other properties without providing pipeline stage details for every
instruction.
llvm-svn: 157979
valid itinerary but no pipeline stages.
An itinerary can contain useful scheduling information without specifying pipeline stages for each instruction.
llvm-svn: 157977
It is an old function that does a lot more than required by
CalcSpillWeights, which was the only remaining caller.
The isRematerializable() function never actually sets the isLoad
argument, so don't try to compute that.
llvm-svn: 157973
IntegersSubsetGeneric, IntegersSubsetMapping: added IntTy template parameter, that allows use either APInt or IntItem. This change allows to write unittest for these classes.
llvm-svn: 157880
No functional change intended.
Sorry for the churn. The iterator classes are supposed to help avoid
giant commits like this one in the future. The TableGen-produced
register lists are getting quite large, and it may be necessary to
change the table representation.
This makes it possible to do so without changing all clients (again).
llvm-svn: 157854
IntegersSubset devided into IntegersSubsetGeneric and into IntegersSubset itself. The first has no references to ConstantInt and works with IntItem only.
IntegersSubsetMapping also made generic. Here added second template parameter "IntegersSubsetTy" that allows to use on of two IntegersSubset types described below.
llvm-svn: 157815
types, as well as int<->ptr casts. This allows us to tailcall functions
with some trivial casts between the call and return (i.e. because the
return types disagree).
llvm-svn: 157798
This patch will optimize the following
movq %rdi, %rax
subq %rsi, %rax
cmovsq %rsi, %rdi
movq %rdi, %rax
to
cmpq %rsi, %rdi
cmovsq %rsi, %rdi
movq %rdi, %rax
Perform this optimization if the actual result of SUB is not used.
rdar: 11540023
llvm-svn: 157755
It helps compile exotic inline asm. In the test case, normal GR32
virtual registers use up eax-edx so the final GR32_ABCD live range has
no registers left. Since all the live ranges were tiny, we had no way of
prioritizing the smaller register class.
This patch allows tiny unspillable live ranges to be evicted by tiny
unspillable live ranges from a smaller register class.
<rdar://problem/11542429>
llvm-svn: 157715
Besides adding the new insertPass function, this patch uses it to
enhance the existing -print-machineinstrs so that the MachineInstrs
after a specific pass can be printed.
Patch by Bin Zeng!
llvm-svn: 157655
ranges for the instruction about to be bundled. This fixes a bug in an external
project where an assertion was triggered due to spurious 'multiple defs' within
the bundle.
Patch by Ivan Llopard. Thanks Ivan!
llvm-svn: 157632
Implemented IntItem - the wrapper around APInt. Why not to use APInt item directly right now?
1. It will very difficult to implement case ranges as series of small patches. We got several large and heavy patches. Each patch will about 90-120 kb. If you replace ConstantInt with APInt in SwitchInst you will need to changes at the same time all Readers,Writers and absolutely all passes that uses SwitchInst.
2. We can implement APInt pool inside and save memory space. E.g. we use several switches that works with 256 bit items (switch on signatures, or strings). We can avoid value duplicates in this case.
3. IntItem can be easyly easily replaced with APInt.
4. Currenly we can interpret IntItem both as ConstantInt and as APInt. It allows to provide SwitchInst methods that works with ConstantInt for non-updated passes.
Why I need it right now? Currently I need to update SimplifyCFG pass (EqualityComparisons). I need to work with APInts directly a lot, so peaces of code
ConstantInt *V = ...;
if (V->getValue().ugt(AnotherV->getValue()) {
...
}
will look awful. Much more better this way:
IntItem V = ConstantIntVal->getValue();
if (AnotherV < V) {
}
Of course any reviews are welcome.
P.S.: I'm also going to rename ConstantRangesSet to IntegersSubset, and CRSBuilder to IntegersSubsetMapping (allows to map individual subsets of integers to the BasicBlocks).
Since in future these classes will founded on APInt, it will possible to use them in more generic ways.
llvm-svn: 157576
definition in the map before calling itself to retrieve the
DIE for the declaration. Without this change, if this causes
getOrCreateSubprogramDIE to be recursively called on the definition,
it will create multiple DIEs for that definition. Fixes PR12831.
llvm-svn: 157541
SimplifyCFG tends to form a lot of 2-3 case switches when merging branches. Move
the most likely condition to the front so it is checked first and the others can
be skipped. This is currently not as effective as it could be because SimplifyCFG
destroys profiling metadata when merging branches and switches. Merging branch
weight metadata is tricky though.
This code touches at most 3 cases so I didn't use a proper sorting algorithm.
llvm-svn: 157521
to pass around a struct instead of a large set of individual values. This
cleans up the interface and allows more information to be added to the struct
for future targets without requiring changes to each and every target.
NV_CONTRIB
llvm-svn: 157479
The Hazard checker implements in-order contraints, or interlocked
resources. Ready instructions with hazards do not enter the available
queue and are not visible to other heuristics.
The major code change is the addition of SchedBoundary to encapsulate
the state at the top or bottom of the schedule, including both a
pending and available queue.
The scheduler now counts cycles in sync with the hazard checker. These
are minimum cycle counts based on known hazards.
Targets with no itinerary (x86_64) currently remain at cycle 0. To fix
this, we need to provide some maximum issue width for all targets. We
also need to add the concept of expected latency vs. minimum latency.
llvm-svn: 157427
Live ranges with a constrained register class may benefit from splitting
around individual uses. It allows the remaining live range to use a
larger register class where it may allocate. This is like spilling to a
different register class.
This is only attempted on constrained register classes.
<rdar://problem/11438902>
llvm-svn: 157354
Now that the coalescer keeps live intervals and machine code in sync at
all times, it needs to deal with identity copies differently.
When merging two virtual registers, all identity copies are removed
right away. This means that other identity copies must come from
somewhere else, and they are going to have a value number.
Deal with such copies by merging the value numbers before erasing the
copy instruction. Otherwise, we leave dangling value numbers in the live
interval.
This fixes PR12927.
llvm-svn: 157340
This helps compile time when the greedy register allocator splits live
ranges in giant functions. Without the bias, we would try to grow
regions through the giant edge bundles, usually to find out that the
region became too big and expensive.
If a live range has many uses in blocks near the giant bundle, the small
negative bias doesn't make a big difference, and we still consider
regions including the giant edge bundle.
Giant edge bundles are usually connected to landing pads or indirect
branches.
llvm-svn: 157174
With physreg joining out of the way, it is easy to recognize the
instructions that need their kill flags cleared while testing for
interference.
This allows us to skip the final scan of all instructions for an 11%
speedup of the coalescer pass.
llvm-svn: 157169
may be RAUW'd by the recursive call to LegalizeOps; instead, retrieve
the other operands when calling UpdateNodeOperands. Fixes PR12889.
llvm-svn: 157162
Dead code elimination during coalescing could cause a virtual register
to be split into connected components. The following rewriting would be
confused about the already joined copies present in the code, but
without a corresponding value number in the live range.
Erase all joined copies instantly when joining intervals such that the
MI and LiveInterval representations are always in sync.
llvm-svn: 157135
Dead code and joined copies are now eliminated on the fly, and there is
no need for a post pass.
This makes the coalescer work like other modern register allocator
passes: Code is changed on the fly, there is no pending list of changes
to be committed.
llvm-svn: 157132
The late dead code elimination is no longer necessary.
The test changes are cause by a register hint that can be either %rdi or
%rax. The choice depends on the use list order, which this patch changes.
llvm-svn: 157131
Before rewriting uses of one value in A to register B, check that there
are no tied uses. That would require multiple A values to be rewritten.
This bug can't bite in the current version of the code for a fairly
subtle reason: A tied use would have caused 2-addr to insert a copy
before the use. If the copy has been coalesced, it will be found by the
same loop changed by this patch, and the optimization is aborted.
This was exposed by 400.perlbench and lua after applying a patch that
deletes joined copies aggressively.
llvm-svn: 157130
Remaining virtreg->physreg copies were rematerialized during
updateRegDefsUses(), but we already do the same thing in joinCopy() when
visiting the physreg copy instruction.
Eliminate the preserveSrcInt argument to reMaterializeTrivialDef(). It
is now always true.
llvm-svn: 157103
Dead copies cause problems because they are trivial to coalesce, but
removing them gived the live range a dangling end point. This patch
enables full dead code elimination which trims live ranges to their uses
so end points don't dangle.
DCE may erase multiple instructions. Put the pointers in an ErasedInstrs
set so we never risk visiting erased instructions in the work list.
There isn't supposed to be any dead copies entering RegisterCoalescer,
but they do slip by as evidenced by test/CodeGen/X86/coalescer-dce.ll.
llvm-svn: 157101
Use a dedicated MachO load command to annotate data-in-code regions.
This is the same format the linker produces for final executable images,
allowing consistency of representation and use of introspection tools
for both object and executable files.
Data-in-code regions are annotated via ".data_region"/".end_data_region"
directive pairs, with an optional region type.
data_region_directive := ".data_region" { region_type }
region_type := "jt8" | "jt16" | "jt32" | "jta32"
end_data_region_directive := ".end_data_region"
The previous handling of ARM-style "$d.*" labels was broken and has
been removed. Specifically, it didn't handle ARM vs. Thumb mode when
marking the end of the section.
rdar://11459456
llvm-svn: 157062
It is no longer necessary to separate VirtCopies, PhysCopies, and
ImpDefCopies. Implicitly defined copies are extremely rare after we
added the ProcessImplicitDefs pass, and physical register copies are not
joined any longer.
llvm-svn: 157059
This has been disabled for a while, and it is not a feature we want to
support. Copies between physical and virtual registers are eliminated by
good hinting support in the register allocator. Joining virtual and
physical registers is really a form of register allocation, and the
coalescer is not properly equipped to do that. In particular, it cannot
backtrack coalescing decisions, and sometimes that would cause it to
create programs that were impossible to register allocate, by exhausting
a small register class.
It was also very difficult to keep track of the live ranges of aliasing
registers when extending the live range of a physreg. By disabling
physreg joining, we can let fixed physreg live ranges remain constant
throughout the register allocator super-pass.
One type of physreg joining remains: A virtual register that has a
single value which is a copy of a reserved register can be merged into
the reserved physreg. This always lowers register pressure, and since we
don't compute live ranges for reserved registers, there are no problems
with aliases.
llvm-svn: 157055
SelectionDAGBuilder::Clusterify : main functinality was replaced with CRSBuilder::optimize, so big part of Clusterify's code was reduced.
llvm-svn: 157046
non-profitable commute using outdated info. The test case would still fail
because of poor pre-RA schedule. That will be fixed by MI scheduler.
rdar://11472010
llvm-svn: 157038
Introduce the basic strategy for register pressure scheduling.
1) Respect target limits at all times.
2) Indentify critical register classes (pressure sets).
Track pressure within the scheduled region.
Avoid increasing scheduled pressure for critical registers.
3) Avoid exceeding the max pressure of the region prior to scheduling.
Added logic for picking between the top and bottom ready Q's based on
regpressure heuristics.
Status: functional but needs to be asjusted to achieve good results.
llvm-svn: 157006
RegisterCoalescer set <undef> flags on all operands of copy instructions
that are scheduled to be removed. This is so they won't affect
shrinkToUses() by introducing false register reads.
Make sure those <undef> flags are never cleared, or shrinkToUses() could
cause live intervals to end at instructions about to be deleted.
This would be a lot simpler if RegisterCoalescer could just erase joined
copies immediately instead of keeping all the to-be-deleted instructions
around.
This fixes PR12862. Unfortunately, bugpoint can't create a sane test
case for this. Like many other coalescer problems, this failure depends
of a very fragile series of events.
<rdar://problem/11474428>
llvm-svn: 157001
When widening an existing <def,reads-undef> operand to a super-register,
it may be necessary to clear the <undef> flag because the wider register
is now read-modify-write through the instruction.
Conversely, it may be necessary to add an <undef> flag when the
coalescer turns a full-register def into a sub-register def, but the
larger register wasn't live before the instruction.
This happens in test/CodeGen/ARM/coalesce-subregs.ll, but the test
is too small for the <undef> flags to affect the generated code.
llvm-svn: 156951
It is now possible to coalesce weird skewed sub-register copies by
picking a super-register class larger than both original registers. The
included test case produces code like this:
vld2.32 {d16, d17, d18, d19}, [r0]!
vst2.32 {d18, d19, d20, d21}, [r0]
We still perform interference checking as if it were a normal full copy
join, so this is still quite conservative. In particular, the f1 and f2
functions in the included test case still have remaining copies because
of false interference.
llvm-svn: 156878
It is possible to coalesce two overlapping registers to a common
super-register that it larger than both of the original registers.
The important difference is that it may be necessary to rewrite DstReg
operands as well as SrcReg operands because the sub-register index has
changed.
This behavior is still disabled by CoalescerPair.
llvm-svn: 156869
Now both SrcReg and DstReg can be sub-registers of the final coalesced
register.
CoalescerPair::setRegisters still rejects such copies because
RegisterCoalescer doesn't yet handle them.
llvm-svn: 156848
This feature avoids creating edges in the scheduler's dependence graph
for non-aliasing memory operations according to whichever alias
analysis is available. It has been fully tested in Hexagon. Before
making this default, it needs to be extended to handle multiple
MachineMemOperands, compile time needs more evaluation, and
benchmarking on X86 and ARM is needed.
Patch by Sergei Larin!
llvm-svn: 156842
This patch will optimize the following cases:
sub r1, r3 | sub r1, imm
cmp r3, r1 or cmp r1, r3 | cmp r1, imm
bge L1
TO
subs r1, r3
bge L1 or ble L1
If the branch instruction can use flag from "sub", then we can replace
"sub" with "subs" and eliminate the "cmp" instruction.
rdar: 10734411
llvm-svn: 156599
Prioritize the instruction that comes closest to keeping pressure
under the target's limit. Then prioritize instructions that avoid
increasing the max pressure in the scheduled region. The max pressure
heuristic is a tad aggressive. Later I'll fix it to consider the
unscheduled pressure as well.
WIP: This is mostly functional but untested and not likely to do much good yet.
llvm-svn: 156574
Added getMaxExcessUpward/DownwardPressure. They somewhat abuse the
tracker by speculatively handling an instruction out of order. But it
is convenient for now. In the future, we will cache each instruction's
pressure contribution to make this efficient.
llvm-svn: 156561
This patch will optimize the following cases:
sub r1, r3 | sub r1, imm
cmp r3, r1 or cmp r1, r3 | cmp r1, imm
bge L1
TO
subs r1, r3
bge L1 or ble L1
If the branch instruction can use flag from "sub", then we can replace
"sub" with "subs" and eliminate the "cmp" instruction.
rdar: 10734411
llvm-svn: 156550
When a combine twiddles an extract_vector, care should be take to preserve
the type of the index operand. No luck extracting a reasonable testcase,
unfortunately.
rdar://11391009
llvm-svn: 156419
At least some of them:
%vreg1:sub_16bit = COPY %vreg2:sub_16bit; GR64:%vreg1, GR32: %vreg2
Previously, we couldn't figure out that the above copy could be
eliminated by coalescing %vreg2 with %vreg1:sub_32bit.
The new getCommonSuperRegClass() hook makes it possible.
This is not very useful yet since the unmodified part of the destination
register usually interferes with the source register. The coalescer
needs to understand sub-register interference checking first.
llvm-svn: 156334
The getPointerRegClass() hook can return register classes that depend on
the calling convention of the current function (ptr_rc_tailcall).
So far, we have been able to infer the calling convention from the
subtarget alone, but as we add support for multiple calling conventions
per target, that no longer works.
Patch by Yiannis Tsiouris!
llvm-svn: 156328
This will be used to determine whether it's profitable to turn a select into a
branch when the branch is likely to be predicted.
Currently enabled for everything but Atom on X86 and Cortex-A9 devices on ARM.
I'm not entirely happy with the name of this flag, suggestions welcome ;)
llvm-svn: 156233
We want the representative register class to contain the largest
super-registers available. This makes the function less sensitive to the
register class numbering.
llvm-svn: 156220
The masks returned by SuperRegClassIterator are computed automatically
by TableGen. This is better than depending on the manually specified
SuperRegClasses.
llvm-svn: 156147
to catch cases like:
%reg1024<def> = MOV r1
%reg1025<def> = MOV r0
%reg1026<def> = ADD %reg1024, %reg1025
r0 = MOV %reg1026
By commuting ADD, it let coalescer eliminate all of the copies. However, there
was a bug in the heuristics where it ended up commuting the ADD in:
%reg1024<def> = MOV r0
%reg1025<def> = MOV 0
%reg1026<def> = ADD %reg1024, %reg1025
r0 = MOV %reg1026
That did no benefit but rather ensure the last MOV would not be coalesced.
rdar://11355268
llvm-svn: 156048
The ensures that virtual registers always belong to an allocatable class.
If your target attempts to create a vreg for an operand that has no
allocatable register subclass, you will crash quickly.
This ensures that targets define register classes as intended.
llvm-svn: 156046
The TargetPassManager's default constructor wants to initialize the PassManager
to 'null'. But it's illegal to bind a null reference to a null l-value. Make the
ivar a pointer instead.
PR12468
llvm-svn: 155902
This time, also fix the caller of AddGlue to properly handle
incomplete chains. AddGlue had failure modes, but shamefully hid them
from its caller. It's luck ran out.
Fixes rdar://11314175: BuildSchedUnits assert.
llvm-svn: 155749
DAGCombine strangeness may result in multiple loads from the same
offset. They both may try to glue themselves to another load. We could
insist that the redundant loads glue themselves to each other, but the
beter fix is to bail out from bad gluing at the time we detect it.
Fixes rdar://11314175: BuildSchedUnits assert.
llvm-svn: 155668
Cross-class joins have been normal and fully supported for a while now.
With TableGen generating the getMatchingSuperRegClass() hook, they are
unlikely to cause problems again.
llvm-svn: 155552
Remove the heuristic for disabling cross-class joins. The greedy
register allocator can handle the narrow register classes, and when it
splits a live range, it can pick a larger register class.
Benchmarks were unaffected by this change.
<rdar://problem/11302212>
llvm-svn: 155551
MachineInstr sequence.
This uses the new target interface for tracking register pressure
using pressure sets to model overlapping register classes and
subregisters.
RegisterPressure results can be tracked incrementally or stored at
region boundaries. Global register pressure can be deduced from local
RegisterPressure results if desired.
This is an early, somewhat untested implementation. I'm working on
testing it within the context of a register pressure reducing
MachineScheduler.
llvm-svn: 155454
on X86 Atom. Some of our tests failed because the tail merging part of
the BranchFolding pass was creating new basic blocks which did not
contain live-in information. When the anti-dependency code in the Post-RA
scheduler ran, it would sometimes rename the register containing
the function return value because the fact that the return value was
live-in to the subsequent block had been lost. To fix this, it is necessary
to run the RegisterScavenging code in the BranchFolding pass.
This patch makes sure that the register scavenging code is invoked
in the X86 subtarget only when post-RA scheduling is being done.
Post RA scheduling in the X86 subtarget is only done for Atom.
This patch adds a new function to the TargetRegisterClass to control
whether or not live-ins should be preserved during branch folding.
This is necessary in order for the anti-dependency optimizations done
during the PostRASchedulerList pass to work properly when doing
Post-RA scheduling for the X86 in general and for the Intel Atom in particular.
The patch adds and invokes the new function trackLivenessAfterRegAlloc()
instead of using the existing requiresRegisterScavenging().
It changes BranchFolding.cpp to call trackLivenessAfterRegAlloc() instead of
requiresRegisterScavenging(). It changes the all the targets that
implemented requiresRegisterScavenging() to also implement
trackLivenessAfterRegAlloc().
It adds an assertion in the Post RA scheduler to make sure that post RA
liveness information is available when it is needed.
It changes the X86 break-anti-dependencies test to use –mcpu=atom, in order
to avoid running into the added assertion.
Finally, this patch restores the use of anti-dependency checking
(which was turned off temporarily for the 3.1 release) for
Intel Atom in the Post RA scheduler.
Patch by Andy Zhang!
Thanks to Jakob and Anton for their reviews.
llvm-svn: 155395
The X86 target is editing the selection DAG while isel is selecting
nodes following a topological ordering. When the DAG hacking triggers
CSE, nodes can be deleted and bad things happen.
llvm-svn: 155257
Now that multiple DAGUpdateListeners can be active at the same time,
ISelPosition can become a local variable in DoInstructionSelection.
We simply register an ISelUpdater with CurDAG while ISelPosition exists.
llvm-svn: 155249
Instead of passing listener pointers to RAUW, let SelectionDAG itself
keep a linked list of interested listeners.
This makes it possible to have multiple listeners active at once, like
RAUWUpdateListener was already doing. It also makes it possible to
register listeners up the call stack without controlling all RAUW calls
below.
DAGUpdateListener uses an RAII pattern to add itself to the SelectionDAG
list of active listeners.
llvm-svn: 155248
The <undef> flag on a def operand only applies to partial register
redefinitions. Only print the flag when relevant, and print it as
<def,read-undef> to make it clearer what it means.
llvm-svn: 155239
This nicely handles the most common case of virtual register sets, but
also handles anticipated cases where we will map pointers to IDs.
The goal is not to develop a completely generic SparseSet
template. Instead we want to handle the expected uses within llvm
without any template antics in the client code. I'm adding a bit of
template nastiness here, and some assumption about expected usage in
order to make the client code very clean.
The expected common uses cases I'm designing for:
- integer keys that need to be reindexed, and may map to additional
data
- densely numbered objects where we want pointer keys because no
number->object map exists.
llvm-svn: 155227
commits have had several major issues pointed out in review, and those
issues are not being addressed in a timely fashion. Furthermore, this
was all committed leading up to the v3.1 branch, and we don't need piles
of code with outstanding issues in the branch.
It is possible that not all of these commits were necessary to revert to
get us back to a green state, but I'm going to let the Hexagon
maintainer sort that out. They can recommit, in order, after addressing
the feedback.
Reverted commits, with some notes:
Primary commit r154616: HexagonPacketizer
- There are lots of review comments here. This is the primary reason
for reverting. In particular, it introduced large amount of warnings
due to a bad construct in tablegen.
- Follow-up commits that should be folded back into this when
reposting:
- r154622: CMake fixes
- r154660: Fix numerous build warnings in release builds.
- Please don't resubmit this until the three commits above are
included, and the issues in review addressed.
Primary commit r154695: Pass to replace transfer/copy ...
- Reverted to minimize merge conflicts. I'm not aware of specific
issues with this patch.
Primary commit r154703: New Value Jump.
- Primarily reverted due to merge conflicts.
- Follow-up commits that should be folded back into this when
reposting:
- r154703: Remove iostream usage
- r154758: Fix CMake builds
- r154759: Fix build warnings in release builds
- Please incorporate these fixes and and review feedback before
resubmitting.
Primary commit r154829: Hexagon V5 (floating point) support.
- Primarily reverted due to merge conflicts.
- Follow-up commits that should be folded back into this when
reposting:
- r154841: Remove unused variable (fixing build warnings)
There are also accompanying Clang commits that will be reverted for
consistency.
llvm-svn: 155047
transformation:
(X op C1) ^ C2 --> (X op C1) & ~C2 iff (C1&C2) == C2
should be done.
This change has been tested:
Using a debug+asserts build:
on the specific test case that brought this bug to light
make check-all
lnt nt
using this clang to build a release version of clang
Using the release+asserts clang-with-clang build:
on the specific test case that brought this bug to light
make check-all
lnt nt
Checking in because Evan wants it checked in. Test case forthcoming after
scrubbing.
llvm-svn: 154955
for the life of me remember why I wrote it this way, but I can't see any good
reason for it now. This patch replaces the custom linked list with an ilist.
This change should preserve the existing numberings exactly, so no generated code
should change (if it does, file a bug!).
llvm-svn: 154904
both fallthrough and a conditional branch target the same successor.
Gracefully delete the conditional branch and introduce any unconditional
branch needed to reach the actual successor. This fixes memory
corruption in 2009-06-15-RegScavengerAssert.ll and possibly other tests.
Also, while I'm here fix a latent bug I spotted by inspection. I never
applied the same fundamental fix to this fallthrough successor finding
logic that I did to the logic used when there are no conditional
branches. As a consequence it would have selected landing pads had they
be aligned in just the right way here. I don't have a test case as
I spotted this by inspection, and the previous time I found this
required have of TableGen's source code to produce it. =/ I hate backend
bugs. ;]
Thanks to Jim Grosbach for helping me reason through this and reviewing
the fix.
llvm-svn: 154867
This is mostly to test the waters. I'd like to get results from FNT
build bots and other bots running on non-x86 platforms.
This feature has been pretty heavily tested over the last few months by
me, and it fixes several of the execution time regressions caused by the
inlining work by preventing inlining decisions from radically impacting
block layout.
I've seen very large improvements in yacr2 and ackermann benchmarks,
along with the expected noise across all of the benchmark suite whenever
code layout changes. I've analyzed all of the regressions and fixed
them, or found them to be impossible to fix. See my email to llvmdev for
more details.
I'd like for this to be in 3.1 as it complements the inliner changes,
but if any failures are showing up or anyone has concerns, it is just
a flag flip and so can be easily turned off.
I'm switching it on tonight to try and get at least one run through
various folks' performance suites in case SPEC or something else has
serious issues with it. I'll watch bots and revert if anything shows up.
llvm-svn: 154816
rotation. When there is a loop backedge which is an unconditional
branch, we will end up with a branch somewhere no matter what. Try
placing this backedge in a fallthrough position above the loop header as
that will definitely remove at least one branch from the loop iteration,
where whole loop rotation may not.
I haven't seen any benchmarks where this is important but loop-blocks.ll
tests for it, and so this will be covered when I flip the default.
llvm-svn: 154812
laid out in a form with a fallthrough into the header and a fallthrough
out of the bottom. In that case, leave the loop alone because any
rotation will introduce unnecessary branches. If either side looks like
it will require an explicit branch, then the rotation won't add any, do
it to ensure the branch occurs outside of the loop (if possible) and
maximize the benefit of the fallthrough in the bottom.
llvm-svn: 154806
This is a complex change that resulted from a great deal of
experimentation with several different benchmarks. The one which proved
the most useful is included as a test case, but I don't know that it
captures all of the relevant changes, as I didn't have specific
regression tests for each, they were more the result of reasoning about
what the old algorithm would possibly do wrong. I'm also failing at the
moment to craft more targeted regression tests for these changes, if
anyone has ideas, it would be welcome.
The first big thing broken with the old algorithm is the idea that we
can take a basic block which has a loop-exiting successor and a looping
successor and use the looping successor as the layout top in order to
get that particular block to be the bottom of the loop after layout.
This happens to work in many cases, but not in all.
The second big thing broken was that we didn't try to select the exit
which fell into the nearest enclosing loop (to which we exit at all). As
a consequence, even if the rotation worked perfectly, it would result in
one of two bad layouts. Either the bottom of the loop would get
fallthrough, skipping across a nearer enclosing loop and thereby making
it discontiguous, or it would be forced to take an explicit jump over
the nearest enclosing loop to earch its successor. The point of the
rotation is to get fallthrough, so we need it to fallthrough to the
nearest loop it can.
The fix to the first issue is to actually layout the loop from the loop
header, and then rotate the loop such that the correct exiting edge can
be a fallthrough edge. This is actually much easier than I anticipated
because we can handle all the hard parts of finding a viable rotation
before we do the layout. We just store that, and then rotate after
layout is finished. No inner loops get split across the post-rotation
backedge because we check for them when selecting the rotation.
That fix exposed a latent problem with our exitting block selection --
we should allow the backedge to point into the middle of some inner-loop
chain as there is no real penalty to it, the whole point is that it
*won't* be a fallthrough edge. This may have blocked the rotation at all
in some cases, I have no idea and no test case as I've never seen it in
practice, it was just noticed by inspection.
Finally, all of these fixes, and studying the loops they produce,
highlighted another problem: in rotating loops like this, we sometimes
fail to align the destination of these backwards jumping edges. Fix this
by actually walking the backwards edges rather than relying on loopinfo.
This fixes regressions on heapsort if block placement is enabled as well
as lots of other cases where the previous logic would introduce an
abundance of unnecessary branches into the execution.
llvm-svn: 154783
This is a special flag for targets that really want their block
terminators in the DAG. The default scheduler cannot handle this
correctly, so it becomes the specialized scheduler's responsibility to
schedule terminators.
llvm-svn: 154712
- Don't copy offsets into HashData, the underlying vector won't change once the table is finalized.
- Allocate HashData and HashDataContents in a BumpPtrAllocator.
- Allocate string map entries in the same allocator.
- Random cleanups.
llvm-svn: 154694
Fix a dagcombine optimization which assumes that the vsetcc result type is always
of the same size as the compared values. This is ture for SSE/AVX/NEON but not
for all targets.
llvm-svn: 154490
Allow cheap instructions to be hoisted if they are register pressure
neutral or better. This happens if the instruction is the last loop use
of another virtual register.
Only expensive instructions are allowed to increase loop register
pressure.
llvm-svn: 154455
Hoisting a value that is used by a PHI in the loop will introduce a
copy because the live range is extended to cross the PHI.
The same applies to PHIs in exit blocks.
Also use this opportunity to make HasLoopPHIUse() non-recursive.
llvm-svn: 154454
the loop header has a non-loop predecessor which has been pre-fused into
its chain due to unanalyzable branches. In this case, rotating the
header into the body of the loop in order to place a loop exit at the
bottom of the loop is a Very Bad Idea as it makes the loop
non-contiguous.
I'm working on a good test case for this, but it's a bit annoynig to
craft. I should get one shortly, but I'm submitting this now so I can
begin the (lengthy) performance analysis process. An initial run of LNT
looks really, really good, but there is too much noise there for me to
trust it much.
llvm-svn: 154395
legalizer always use the DAG entry node. This is wrong when the libcall is
emitted as a tail call since it effectively folds the return node. If
the return node's input chain is not the entry (i.e. call, load, or store)
use that as the tail call input chain.
PR12419
rdar://9770785
rdar://11195178
llvm-svn: 154370
This patch restores TwoAddressInstructionPass's pre-r153892 behaviour when
rescheduling instructions in TryInstructionTransform. Hopefully this will fix
PR12493. To refix PR11861, lowering of INSERT_SUBREGS is deferred until after
the copy that unties the operands is emitted (this seems to be a more
appropriate fix for that issue anyway).
llvm-svn: 154338
when -ffast-math, i.e. don't just always do it if the reciprocal can
be formed exactly. There is already an IR level transform that does
that, and it does it more carefully.
llvm-svn: 154296
in TargetLowering. There was already a FIXME about this location being
odd. The interface is simplified as a consequence. This will also make
it easier to change TLS models when compiling with PIE.
llvm-svn: 154292
where a chain outside of the loop block-set ended up in the worklist for
scheduling as part of the contiguous loop. However, asserting the first
block in the chain is in the loop-set isn't a valid check -- we may be
forced to drag a chain into the worklist due to one block in the chain
being part of the loop even though the first block is *not* in the loop.
This occurs when we have been forced to form a chain early due to
un-analyzable branches.
No test case here as I have no idea how to even begin reducing one, and
it will be hopelessly fragile. We have to somehow end up with a loop
header of an inner loop which is a successor of a basic block with an
unanalyzable pair of branch instructions. Ow. Self-host triggers it so
it is unlikely it will regress.
This at least gets block placement back to passing selfhost and the test
suite. There are still a lot of slowdown that I don't like coming out of
block placement, although there are now also a lot of speedups. =[ I'm
seeing swings in both directions up to 10%. I'm going to try to find
time to dig into this and see if we can turn this on for 3.1 as it does
a really good job of cleaning up after some loops that degraded with the
inliner changes.
llvm-svn: 154287
shuffle node because it could introduce new shuffle nodes that were not
supported efficiently by the target.
2. Add a more restrictive shuffle-of-shuffle optimization for cases where the
second shuffle reverses the transformation of the first shuffle.
llvm-svn: 154266
reciprocal if converting to the reciprocal is exact. Do it even if inexact
if -ffast-math. This substantially speeds up ac.f90 from the polyhedron
benchmarks.
llvm-svn: 154265
LSR always tries to make the ICmp in the loop latch use the incremented
induction variable. This allows the induction variable to be kept in a
single register.
When the induction variable limit is equal to the stride,
SimplifySetCC() would break LSR's hard work by transforming:
(icmp (add iv, stride), stride) --> (cmp iv, 0)
This forced us to use lea for the IC update, preventing the simpler
incl+cmp.
<rdar://problem/7643606>
<rdar://problem/11184260>
llvm-svn: 154119
This allows us to keep passing reduced masks to SimplifyDemandedBits, but
know about all the bits if SimplifyDemandedBits fails. This allows instcombine
to simplify cases like the one in the included testcase.
llvm-svn: 154011
When folding X == X we need to check getBooleanContents() to determine if the
result is a vector of ones or a vector of negative ones.
I tried creating a test case, but the problem seems to only be exposed on a
much older version of clang (around r144500).
rdar://10923049
llvm-svn: 153966
brace) so that we get more accurate line number information about the
declaration of a given function and the line where the function
first starts.
Part of rdar://11026482
llvm-svn: 153916
Do not try to optimize swizzles of shuffles if the source shuffle has more than
a single user, except when the source shuffle is also a swizzle.
llvm-svn: 153864
This is the CodeGen equivalent of r153747. I tested that there is not noticeable
performance difference with any combination of -O0/-O2 /-g when compiling
gcc as a single compilation unit.
llvm-svn: 153817
here but it has no other uses, then we have a problem. E.g.,
int foo (const int *x) {
char a[*x];
return 0;
}
If we assign 'a' a vreg and fast isel later on has to use the selection
DAG isel, it will want to copy the value to the vreg. However, there are
no uses, which goes counter to what selection DAG isel expects.
<rdar://problem/11134152>
llvm-svn: 153705
Some targets still mess up the liveness information, but that isn't
verified after MRI->invalidateLiveness().
The verifier can still check other useful things like register classes
and CFG, so it should be enabled after all passes.
llvm-svn: 153615
The late scheduler depends on accurate liveness information if it is
breaking anti-dependencies, so we should be able to verify it.
Relax the terminator checking in the machine code verifier so it can
handle the basic blocks created by if conversion.
llvm-svn: 153614
Extract the liveness verification into its own method.
This makes it possible to run the machine code verifier after liveness
information is no longer required to be valid.
llvm-svn: 153596
Branch folding can use a register scavenger to update liveness
information when required. Don't do that if liveness information is
already invalid.
llvm-svn: 153517
Late optimization passes like branch folding and tail duplication can
transform the machine code in a way that makes it expensive to keep the
register liveness information up to date. There is a fuzzy line between
register allocation and late scheduling where the liveness information
degrades.
The MRI::tracksLiveness() flag makes the line clear: While true,
liveness information is accurate, and can be used for register
scavenging. Once the flag is false, liveness information is not
accurate, and can only be used as a hint.
Late passes generally don't need the liveness information, but they will
sometimes use the register scavenger to help update it. The scavenger
enforces strict correctness, and we have to spend a lot of code to
update register liveness that may never be used.
llvm-svn: 153511
copies being considered for removal. Make sure to track all of the copies,
rather than just the most recent encountered, by holding a DenseSet instead of
an unsigned in SrcMap.
No test case - couldn't reduce something with a sane size.
llvm-svn: 153487
execution-time regression for nsieve-bits on the ARMv7 -O0 -g nightly tester.
This may also improve compile-time on architectures that would otherwise
generate a libcall for urem (e.g., ARM) or fall back to the DAG selector.
rdar://10810716
llvm-svn: 153230
Type legalization can zero-extend the elements of the build_vector node, so,
for example, we may have an <8 x i8> with i32 elements of value 255. That
should return 'true' for the vector being all ones.
llvm-svn: 153203
i128). In that case, we may not be able to print out the MCExpr as an
expression. For instance, we could have an MCExpr like this:
0xBEEF0000BEEF0000 | (0xBEEF0000BEEF0000 << 64)
The MCExpr printer handles sizes up to 64-bits, but this expression would
require 128-bits. In this situation, try to evaluate the constant expression and
emit that as the value into 64-bit chunks.
<rdar://problem/11070338>
llvm-svn: 153081
a variable. The previous code would break the debug info changing
code invariant. This will regress debug info for arguments where
we elide the alloca created.
Fixes rdar://11066468
llvm-svn: 153074
These edges are not really necessary, but it is consistent with the
way we currently create physreg edges. Scheduler heuristics that
expect a DAG edge to the block terminator could benefit from this
change. Although in the future I hope we have a better mechanism for
modeling latency across scheduling regions.
llvm-svn: 152895
on our internal nightly testers. So, basically revert r152486 again.
Abbreviated original commit message:
Implement a more intelligent way of spilling uses across an invoke boundary.
It looks as if Chander's inlining work, r152737, exposed an issue.
llvm-svn: 152887
It caused MSP430DAGToDAGISel::SelectIndexedBinOp() to be miscompiled.
When two ReplaceUses()'s are expanded as inline, vtable in base class is stored to latter (ISelUpdater)ISU.
llvm-svn: 152877
out the DW_AT_name. Older gdbs unfortunately still use it to
disambiguate member functions in templated classes (gdb.cp/templates.exp).
rdar://11043421 (which is now deferred for a bit)
llvm-svn: 152782
There were cases where a value could be used and it's both crossing an invoke
and NOT crossing an invoke. This could happen in the landing pads. In that case,
we will demote the value to the stack like we did before.
<rdar://problem/10609139>
llvm-svn: 152705
New flags: -misched-topdown, -misched-bottomup. They can be used with
the default scheduler or with -misched=shuffle. Without either
topdown/bottomup flag -misched=shuffle now alternates scheduling
direction.
LiveIntervals update is unimplemented with bottom-up scheduling, so
only -misched-topdown currently works.
Capped the ScheduleDAG hierarchy with a concrete ScheduleDAGMI class.
ScheduleDAGMI is aware of the top and bottom of the unscheduled zone
within the current region. Scheduling policy can be plugged into
the ScheduleDAGMI driver by implementing MachineSchedStrategy.
ConvergingScheduler is now the default scheduling algorithm.
It exercises the new driver but still does no reordering.
llvm-svn: 152700
output (we're emitting a specification already and the information
isn't changing).
Saves 1% on the debug information for a build of llvm.
Fixes rdar://11043421
llvm-svn: 152697
(i16 load $addr+c*sizeof(i16)) and replace uses of (i32 vextract) with the
i16 load. It should issue an extload instead: (i32 extload $addr+c*sizeof(i16)).
rdar://11035895
llvm-svn: 152675
Renamed methods caseBegin, caseEnd and caseDefault with case_begin, case_end, and case_default.
Added some notes relative to case iterators.
llvm-svn: 152532
The old way of determine when and where to spill a value that was used inside of
a landing pad resulted in spilling that value everywhere and not just at the
invoke edge.
This algorithm determines which values are used within a landing pad. It then
spills those values before the invoke and reloads them before the uses. This
should prevent excessive spilling in many cases, e.g. inside of loops.
<rdar://problem/10609139>
llvm-svn: 152486
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20120130/136146.html
Implemented CaseIterator and it solves almost all described issues: we don't need to mix operand/case/successor indexing anymore. Base iterator class is implemented as a template since it may be initialized either from "const SwitchInst*" or from "SwitchInst*".
ConstCaseIt is just a read-only iterator.
CaseIt is read-write iterator; it allows to change case successor and case value.
Usage of iterator allows totally remove resolveXXXX methods. All indexing convertions done automatically inside the iterator's getters.
Main way of iterator usage looks like this:
SwitchInst *SI = ... // intialize it somehow
for (SwitchInst::CaseIt i = SI->caseBegin(), e = SI->caseEnd(); i != e; ++i) {
BasicBlock *BB = i.getCaseSuccessor();
ConstantInt *V = i.getCaseValue();
// Do something.
}
If you want to convert case number to TerminatorInst successor index, just use getSuccessorIndex iterator's method.
If you want initialize iterator from TerminatorInst successor index, use CaseIt::fromSuccessorIndex(...) method.
There are also related changes in llvm-clients: klee and clang.
llvm-svn: 152297
Allow targets to provide their own schedulers (subclass of
ScheduleDAGInstrs) to the misched pass. Select schedulers using
-misched=...
llvm-svn: 152278
ScheduleDAGInstrs will be the main interface for MI-level
schedulers. Make sure it's readable: one page of protected fields, one
page of public methids.
llvm-svn: 152258
This one is particularly annoying because the hashing algorithm is
highly specialized, with a strange "equivalence" definition that subsets
the fields involved.
Still, this looks at the exact same set of data as the old code, but
without bitwise or-ing over parts of it and other mixing badness. No
functionality changed here. I've left a substantial fixme about the fact
that there is a cleaner and more principled way to do this, but it
requires making the equality definition actual stable for particular
types...
llvm-svn: 152218
the DebugLoc information can be maintained throughout by grabbing the DebugLoc
before the RemoveBranch and then passing the result to the InsertBranch.
Patch by Andrew Stanford-Jason!
llvm-svn: 152212
ScheduleDAG is responsible for the DAG: SUnits and SDeps. It provides target hooks for latency computation.
ScheduleDAGInstrs extends ScheduleDAG and defines the current scheduling region in terms of MachineInstr iterators. It has access to the target's scheduling itinerary data. ScheduleDAGInstrs provides the logic for building the ScheduleDAG for the sequence of MachineInstrs in the current region. Target's can implement highly custom schedulers by extending this class.
ScheduleDAGPostRATDList provides the driver and diagnostics for current postRA scheduling. It maintains a current Sequence of scheduled machine instructions and logic for splicing them into the block. During scheduling, it uses the ScheduleHazardRecognizer provided by the target.
Specific changes:
- Removed driver code from ScheduleDAG. clearDAG is the only interface needed.
- Added enterRegion/exitRegion hooks to ScheduleDAGInstrs to delimit the scope of each scheduling region and associated DAG. They should be used to setup and cleanup any region-specific state in addition to the DAG itself. This is necessary because we reuse the same ScheduleDAG object for the entire function. The target may extend these hooks to do things at regions boundaries, like bundle terminators. The hooks are called even if we decide not to schedule the region. So all instructions in a block are "covered" by these calls.
- Added ScheduleDAGInstrs::begin()/end() public API.
- Moved Sequence into the driver layer, which is specific to the scheduling algorithm.
llvm-svn: 152208
The first def of a virtual register cannot also read the register.
Assert on such bad machine code instead of trying to fix it.
TwoAddressInstructionPass should never create code like that.
llvm-svn: 152010
The inline table needs to be constructed ahead of time so that it doesn't try to
create new strings while we're emitting everything.
This reverts commit a8ff9bccb399183cdd5f1c3cec2bda763664b4b0.
llvm-svn: 151864
fixups that are being used to determine section offsets. Reduces
the total number of fixups by 50% for a non-trivial testcase.
Part of rdar://10413936
llvm-svn: 151852
Simply treat bundles as instructions. Spill code is inserted between
bundles, never inside a bundle. Rewrite all operands in a bundle at
once.
Don't attempt and memory operand folding inside bundles.
llvm-svn: 151787
This allows the function to be inlined, and makes it suitable for use in
getInstructionIndex().
Also provide a const version. C++ is great for touch typing practice.
llvm-svn: 151782
This function does more or less the same as
MI::readsWritesVirtualRegister(), but it supports bundles as well.
It also determines if any constraint requires reading and writing
operands to use the same register. Most clients want to know.
Use the more modern MO.readsReg() instead of trying to sort out undefs
and partial redefines. Stop supporting the extra full <imp-def> operand
as an alternative to <def,undef> sub-register defines.
llvm-svn: 151690
Extract a base class and provide four specific sub-classes for iterating
over const/non-const bundles/instructions.
This eliminates the mystery bool constructor argument.
llvm-svn: 151684
methods are no longer needed now that LinearScan has gone away.
(Contains tweaks trivialSpillEverywhere to enable the removal of getNewVRegs).
llvm-svn: 151658
To avoid problems with zero shifts when getting the bits that move between words
we use a trick: first shift the by amount-1, then do another shift by one. When
amount is 0 (and size 32) we first shift by 31, then by one, instead of by 32.
Also fix a latent bug that emitted the low and high words in the wrong order
when shifting right.
Fixes PR12113.
llvm-svn: 151637
When the GEP index is a vector of pointers, the code that calculated the size
of the element started from the vector type, and not the contained pointer type.
As a result, instead of looking at the data element pointed by the vector, this
code used the size of the vector. This works for 32bit members (on 32bit
systems), but not for other types. Added code to peel the vector type and
added a test.
llvm-svn: 151626
the processor keeps a return addresses stack (RAS) which stores the address
and the instruction execution state of the instruction after a function-call
type branch instruction.
Calling a "noreturn" function with normal call instructions (e.g. bl) can
corrupt RAS and causes 100% return misprediction so LLVM should use a
unconditional branch instead. i.e.
mov lr, pc
b _foo
The "mov lr, pc" is issued in order to get proper backtrace.
rdar://8979299
llvm-svn: 151623
Richard Smith