This change cleans up two issues with Altivec register spilling:
1. The spilling code was inefficient (using two instructions, and add and a
load, when just one would do)
2. The code assumed that r0 would always be available (true for now, but this
will change)
The new code handles VR spilling just like GPR spills but forced into r+r mode.
As a result, when any VR spills are present, we must now always allocate the
register-scavenger spill slot.
llvm-svn: 177231
As a follow-up to r158719, remove PPCRegisterInfo::avoidWriteAfterWrite.
Jakob pointed out in response to r158719 that this callback is currently unused
and so this has no effect (and the speedups that I thought that I had observed
as a result of implementing this function must have been noise).
llvm-svn: 177228
Implicit defs are not currently positional and not modeled by the
per-operand machine model. Unfortunately, we treat defs that are part
of the architectural instruction description, like flags, the same as
other implicit defs. Really, they should have a fixed MachineInstr
layout and probably shouldn't be "implicit" at all.
For now, we'll change the default latency to be the max operand
latency. That will give flag setting operands full latency for x86
folded loads. Other kinds of "fake" implicit defs don't occur prior to
regalloc anyway, and we would like them to go away postRegAlloc as
well.
llvm-svn: 177227
Since almost all X86 instructions can fold loads, use a multiclass to
define register/memory pairs of SchedWrites.
An X86FoldableSchedWrite represents the register version of an
instruction. It holds a reference to the SchedWrite to use when the
instruction folds a load.
This will be used inside multiclasses that define rr and rm instruction
versions together.
llvm-svn: 177210
I was too pessimistic in r177105. Vector selects that fit into a legal register
type lower just fine. I was mislead by the code fragment that I was using. The
stores/loads that I saw in those cases came from lowering the conditional off
an address.
Changing the code fragment to:
%T0_3 = type <8 x i18>
%T1_3 = type <8 x i1>
define void @func_blend3(%T0_3* %loadaddr, %T0_3* %loadaddr2,
%T1_3* %blend, %T0_3* %storeaddr) {
%v0 = load %T0_3* %loadaddr
%v1 = load %T0_3* %loadaddr2
==> FROM:
;%c = load %T1_3* %blend
==> TO:
%c = icmp slt %T0_3 %v0, %v1
==> USE:
%r = select %T1_3 %c, %T0_3 %v0, %T0_3 %v1
store %T0_3 %r, %T0_3* %storeaddr
ret void
}
revealed this mistake.
radar://13403975
llvm-svn: 177170
Unaligned access is supported on PPC for non-vector types, and is generally
more efficient than manually expanding the loads and stores.
A few of the existing test cases were using expanded unaligned loads and stores
to test other features (like load/store with update), and for these test cases,
unaligned access remains disabled.
llvm-svn: 177160
In preparation for the addition of other SIMD ISA extensions (such as QPX) we
need to make sure that all Altivec patterns are properly predicated on having
Altivec support.
No functionality change intended (one test case needed to be updated b/c it
assumed that Altivec intrinsics would be supported without enabling Altivec
support).
llvm-svn: 177152
This is a very late complement to r130637 which fixed this on x86_64. Fixes
pr15448.
Since it looks like that every elf architecture uses this encoding when using
cfi, make it the default for elf. Just exclude mips64el. It has a lovely
.ll -> .o test (ef_frame.ll) that tests that nothing changes in the binary
content of the .eh_frame produced by llc. Oblige it.
llvm-svn: 177141
For spills into a large stack frame, the FI-elimination code uses the register
scavenger to obtain a free GPR for use with an r+r-addressed load or store.
When there are no available GPRs, the scavenger gets one by using its spill
slot. Previously, we were not always allocating that spill slot and the RS
would assert when the spill slot was needed.
I don't currently have a small test that triggered the assert, but I've
created a small regression test that verifies that the spill slot is now
added when the stack frame is sufficiently large.
llvm-svn: 177140
The new InstrSchedModel is easier to use than the instruction
itineraries. It will be used to model instruction latency and throughput
in modern Intel microarchitectures like Sandy Bridge.
InstrSchedModel should be able to coexist with instruction itinerary
classes, but for cleanliness we should switch the Atom processor model
to the new InstrSchedModel as well.
llvm-svn: 177122
See the Mips16ISetLowering.cpp patch to see a use of this.
For now now the extra code in Mips16ISetLowering.cpp is a nop but is
used for test purposes. Mips32 registers are setup and then removed and
then the Mips16 registers are setup.
Normally you need to add register classes and then call
computeRegisterProperties.
llvm-svn: 177120
This is a generic function (derived from PEI); moving it into
MachineFrameInfo eliminates a current redundancy between the ARM and AArch64
backends, and will allow it to be used by the PowerPC target code.
No functionality change intended.
llvm-svn: 177111
Add the current PEI register scavenger as a parameter to the
processFunctionBeforeFrameFinalized callback.
This change is necessary in order to allow the PowerPC target code to
set the register scavenger frame index after the save-area offset
adjustments performed by processFunctionBeforeFrameFinalized. Only
after these adjustments have been made is it possible to estimate
the size of the stack frame.
llvm-svn: 177108
Make requiresFrameIndexScavenging return true, and create virtual registers in
the spilling code instead of using the register scavenger directly. This makes
the target-level code simpler, and importantly, delays the scavenging until
after callee-saved register processing (which will be important for later
changes).
Also cleans up trackLivenessAfterRegAlloc (makes it inline in the header with
the other related functions). This makes it clear that it always returns true.
No functionality change intended.
llvm-svn: 177107
We used to add a spill slot for the register scavenger whenever the function
has a frame pointer. This is unnecessarily conservative: We may need the spill
slot for dynamic stack allocations, and functions with dynamic stack
allocations always have a FP, but we might also have a FP for other reasons
(such as the user explicitly disabling frame-pointer elimination), and we don't
necessarily need a spill slot for those functions.
The structsinregs test needed adjustment because it disables FP elimination.
llvm-svn: 177106
By terrible I mean we store/load from the stack.
This matters on PAQp8 in _Z5trainPsS_ii (which is inlined into Mixer::update)
where we decide to vectorize a loop with a VF of 8 resulting in a 25%
degradation on a cortex-a8.
LV: Found an estimated cost of 2 for VF 8 For instruction: icmp slt i32
LV: Found an estimated cost of 2 for VF 8 For instruction: select i1, i32, i32
The bug that tracks the CodeGen part is PR14868.
radar://13403975
llvm-svn: 177105
I don't think that it is otherwise clear how the overlapping offsets
are processed into distinct spill slots. Comment that this is done
in processFunctionBeforeFrameFinalized.
llvm-svn: 177094
Rules include:
1)1 x*y +/- x*z => x*(y +/- z)
(the order of operands dosen't matter)
2) y/x +/- z/x => (y +/- z)/x
The transformation is disabled if the new add/sub expr "y +/- z" is a
denormal/naz/inifinity.
rdar://12911472
llvm-svn: 177088
The fundamental problem is that SROA didn't allow for overly wide loads
where the bits past the end of the alloca were masked away and the load
was sufficiently aligned to ensure there is no risk of page fault, or
other trapping behavior. With such widened loads, SROA would delete the
load entirely rather than clamping it to the size of the alloca in order
to allow mem2reg to fire. This was exposed by a test case that neatly
arranged for GVN to run first, widening certain loads, followed by an
inline step, and then SROA which miscompiles the code. However, I see no
reason why this hasn't been plaguing us in other contexts. It seems
deeply broken.
Diagnosing all of the above took all of 10 minutes of debugging. The
really annoying aspect is that fixing this completely breaks the pass.
;] There was an implicit reliance on the fact that no loads or stores
extended past the alloca once we decided to rewrite them in the final
stage of SROA. This was used to encode information about whether the
loads and stores had been split across multiple partitions of the
original alloca. That required threading explicit tracking of whether
a *use* of a partition is split across multiple partitions.
Once that was done, another problem arose: we allowed splitting of
integer loads and stores iff they were loads and stores to the entire
alloca. This is a really arbitrary limitation, and splitting at least
some integer loads and stores is crucial to maximize promotion
opportunities. My first attempt was to start removing the restriction
entirely, but currently that does Very Bad Things by causing *many*
common alloca patterns to be fully decomposed into i8 operations and
lots of or-ing together to produce larger integers on demand. The code
bloat is terrifying. That is still the right end-goal, but substantial
work must be done to either merge partitions or ensure that small i8
values are eagerly merged in some other pass. Sadly, figuring all this
out took essentially all the time and effort here.
So the end result is that we allow splitting only when the load or store
at least covers the alloca. That ensures widened loads and stores don't
hurt SROA, and that we don't rampantly decompose operations more than we
have previously.
All of this was already fairly well tested, and so I've just updated the
tests to cover the wide load behavior. I can add a test that crafts the
pass ordering magic which caused the original PR, but that seems really
brittle and to provide little benefit. The fundamental problem is that
widened loads should Just Work.
llvm-svn: 177055
constructs default arguments. It can now take default arguments from
cl::opt'ions. Add a new -default-gcov-version=... option, and actually test it!
Sink the reverse-order of the version into GCOVProfiling, hiding it from our
users.
llvm-svn: 177002
emitProfileNotes(), similar to emitProfileArcs(). Also update its comment.
Also add a comment on Version[4] (there will be another comment in clang later),
and compress lines that exceeded 80 columns.
llvm-svn: 176994
This doesn't reset all of the target options within the TargetOptions
object. This is because some of those are ABI-specific and must be determined if
it's okay to change those on the fly.
llvm-svn: 176986
This is the next step towards making the metadata for DIScopes have a common
prefix rather than having to delegate based on their tag type.
llvm-svn: 176913
This could be 'null' or the empty string, DIDescriptor::getStringField
coalesces the two cases anyway so it's just a matter of legible/efficient
representation.
The change in behavior of the DICompileUnit::get* functions could be
subsumed by the full verification check - but ideally that should just be an
assertion if we could front-load the actual debug info metadata failure paths.
llvm-svn: 176907
Increase the cost of v8/v16-i8 to v8/v16-i32 casts and truncates as the backend
currently lowers those using stack accesses.
This was responsible for a significant degradation on
MultiSource/Benchmarks/Trimaran/enc-pc1/enc-pc1
where we vectorize one loop to a vector factor of 16. After this patch we select
a vector factor of 4 which will generate reasonable code.
unsigned char cle[32];
void test(short c) {
unsigned short compte;
for (compte = 0; compte <= 31; compte++) {
cle[compte] = cle[compte] ^ c;
}
}
radar://13220512
llvm-svn: 176898
belongs to a different compile unit.
DW_FORM_ref_addr should be used for cross compile-unit reference.
When compiling a large application, we got a dwarfdump verification error where
abstract_origin points to nowhere.
This error can't be reproduced on any testing case in MultiSource.
We may have other cases where we use DW_FORM_ref4 unconditionally.
rdar://problem/13370501
llvm-svn: 176882
return 0 to indicate failure to create the disassembler. A library routine
should not assert and just let the caller handler the error. For example
darwin's otool(1) will simply print an error if it ends up using a library
that is not configured for a target it wants:
% otool -tv ViewController.o
ViewController.o:
(__TEXT,__text) section
can't create arm llvm disassembler
This is much better than an abort which appears as a crash to the user or
even the assert when using a Debug+Asserts built library:
Assertion failed: (MAI && "Unable to create target asm info!"), function LLVMCreateDisasmCPU, file /Volumes/SandBox/llvm/lib/MC/MCDisassembler/Disassembler.cpp, line 47.
radr://12539918
llvm-svn: 176880
This pass is meant to be immutable, however it holds mutable state to cache StructLayouts.
This method will allow the pass manager to clear the mutable state between runs.
Note that unfortunately it is still necessary to have the destructor, even though it does the
same thing as doFinalization. This is because most TargetMachines embed a DataLayout on which
doFinalization isn't run as its never added to the pass manager.
I also didn't think it was necessary to complication things with a deInit method for which
doFinalization and ~DataLayout both call as there's only one field of mutable state. If we had
more fields to finalize i'd have added this.
llvm-svn: 176877
Now that only the register-scavenger version of the CR spilling code remains,
we no longer need the Darwin R2 hack. Darwin can use R0 as a spare register in
any case where the System V ABI uses it (R0 is special architecturally, and so
is reserved under all common ABIs).
A few test cases needed to be updated to reflect the register-allocation changes.
llvm-svn: 176868
This removes the -disable-ppc[32|64]-regscavenger options; the code
that uses the register scavenger has been working well (and has been the default)
for some time, and we don't need options to enable the old (broken) CR spilling code.
llvm-svn: 176865
Nadav reported a performance regression due to the work I did to
merge the library call simplifier into instcombine [1]. The issue
is that a new LibCallSimplifier object is being created whenever
InstCombiner::runOnFunction is called. Every time a LibCallSimplifier
object is used to optimize a call it creates a hash table to map from
a function name to an object that optimizes functions of that name.
For short-lived LibCallSimplifier instances this is quite inefficient.
Especially for cases where no calls are actually simplified.
This patch fixes the issue by dropping the hash table and implementing
an explicit lookup function to correlate the function name to the object
that optimizes functions of that name. This avoids the cost of always
building and destroying the hash table in cases where the LibCallSimplifier
object is short-lived and avoids the cost of building the table when no
simplifications are actually preformed.
On a benchmark containing 100,000 calls where none of them are simplified
I noticed a 30% speedup. On a benchmark containing 100,000 calls where
all of them are simplified I noticed an 8% speedup.
[1] http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20130304/167639.html
llvm-svn: 176840
Versioned debug info support has been a burden to maintain & also compromised
current debug info verification by causing test cases testing old debug info to
remain rather than being updated to the latest. It also makes it hard to add or
change the metadata schema by requiring various backwards-compatibility in the
DI* hierarchy.
So it's being removed in preparation for new changes to the schema to tidy up
old/unnecessary fields and add new fields needed for new debug info (well, new
to LLVM at least).
The more surprising part of this is the changes to DI*::Verify - this became
necessary due to the changes to AsmWriter. AsmWriter was relying on the version
test to decide which bits of metadata were actually debug info when printing
the comment annotations. Without the version information the tag numbers were
too common & it would print debug info on random metadata that happened to
start with an integer that matched a tag number. Instead this change makes the
Verify functions more precise (just adding "number of operands" checks - not
type checking those operands yet) & relies on that to decide which metadata is
debug info metadata.
llvm-svn: 176838
We want vectorization to happen at -g. Ignore calls to the dbg.value intrinsic
and don't transfer them to the vectorized code.
radar://13378964
llvm-svn: 176768
The strlen+memcmp was hidden in a call to StringRef::operator==. We check if
there are any null bytes in the string upfront so we can simplify the comparison
Small speedup when compiling code with many function calls.
llvm-svn: 176766
Sylvestre Ledru