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Internally TargetOptions::AllowExcessFPPrecision has been replaced by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
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
This cleans up the method used to find trip counts in order to form CTR loops on PPC.
This refactoring allows the pass to find loops which have a constant trip count but also
happen to end with a comparison to zero. This also adds explicit FIXMEs to mark two different
classes of loops that are currently ignored.
In addition, we now search through all potential induction operations instead of just the first.
Also, we check the predicate code on the conditional branch and abort the transformation if the
code is not EQ or NE, and we then make sure that the branch to be transformed matches the
condition register defined by the comparison (multiple possible comparisons will be considered).
llvm-svn: 158607
Over the entire test-suite, this has an insignificantly negative average
performance impact, but reduces some of the worst slowdowns from the
anti-dep. change (r158294).
Largest speedups:
SingleSource/Benchmarks/Stanford/Quicksort - 28%
SingleSource/Benchmarks/Stanford/Towers - 24%
SingleSource/Benchmarks/Shootout-C++/matrix - 23%
MultiSource/Benchmarks/SciMark2-C/scimark2 - 19%
MultiSource/Benchmarks/MiBench/automotive-bitcount/automotive-bitcount - 15%
(matrix and automotive-bitcount were both in the top-5 slowdown list from the
anti-dep. change)
Largest slowdowns:
MultiSource/Benchmarks/McCat/03-testtrie/testtrie - 28%
MultiSource/Benchmarks/mediabench/gsm/toast/toast - 26%
MultiSource/Benchmarks/MiBench/automotive-susan/automotive-susan - 21%
SingleSource/Benchmarks/CoyoteBench/lpbench - 20%
MultiSource/Applications/d/make_dparser - 16%
llvm-svn: 158296
The PPC64 backend had patterns for i32 <-> i64 extensions and truncations that
would leave self-moves in the final assembly. Replacing those patterns with ones
based on the SUBREG builtins yields better-looking code.
Thanks to Jakob and Owen for their suggestions in this matter.
llvm-svn: 158283
Tail merging had been disabled on PPC because it would disturb bundling decisions
made during pre-RA scheduling on the 970 cores. Now, however, all bundling decisions
are made during post-RA scheduling, and tail merging is generally beneficial (the
average test-suite speedup is insignificantly positive).
Largest test-suite speedups:
MultiSource/Benchmarks/mediabench/gsm/toast/toast - 30%
MultiSource/Benchmarks/BitBench/uuencode/uuencode - 23%
SingleSource/Benchmarks/Shootout-C++/ary - 21%
SingleSource/Benchmarks/Stanford/Queens - 17%
Largest slowdowns:
MultiSource/Benchmarks/MiBench/security-sha/security-sha - 24%
MultiSource/Benchmarks/McCat/03-testtrie/testtrie - 22%
MultiSource/Applications/JM/ldecod/ldecod - 14%
MultiSource/Benchmarks/mediabench/g721/g721encode/encode - 9%
This is improved by using full (instead of just critical) anti-dependency breaking,
but doing so still causes miscompiles and so cannot yet be enabled by default.
llvm-svn: 158259
The fast register allocator is not supposed to work in the optimizing
pipeline. It doesn't make sense to compute live intervals, run full copy
coalescing, and then run RAFast.
Fast register allocation in the optimizing pipeline is better done by
RABasic.
llvm-svn: 158242
Thanks to Jakob's help, this now causes no new test suite failures!
Over the entire test suite, this gives an average 1% speedup. The largest speedups are:
SingleSource/Benchmarks/Misc/pi - 108%
SingleSource/Benchmarks/CoyoteBench/lpbench - 54%
MultiSource/Benchmarks/Prolangs-C/unix-smail/unix-smail - 50%
SingleSource/Benchmarks/Shootout/ary3 - 32%
SingleSource/Benchmarks/Shootout-C++/matrix - 30%
The largest slowdowns are:
MultiSource/Benchmarks/mediabench/gsm/toast/toast - -30%
MultiSource/Benchmarks/Prolangs-C/bison/mybison - -25%
MultiSource/Benchmarks/BitBench/uuencode/uuencode - -22%
MultiSource/Applications/d/make_dparser - -14%
SingleSource/Benchmarks/Shootout-C++/ary - -13%
In light of these slowdowns, additional profiling work is obviously needed!
llvm-svn: 158223
The pass itself works well, but the something in the Machine* infrastructure
does not understand terminators which define registers. Without the ability
to use the block-placement pass, etc. this causes performance regressions (and
so is turned off by default). Turning off the analysis turns off the problems
with the Machine* infrastructure.
llvm-svn: 158206
The code which tests for an induction operation cannot assume that any
ADDI instruction will have a register operand because the operand could
also be a frame index; for example:
%vreg16<def> = ADDI8 <fi#0>, 0; G8RC:%vreg16
llvm-svn: 158205
This pass is derived from the Hexagon HardwareLoops pass. The only significant enhancement over the Hexagon
pass is that PPCCTRLoops will also attempt to delete the replaced add and compare operations if they are
no longer otherwise used. Also, invalid preheader DebugLoc is not used.
llvm-svn: 158204
It seems that this no longer causes test suite failures on PPC64 (after r157159),
and often gives a performance benefit, so it can be enabled by default.
llvm-svn: 157911
This option has been disabled for a while, and it is going away so I can
clean up the coalescer code.
The tests that required physreg joining to be enabled were almost all of
the form "tiny function with interference between arguments and return
value". Such functions are usually inlined in the real world.
The problem exposed by phys_subreg_coalesce-3.ll is real, but fairly
rare.
llvm-svn: 157027
This adds a full itinerary for IBM's PPC64 A2 embedded core. These
cores form the basis for the CPUs in the new IBM BG/Q supercomputer.
llvm-svn: 153842
* Removed test/lib/llvm.exp - it is no longer needed
* Deleted the dg.exp reading code from test/lit.cfg. There are no dg.exp files
left in the test suite so this code is no longer required. test/lit.cfg is
now much shorter and clearer
* Removed a lot of duplicate code in lit.local.cfg files that need access to
the root configuration, by adding a "root" attribute to the TestingConfig
object. This attribute is dynamically computed to provide the same
information as was previously provided by the custom getRoot functions.
* Documented the config.root attribute in docs/CommandGuide/lit.pod
llvm-svn: 153408
The PPC64 SVR4 ABI requires integer stack arguments, and thus the var. args., that
are smaller than 64 bits be zero extended to 64 bits.
llvm-svn: 153373
Reverting this because it breaks static linking on ppc64. Specifically, it may be linkonce_odr functions that are the problem.
With this patch, if you link statically, calls to some functions end up calling their descriptor addresses instead
of calling to their entry points. This causes the execution to fail with SIGILL (b/c the descriptor address just
has some pointers, not code).
llvm-svn: 151433
The standard function epilog includes a .size directive, but ppc64 uses
an alternate local symbol to tag the actual start of each function.
Until recently, binutils accepted the .size directive as:
.size test1, .Ltmp0-test1
however, using this directive with recent binutils will result in the error:
.size expression for XXX does not evaluate to a constant
so we must use the label which actually tags the start of the function.
llvm-svn: 151200
This test case was way too strict, matching the entire assembly output.
Every non-trivial change to the ppc backend or -O0 pipeline required
the test to be updated.
It should be replaced with a test of the specific vaarg feature.
llvm-svn: 151105
1. The ST*UX instructions that store and update the stack pointer did not set define/kill on R1. This became a problem when I activated post-RA scheduling (and had incorrectly adjusted the Frames-large test).
2. eliminateFrameIndex did not kill its scavenged temporary register, and this could cause the scavenger to exhaust all available registers (and its emergency spill slot) when there were a lot of CR values to spill. The 2010-02-12-saveCR test has been adjusted to check for this.
llvm-svn: 147359
I followed three heuristics for deciding whether to set 'true' or
'false':
- Everything target independent got 'true' as that is the expected
common output of the GCC builtins.
- If the target arch only has one way of implementing this operation,
set the flag in the way that exercises the most of codegen. For most
architectures this is also the likely path from a GCC builtin, with
'true' being set. It will (eventually) require lowering away that
difference, and then lowering to the architecture's operation.
- Otherwise, set the flag differently dependending on which target
operation should be tested.
Let me know if anyone has any issue with this pattern or would like
specific tests of another form. This should allow the x86 codegen to
just iteratively improve as I teach the backend how to differentiate
between the two forms, and everything else should remain exactly the
same.
llvm-svn: 146370
I did not convert Atomics-32.ll and Atomics-64.ll by hand; the diff is autoupgrade output.
The wmb test is gone because there isn't any way to express wmb with the new atomic instructions; if someone really needs a non-asm way to write a wmb on Alpha, a platform-specific intrisic could be added.
llvm-svn: 140566
init.trampoline and adjust.trampoline intrinsics, into two intrinsics
like in GCC. While having one combined intrinsic is tempting, it is
not natural because typically the trampoline initialization needs to
be done in one function, and the result of adjust trampoline is needed
in a different (nested) function. To get around this llvm-gcc hacks the
nested function lowering code to insert an additional parent variable
holding the adjust.trampoline result that can be accessed from the child
function. Dragonegg doesn't have the luxury of tweaking GCC code, so it
stored the result of adjust.trampoline in the memory GCC set aside for
the trampoline itself (this is always available in the child function),
and set up some new memory (using an alloca) to hold the trampoline.
Unfortunately this breaks Go which allocates trampoline memory on the
heap and wants to use it even after the parent has exited (!). Rather
than doing even more hacks to get Go working, it seemed best to just use
two intrinsics like in GCC. Patch mostly by Sanjoy Das.
llvm-svn: 139140
for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the
target indep prefetch change.
As usual, updating the testsuite is a PITA.
llvm-svn: 133337
- Check for MTCTR8 in addition to MTCTR when looking up a hazard.
- When lowering an indirect call use CTR8 when targeting 64bit.
- Introduce BCTR8 that uses CTR8 and use it on 64bit when expanding ISD::BRIND.
The last change fixes PR8487. With those changes, we are able to compile a
running "ls" and "sh" on FreeBSD/PowerPC64.
llvm-svn: 132552
Currently the output should be almost identical to the one produced by CodeGen
to make the transition easier.
The only two differences I know of are:
* Some files get an extra advance loc of size 0. This will be fixed when
relaxations are enabled.
* The optimization of declaring an EH symbol as an external variable is not
implemented. This is a subset of adding the nounwind attribute, so we if really
this at -O0 we should probably do it at the IL level.
llvm-svn: 130623
The code inserted by PPCTargetLowering::EmitInstrWithCustomInserter for ppc64 is
wrong, and I don't know how to fix it. It seems to be using the correct register
classes for pointers, but it inserts all 32-bit instructions.
llvm-svn: 128835
It turns out that ppc backend has really weird interdependencies
over different hooks and all stuff is fragile wrt small changes.
This should fix PR8749
llvm-svn: 122155
alignment for PPC32/64, avoiding some masking operations.
llvm-gcc expands vaarg inline instead of using the instruction
so it has never hit this.
llvm-svn: 116168
void foo() { __builtin_unreachable(); }
It will output the following on Darwin X86:
_func1:
Leh_func_begin0:
pushq %rbp
Ltmp0:
movq %rsp, %rbp
Ltmp1:
Leh_func_end0:
This prolog adds a new Call Frame Information (CFI) row to the FDE with an
address that is not within the address range of the code it describes -- part is
equal to the end of the function -- and therefore results in an invalid EH
frame. If we emit a nop in this situation, then the CFI row is now within the
address range.
llvm-svn: 108568
the function. We'll just turn it into a "trap" instruction instead.
The problem with not handling this is that it might generate a prologue without
the equivalent epilogue to go with it:
$ cat t.ll
define void @foo() {
entry:
unreachable
}
$ llc -o - t.ll -relocation-model=pic -disable-fp-elim -unwind-tables
.section __TEXT,__text,regular,pure_instructions
.globl _foo
.align 4, 0x90
_foo: ## @foo
Leh_func_begin0:
## BB#0: ## %entry
pushq %rbp
Ltmp0:
movq %rsp, %rbp
Ltmp1:
Leh_func_end0:
...
The unwind tables then have bad data in them causing all sorts of problems.
Fixes <rdar://problem/8096481>.
llvm-svn: 108473
Objective-C metadata types which should be marked as "weak", but which the
linker will remove upon final linkage. However, this linkage isn't specific to
Objective-C.
For example, the "objc_msgSend_fixup_alloc" symbol is defined like this:
.globl l_objc_msgSend_fixup_alloc
.weak_definition l_objc_msgSend_fixup_alloc
.section __DATA, __objc_msgrefs, coalesced
.align 3
l_objc_msgSend_fixup_alloc:
.quad _objc_msgSend_fixup
.quad L_OBJC_METH_VAR_NAME_1
This is different from the "linker_private" linkage type, because it can't have
the metadata defined with ".weak_definition".
Currently only supported on Darwin platforms.
llvm-svn: 107433
when the condition is constant. This optimization shouldn't be
necessary, because codegen shouldn't be able to find dead control
paths that the IR-level optimizer can't find. And it's undesirable,
because it encourages bugpoint to leave "br i1 false" branches
in its output. And it wasn't updating the CFG.
I updated all the tests I could, but some tests are too reduced
and I wasn't able to meaningfully preserve them.
llvm-svn: 106748
otherwise labels get incorrectly merged. We handled this by emitting a
".byte 0", but this isn't correct on thumb/arm targets where the text segment
needs to be a multiple of 2/4 bytes. Handle this by emitting a noop. This
is more gross than it should be because arm/ppc are not fully mc'ized yet.
This fixes rdar://7908505
llvm-svn: 102400
in other types. fix this by only bumping zero-byte globals
up to a single byte if the *entire global* is zero size,
fixing PR6340.
This also fixes empty arrays etc to be handled correctly,
and only does this on subsection-via-symbols targets (aka
darwin) which is the only place where this matters.
llvm-svn: 101879
Make it so. (This patch is in LowerCall_Darwin, which seems
to be used by SVR4 code as well; since that doesn't belong here,
I haven't worried about this case.)
llvm-svn: 98077
The PowerPC floating point registers can represent both f32 and f64 via the
two register classes F4RC and F8RC. F8RC is considered a subclass of F4RC to
allow cross-class coalescing. This coalescing only affects whether registers
are spilled as f32 or f64.
Spill slots must be accessed with load/store instructions corresponding to the
class of the spilled register. PPCInstrInfo::foldMemoryOperandImpl was looking
at the instruction opcode which is wrong.
X86 has similar floating point register classes, but doesn't try to fold
memory operands, so there is no problem there.
llvm-svn: 97262
to adding them in a determinstic order (bottom up from
the root) based on the structure of the graph itself.
This updates tests for some random changes, interesting
bits: CodeGen/Blackfin/promote-logic.ll no longer crashes.
I have no idea why, but that's good right?
CodeGen/X86/2009-07-16-LoadFoldingBug.ll also fails, but
now compiles to have one fewer constant pool entry, making
the expected load that was being folded disappear. Since it
is an unreduced mass of gnast, I just removed it.
This fixes PR6370
llvm-svn: 97023
induction variable value and a loop-variant value, don't force the
insert position to be at the post-increment position, because it may
not be dominated by the loop-variant value. This fixes a
use-before-def problem noticed on PPC.
llvm-svn: 96774
stack frame, the prolog/epilog code was using the same
register for the copy of CR and the address of the save slot. Oops.
This is fixed here for Darwin, sort of, by reserving R2 for this case.
A better way would be to do the store before the decrement of SP,
which is safe on Darwin due to the red zone.
SVR4 probably has the same problem, but I don't know how to fix it;
there is no red zone and R2 is already used for something else.
I'm going to leave it to someone interested in that target.
Better still would be to rewrite the CR-saving code completely;
spilling each CR subregister individually is horrible code.
llvm-svn: 96015
following it. However, the EmitGlobalConstant method wasn't emitting a body for
the constant. The assembler doesn't like that. Before, we were generating this:
.zerofill __DATA, __common, __cmd, 1, 3
This fix puts us back to that semantic.
llvm-svn: 95336
doing global variable classification anymore) and hookized, sink almost
all target targets global variable emission code into AsmPrinter and out
of each target.
Some notes:
1. PIC16 does completely custom and crazy stuff, so it is not changed.
2. XCore has some custom handling for extra directives. I'll look at it next.
3. This switches linux/ppc to use .globl instead of .global. If .globl is
actually wrong, let me know and I'll fix it.
4. This makes linux/ppc get a lot of random cases right which were obviously
wrong before, it is probably now a bit healthier.
5. Blackfin will probably start getting .comm and other things that it didn't
before. If this is undesirable, it should explicitly opt out of these
things by clearing the relevant fields of MCAsmInfo.
This leads to a nice diffstat:
14 files changed, 127 insertions(+), 830 deletions(-)
llvm-svn: 93858
different BlockAddress labels, but nothing semantically important.
Add a FIXME that BlockAddress codegen is broken if the LLVM BB has
an empty name (e.g. strip was run).
llvm-svn: 93303
in local register allocator. If a reg-reg copy has a phys reg
input and a virt reg output, and this is the last use of the phys
reg, assign the phys reg to the virt reg. If a reg-reg copy has
a phys reg output and we need to reload its spilled input, reload
it directly into the phys reg than passing it through another reg.
Following 76208, there is sometimes no dependency between the def of
a phys reg and its use; this creates a window where that phys reg
can be used for spilling (this is true in linear scan also). This
is bad and needs to be fixed a better way, although 76208 works too
well in practice to be reverted. However, there should normally be
no spilling within inline asm blocks. The patch here goes a long way
towards making this actually be true.
llvm-svn: 91485
This violates the ABI (that area is "reserved"), and
while it is safe if all code is generated with current
compilers, there is some very old code around that uses
that slot for something else, and breaks if it is stored
into. Adjust testcases looking for current behavior.
I've verified that the stack frame size is right in all
testcases, whether it changed or not. 7311323.
llvm-svn: 89811
constant whose component type is not a legal type for the target.
(If the target ConstantPool cannot handle this type either, it has
an opportunity to merge elements. In practice any target with
8-bit bytes must support i8 *as data*). 7320806 (partial).
llvm-svn: 86751
generates a sequence similar to this:
__Z4funci:
LFB2:
mflr r0
LCFI0:
stmw r30,-8(r1)
LCFI1:
stw r0,8(r1)
LCFI2:
stwu r1,-80(r1)
LCFI3:
mr r30,r1
LCFI4:
where LCFI3 and LCFI4 are used by the FDE to indicate what the FP, LR, and other
things are. We generated something more like this:
Leh_func_begin1:
mflr r0
stw r31, 20(r1)
stw r0, 8(r1)
Llabel1:
stwu r1, -80(r1)
Llabel2:
mr r31, r1
Note that we are missing the "mr" instruction. This patch makes it more like the
GCC output.
llvm-svn: 86729
input filename so that opt doesn't print the input filename in the
output so that grep lines in the tests don't unintentionally match
strings in the input filename.
llvm-svn: 81537
for a single "m" constraint; this is wrong because the
opcode of a load or store would have to change in parallel.
This patch makes it always compute addresses into a register,
which is correct but not as efficient as possible. 7144566.
llvm-svn: 79292
x86_64-apple-darwin10.
--- Reverse-merging r78895 into '.':
U test/CodeGen/PowerPC/2008-12-12-EH.ll
U lib/Target/DarwinTargetAsmInfo.cpp
--- Reverse-merging r78892 into '.':
U include/llvm/Target/DarwinTargetAsmInfo.h
U lib/Target/X86/X86TargetAsmInfo.cpp
U lib/Target/X86/X86TargetAsmInfo.h
U lib/Target/ARM/ARMTargetAsmInfo.h
U lib/Target/ARM/ARMTargetMachine.cpp
U lib/Target/ARM/ARMTargetAsmInfo.cpp
U lib/Target/PowerPC/PPCTargetAsmInfo.cpp
U lib/Target/PowerPC/PPCTargetAsmInfo.h
U lib/Target/PowerPC/PPCTargetMachine.cpp
G lib/Target/DarwinTargetAsmInfo.cpp
llvm-svn: 78919
instead of syntactically as a string. This means that it keeps track of the
segment, section, flags, etc directly and asmprints them in the right format.
This also includes parsing and validation support for llvm-mc and
"attribute(section)", so we should now start getting errors about invalid
section attributes from the compiler instead of the assembler on darwin.
Still todo:
1) Uniquing of darwin mcsections
2) Move all the Darwin stuff out to MCSectionMachO.[cpp|h]
3) there are a few FIXMEs, for example what is the syntax to get the
S_GB_ZEROFILL segment type?
llvm-svn: 78547
The inline asm operands must be parsed from the first flag, you cannot assume
that an immediate operand preceeding a register use operand is the flag.
PowerPC "m" operands are represented as (flag, imm, reg) triples.
isRegTiedToDefOperand() would incorrectly interpret the imm as the flag.
llvm-svn: 76101
additional bug fixes:
1. The bug that everyone hit was a problem in the asmprinter where it
would remove $stub but keep the L prefix on a name when emitting the
indirect symbol. This is easy to fix by keeping the name of the stub
and the name of the symbol in a StringMap instead of just keeping a
StringSet and trying to reconstruct it late.
2. There was a problem printing the personality function. The current
logic to print out the personality function from the DWARF information
is a bit of a cesspool right now that duplicates a bunch of other
logic in the asm printer. The short version of it is that it depends
on emitting both the L and _ prefix for symbols (at least on darwin)
and until I can untangle it, it is best to switch the mangler back to
emitting both prefixes.
llvm-svn: 75646
indicates whether the label is private or not, instead of taking
prefix stuff. One effect of this is that symbols will be generated
with *just* the private prefix, instead of both the private prefix
*and* the user-label-prefix, but this doesn't matter as long as it
is consistent. For example we'll now get "Lfoo" instead of "L_foo".
These are just assembler temporary labels anyway, so they never even
make it into the .o file.
llvm-svn: 75607
support for x86, and UMULO/SMULO for many architectures, including PPC
(PR4201), ARM, and Cell. The resulting expansion isn't perfect, but it's
not bad.
llvm-svn: 73477
integer and floating-point opcodes, introducing
FAdd, FSub, and FMul.
For now, the AsmParser, BitcodeReader, and IRBuilder all preserve
backwards compatability, and the Core LLVM APIs preserve backwards
compatibility for IR producers. Most front-ends won't need to change
immediately.
This implements the first step of the plan outlined here:
http://nondot.org/sabre/LLVMNotes/IntegerOverflow.txt
llvm-svn: 72897
When a test fails with more than a pipeful of output on stdout AND stderr, one
of the DejaGnu programs blocks. The problem can be avoided by redirecting
stdout to a file.
llvm-svn: 71919
Massive check in. This changes the "-fast" flag to "-O#" in llc. If you want to
use the old behavior, the flag is -O0. This change allows for finer-grained
control over which optimizations are run at different -O levels.
Most of this work was pretty mechanical. The majority of the fixes came from
verifying that a "fast" variable wasn't used anymore. The JIT still uses a
"Fast" flag. I'll change the JIT with a follow-up patch.
llvm-svn: 70343
use the old behavior, the flag is -O0. This change allows for finer-grained
control over which optimizations are run at different -O levels.
Most of this work was pretty mechanical. The majority of the fixes came from
verifying that a "fast" variable wasn't used anymore. The JIT still uses a
"Fast" flag. I'm not 100% sure if it's necessary to change it there...
llvm-svn: 70270
type as the vector element type: allow them to be of
a wider integer type than the element type all the way
through the system, and not just as far as LegalizeDAG.
This should be safe because it used to be this way
(the old type legalizer would produce such nodes), so
backends should be able to handle it. In fact only
targets which have legal vector types with an illegal
promoted element type will ever see this (eg: <4 x i16>
on ppc). This fixes a regression with the new type
legalizer (vec_splat.ll). Also, treat SCALAR_TO_VECTOR
the same as BUILD_VECTOR. After all, it is just a
special case of BUILD_VECTOR.
llvm-svn: 69467
172 %ECX<def> = MOV32rr %reg1039<kill>
180 INLINEASM <es:subl $5,$1
sbbl $3,$0>, 10, %EAX<def>, 14, %ECX<earlyclobber,def>, 9, %EAX<kill>,
36, <fi#0>, 1, %reg0, 0, 9, %ECX<kill>, 36, <fi#1>, 1, %reg0, 0
188 %EAX<def> = MOV32rr %EAX<kill>
196 %ECX<def> = MOV32rr %ECX<kill>
204 %ECX<def> = MOV32rr %ECX<kill>
212 %EAX<def> = MOV32rr %EAX<kill>
220 %EAX<def> = MOV32rr %EAX
228 %reg1039<def> = MOV32rr %ECX<kill>
The early clobber operand ties ECX input to the ECX def.
The live interval of ECX is represented as this:
%reg20,inf = [46,47:1)[174,230:0) 0@174-(230) 1@46-(47)
The right way to represent this is something like
%reg20,inf = [46,47:2)[174,182:1)[181:230:0) 0@174-(182) 1@181-230 @2@46-(47)
Of course that won't work since that means overlapping live ranges defined by two val#.
The workaround for now is to add a bit to val# which says the val# is redefined by a early clobber def somewhere. This prevents the move at 228 from being optimized away by SimpleRegisterCoalescing::AdjustCopiesBackFrom.
llvm-svn: 61259
The EH_frame and .eh symbols are now private, except for darwin9 and earlier.
The patch also fixes the definition of PrivateGlobalPrefix on pcc linux.
llvm-svn: 61242
1. ppcf128 select is expanded to f64 select's.
2. f64 select operand 0 is an i1 truncate, it's promoted to i32 zero_extend.
3. f64 select is updated. It's changed back to a "NewNode" and being re-analyzed.
4. f64 select operands are being processed. Operand 0 is a "NewNode". It's being expunged out of ReplacedValues map.
5. ExpungeNode tries to remap f64 select and notice it's a "NewNode" and assert.
Duncan, please take a look. Thanks.
llvm-svn: 60443
is noticeably worse than previous PPC-specific code.
Since the latter was also wrong in some cases and
correctness is more important than efficiency, I'm
disabling this test temporarily while I fix it.
llvm-svn: 58876
ppcf128 to i32 conversion and expand it into a code
sequence like in LegalizeDAG. This needs custom
ppc lowering of FP_ROUND_INREG, so turn that on and
make it work with LegalizeTypes. Probably PPC should
simply custom lower the original conversion.
llvm-svn: 58329
the previous patch this one actually passes make check.
"Fix PR2356 on PowerPC: if we have an input and output that are tied together
that have different sizes (e.g. i32 and i64) make sure to reserve registers for
the bigger operand."
llvm-svn: 57771
vr2 = OR vr0, vr1
=>
vr2 = OR vr1, vr1 // after coalescing vr0 with vr1
Update the value# of the destination register with the copy instruction if that happens.
llvm-svn: 56165
replacement of multiple values. This is slightly more efficient
than doing multiple ReplaceAllUsesOfValueWith calls, and theoretically
could be optimized even further. However, an important property of this
new function is that it handles the case where the source value set and
destination value set overlap. This makes it feasible for isel to use
SelectNodeTo in many very common cases, which is advantageous because
SelectNodeTo avoids a temporary node and it doesn't require CSEMap
updates for users of values that don't change position.
Revamp MorphNodeTo, which is what does all the work of SelectNodeTo, to
handle operand lists more efficiently, and to correctly handle a number
of corner cases to which its new wider use exposes it.
This commit also includes a change to the encoding of post-isel opcodes
in SDNodes; now instead of being sandwiched between the target-independent
pre-isel opcodes and the target-dependent pre-isel opcodes, post-isel
opcodes are now represented as negative values. This makes it possible
to test if an opcode is pre-isel or post-isel without having to know
the size of the current target's post-isel instruction set.
These changes speed up llc overall by 3% and reduce memory usage by 10%
on the InstructionCombining.cpp testcase with -fast and -regalloc=local.
llvm-svn: 53728
simply does the atomic.cmp.swap on the larger type,
which means it blows away whatever is sitting in
the bytes just after the memory location, i.e.
causes a buffer overflow. This really requires
target specific code, which is why LegalizeTypes
doesn't try to handle this case generically. The
existing (wrong) code in LegalizeDAG will go away
automatically once the type legalization code is
removed from LegalizeDAG so I'm leaving it there
for the moment. Meanwhile, don't test for this
feature.
llvm-svn: 53669
In LegalizeDAG the value is zero-extended to
the new type before byte swapping. It doesn't
matter how the extension is done since the new
bits are shifted off anyway after the swap, so
extend by any old rubbish bits. This results
in the final assembler for the testcase being
one line shorter.
llvm-svn: 53604
Lang Hames