transformation if we can't legally create a build vector of the correct
type. Check that we can make the transformation first, and add a TODO to
refactor this code with similar cases.
Fixes: PR9223 and rdar://9000350
llvm-svn: 125631
have their low bits set to zero. This allows us to optimize
out explicit stack alignment code like in stack-align.ll:test4 when
it is redundant.
Doing this causes the code generator to start turning FI+cst into
FI|cst all over the place, which is general goodness (that is the
canonical form) except that various pieces of the code generator
don't handle OR aggressively. Fix this by introducing a new
SelectionDAG::isBaseWithConstantOffset predicate, and using it
in places that are looking for ADD(X,CST). The ARM backend in
particular was missing a lot of addressing mode folding opportunities
around OR.
llvm-svn: 125470
generating i8 shift amounts for things like i1024 types. Add
an assert in getNode to prevent this from occuring in the future,
fix the buggy transformation, revert my previous patch, and
document this gotcha in ISDOpcodes.h
llvm-svn: 125465
The DAGCombiner created illegal BUILD_VECTOR operations.
The patch added a check that either illegal operations are
allowed or that the created operation is legal.
llvm-svn: 125435
The bug happens when the DAGCombiner attempts to optimize one of the patterns
of the SUB opcode. It tries to create a zero of type v2i64. This type is legal
on 32bit machines, but the initializer of this vector (i64) is target dependent.
Currently, the initializer attempts to create an i64 zero constant, which fails.
Added a flag to tell the DAGCombiner to create a legal zero, if we require that
the pass would generate legal types.
llvm-svn: 125391
precisely track pressure on a selection DAG, but we can at least keep
it balanced. This design accounts for various interesting aspects of
selection DAGS: register and subregister copies, glued nodes, dead
nodes, unused registers, etc.
Added SUnit::NumRegDefsLeft and ScheduleDAGSDNodes::RegDefIter.
Note: I disabled PrescheduleNodesWithMultipleUses when register
pressure is enabled, based on no evidence other than I don't think it
makes sense to have both enabled.
llvm-svn: 124853
the load, then it may be legal to transform the load and store to integer
load and store of the same width.
This is done if the target specified the transformation as profitable. e.g.
On arm, this can transform:
vldr.32 s0, []
vstr.32 s0, []
to
ldr r12, []
str r12, []
rdar://8944252
llvm-svn: 124708
This happens all the time when a smul is promoted to a larger type.
On x86-64 we now compile "int test(int x) { return x/10; }" into
movslq %edi, %rax
imulq $1717986919, %rax, %rax
movq %rax, %rcx
shrq $63, %rcx
sarq $34, %rax <- used to be "shrq $32, %rax; sarl $2, %eax"
addl %ecx, %eax
This fires 96 times in gcc.c on x86-64.
llvm-svn: 124559
This happens e.g. for code like "X - X%10" where we lower the modulo operation
to a series of multiplies and shifts that are then subtracted from X, leading to
this missed optimization.
llvm-svn: 124532
default implementation for x86, going through the stack in a similr
fashion to how the codegen implements BUILD_VECTOR. Eventually this
will get matched to VINSERTF128 if AVX is available.
llvm-svn: 124307
implementation of EXTRACT_SUBVECTOR for x86, going through the stack
in a similr fashion to how the codegen implements BUILD_VECTOR.
Eventually this will get matched to VEXTRACTF128 if AVX is available.
llvm-svn: 124292
clang's -Wuninitialized-experimental warning.
While these don't look like real bugs, clang's
-Wuninitialized-experimental analysis is stricter
than GCC's, and these fixes have the benefit
of being general nice cleanups.
llvm-svn: 124073
DAG. Disable using "-disable-sched-cycles".
For ARM, this enables a framework for modeling the cpu pipeline and
counting stalls. It also activates several heuristics to drive
scheduling based on the model. Scheduling is inherently imprecise at
this stage, and until spilling is improved it may defeat attempts to
schedule. However, this framework provides greater control over
tuning codegen.
Although the flag is not target-specific, it should have very little
affect on the default scheduler used by x86. The only two changes that
affect x86 are:
- scheduling a high-latency operation bumps the current cycle so independent
operations can have their latency covered. i.e. two independent 4
cycle operations can produce results in 4 cycles, not 8 cycles.
- Two operations with equal register pressure impact and no
latency-based stalls on their uses will be prioritized by depth before height
(height is irrelevant if no stalls occur in the schedule below this point).
llvm-svn: 123971
flags. They are still not enable in this revision.
Added TargetInstrInfo::isZeroCost() to fix a fundamental problem with
the scheduler's model of operand latency in the selection DAG.
Generalized unit tests to work with sched-cycles.
llvm-svn: 123969
to add/sub by doing the normal operation and then checking for overflow
afterwards. This generally relies on the DAG handling the later invalid
operations as well.
Fixes the 64-bit part of rdar://8622122 and rdar://8774702.
llvm-svn: 123908
Added a check for already live regs before claiming HighRegPressure.
Fixed a few cases of checking the wrong number of successors.
Added some tracing until these heuristics are better understood.
llvm-svn: 123892
with an invalid type then split the result and perform the overflow check
normally.
Fixes the 32-bit parts of rdar://8622122 and rdar://8774702.
llvm-svn: 123864
This shaves off 4 popcounts from the hacked 186.crafty source.
This is enabled even when a native popcount instruction is available. The
combined code is one operation longer but it should be faster nevertheless.
llvm-svn: 123621
http://graphics.stanford.edu/~seander/bithacks.html#CountBitsSetParallel
In a silly microbenchmark on a 65 nm core2 this is 1.5x faster than the old
code in 32 bit mode and about 2x faster in 64 bit mode. It's also a lot shorter,
especially when counting 64 bit population on a 32 bit target.
I hope this is fast enough to replace Kernighan-style counting loops even when
the input is rather sparse.
llvm-svn: 123547
disabled in this checkin. Sorry for the large diffs due to
refactoring. New functionality is all guarded by EnableSchedCycles.
Scheduling the isel DAG is inherently imprecise, but we give it a best
effort:
- Added MayReduceRegPressure to allow stalled nodes in the queue only
if there is a regpressure need.
- Added BUHasStall to allow checking for either dependence stalls due to
latency or resource stalls due to pipeline hazards.
- Added BUCompareLatency to encapsulate and standardize the heuristics
for minimizing stall cycles (vs. reducing register pressure).
- Modified the bottom-up heuristic (now in BUCompareLatency) to
prioritize nodes by their depth rather than height. As long as it
doesn't stall, height is irrelevant. Depth represents the critical
path to the DAG root.
- Added hybrid_ls_rr_sort::isReady to filter stalled nodes before
adding them to the available queue.
Related Cleanup: most of the register reduction routines do not need
to be templates.
llvm-svn: 123468
There's an inherent tension in DAGCombine between assuming
that things will be put in canonical form, and the Depth
mechanism that disables transformations when recursion gets
too deep. It would not surprise me if there's a lot of little
bugs like this one waiting to be discovered. The mechanism
seems fragile and I'd suggest looking at it from a design viewpoint.
llvm-svn: 123191
Print virtual registers numbered from 0 instead of the arbitrary
FirstVirtualRegister. The first virtual register is printed as %vreg0.
TRI::NoRegister is printed as %noreg.
llvm-svn: 123107
Instead encode llvm IR level property "HasSideEffects" in an operand (shared
with IsAlignStack). Added MachineInstrs::hasUnmodeledSideEffects() to check
the operand when the instruction is an INLINEASM.
This allows memory instructions to be moved around INLINEASM instructions.
llvm-svn: 123044
Also fix an off-by-one in SelectionDAGBuilder that was preventing shuffle
vectors from being translated to EXTRACT_SUBVECTOR.
Patch by Tim Northover.
The test changes are needed to keep those spill-q tests from testing aligned
spills and restores. If the only aligned stack objects are spill slots, we
no longer realign the stack frame. Prior to this patch, an EXTRACT_SUBVECTOR
was legalized by loading from the stack, which created an aligned frame index.
Now, however, there is nothing except the spill slot in the stack frame, so
I added an aligned alloca.
llvm-svn: 122995
We were never generating any of these nodes with variable indices, and there
was one legalizer function asserting on a non-constant index. If we ever have
a need to support variable indices, we can add this back again.
llvm-svn: 122993
up freebsd bootloader. However, this doesn't make much sense for Darwin, whose
-Os is meant to optimize for size only if it doesn't hurt performance.
rdar://8821501
llvm-svn: 122936
This allows us to compile:
void test(char *s, int a) {
__builtin_memset(s, a, 15);
}
into 1 mul + 3 stores instead of 3 muls + 3 stores.
llvm-svn: 122710
We could implement a DAGCombine to turn x * 0x0101 back into logic operations
on targets that doesn't support the multiply or it is slow (p4) if someone cares
enough.
Example code:
void test(char *s, int a) {
__builtin_memset(s, a, 4);
}
before:
_test: ## @test
movzbl 8(%esp), %eax
movl %eax, %ecx
shll $8, %ecx
orl %eax, %ecx
movl %ecx, %eax
shll $16, %eax
orl %ecx, %eax
movl 4(%esp), %ecx
movl %eax, 4(%ecx)
movl %eax, (%ecx)
ret
after:
_test: ## @test
movzbl 8(%esp), %eax
imull $16843009, %eax, %eax ## imm = 0x1010101
movl 4(%esp), %ecx
movl %eax, 4(%ecx)
movl %eax, (%ecx)
ret
llvm-svn: 122707
DAG scheduling during isel. Most new functionality is currently
guarded by -enable-sched-cycles and -enable-sched-hazard.
Added InstrItineraryData::IssueWidth field, currently derived from
ARM itineraries, but could be initialized differently on other targets.
Added ScheduleHazardRecognizer::MaxLookAhead to indicate whether it is
active, and if so how many cycles of state it holds.
Added SchedulingPriorityQueue::HasReadyFilter to allowing gating entry
into the scheduler's available queue.
ScoreboardHazardRecognizer now accesses the ScheduleDAG in order to
get information about it's SUnits, provides RecedeCycle for bottom-up
scheduling, correctly computes scoreboard depth, tracks IssueCount, and
considers potential stall cycles when checking for hazards.
ScheduleDAGRRList now models machine cycles and hazards (under
flags). It tracks MinAvailableCycle, drives the hazard recognizer and
priority queue's ready filter, manages a new PendingQueue, properly
accounts for stall cycles, etc.
llvm-svn: 122541
In the bottom-up selection DAG scheduling, handle two-address
instructions that read/write unspillable registers. Treat
the entire chain of two-address nodes as a single live range.
llvm-svn: 122472
loads properly. We miscompiled the testcase into:
_test: ## @test
movl $128, (%rdi)
movzbl 1(%rdi), %eax
ret
Now we get a proper:
_test: ## @test
movl $128, (%rdi)
movsbl (%rdi), %eax
movzbl %ah, %eax
ret
This fixes PR8757.
llvm-svn: 122392
count operand. These should be the same but apparently are
not always, and this is cleaner anyway. This improves the
code in an existing test.
llvm-svn: 122354
ARM (and other 32-bit-only) targets support for i8 and i16 overflow
multiplies. The generated code isn't great, but this at least fixes
CodeGen/Generic/overflow.ll when running on ARM hosts.
llvm-svn: 122221
Imagine we see:
EFLAGS = inst1
EFLAGS = inst2 FLAGS
gpr = inst3 EFLAGS
Previously, we would refuse to schedule inst2 because it clobbers
the EFLAGS of the predecessor. However, it also uses the EFLAGS
of the predecessor, so it is safe to emit. SDep edges ensure that
the right order happens already anyway.
This fixes 2 testsuite crashes with the X86 patch I'm going to
commit next.
llvm-svn: 122211
BUILD_VECTOR operands where the element type is not legal. I had previously
changed this code to insert TRUNCATE operations, but that was just wrong.
llvm-svn: 122102
when the wider type is legal. This allows us to compile:
define zeroext i16 @test1(i16 zeroext %x) nounwind {
entry:
%div = udiv i16 %x, 33
ret i16 %div
}
into:
test1: # @test1
movzwl 4(%esp), %eax
imull $63551, %eax, %eax # imm = 0xF83F
shrl $21, %eax
ret
instead of:
test1: # @test1
movw $-1985, %ax # imm = 0xFFFFFFFFFFFFF83F
mulw 4(%esp)
andl $65504, %edx # imm = 0xFFE0
movl %edx, %eax
shrl $5, %eax
ret
Implementing rdar://8760399 and example #4 from:
http://blog.regehr.org/archives/320
We should implement the same thing for [su]mul_hilo, but I don't
have immediate plans to do this.
llvm-svn: 121696
zextOrTrunc(), and APSInt methods extend(), extOrTrunc() and new method
trunc(), to be const and to return a new value instead of modifying the
object in place.
llvm-svn: 121120
legalization time. Since at legalization time there is no mapping from
SDNode back to the corresponding LLVM instruction and the return
SDNode is target specific, this requires a target hook to check for
eligibility. Only x86 and ARM support this form of sibcall optimization
right now.
rdar://8707777
llvm-svn: 120501
so don't claim they are. They are allocated using DAG.getNode, so attempts
to access MemSDNode fields results in reading off the end of the allocated
memory. This fixes crashes with "llc -debug" due to debug code trying to
print MemSDNode fields for these barrier nodes (since the crashes are not
deterministic, use valgrind to see this). Add some nasty checking to try
to catch this kind of thing in the future.
llvm-svn: 119901
if the extension types were not the same. The result was that if you
fed a select with sext and zext loads, as in the testcase, then it
would get turned into a zext (or sext) of the select, which is wrong
in the cases when it should have been an sext (resp. zext). Reported
and diagnosed by Sebastien Deldon.
llvm-svn: 119728
catastrophic compilation time in the event of unreasonable LLVM
IR. Code quality is a separate issue--someone upstream needs to do a
better job of reducing to llvm.memcpy. If the situation can be reproduced with
any supported frontend, then it will be a separate bug.
llvm-svn: 118904
to perform the copy, which may be of lots of memory [*]. It would be good if the
fall-back code generated something reasonable, i.e. did the copy in a loop, rather
than vast numbers of loads and stores. Add a note about this. Currently target
specific code seems to always kick in so this is more of a theoretical issue rather
than a practical one now that X86 has been fixed.
[*] It's amazing how often people pass mega-byte long arrays by copy...
llvm-svn: 118275
value type, so there is no point in passing it around using
an EVT. Use the simpler MVT everywhere. Rather than trying
to propagate this information maximally in all the code that
using the calling convention stuff, I chose to do a mainly
low impact change instead.
llvm-svn: 118167
1. Fix pre-ra scheduler so it doesn't try to push instructions above calls to
"optimize for latency". Call instructions don't have the right latency and
this is more likely to use introduce spills.
2. Fix if-converter cost function. For ARM, it should use instruction latencies,
not # of micro-ops since multi-latency instructions is completely executed
even when the predicate is false. Also, some instruction will be "slower"
when they are predicated due to the register def becoming implicit input.
rdar://8598427
llvm-svn: 118135
operand and one of them has a single use that is a live out copy, favor the
one that is live out. Otherwise it will be difficult to eliminate the copy
if the instruction is a loop induction variable update. e.g.
BB:
sub r1, r3, #1
str r0, [r2, r3]
mov r3, r1
cmp
bne BB
=>
BB:
str r0, [r2, r3]
sub r3, r3, #1
cmp
bne BB
This fixed the recent 256.bzip2 regression.
llvm-svn: 117675
must be called in the pass's constructor. This function uses static dependency declarations to recursively initialize
the pass's dependencies.
Clients that only create passes through the createFooPass() APIs will require no changes. Clients that want to use the
CommandLine options for passes will need to manually call the appropriate initialization functions in PassInitialization.h
before parsing commandline arguments.
I have tested this with all standard configurations of clang and llvm-gcc on Darwin. It is possible that there are problems
with the static dependencies that will only be visible with non-standard options. If you encounter any crash in pass
registration/creation, please send the testcase to me directly.
llvm-svn: 116820
allow target to correctly compute latency for cases where static scheduling
itineraries isn't sufficient. e.g. variable_ops instructions such as
ARM::ldm.
This also allows target without scheduling itineraries to compute operand
latencies. e.g. X86 can return (approximated) latencies for high latency
instructions such as division.
- Compute operand latencies for those defined by load multiple instructions,
e.g. ldm and those used by store multiple instructions, e.g. stm.
llvm-svn: 115755
having to do a double cast (uint64_t --> double --> float). This is based on the algorithm from compiler_rt's __floatundisf
for X86-64.
llvm-svn: 115634
// %a = ...
// %b = and i32 %a, 2
// %c = srl i32 %b, 1
// brcond i32 %c ...
//
// into
//
// %a = ...
// %b = and i32 %a, 2
// %c = setcc eq %b, 0
// brcond %c ...
Make sure it restores local variable N1, which corresponds to the condition operand if it fails to match.
This apparently breaks TCE but since that backend isn't in the tree I don't have a test for it.
llvm-svn: 115571
The x86_mmx type is used for MMX intrinsics, parameters and
return values where these use MMX registers, and is also
supported in load, store, and bitcast.
Only the above operations generate MMX instructions, and optimizations
do not operate on or produce MMX intrinsics.
MMX-sized vectors <2 x i32> etc. are lowered to XMM or split into
smaller pieces. Optimizations may occur on these forms and the
result casted back to x86_mmx, provided the result feeds into a
previous existing x86_mmx operation.
The point of all this is prevent optimizations from introducing
MMX operations, which is unsafe due to the EMMS problem.
llvm-svn: 115243
edited during emission.
If the basic block ends in a switch that gets lowered to a jump table, any
phis at the default edge were getting updated wrong. The jump table data
structure keeps a pointer to the header blocks that wasn't getting updated
after the MBB is split.
This bug was exposed on 32-bit Linux when disabling critical edge splitting in
codegen prepare.
The fix is to uipdate stale MBB pointers whenever a block is split during
emission.
llvm-svn: 115191
when the unconditional branch destination is the fallthrough block. The
canonicalization makes it easier to allow optimizations on DAGs to invert
conditional branches. The branch folding pass (and AnalyzeBranch) will clean up
the unnecessary unconditional branches later.
This is one of the patches leading up to disabling codegen prepare critical edge
splitting.
llvm-svn: 114630
that complex patterns are matched after the entire pattern has
a structural match, therefore the NodeStack isn't in a useful
state when the actual call to the matcher happens.
llvm-svn: 114489
I think I've audited all uses, so it should be dependable for address spaces,
and the pointer+offset info should also be accurate when there.
llvm-svn: 114464
CombinerAA cannot assume that different FrameIndex's never alias, but can instead use
MachineFrameInfo to get the actual offsets of these slots and check for actual aliasing.
This fixes CodeGen/X86/2010-02-19-TailCallRetAddrBug.ll and CodeGen/X86/tailcallstack64.ll
when CombinerAA is enabled, modulo a different register allocation sequence.
llvm-svn: 114348
NO path to the destination containing side effects, not that SOME path contains no side effects.
In practice, this only manifests with CombinerAA enabled, because otherwise the chain has little
to no branching, so "any" is effectively equivalent to "all".
llvm-svn: 114268
take multiple cycles to decode.
For the current if-converter clients (actually only ARM), the instructions that
are predicated on false are not nops. They would still take machine cycles to
decode. Micro-coded instructions such as LDM / STM can potentially take multiple
cycles to decode. If-converter should take treat them as non-micro-coded
simple instructions.
llvm-svn: 113570
Since mem2reg isn't run at -O0, we get a ton of reloads from the stack,
for example, before, this code:
int foo(int x, int y, int z) {
return x+y+z;
}
used to compile into:
_foo: ## @foo
subq $12, %rsp
movl %edi, 8(%rsp)
movl %esi, 4(%rsp)
movl %edx, (%rsp)
movl 8(%rsp), %edx
movl 4(%rsp), %esi
addl %edx, %esi
movl (%rsp), %edx
addl %esi, %edx
movl %edx, %eax
addq $12, %rsp
ret
Now we produce:
_foo: ## @foo
subq $12, %rsp
movl %edi, 8(%rsp)
movl %esi, 4(%rsp)
movl %edx, (%rsp)
movl 8(%rsp), %edx
addl 4(%rsp), %edx ## Folded load
addl (%rsp), %edx ## Folded load
movl %edx, %eax
addq $12, %rsp
ret
Fewer instructions and less register use = faster compiles.
llvm-svn: 113102
solve the root problem, but it corrects the bug in the code I added to
support legalizing in the case where the non-extended type is also legal.
llvm-svn: 112997
there are clearly no stores between the load and the store. This fixes
this miscompile reported as PR7833.
This breaks the test/CodeGen/X86/narrow_op-2.ll optimization, which is
safe, but awkward to prove safe. Move it to X86's README.txt.
llvm-svn: 112861
expanding: e.g. <2 x float> -> <4 x float> instead of -> 2 floats. This
affects two places in the code: handling cross block values and handling
function return and arguments. Since vectors are already widened by
legalizetypes, this gives us much better code and unblocks x86-64 abi
and SPU abi work.
For example, this (which is a silly example of a cross-block value):
define <4 x float> @test2(<4 x float> %A) nounwind {
%B = shufflevector <4 x float> %A, <4 x float> undef, <2 x i32> <i32 0, i32 1>
%C = fadd <2 x float> %B, %B
br label %BB
BB:
%D = fadd <2 x float> %C, %C
%E = shufflevector <2 x float> %D, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
ret <4 x float> %E
}
Now compiles into:
_test2: ## @test2
## BB#0:
addps %xmm0, %xmm0
addps %xmm0, %xmm0
ret
previously it compiled into:
_test2: ## @test2
## BB#0:
addps %xmm0, %xmm0
pshufd $1, %xmm0, %xmm1
## kill: XMM0<def> XMM0<kill> XMM0<def>
insertps $0, %xmm0, %xmm0
insertps $16, %xmm1, %xmm0
addps %xmm0, %xmm0
ret
This implements rdar://8230384
llvm-svn: 112101
hierarchy with virtual methods and using llvm_unreachable to properly indicate
unreachable states which would otherwise leave variables uninitialized.
llvm-svn: 111803
it involves specific floating-point types, legalize should expand an
extending load to a non-extending load followed by a separate extend operation.
For example, we currently expand SEXTLOAD to EXTLOAD+SIGN_EXTEND_INREG (and
assert that EXTLOAD should always be supported). Now we can expand that to
LOAD+SIGN_EXTEND. This is needed to allow vector SIGN_EXTEND and ZERO_EXTEND
to be used for NEON.
llvm-svn: 111586
protectors, to be near the stack protectors on the stack. Accomplish this by
tagging the stack object with a predicate that indicates that it would trigger
this. In the prolog-epilog inserter, assign these objects to the stack after the
stack protector but before the other objects.
llvm-svn: 109481
appropriate for targets without detailed instruction iterineries.
The scheduler schedules for increased instruction level parallelism in
low register pressure situation; it schedules to reduce register pressure
when the register pressure becomes high.
On x86_64, this is a win for all tests in CFP2000. It also sped up 256.bzip2
by 16%.
llvm-svn: 109300
it's too late to start backing off aggressive latency scheduling when most
of the registers are in use so the threshold should be a bit tighter.
- Correctly handle live out's and extract_subreg etc.
- Enable register pressure aware scheduling by default for hybrid scheduler.
For ARM, this is almost always a win on # of instructions. It's runtime
neutral for most of the tests. But for some kernels with high register
pressure it can be a huge win. e.g. 464.h264ref reduced number of spills by
54 and sped up by 20%.
llvm-svn: 109279
Stuart Hastings