ISD::VSELECT mask uses 1 to identify the first argument and 0 to identify the
second argument.
On the other hand, BLENDI uses 0 to identify the first argument and 1 to
identify the second argument.
Fix the generation of the blend mask to account for this difference.
The bug did not show up with r209043, because we were not checking for the
actual arguments of the blend instruction!
This commit also fixes the test cases.
Note: The same mask works for the BLENDr variant because the arguments are
swapped during instruction selection (see the BLENDXXrr patterns).
<rdar://problem/16975435>
llvm-svn: 209324
According to Intel Software Optimization Manual on Silvermont in some cases LEA
is better to be replaced with ADD instructions:
"The rule of thumb for ADDs and LEAs is that it is justified to use LEA
with a valid index and/or displacement for non-destructive destination purposes
(especially useful for stack offset cases), or to use a SCALE.
Otherwise, ADD(s) are preferable."
Differential Revision: http://reviews.llvm.org/D3826
llvm-svn: 209198
Currently the X86 backend doesn't support types larger than i128 very well. For
example an i192 multiply will assert in codegen when the 2nd argument is a constant and the constant got hoisted.
This fix changes the cost model to never hoist constants for types larger than
i128. Once the codegen issues have been resolved, the cost model can be updated
to allow also larger types.
This is related to <rdar://problem/16954938>
llvm-svn: 209162
Instructions TZCNT (requires BMI1) and LZCNT (requires LZCNT), always
provide the operand size as output if the input operand is zero.
We can take advantage of this knowledge during instruction selection
stage in order to simplify a few corner case.
llvm-svn: 209159
Summary:
When inserting an element that's coming from a vector load or a broadcast
of a vector (or scalar) load, combine the load into the insertps
instruction.
Added PerformINSERTPSCombine for the case where we need to fix the load
(load of a vector + insertps with a non-zero CountS).
Added patterns for the broadcasts.
Also added tests for SSE4.1, AVX, and AVX2.
Reviewers: delena, nadav, craig.topper
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3581
llvm-svn: 209156
- On ARM/ARM64 we get a vrev because the shuffle matching code is really smart. We still unroll anything that's not v4i32 though.
- On X86 we get a pshufb with SSSE3. Required more cleverness in isShuffleMaskLegal.
- On PPC we get a vperm for v8i16 and v4i32. v2i64 is unrolled.
llvm-svn: 209123
This is mostly a mechanical change changing all the call sites to the newer
chained-function construction pattern. This removes the horrible 15-parameter
constructor for the CallLoweringInfo in favour of setting properties of the call
via chained functions. No functional change beyond the removal of the old
constructors are intended.
llvm-svn: 209082
were added in SSE2, no SSSE3. Found this while auditing all uses of
SSSE3 in the X86 target. I don't actually expect this to make
a significant difference on anything and I don't have any detailed test
cases but I updated the existing test cases that already covered some of
this code path.
llvm-svn: 209056
vselects with constant masks, after legalization, will get turned into
specialized shuffle_vectors so they can be matched to blend+imm
instructions.
Fixed some tests.
llvm-svn: 209044
LowerVSELECT will, if possible, generate a X86ISD::BLENDI DAG node if the
condition is constant and we can emit that instruction, given the
subtarget.
This is not enough for all cases. An additional SELECTCombine optimization
will be committed.
Fixed tests that were expecting variable blends but where a blend+imm can
be generated.
Added test where we can't emit blend+immediate.
Added avx2 blend+imm tests.
llvm-svn: 209043
No functionality change intended. The types that previously were set to
lower as Expand or Legal are doing the same thing with this lowering
function.
llvm-svn: 209042
In AT&T syntax, we should probably print the full "movl" or "movw". TableGen
used to ignore these aliases because it was miscounting the number of operands.
This fixes the issue.
This will be tested when the TableGen "should I print this Alias"
heuristic is fixed (very soon).
llvm-svn: 208963
Added target specific combine rules to fold blend intrinsics according
to the following rules:
1) fold(blend A, A, Mask) -> A;
2) fold(blend A, B, <allZeros>) -> A;
3) fold(blend A, B, <allOnes>) -> B.
Added two new tests to verify that the new folding rules work for all
the optimized blend intrinsics.
llvm-svn: 208895
Previously, TableGen assumed that every aliased operand consumed precisely 1
MachineInstr slot (this was reasonable because until a couple of days ago,
nothing more complicated was eligible for printing).
This allows a couple more ARM64 aliases to print so we can remove the special
code.
On the X86 side, I've gone for explicit AT&T size specifiers as the default, so
turned off a few of the aliases that would have just started printing.
llvm-svn: 208880
For example
tzcntl %edi, %ebx
testl %edi, %edi
je .label
can be rewritten into
tzcntl %edi, %ebx
jb .label
A minor complication is that tzcnt sets CF instead of ZF when the input
is zero, we have to rewrite users of the flags from ZF to CF. Currently
we recognize patterns using lzcnt, tzcnt and popcnt.
Differential Revision: http://reviews.llvm.org/D3454
llvm-svn: 208788
r208453 added support for having sret on the second parameter. In that
change, the code for copying sret into a virtual register was hoisted
into the loop that lowers formal parameters. This caused a "Wrong
topological sorting" assertion failure during scheduling when a
parameter is passed in memory. This change undoes that by creating a
second loop that deals with sret.
I'm worried that this fix is incomplete. I don't fully understand the
dependence issues. However, with this change we produce the same DAGs
we used to produce, so if they are broken, they are just as broken as
they have always been.
llvm-svn: 208637
1) Changed gather and scatter intrinsics. Now they are aligned with GCC built-ins. There is no more non-masked form. Masked intrinsic receives -1 if all lanes are executed.
2) I changed the function that works with intrinsics inside X86ISelLowering.cpp. I put all intrinsics in one table. I did it for INTRINSICS_W_CHAIN and plan to put all intrinsics from WO_CHAIN set to the same table in order to avoid the long-long "switch". (I wanted to use static map initialization that allowed by C++11 but I wasn't able to compile it on VS2012).
3) I added gather/scatter prefetch intrinsics.
4) I fixed MRMm encoding for masked instructions.
llvm-svn: 208522
We must validate the value type in TLI::getRegisterByName, because if we
don't and the wrong type was used with the IR intrinsic, then we'll assert
(because we won't be able to find a valid register class with which to
construct the requested copy operation). For PPC64, additionally, the type
information is necessary to decide between the 64-bit register and the 32-bit
subregister.
No functionality change.
llvm-svn: 208508
When lowering build_vector to an insertps, we would still lower it, even
if the source vectors weren't v4x32. This would break on avx if the source
was a v8x32. We now check the type of the source vectors.
llvm-svn: 208487
This reverts commit r200561.
This calling convention was an attempt to match the MSVC C++ ABI for
methods that return structures by value. This solution didn't scale,
because it would have required splitting every CC available on Windows
into two: one for methods and one for free functions.
Now that we can put sret on the second arg (r208453), and Clang does
that (r208458), revert this hack.
llvm-svn: 208459
MSVC always places the implicit sret parameter after the implicit this
parameter of instance methods. We used to handle this for
x86_thiscallcc by allocating the sret parameter on the stack and leaving
the this pointer in ecx, but that doesn't handle alternative calling
conventions like cdecl, stdcall, fastcall, or the win64 convention.
Instead, change the verifier to allow sret on the second parameter.
This also requires changing the Mips and X86 backends to return the
argument with the sret parameter, instead of assuming that the sret
parameter comes first.
The Sparc backend also returns sret parameters in a register, but I
wasn't able to update it to handle secondary sret parameters. It
currently calls report_fatal_error if you feed it an sret in the second
parameter.
Reviewers: rafael.espindola, majnemer
Differential Revision: http://reviews.llvm.org/D3617
llvm-svn: 208453
Summary:
Also ran clang-format on the function. The code added is the last else
if block.
Reviewers: nadav, craig.topper, delena
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3518
llvm-svn: 208372
This patch teaches the backend how to combine packed SSE2/AVX2 arithmetic shift
intrinsics.
The rules are:
- Always fold a packed arithmetic shift by zero to its first operand;
- Convert a packed arithmetic shift intrinsic dag node into a ISD::SRA only if
the shift count is known to be smaller than the vector element size.
This patch also teaches to function 'getTargetVShiftByConstNode' how fold
target specific vector shifts by zero.
Added two new tests to verify that the DAGCombiner is able to fold
sequences of SSE2/AVX2 packed arithmetic shift calls.
llvm-svn: 208342
The old method used by X86TTI to determine partial-unrolling thresholds was
messy (because it worked by testing target features), and also would not
correctly identify the target CPU if certain target features were disabled.
After some discussions on IRC with Chandler et al., it was decided that the
processor scheduling models were the right containers for this information
(because it is often tied to special uop dispatch-buffer sizes).
This does represent a small functionality change:
- For generic x86-64 (which uses the SB model and, thus, will get some
unrolling).
- For AMD cores (because they still currently use the SB scheduling model)
- For Haswell (based on benchmarking by Louis Gerbarg, it was decided to bump
the default threshold to 50; we're working on a test case for this).
Otherwise, nothing has changed for any other targets. The logic, however, has
been moved into BasicTTI, so other targets may now also opt-in to this
functionality simply by setting LoopMicroOpBufferSize in their processor
model definitions.
llvm-svn: 208289
Summary:
Vectors built with zeros and elements in the same order as another
(source) vector are optimized to be built using a single insertps
instruction.
Also optimize when we move one element in a vector to a different place
in that vector while zeroing out some of the other elements.
Further optimizations are possible, described in TODO comments.
I will be implementing at least some of them in the near future.
Added some tests for different cases where this optimization triggers.
Reviewers: nadav, delena, craig.topper
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3521
llvm-svn: 208271
The loop stream detector (LSD) on modern Intel cores, which optimizes the
execution of small loops, has limits on the number of taken branches in
addition to uop-count limits (modern AMD cores have similar limits).
Unfortunately, at the IR level, estimating the number of branches that will be
taken is difficult. For one thing, it strongly depends on later passes (block
placement, etc.). The original implementation took a conservative approach and
limited the maximal BB DFS depth of the loop. However, fairly-extensive
benchmarking by several of us has revealed that this is the wrong approach. In
fact, there are zero known cases where the branch limit prevents a detrimental
unrolling (but plenty of cases where it does prevent beneficial unrolling).
While we could improve the current branch counting logic by incorporating
branch probabilities, this further complication seems unjustified without a
motivating regression. Instead, unless and until a regression appears, the
branch counting will be removed.
llvm-svn: 208255
Given a FMA family (e.g., 213, 231), not all the variants (i.e., register or
memory) are commutable.
E.g., for the 213 family (with the syntax src1, src2, src3):
fmaXXX213 A, B, reg3/mem3 == fmaXXX213 B, A, reg3/mem3
Now consider the 231 family:
fmaXXX231 A, B, reg3 == fmaXXX231 A, reg3, B
But
fmaXXX231 A, B, mem3 != fmaXXX231 A, mem3, B
Indeed, mem3 cannot be the second argument of the memory variant of fmaXXX231.
Working on a reduced test case!
<rdar://problem/16800495>
llvm-svn: 208252
default architecture for reasonable modern x86 processors, actually be
modern. This processor model should essentially be "tuned" for modern
x86 chips as much as possible without undue penalties on any specific
architecture. Previously we weren't even using the nice scheduling
models. There are a few other tweaks needed here, but this change at
least I have benchmarked across a decent swatch of chips (intel's
clovertown, westmere, and sandybridge; amd's istanbul) and seen no
significant regressions.
If anyone has suggested ways to test this, just let me know. Somewhat
alarmingly, no existing tests failed.
llvm-svn: 208230
Before this patch, the backend always emitted a store+load sequence to
bitconvert from f64 to i64 the input operand of a ISD::BITCAST dag node that
performed a bitconvert from type MVT::f64 to type MVT::v2i32. The resulting
i64 node was then used to build a v2i32 vector.
With this patch, the backend now produces a cheaper SCALAR_TO_VECTOR from
MVT::f64 to MVT::v2f64. That SCALAR_TO_VECTOR is then followed by a "free"
bitcast to type MVT::v4i32. The elements of the resulting
v4i32 are then extracted to build a v2i32 vector (which is illegal and
therefore promoted to MVT::v2i64).
This is in general cheaper than emitting a stack store+load sequence
to bitconvert the operand from type f64 to type i64.
llvm-svn: 208107
This patch implements the infrastructure to use named register constructs in
programs that need access to specific registers (bare metal, kernels, etc).
So far, only the stack pointer is supported as a technology preview, but as it
is, the intrinsic can already support all non-allocatable registers from any
architecture.
llvm-svn: 208104
The Win64 docs are very clear that anything larger than 8 bytes is
passed by reference, and GCC MinGW64 honors that for __modti3 and
friends.
Patch by Jameson Nash!
llvm-svn: 208029
Summary:
Also ran clang-format on the function. The code added is the last else
if block.
Reviewers: nadav, craig.topper
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3518
llvm-svn: 207992
Both MinGW and cygwin (i686) construct export directives without the global
leader prefix. This is mostly due to the fact that they use GNU ld which does
not correctly handle the export directive. This apparently has been been broken
for a while. However, this was recently reported as being broken by
mingwandroid and diorcety of the msys2 project.
Remove the global leader prefix if targeting MinGW or cygwin, otherwise, retain
the global leader prefix. Add an explicit test for cygwin's behaviour of export
directives.
llvm-svn: 207926
Create a helper function to generate the export directive. This was previously
duplicated inline to handle export directives for variables and functions. This
also enables the use of range-based iterators for the generation of the
directive rather than the traditional loops. NFC.
llvm-svn: 207925
The fix itself is fairly simple: move getAccessVariant to MCValue so that we
replace the old weak expression evaluation with the far more general
EvaluateAsRelocatable.
This then requires that EvaluateAsRelocatable stop when it finds a non
trivial reference kind. And that in turn requires the ELF writer to look
harder for weak references.
Last but not least, this found a case where we were being bug by bug
compatible with gas and accepting an invalid input. I reported pr19647
to track it.
llvm-svn: 207920
There is no need to check if we want to hoist the immediate value of an
shift instruction. Simply return TCC_Free right away.
This change is like r206101, but for X86.
rdar://problem/16190769
llvm-svn: 207692
Only emit calls to compiler-rt asm routines on platforms where they are
present (currently limited to linux i386/x86_64).
Patch by Yuri Gorshenin.
llvm-svn: 207651
Currently, musttail codegen is relying on sibcall optimization, and
reporting a fatal error if fails. Sibcall optimization fails when stack
arguments need to be modified, which is insufficient for musttail.
The logic for moving arguments in memory safely is already implemented
for GuaranteedTailCallOpt. This change merely arranges for musttail
calls to use it.
No functional change for GuaranteedTailCallOpt.
Reviewers: espindola
Differential Revision: http://reviews.llvm.org/D3493
llvm-svn: 207598
X86_MAX_OPERANDS is changed to unsigned.
Also, add range-based for loops for affected loops. This in turn
needed an ArrayRef instead of a pointer-to-array in
InternalInstruction.
llvm-svn: 207413
This introduces a target specific streamer, X86WinCOFFStreamer, which handles
the target specific behaviour (e.g. WinEH). This is mostly to ensure that
differences between ARM and X86 remain disjoint and do not accidentally cross
boundaries. This is the final staging change for enabling object emission for
Windows on ARM.
llvm-svn: 207344
Scaling factors are not free on X86 because every "complex" addressing mode
breaks the related instruction into 2 allocations instead of 1.
<rdar://problem/16730541>
llvm-svn: 207301
Summary:
If we're doing a v4f32/v4i32 shuffle on x86 with SSE4.1, we can lower
certain shufflevectors to an insertps instruction:
When most of the shufflevector result's elements come from one vector (and
keep their index), and one element comes from another vector or a memory
operand.
Added tests for insertps optimizations on shufflevector.
Added support and tests for v4i32 vector optimization.
Reviewers: nadav
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3475
llvm-svn: 207291
Change the object streamer selection to a switch from a series of if conditions.
Rather than defaulting to ELF, require that an ELF format is requested. The
Windows/!ELF is maintained as MachO would have been selected first and will
still provide a MachO format. Add an assertion that if COFF is requested that
the target platform is Windows as only WinCOFF object emission is currently
supported.
llvm-svn: 207200
This is similar to the 'tail' marker, except that it guarantees that
tail call optimization will occur. It also comes with convervative IR
verification rules that ensure that tail call optimization is possible.
Reviewers: nicholas
Differential Revision: http://llvm-reviews.chandlerc.com/D3240
llvm-svn: 207143
This patch:
- Adds two new X86 builtin intrinsics ('int_x86_rdtsc' and
'int_x86_rdtscp') as GCCBuiltin intrinsics;
- Teaches the backend how to lower the two new builtins;
- Introduces a common function to lower READCYCLECOUNTER dag nodes
and the two new rdtsc/rdtscp intrinsics;
- Improves (and extends) the existing x86 test 'rdtsc.ll'; now test 'rdtsc.ll'
correctly verifies that both READCYCLECOUNTER and the two new intrinsics
work fine for both 64bit and 32bit Subtargets.
llvm-svn: 207127
I discovered this const-hole while attempting to coalesnce the Symbol
and SymbolMap data structures. There's some pending issues with that,
but I figured this change was easy to flush early.
llvm-svn: 207124
This allows us to compile
return (mask & 0x8 ? a : b);
into
testb $8, %dil
cmovnel %edx, %esi
instead of
andl $8, %edi
shrl $3, %edi
cmovnel %edx, %esi
which we formed previously because dag combiner canonicalizes setcc of and into shift.
llvm-svn: 207088
