Merges equivalent loads on both sides of a hammock/diamond
and hoists into into the header.
Merges equivalent stores on both sides of a hammock/diamond
and sinks it to the footer.
Can enable if conversion and tolerate better load misses
and store operand latencies.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213396 91177308-0d34-0410-b5e6-96231b3b80d8
On AArch64 the pseudo instruction ldr <reg>, =... supports both
32-bit and 64-bit constants. Add support for 64 bit constants for
the pools to support the pseudo instruction fully.
Changes the AArch64 ldr-pseudo tests to use 32-bit registers and
adds tests with 64-bit registers.
Patch by Janne Grunau!
Differential Revision: http://reviews.llvm.org/D4279
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213387 91177308-0d34-0410-b5e6-96231b3b80d8
This attribute indicates that the parameter or return pointer is
dereferenceable. Practically speaking, loads from such a pointer within the
associated byte range are safe to speculatively execute. Such pointer
parameters are common in source languages (C++ references, for example).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213385 91177308-0d34-0410-b5e6-96231b3b80d8
Because i16 is illegal, there's no native DAG method to
represent a bitcast to or from an f16 type. This meant LLVM was
inserting a stack store/load pair which is really not ideal.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213378 91177308-0d34-0410-b5e6-96231b3b80d8
Actual support for softening f16 operations is still limited, and can be added
when it's needed. But Soften is much closer to being a useful thing to try
than keeping it Legal when no registers can actually hold such values.
Longer term, we probably want something between Soften and Promote semantics
for most targets, it'll be more efficient to promote the 4 basic operations to
f32 than libcall them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213372 91177308-0d34-0410-b5e6-96231b3b80d8
The post-indexed instructions were missing the constraint, causing unpredictable STR instructions to be emitted.
The earlyclobber constraint on the pre-indexed STR instructions is not strictly necessary, as the instruction selection for pre-indexed STR instructions goes through an additional layer of pseudo instructions which have the constraint defined, however it doesn't hurt to specify the constraint directly on the pre-indexed instructions as well, since at some point someone might create instances of them programmatically and then the constraint is definitely needed.
This fixes PR20323.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213369 91177308-0d34-0410-b5e6-96231b3b80d8
This test is actually going in the opposite direction to what the
filename and function name suggested.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213358 91177308-0d34-0410-b5e6-96231b3b80d8
Unfortunately, we don't seem to have a direct truncation, but the
extension can be legally split into two operations so we should
support that.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213357 91177308-0d34-0410-b5e6-96231b3b80d8
Clang may well start emitting these soon, and while it may not be
directly relevant for OpenCL or GLSL, the instructions were just
sitting there waiting to be used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213356 91177308-0d34-0410-b5e6-96231b3b80d8
Since the result of a SETCC for X86 is 0 or -1 in each lane, we can
move unary operations, in this case [su]int_to_fp through the mask
operation and constant fold the operation away. Generally speaking:
UNARYOP(AND(VECTOR_CMP(x,y), constant))
--> AND(VECTOR_CMP(x,y), constant2)
where constant2 is UNARYOP(constant).
This implements the transform where UNARYOP is [su]int_to_fp.
For example, consider the simple function:
define <4 x float> @foo(<4 x float> %val, <4 x float> %test) nounwind {
%cmp = fcmp oeq <4 x float> %val, %test
%ext = zext <4 x i1> %cmp to <4 x i32>
%result = sitofp <4 x i32> %ext to <4 x float>
ret <4 x float> %result
}
Before this change, the SSE code is generated as:
LCPI0_0:
.long 1 ## 0x1
.long 1 ## 0x1
.long 1 ## 0x1
.long 1 ## 0x1
.section __TEXT,__text,regular,pure_instructions
.globl _foo
.align 4, 0x90
_foo: ## @foo
cmpeqps %xmm1, %xmm0
andps LCPI0_0(%rip), %xmm0
cvtdq2ps %xmm0, %xmm0
retq
After, the code is improved to:
LCPI0_0:
.long 1065353216 ## float 1.000000e+00
.long 1065353216 ## float 1.000000e+00
.long 1065353216 ## float 1.000000e+00
.long 1065353216 ## float 1.000000e+00
.section __TEXT,__text,regular,pure_instructions
.globl _foo
.align 4, 0x90
_foo: ## @foo
cmpeqps %xmm1, %xmm0
andps LCPI0_0(%rip), %xmm0
retq
The cvtdq2ps has been constant folded away and the floating point 1.0f
vector lanes are materialized directly via the ModRM operand of andps.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213342 91177308-0d34-0410-b5e6-96231b3b80d8
Since the result of a SETCC for AArch64 is 0 or -1 in each lane, we can
move unary operations, in this case [su]int_to_fp through the mask
operation and constant fold the operation away. Generally speaking:
UNARYOP(AND(VECTOR_CMP(x,y), constant))
--> AND(VECTOR_CMP(x,y), constant2)
where constant2 is UNARYOP(constant).
This implements the transform where UNARYOP is [su]int_to_fp.
For example, consider the simple function:
define <4 x float> @foo(<4 x float> %val, <4 x float> %test) nounwind {
%cmp = fcmp oeq <4 x float> %val, %test
%ext = zext <4 x i1> %cmp to <4 x i32>
%result = sitofp <4 x i32> %ext to <4 x float>
ret <4 x float> %result
}
Before this change, the code is generated as:
fcmeq.4s v0, v0, v1
movi.4s v1, #0x1 // Integer splat value.
and.16b v0, v0, v1 // Mask lanes based on the comparison.
scvtf.4s v0, v0 // Convert each lane to f32.
ret
After, the code is improved to:
fcmeq.4s v0, v0, v1
fmov.4s v1, #1.00000000 // f32 splat value.
and.16b v0, v0, v1 // Mask lanes based on the comparison.
ret
The svvtf.4s has been constant folded away and the floating point 1.0f
vector lanes are materialized directly via fmov.4s.
Rather than do the folding manually in the target code, teach getNode()
in the generic SelectionDAG to handle folding constant operands of
vector [su]int_to_fp nodes. It is reasonable (as noted in a FIXME) to do
additional constant folding there as well, but I don't have test cases
for those operations, so leaving them for another time when it becomes
appropriate.
rdar://17693791
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213341 91177308-0d34-0410-b5e6-96231b3b80d8
There's a bug where this can create cycles in the DAG. It will take a bit
to fix, so I'm backing it out for now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213339 91177308-0d34-0410-b5e6-96231b3b80d8
We now consider the FPOpFusion flag when determining whether
to fuse ops. We also explicitly emit add.rn when fusion is
disabled to prevent ptxas from fusing the operations on its
own.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213287 91177308-0d34-0410-b5e6-96231b3b80d8
There are two parts here. First is to modify tablegen to adjust the encoding
type ENCODING_RM with the scaling factor.
The second is to use the new encoding types to compute the correct
displacement in the decoder.
Fixes <rdar://problem/17608489>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213281 91177308-0d34-0410-b5e6-96231b3b80d8
Convert the operand to int if possible, i.e. if the value is properly
initialized. (I suppose there is further room for improvement here to also
peform the shift if the uninitialized bits are shifted out.)
With this little change we can now compute the scaling factor for compressed
displacement with pure tablegen code in the X86 backend. This is useful
because both the X86-disassembler-specific part of tablegen and the assembler
need this and TD is the natural sharing place.
The patch also adds the missing documentation for the shift and add operator.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213277 91177308-0d34-0410-b5e6-96231b3b80d8
The header contains an offset to the DWARF abbreviations for the CU. The offset
must be section relative for COFF and absolute for others. The non-assembly
code path for the DWARF header generation already had the correct emission for
the headers. This corrects just the assembly path. Due to the invalid
relocation, processing of the debug information would halt previously on the
first assembly input as the associated abbreviations would be out of range as
they would have the location increased by image base and the section offset.
This address PR20332.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213275 91177308-0d34-0410-b5e6-96231b3b80d8
This also uses TSFlags to mark machine instructions that are surface/texture
accesses, as well as the vector width for surface operations. This is used
to simplify some of the switch statements that need to detect surface/texture
instructions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213256 91177308-0d34-0410-b5e6-96231b3b80d8
Previously we asserted on this code. Currently compiler-rt doesn't
actually implement any of these new libcalls, but external help is
pretty much the only viable option for LLVM.
I've followed the much more generic "__truncST2" naming, as opposed to
the odd name for f32 -> f16 truncation. This can obviously be changed
later, or overridden by any targets that need to.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213252 91177308-0d34-0410-b5e6-96231b3b80d8
x86 has no native ability to extend an f16 to f64, but the same result
is obtained if we expand it into two separate extensions: f16 -> f32
-> f64.
Unfortunately the same is not true for truncate, so that still results
in a compilation failure.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213251 91177308-0d34-0410-b5e6-96231b3b80d8
This makes the two intrinsics @llvm.convert.from.f16 and
@llvm.convert.to.f16 accept types other than simple "float". This is
only strictly needed for the truncate operation, since otherwise
double rounding occurs and there's no way to represent the strict IEEE
conversion. However, for symmetry we allow larger types in the extend
too.
During legalization, we can expand an "fp16_to_double" operation into
two extends for convenience, but abort when the truncate isn't legal. A new
libcall is probably needed here.
Even after this commit, various target tweaks are needed to actually use the
extended intrinsics. I've put these into separate commits for clarity, so there
are no actual tests of f64 conversion here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213248 91177308-0d34-0410-b5e6-96231b3b80d8
Memory barrier __builtin_arm_[dmb, dsb, isb] intrinsics are required to
implement their corresponding ACLE and MSVC intrinsics.
This patch ports ARM dmb, dsb, isb intrinsic to AArch64.
Differential Revision: http://reviews.llvm.org/D4520
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213247 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Generally speaking, mips-* vs mips64-* should not be used to make decisions
about the content or format of the ELF. This should be based on the ABI
and CPU in use. For example, `mips-linux-gnu-clang -mips64r2 -mabi=64`
should produce an ELF64 as should `mips64-linux-gnu-clang -mabi=64`.
Conversely, `mips64-linux-gnu-clang -mabi=n32` should produce an ELF32 as
should `mips-linux-gnu-clang -mips64r2 -mabi=n32`.
This patch fixes the e_flags but leaves the ELF32 vs ELF64 issue for now
since there is no apparent way to base this decision on the ABI and CPU.
Differential Revision: http://reviews.llvm.org/D4539
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213244 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
The cpr1_size field describes the minimum register width to run the program
rather than the size of the registers on the target. MIPS32r6 was acting
as if -mfp64 has been given because it starts off with 64-bit FPU registers.
Differential Revision: http://reviews.llvm.org/D4538
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213243 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
These options are not implemented yet but we act as if they are always
given.
The integrated assembler is driven by the clang driver so the e_flag test
cases should match the e_flags emitted by GCC+GAS rather than GAS
by itself.
Differential Revision: http://reviews.llvm.org/D4536
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213242 91177308-0d34-0410-b5e6-96231b3b80d8
Origin is meaningless for fully initialized values. Avoid
storing origin for function arguments that are known to
be always initialized (i.e. shadow is a compile-time null
constant).
This is not about correctness, but purely an optimization.
Seems to affect compilation time of blacklisted functions
significantly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213239 91177308-0d34-0410-b5e6-96231b3b80d8
This reverts, "r213024 - Revert r212572 "improve BasicAA CS-CS queries", it
causes PR20303." with a fix for the bug in pr20303. As it turned out, the
relevant code was both wrong and over-conservative (because, as with the code
it replaced, it would return the overall ModRef mask even if just Ref had been
implied by the argument aliasing results). Hopefully, this correctly fixes both
problems.
Thanks to Nick Lewycky for reducing the test case for pr20303 (which I've
cleaned up a little and added in DSE's test directory). The BasicAA test has
also been updated to check for this error.
Original commit message:
BasicAA contains knowledge of certain intrinsics, such as memcpy and memset,
and uses that information to form more-accurate answers to CallSite vs. Loc
ModRef queries. Unfortunately, it did not use this information when answering
CallSite vs. CallSite queries.
Generically, when an intrinsic takes one or more pointers and the intrinsic is
marked only to read/write from its arguments, the offset/size is unknown. As a
result, the generic code that answers CallSite vs. CallSite (and CallSite vs.
Loc) queries in AA uses UnknownSize when forming Locs from an intrinsic's
arguments. While BasicAA's CallSite vs. Loc override could use more-accurate
size information for some intrinsics, it did not do the same for CallSite vs.
CallSite queries.
This change refactors the intrinsic-specific logic in BasicAA into a generic AA
query function: getArgLocation, which is overridden by BasicAA to supply the
intrinsic-specific knowledge, and used by AA's generic implementation. This
allows the intrinsic-specific knowledge to be used by both CallSite vs. Loc and
CallSite vs. CallSite queries, and simplifies the BasicAA implementation.
Currently, only one function, Mac's memset_pattern16, is handled by BasicAA
(all the rest are intrinsics). As a side-effect of this refactoring, BasicAA's
getModRefBehavior override now also returns OnlyAccessesArgumentPointees for
this function (which is an improvement).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213219 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Converting outermost zext(a) to sext(a) causes worse code when the
computation of zext(a) could be reused. For example, after converting
... = array[zext(a)]
... = array[zext(a) + 1]
to
... = array[sext(a)]
... = array[zext(a) + 1],
the program computes sext(a), which is actually unnecessary. I added one
test in split-gep-and-gvn.ll to illustrate this scenario.
Also, with r211281 and r211084, we annotate more "nuw" tags to
computation involving CUDA intrinsics such as threadIdx.x. These
annotations help with splitting GEP a lot, rendering the benefit we get
from this reverted optimization only marginal.
Test Plan: make check-all
Reviewers: eliben, meheff
Reviewed By: meheff
Subscribers: jholewinski, llvm-commits
Differential Revision: http://reviews.llvm.org/D4542
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213209 91177308-0d34-0410-b5e6-96231b3b80d8
This fixes an issue where a local value is defined before and used after an
inline asm call with side effects.
This fix simply flushes the local value map, which updates the insertion point
for the inline asm call to be above any previously defined local values.
This fixes <rdar://problem/17694203>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213203 91177308-0d34-0410-b5e6-96231b3b80d8
We were not considering the stated alignment on vector loads/stores,
leading us to generate vector instructions even when we do not have
sufficient alignment.
Now, for IR like:
%1 = load <4 x float>, <4 x float>* %ptr, align 4
we will generate correct, conservative PTX like:
ld.f32 ... [%ptr]
ld.f32 ... [%ptr+4]
ld.f32 ... [%ptr+8]
ld.f32 ... [%ptr+12]
Or if we have an alignment of 8 (for example), we can
generate code like:
ld.v2.f32 ... [%ptr]
ld.v2.f32 ... [%ptr+8]
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213186 91177308-0d34-0410-b5e6-96231b3b80d8
