Given the following C code llvm currently generates suboptimal code for
x86-64:
__m128 bss4( const __m128 *ptr, size_t i, size_t j )
{
float f = ptr[i][j];
return (__m128) { f, f, f, f };
}
=================================================
define <4 x float> @_Z4bss4PKDv4_fmm(<4 x float>* nocapture readonly %ptr, i64 %i, i64 %j) #0 {
%a1 = getelementptr inbounds <4 x float>* %ptr, i64 %i
%a2 = load <4 x float>* %a1, align 16, !tbaa !1
%a3 = trunc i64 %j to i32
%a4 = extractelement <4 x float> %a2, i32 %a3
%a5 = insertelement <4 x float> undef, float %a4, i32 0
%a6 = insertelement <4 x float> %a5, float %a4, i32 1
%a7 = insertelement <4 x float> %a6, float %a4, i32 2
%a8 = insertelement <4 x float> %a7, float %a4, i32 3
ret <4 x float> %a8
}
=================================================
shlq $4, %rsi
addq %rdi, %rsi
movslq %edx, %rax
vbroadcastss (%rsi,%rax,4), %xmm0
retq
=================================================
The movslq is uneeded, but is present because of the trunc to i32 and then
sext back to i64 that the backend adds for vbroadcastss.
We can't remove it because it changes the meaning. The IR that clang
generates is already suboptimal. What clang really should emit is:
%a4 = extractelement <4 x float> %a2, i64 %j
This patch makes that legal. A separate patch will teach clang to do it.
Differential Revision: http://reviews.llvm.org/D3519
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207801 91177308-0d34-0410-b5e6-96231b3b80d8
This creates a lot of core infrastructure in which to add, with little
effort, quite a bit more to mips fast-isel
Test Plan: simplestore.ll
Reviewers: dsanders
Reviewed By: dsanders
Differential Revision: http://reviews.llvm.org/D3527
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207790 91177308-0d34-0410-b5e6-96231b3b80d8
Use i32 instead of specifying SReg_32. When this is
the pseudo INDIRECT_BASE_ADDR, this would give a bogus
verifier error.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207770 91177308-0d34-0410-b5e6-96231b3b80d8
The canonical form of the BFM instruction is always one of the more explicit
extract or insert operations, which makes reading output much easier.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207752 91177308-0d34-0410-b5e6-96231b3b80d8
For pattern like ((x >> C1) & Mask) << C2, DAG combiner may convert it
into (x >> (C1-C2)) & (Mask << C2), which makes pattern matching of ubfx
more difficult.
For example:
Given
%shr = lshr i64 %x, 4
%and = and i64 %shr, 15
%arrayidx = getelementptr inbounds [8 x [64 x i64]]* @arr, i64 0, %i64 2, i64 %and
%0 = load i64* %arrayidx
With current shift folding, it takes 3 instrs to compute base address:
lsr x8, x0, #1
and x8, x8, #0x78
add x8, x9, x8
If using ubfx, it only needs 2 instrs:
ubfx x8, x0, #4, #4
add x8, x9, x8, lsl #3
This fixes bug 19589
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207702 91177308-0d34-0410-b5e6-96231b3b80d8
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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207692 91177308-0d34-0410-b5e6-96231b3b80d8
We can't use SALU instructions for this since they ignore the EXEC mask
and are always executed.
This fixes several OpenCV tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207661 91177308-0d34-0410-b5e6-96231b3b80d8
target cannot be determined accurately. This is the case for NaCl where the
sandboxing instructions are added in MC layer, after the MipsLongBranch pass.
It is also the case when the code has inline assembly. Instead of calculating
offset in the MipsLongBranch pass, use %hi(sym1 - sym2) and %lo(sym1 - sym2)
expressions that are resolved during the fixup.
This patch also deletes microMIPS test file test/CodeGen/Mips/micromips-long-branch.ll
and implements microMIPS CHECKs in a much simpler way in a file
test/CodeGen/Mips/longbranch.ll, together with MIPS32 and MIPS64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207656 91177308-0d34-0410-b5e6-96231b3b80d8
The canonical syntax for shifts by a variable amount does not end with 'v', but
that syntax should be supported as an alias (presumably for legacy reasons).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207649 91177308-0d34-0410-b5e6-96231b3b80d8
On instructions using the NZCV register, a couple of conditions have dual
representations: HS/CS and LO/CC (meaning unsigned-higher-or-same/carry-set and
unsigned-lower/carry-clear). The first of these is more descriptive in most
circumstances, so we should print it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207644 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
This isn't supported directly so we rotate the vector by the desired number of
elements, insert to element zero, then rotate back.
The i64 case generates rather poor code on MIPS32. There is an obvious
optimisation to be made in future (do both insert.w's inside a shared
rotate/unrotate sequence) but for now it's sufficient to select valid code
instead of aborting.
Depends on D3536
Reviewers: matheusalmeida
Reviewed By: matheusalmeida
Differential Revision: http://reviews.llvm.org/D3537
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207640 91177308-0d34-0410-b5e6-96231b3b80d8
Since these are mostly used in "lsl #16", "lsl #32", "lsl #48" combinations to
piece together an immediate in 16-bit chunks, hex is probably the most
appropriate format.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207635 91177308-0d34-0410-b5e6-96231b3b80d8
This is mostly aimed at the NEON logical operations and MOVI/MVNI (since they
accept weird shifts which are more naturally understandable in hex notation).
Also changes BRK/HINT etc, which is probably a neutral change, but easier than
the alternative.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207634 91177308-0d34-0410-b5e6-96231b3b80d8
Since these instructions only accept a 12-bit immediate, possibly shifted left
by 12, the canonical syntax used by the architecture reference manual is "#N {,
lsl #12 }". We should accept an immediate that has already been shifted, (e.g.
Also, print a comment giving the full addend since it can be helpful.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207633 91177308-0d34-0410-b5e6-96231b3b80d8
This is a partial port of r204816 (cpirker "Elf support for MC-JIT
runtime dynamic linker") from AArch64 to ARM64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207625 91177308-0d34-0410-b5e6-96231b3b80d8
This introduces the stack lowering emission of the stack probe function for
Windows on ARM. The stack on Windows on ARM is a dynamically paged stack where
any page allocation which crosses a page boundary of the following guard page
will cause a page fault. This page fault must be handled by the kernel to
ensure that the page is faulted in. If this does not occur and a write access
any memory beyond that, the page fault will go unserviced, resulting in an
abnormal program termination.
The watermark for the stack probe appears to be at 4080 bytes (for
accommodating the stack guard canaries and stack alignment) when SSP is
enabled. Otherwise, the stack probe is emitted on the page size boundary of
4096 bytes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207615 91177308-0d34-0410-b5e6-96231b3b80d8
IMAGE_REL_ARM_MOV32T relocations require that the movw/movt pair-wise
relocation is not split up and reordered. When expanding the mov32imm
pseudo-instruction, create a bundle if the machine operand is referencing an
address. This helps ensure that the relocatable address load is not reordered
by subsequent passes.
Unfortunately, this only partially handles the case as the Constant Island Pass
occurs after the instructions are unbundled and does not properly handle
bundles. That is a more fundamental issue with the pass itself and beyond the
scope of this change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207608 91177308-0d34-0410-b5e6-96231b3b80d8
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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207598 91177308-0d34-0410-b5e6-96231b3b80d8
SI_IF and SI_ELSE are terminators which also produce a value. For
these instructions ISel always inserts a COPY to move their value
to another basic block. This COPY ends up between SI_(IF|ELSE)
and the S_BRANCH* instruction at the end of the block.
This breaks MachineBasicBlock::getFirstTerminator() and also the
machine verifier which assumes that terminators are grouped together at
the end of blocks.
To solve this we coalesce the copy away right after ISel to make sure
there are no instructions in between terminators at the end of blocks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207591 91177308-0d34-0410-b5e6-96231b3b80d8
SALU instructions ignore control flow, so it is not always safe to use
them within branches. This is a partial solution to this problem
until we can come up with something better.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207590 91177308-0d34-0410-b5e6-96231b3b80d8
This is a squash of several optimization commits:
- calculate DIV_Lo and DIV_Hi separately
- use BFE_U32 if we are operating on 32bit values
- use precomputed constants instead of shifting in UDVIREM
- skip the first 32 iterations of udivrem
v2: Check whether BFE is supported before using it
Patch by: Jan Vesely
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207589 91177308-0d34-0410-b5e6-96231b3b80d8
There are no patterns for this. This was already fixed for ARM64 but I forgot
to apply it to AArch64 too.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207515 91177308-0d34-0410-b5e6-96231b3b80d8
This gets us pretty code for divs of i16 vectors. Turn the existing
intrinsics into the corresponding nodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207317 91177308-0d34-0410-b5e6-96231b3b80d8
Otherwise the legalizer would just scalarize everything. Support for
mulhi in the targets isn't that great yet so on most targets we get
exactly the same scalarized output. Add a test for x86 vector udiv.
I had to disable the mulhi nodes on ARM because there aren't any patterns
for it. As far as I know ARM has instructions for getting the high part of
a multiply so this should be fixed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207315 91177308-0d34-0410-b5e6-96231b3b80d8
The included test case would return the incorrect results, because the expansion
of an shift with a constant shift amount of 0 would generate undefined behavior.
This is because ExpandShiftByConstant assumes that all shifts by constants with
a value of 0 have already been optimized away. This doesn't happen for opaque
constants and usually this isn't a problem, because opaque constants won't take
this code path - they are not supposed to. In the case that the opaque constant
has to be expanded by the legalizer, the legalizer would drop the opaque flag.
In this case we hit the limitations of ExpandShiftByConstant and create incorrect
code.
This commit fixes the legalizer by not dropping the opaque flag when expanding
opaque constants and adding an assertion to ExpandShiftByConstant to catch this
not supported case in the future.
This fixes <rdar://problem/16718472>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207304 91177308-0d34-0410-b5e6-96231b3b80d8
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>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207301 91177308-0d34-0410-b5e6-96231b3b80d8
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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207291 91177308-0d34-0410-b5e6-96231b3b80d8
