This patch allows the same implicit conversions for vector-scalar
operations in SVE that are allowed for NEON.
Depends on D126377
Reviewed By: c-rhodes
Differential Revision: https://reviews.llvm.org/D126380
According D125377, we order STP Q's by ascending address. While on some
targets, paired 128 bit loads and stores are slow, so the STP will split
into STRQ and STUR, so I hope these stores will also be ordered.
Also add subtarget feature ascend-store-address to control the aggressive order.
Reviewed By: dmgreen, fhahn
Differential Revision: https://reviews.llvm.org/D126700
Currently the a AAPCS compliant frame record is not always created for
functions when it should. Although a consistent frame record might not
be required in some cases, there are still scenarios where applications
may want to make use of the call hierarchy made available trough it.
In order to enable the use of AAPCS compliant frame records whilst keep
backwards compatibility, this patch introduces a new command-line option
(`-mframe-chain=[none|aapcs|aapcs+leaf]`) for Aarch32 and Thumb backends.
The option allows users to explicitly select when to use it, and is also
useful to ensure the extra overhead introduced by the frame records is
only introduced when necessary, in particular for Thumb targets.
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D125094
Simiarly to what's done on both ARM's and AArch64's frame lowering code,
this updates Thumb1FrameLowering to use the FrameDestroy Machine
Instruction flag to identify instructions inserted as part of the epilog
instead of relying on assumptions about specific machine instructions.
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D126285
When pushing an operation across a phi node, we should avoid doing
so across a loop backedge. This is generally non-profitable, because
it does not reduce the number of times the operation is executed,
and could lead to an infinite combine loop.
The code was already guarding against this, but using an
insufficiently strong condition, which did not cover the case where
the operation was originally outside the loop (in which case the
transform moves the operation from outside the loop into the loop,
which is particularly undesirable).
Differential Revision: https://reviews.llvm.org/D127499
With opaque pointers, we end up merging these GEPs and dropping
the inrange attribute (in the last two cases). This did not happen
previously, because typed pointers use less powerful GEP folding logic.
I'm a bit unsure whether this is something we need to be concerned
about or not. I believe that generally our stance is that we should
perform folds even if this requires losing poison-generating flags
like inrange.
We can either a) accept this as-is, b) try to inhibit folding if it
requires dropping inrange or c) try to fold to poison if we know
that inrange is going to be violated.
For now, we accept it as-is.
Differential Revision: https://reviews.llvm.org/D127503
On 64-bit X86, 0x66 operand-size override prefix will change the size of
the instruction operand, e.g. from 32 bits to 16 bits, but it will not
modify the size of the displacement operand used for memory addressing,
which will always be 32 bits.
Reviewed By: skan, rafauler
Differential Revision: https://reviews.llvm.org/D126726
Just matter of enabling the config option.
(Also changed the platform of the input test file to macOS, since that's
the default that we specify in the `%lld` substitution. The conflict was
causing errors when linking with LTO.)
Reviewed By: #lld-macho, thakis
Differential Revision: https://reviews.llvm.org/D127600
D125496 changed the string layout on windows. Change `__is_long_` and `__size_` back to using `unsigned char` to fix the issue.
Reviewed By: Mordante, #libc
Spies: jloser, libcxx-commits, ayzhao
Differential Revision: https://reviews.llvm.org/D127566
This patch allows SimplifyDemandedBits to call SimplifyMultipleUseDemandedBits in cases where the source operand has other uses, enabling us to peek through the shifted value if we don't demand all the bits/elts.
This helps with several of the regressions from D125836
These methods don't access any state from RISCVInstrInfo. Make them
free functions in the RISCV namespace.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D127583
If the LHS/RHS selection operands can be cheaply concatenated back together then replace 2 x 128-bit selection nodes with 1 x 256-bit node
Addresses the regression introduced in the bug fix from rGd5af6a38082b39ae520a328e44dc29ebcb036bb2
REAPPLIED with for bug identified in rGea8fb3b60196
When convertEndianOfCharForBEmachine is called with elementBitWidth
smaller than CHAR_BIT, the default case is invoked, but this does
nothing at all and leaves the output array unchanged.
Fix DenseIntOrFPElementsAttr::convertEndianOfArrayRefForBEmachine
by not calling convertEndianOfCharForBEmachine in this case, and
instead simply copying the input to the output (for sub-byte types,
endian conversion is in fact a no-op).
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D125676
Previously, omitting unnecessary DWARF unwinds was only done in two
cases:
* For Darwin + aarch64, if no DWARF unwind info is needed for all the
functions in a TU, then the `__eh_frame` section would be omitted
entirely. If any one function needed DWARF unwind, then MC would emit
DWARF unwind entries for all the functions in the TU.
* For watchOS, MC would omit DWARF unwind on a per-function basis, as
long as compact unwind was available for that function.
This diff makes it so that we omit DWARF unwind on a per-function basis
for Darwin + aarch64 as well. In addition, we introduce the flag
`--emit-dwarf-unwind=` which can toggle between `always`,
`no-compact-unwind` (only emit DWARF when CU cannot be emitted for a
given function), and the target platform `default`. `no-compact-unwind`
is particularly useful for newer x86_64 platforms: we don't want to omit
DWARF unwind for x86_64 in general due to possible backwards compat
issues, but we should make it possible for people to opt into this
behavior if they are only targeting newer platforms.
**Motivation:** I'm working on adding support for `__eh_frame` to LLD,
but I'm concerned that we would suffer a perf hit. Processing compact
unwind is already expensive, and that's a simpler format than EH frames.
Given that MC currently produces one EH frame entry for every compact
unwind entry, I don't think processing them will be cheap. I tried to do
something clever on LLD's end to drop the unnecessary EH frames at parse
time, but this made the code significantly more complex. So I'm looking
at fixing this at the MC level instead.
**Addendum:** It turns out that there was a latent bug in the X86
backend when `OmitDwarfIfHaveCompactUnwind` is naively enabled, which is
not too surprising given that this combination has not been heretofore
used.
For functions that have unwind info that cannot be encoded with CU, MC
would end up dropping both the compact unwind entry (OK; existing
behavior) as well as the DWARF entries (not OK). This diff fixes things
so that we emit the DWARF entry, as well as a CU entry with encoding
`UNWIND_X86_MODE_DWARF` -- this basically tells the unwinder to look for
the DWARF entry. I'm not 100% sure the `UNWIND_X86_MODE_DWARF` CU entry
is necessary, this was the simplest fix. ld64 seems to be able to handle
both the absence and presence of this CU entry. Ultimately ld64 (and
LLD) will synthesize `UNWIND_X86_MODE_DWARF` if it is absent, so there
is no impact to the final binary size.
Reviewed By: davide, lhames
Differential Revision: https://reviews.llvm.org/D122258
Since D61806, DAGCombiner has folded subvector_extract(bitcast(..)) to
bitcast(subvector_extract(..)), which would place a bitcast between a
subvector_extract and the operation that could be converted to a high
neon instruction (like smull2). This adds better matching for the
subvector_extract, through the tablegen extract_high PatFrags to
optionally skip the bitcast under little ending, still matchings an
extract of the high half of the input vector.
I didn't update the extract_high of a duplicate patterns, as the
ComplexPattern need names operands. I did add a extract_high_dup_v8i16
PatFrag to abstract away the common code, which can be extended in a
future patch.
Differential Revision: https://reviews.llvm.org/D126782
GCC and Clang/LLVM will support `_Float16` on X86 in C/C++, following
the latest X86 psABI. (https://gitlab.com/x86-psABIs)
_Float16 arithmetic will be performed using native half-precision. If
native arithmetic instructions are not available, it will be performed
at a higher precision (currently always float) and then truncated down
to _Float16 immediately after each single arithmetic operation.
Reviewed By: LuoYuanke
Differential Revision: https://reviews.llvm.org/D107082
