between the two instructions.
If there's a pattern like:
$xA = ADRP foo @PAGE
[some killing use of reg Xb]
$Xb = ADDXri $Xa, 0, @PAGEOFF
CollectLOH would create an AdrpAdd LOH that resulted in the linker optimizing
this sequence into:
$xB = ADR foo
[some killing use of reg $Xb]
... and therefore clobbers the live $Xb register that was used by the
instruction in between.
This was discovered by a GlobalISel patch D78465 which broke up global variable
accesses into two pseudos, which in some cases could be moved apart.
Differential Revision: https://reviews.llvm.org/D80834
For most tables, we already use commas in headers. This set of patches
unifies dumping the remaining ones.
Differential Revision: https://reviews.llvm.org/D80806
For most tables, we already use commas in headers. This set of patches
unifies dumping the remaining ones.
Differential Revision: https://reviews.llvm.org/D80806
When a stack offset was too big to materialize in a single instruction, we were
trying to do it in stages:
adds xD, sp, #imm
adds xD, xD, #imm
Unfortunately, if xD is xzr then the second instruction doesn't exist and
wouldn't do what was needed if it did. Instead we can use a temporary register
for all but the last addition.
Summary: Exploit vabsd* for for absolute difference of vectors on P9,
for example:
void foo (char *restrict p, char *restrict q, char *restrict t)
{
for (int i = 0; i < 16; i++)
t[i] = abs (p[i] - q[i]);
}
this case should be matched to the HW instruction vabsdub.
Reviewed By: steven.zhang
Differential Revision: https://reviews.llvm.org/D80271
Previously we walked the users of any vector binop looking for
more binops with the same opcode or phis that eventually ended up
in a reduction. While this is simple it also means visiting the
same nodes many times since we'll do a forward walk for each
BinaryOperator in the chain. It was also far more general than what
we have tests for or expect to see.
This patch replaces the algorithm with a new method that starts at
extract elements looking for a horizontal reduction. Once we find
a reduction we walk through backwards through phis and adds to
collect leaves that we can consider for rewriting.
We only consider single use adds and phis. Except for a special
case if the Add is used by a phi that forms a loop back to the
Add. Including other single use Adds to support unrolled loops.
Ultimately, I want to narrow the Adds, Phis, and final reduction
based on the partial reduction we're doing. I still haven't
figured out exactly what that looks like yet. But restricting
the types of graphs we expect to handle seemed like a good first
step. As does having all the leaves and the reduction at once.
Differential Revision: https://reviews.llvm.org/D79971
This matches what we do for the full sized vector ops at the start of combineX86ShufflesRecursively, and helps getFauxShuffleMask extract more INSERT_SUBVECTOR patterns.
I inverted the mask when I ported to the new form of G_PTRMASK in
8bc03d2168.
I don't think this really broke anything, since G_VASTART isn't
handled for types with an alignment higher than the stack alignment.
The instruction is defined to only produce high result if both
destinations are the same. We can exploit this to avoid
unnecessarily clobbering a register.
In order to hide this from register allocation we use a pseudo
instruction and expand the result during MCInst creation.
Differential Revision: https://reviews.llvm.org/D80500
Treat it as callee-saved, and always back it up. When windows code calls
entry points in unix code, marked with the windows calling convention,
that unix code can call other functions that isn't compiled with
-ffixed-x18 which may clobber x18 freely. By backing it up and restoring
it on return, we preserve the register across the function call,
fulfilling this part of the windows calling convention on another OS.
This isn't enough for making sure that x18 is preseved when non-windows
code does a callback to windows code, but is a clear improvement over
the current status quo. Additionally, wine is nowadays building many
modules as PE DLLs, which avoids the callback issue altogether for those
DLLs.
Differential Revision: https://reviews.llvm.org/D61892
This is a custom inserter because it was less work than teaching
tablegen a way to indicate that it is sometimes OK to have a no side
effect instruction in the output of a side effecting pattern.
The asm is needed to look like a read of the mode register to prevent
it from being deleted. However, there seems to be a bug where the mode
register def instructions are moved across the asm sideeffect by the
post-RA scheduler.
Another oddity is the immediate is formatted differently between
s_denorm_mode and s_round_mode.
Let the codegen recognized the nomerge attribute and disable branch folding when the attribute is given
Differential Revision: https://reviews.llvm.org/D79537
Large code model doesn't mean anything to 32-bit mode. But nothing
prevents it from being set. Ignore to avoid generating 64-bit mode
only instructions.
Differential Revision: https://reviews.llvm.org/D80768
Only 64-bit bits will be loaded, not the whole 128 bits. We can
just combine it to plain mmx load. This has the side effect of
enabling isel load folding for it.
This part of my desire to get rid of isel patterns that shrink loads.
Since on AIX, our strategy is to not use -u to suppress any undefined
symbols, we need to emit .extern for the symbols with AvailableExternally
linkage.
Differential Revision: https://reviews.llvm.org/D80642
optimizations
As discussed in the thread http://lists.llvm.org/pipermail/llvm-dev/2020-May/141838.html,
some bit field access width can be reduced by ReduceLoadOpStoreWidth, some
can't. If two accesses are very close, and the first access width is reduced,
the second is not. Then the wide load of second access will be stalled for long
time.
This patch add command line options to guard ReduceLoadOpStoreWidth and
ShrinkLoadReplaceStoreWithStore, so users can use them to disable these
store width reduction optimizations.
Differential Revision: https://reviews.llvm.org/D80745
These patterns for i8 and i16 VMLA's were missing. They end up from
legalized vector.reduce.add.v8i16 and vector.reduce.add.v16i8, and
although the instruction works differently (the mul and add are
performed in a higher precision), I believe it is OK because only an
i8/i16 are demanded from them, and so the results will be the same. At
least, they pass any testing I can think to run on them.
There are some tests that end up looking worse, but are quite artificial
due to passing half vector types through a call boundary. I would not
expect the vmull to realistically come up like that, and a vmlava is
likely better a lot of the time.
Differential Revision: https://reviews.llvm.org/D80524
Currently combineInsertEltToShuffle turns insert_vector_elt into a
vector_shuffle, even if the inserted element is a vector with a single
element. In this case, it should be unlikely that the additional shuffle
would be more efficient than a insert_vector_elt.
Additionally, this fixes a infinite cycle in DAGCombine, where
combineInsertEltToShuffle turns a insert_vector_elt into a shuffle,
which gets turned back into a insert_vector_elt/extract_vector_elt by
a custom AArch64 lowering (in visitVECTOR_SHUFFLE).
Such insert_vector_elt and extract_vector_elt combinations can be
lowered efficiently using mov on AArch64.
There are 2 test changes in arm64-neon-copy.ll: we now use one or two
mov instructions instead of a single zip1. The reason that we need a
second mov in ins1f2 is that we have to move the result to the result
register and is not really related to the DAGCombine fold I think.
But in any case, on most uarchs, mov should be cheaper than zip1. On a
Cortex-A75 for example, zip1 is twice as expensive as mov
(https://developer.arm.com/docs/101398/latest/arm-cortex-a75-software-optimization-guide-v20)
Reviewers: spatel, efriedma, dmgreen, RKSimon
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D80710
AddressingModeMatcher::matchScaledValue was calling getSExtValue for a constant before ensuring that we can actually represent the value as int64_t
Fixes OSSFuzz#22723 which is a followup to rGc479052a74b2 (PR46004 / OSSFuzz#22357)
I have tried to ensure that SelectionDAG and DAGCombiner do
sensible things for scalable vectors, and added support for a
limited number of simple folds. Codegen support for the vector
extract patterns have also been added to the AArch64 backend.
New vector extract tests have been added here:
CodeGen/AArch64/sve-extract-element.ll
and I have also added new folds using inserts and extracts here:
CodeGen/AArch64/sve-insert-element.ll
Differential Revision: https://reviews.llvm.org/D80208
Amara Emerson