There is support for intrinsics in Instruction::isCommunative, but there
is no equivalent implementation for isAssociative. This patch builds
support for associative intrinsics with TRE as an application. TRE can
now have associative intrinsics as an accumulator. For example:
```
struct Node {
Node *next;
unsigned val;
}
unsigned maxval(struct Node *n) {
if (!n) return 0;
return std::max(n->val, maxval(n->next));
}
```
Can be transformed into:
```
unsigned maxval(struct Node *n) {
struct Node *head = n;
unsigned max = 0; // Identity of unsigned std::max
while (true) {
if (!head) return max;
max = std::max(max, head->val);
head = head->next;
}
return max;
}
```
This example results in about 5x speedup in local runs.
We conservatively only consider min/max and as associative for this
patch to limit testing scope. There are probably other intrinsics that
could be considered associative. There are a few consumers of
isAssociative() that could be impacted. Testing has only required to
Reassociate pass be updated.
When all the large const offsets masked with the same value from bit-12 to bit-23.
Fold
add x8, x0, #2031, lsl #12
add x8, x8, #960
ldr x9, [x8, x8]
ldr x8, [x8, #2056]
into
add x8, x0, #2031, lsl #12
ldr x9, [x8, #960]
ldr x8, [x8, #3016]
MinBWs contains entries that specify the minimum required bitwidth. In
some cases, the old and new bitwidths can be equal (see test case) and
in those cases no truncations are needed, so skip those cases.
Fixes#74307.
Check the operands of I are used in no more than one place, which can
not be deleted, cause a mul instruction has far more weight than add and
shl instruction in IR, thus this method cannot achieve the goal of
simplifying instructions, just return null.
Clean up sve-tail-folding-option.ll by removing the unused
CHECK-TF-NEOVERSE-V1 prefix (note the use of non-opaque pointers) and
remove IR value references.
In some cases MinBWs may contain entries for live-ins that are not used
by VPWidenRecipe or VPWidenSelectRecipes. In those cases, the live-ins
won't get processed, so make sure we include them in the count when used
as operands in VPWidenCast and VPWidenSelectRecipe.
Fixes https://github.com/llvm/llvm-project/issues/74231
We were able to fold
or (mul X, Y), X --> mul X, (add Y, 1) (when the multiply has no common
bits with X)
This patch makes the transform work if the mul operands are commuted.
The disjoint flag was recently added to IR in #72583
We already set it when we turn an add into an or. This patch sets it on Ors that weren't converted from an Add.
When using `BranchProbabilityPrinterPass`, if a BB has no name, we get pretty unusable information like `edge -> has probability...` (i.e. we have no idea what the vertices of that edge are).
This patch uses `printAsOperand`, which uses the same naming scheme as `Function::dump`, so for example during debugging sessions, the IR obtained from a function and the names used by `BranchProbabilityPrinterPass` will match.
A shortcoming is that `printAsOperand` will result in the numbering algorithm re-running for every edge and every vertex (when `BranchProbabilityPrinterPass` is run on a function). If, for the given scenario, this is a problem, we can revisit this subsequently.
Another nuance is that the entry basic block will be numbered, which may be slightly confusing when it's anonymous, but it's easily identifiable - the first edge would have it as source (and the number should be easily recognizable)
For example, this allows us to peel this loop with a `and`:
```
for (int i = 0; i < N; ++i) {
if (i % 2 == 0 && i < 3) // can peel based on || as well
f1();
f2();
```
into:
```
for (int i = 0; i < 3; ++i) { // peel three iterations
if (i % 2 == 0)
f1();
f2();
}
for (int i = 3; i < N; ++i)
f2();
```
This patch replaces the IR based truncateToMinimalBitwidths with a VPlan
version. This has 3 benefits:
1) the VPlan-based version is simpler; we don't need to implement
special codegen for each supported instruction type like the IR based
one.
2) Removes a dependency on the cost-model after VPlan execution and
3) Removes a use of getVPValue that uses underlying values after VPlan
execution (See removed FIXME).
Depends on D149081.
Depends on D149079.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D149903
After inserting a select for the final value, update the VPlan def-use
chains. At the moment, the incorrect live-out doesn't cause a
mis-compile, as computing the final reduction value is not yet modeled
in VPlan.
Currently we emit gathers for scalars being vectorized in the tree as
a pair of extractelement/insertelement instructions. Instead we can try
to find all required vectors and emit shuffle vector instructions
directly, improving the code and reducing compile time.
Part of non-power-of-2 vectorization.
Differential Revision: https://reviews.llvm.org/D110978
My initial patch contained a typo, resulting in the wrong value
being checked for non-negativeness.
-----
If the lshr operand is non-negative, we can treat it the same
way as an ashr. Ideally we would represent this as "lshr nneg",
but for now just perform the necessary ValueTracking query.
Proof: https://alive2.llvm.org/ce/z/Ahg4ri
If the lshr operand is non-negative, we can treat it the same
way as an ashr. Ideally we would represent this as "lshr nneg",
but for now just perform the necessary ValueTracking query.
Proof: https://alive2.llvm.org/ce/z/Ahg4ri
If both icmps have the same operands and the RHS is constant, we
would currently go into the isImpliedCondMatchingOperands() code
path, instead of the isImpliedCondCommonOperandWithConstants()
path. Both are correct, but the latter can produce more accurate
results if the implication is dependent on the sign.
This reverts commit 96a0d714d58e48c363ee6abbbcdfd7a6ce646ac1.
Avoid assert with dynamic denormal-fp-math We don't recognize compares
with 0 as an exact class test if we don't know the denormal mode. We could
try to do better here, but it's probably not worth it.
Fixes asserts reported after 1adce7d8e47e2438f99f91607760b825e5e3cc37
Now MemProf can't do IR annotation right in the local linkage function
and global initial function __cxx_global_var_init. In llvm-profdata
which convert raw memory profile to memory profile, it uses function
name in dwarf to create GUID. But when llvm consumes memory profile, it
use `getIRPGOFuncName` or `getPGOFuncName` which returns local linkage
function as `FileName;FunctionName` or `FileName:FunctionName` to get
function name and create GUID. So profile creator's GUID is not same as
profile consumer.
So I think MemProf should be used with `unique-internal-linkage-names`
and don't use PGOFuncName.
__cxx_global_var_init is created later than where
UniqueInternalLinkageNames works. So I add uniq suffix to
__cxx_global_var_init additionally.
Co-authored-by: lifengxiang <lifengxiang.1025@bytedance.com>
https://github.com/llvm/llvm-project/issues/70703 pointed out that
cloning LLVM modules could lead to miscompiles when using FatLTO.
This is due to an existing issue when cloning modules with labels (see
#55991 and #47769). Since this can lead to miscompilation, we can avoid
cloning the LLVM modules, which was desirable anyway.
This patch modifies the EmbedBitcodePass to no longer clone the module
or run an input pipeline over it. Further, it make FatLTO always perform
UnifiedLTO, so we can still defer the Thin/Full LTO decision to
link-time. Lastly, it removes dead/obsolete code related to now defunct
options that do not work with the EmbedBitcodePass implementation any
longer.
If a suspend happens in the resume part (this can happen in the case of chained coroutines), and that's part of a loop, the pre-split CFG has the suspend block as an exit of that loop. PGO Counter Promotion will then try to commit the temporary counter to the global in that "exit" block (it also does that in the other loop exit BBs, which also includes
the "destroy" case). This interferes with symmetric transfer.
We don't need to commit the counter in the suspend case - it's not a loop exit from the perspective of the behavior of the program. The regular loop exit, together with the "destroy" case, completely cover any updates that may need to happen to the global counter.
This reverts commit a4d5fd4d2ee9470e55345a9540f6b6fb6faf66e1.
The above commit breaks greendragon lldb bots:
Link to failing builds:
https://green.lab.llvm.org/green/view/LLDB/job/lldb-cmake/63300/https://green.lab.llvm.org/green/view/LLDB/job/as-lldb-cmake/10345/
I found this PR to be the offending one after using git bisect with the
cmake invocation:
cmake -G Ninja ../llvm -DCMAKE_BUILD_TYPE=Release
-DCMAKE_EXPORT_COMPILE_COMMANDS=ON
'-DLLVM_TARGETS_TO_BUILD=X86;ARM;AArch64'
-DLLVM_ENABLE_ASSERTIONS:BOOL=TRUE
-DLLVM_ENABLE_MODULES=On
-DLLVM_ENABLE_PROJECTS='clang;lld;lldb;cross-project-tests'
-DLLVM_VERSION_PATCH=99
'-DLLVM_ENABLE_RUNTIMES=libcxx;libcxxabi;compiler-rt'
and running
ninja lib/CodeGen/CMakeFiles/LLVMCodeGen.dir/CodeGenPassBuilder.cpp.o
This patch updates the last few places in LLVM using findDbgValues that
don't also collect and handle DPValue objects. This largely involves
instcombine and mem2reg changes, and are largely mechanical, calling
existing utilities on collections of DPValues instead of just
DbgValuesInsts.
A variety of tests have had RemoveDIs RUN lines added to them to cover
these behaviours. We have some technical debt of the instcombine sinking
code for DPValues not being implemented yet, so I've left FIXME stubs
indicating that we intend to cover tests with RemoveDIs but haven't yet.
CodeGenPrepare needs to support the maintenence of DPValues, the
non-instruction replacement for dbg.value intrinsics. This means there are
a few functions we need to duplicate or replicate the functionality of:
* fixupDbgValue for setting users of sunk addr GEPs,
* The remains of placeDbgValues needs a DPValue implementation for sinking
* Rollback of RAUWs needs to update DPValues
* Rollback of instruction removal needs supporting (see github #73350)
* A few places where we have to use iterators rather than instructions.
There are three places where we have to use the setHeadBit call on
iterators to indicate which portion of debug-info records we're about to
splice around. This is because CodeGenPrepare, unlike other optimisation
passes, is very much concerned with which block an operation occurs in and
where in the block instructions are because it's preparing things to be in
a format that's good for SelectionDAG.
There isn't a large amount of test coverage for debuginfo behaviours in
this pass, hence I've added some more.
Currently, MinBWs map uses Value* as a key and stores mapping for each
value to be demoted. It make is it hard to get the actual MinBWs value
for the buildvector scalars(constants), since same constant might be
used in different nodes with the different MinBWs values/decisions.
Also, it consumes extra memory for the vectorized values/instructions
from the same nodes.
Better to map actual nodes. It fixes the bitwidth data fetching for
buildvector scalars and improves memory consumption/analysis time for
other instructions.
Because we're storing some extra debug-info information in the iterator
class, we need to insert new LICM-created stores using such iterators.
Switch LICM to storing iterators instead of pointers when it promotes
variables in loops, add a test for the desired behaviour, and enable
RemoveDIs instrumentation on a variety of other LICM tests for good
measure.
(This would appear to be the only pass in LLVM that needs to store
iterators on the heap).
