The concept of a 'program point' in the original data flow framework is
ambiguous. It can refer to either an operation or a block itself. This
representation has different interpretations in forward and backward
data-flow analysis. In forward data-flow analysis, the program point of
an operation represents the state after the operation, while in backward
data flow analysis, it represents the state before the operation. When
using forward or backward data-flow analysis, it is crucial to carefully
handle this distinction to ensure correctness.
This patch refactors the definition of program point, unifying the
interpretation of program points in both forward and backward data-flow
analysis.
How to integrate this patch?
For dense forward data-flow analysis and other analysis (except dense
backward data-flow analysis), the program point corresponding to the
original operation can be obtained by `getProgramPointAfter(op)`, and
the program point corresponding to the original block can be obtained by
`getProgramPointBefore(block)`.
For dense backward data-flow analysis, the program point corresponding
to the original operation can be obtained by
`getProgramPointBefore(op)`, and the program point corresponding to the
original block can be obtained by `getProgramPointAfter(block)`.
NOTE: If you need to get the lattice of other data-flow analyses in
dense backward data-flow analysis, you should still use the dense
forward data-flow approach. For example, to get the Executable state of
a block in dense backward data-flow analysis and add the dependency of
the current operation, you should write:
``getOrCreateFor<Executable>(getProgramPointBefore(op),
getProgramPointBefore(block))``
In case above, we use getProgramPointBefore(op) because the analysis we
rely on is dense backward data-flow, and we use
getProgramPointBefore(block) because the lattice we query is the result
of a non-dense backward data flow computation.
related dsscussion:
https://discourse.llvm.org/t/rfc-unify-the-semantics-of-program-points/80671/8
corresponding PSA:
https://discourse.llvm.org/t/psa-program-point-semantics-change/81479
Base `DataFlowAnalysis::visit` returns `LogicalResult`, but wrappers's
Sparse/Dense/Forward/Backward `visitOperation` doesn't.
Sometimes it's needed to abort solver early if some unrecoverable
condition detected inside analysis.
Update `visitOperation` to return `LogicalResult` and propagate it to
`solver.initializeAndRun()`. Only `visitOperation` is updated for now,
it's possible to update other hooks like `visitNonControlFlowArguments`,
bit it's not needed immediately and let's keep this PR small.
Hijacked `UnderlyingValueAnalysis` test analysis to test it.
This enum is used by dataflow analyses to indicate whether further
propagation is necessary to reach the fix point. Accidentally discarding
such a value will likely lead to propagation stopping early, leading to
incomplete or incorrect results. The most egregious example is the
duality between `join` on the analysis class, which triggers propagation
internally, and `join` on the lattice class that does not and expects
the caller to trigger it depending on the returned `ChangeResult`.
Currently, data in `AbstractSparseBackwardDataFlowAnalysis` is
considered to flow one-to-one, in order, from the operands of an op
implementing `CallOpInterface` to the arguments of the function it is
calling.
This understanding of the data flow is inaccurate. The operands of such
an op that forward to the function arguments are obtained using a
method provided by `CallOpInterface` called `getArgOperands()`.
This commit fixes this bug by using `getArgOperands()` instead of
`getOperands()` to get the mapping from operands to function arguments
because not all operands necessarily forward to the function arguments
and even if they do, they don't necessarily have to be in the order in
which they appear in the op. The operands that don't get forwarded are
handled by the newly introduced `visitCallOperand()` function, which
works analogous to the `visitBranchOperand()` function.
This fix is also propagated to liveness analysis that earlier relied on
this incorrect implementation of the sparse backward dataflow analysis
framework and corrects some incorrect assumptions made in it.
Extra cleanup: Improved a comment and removed an unnecessary code line.
Signed-off-by: Srishti Srivastava <srishtisrivastava.ai@gmail.com>
Reviewed By: matthiaskramm, jcai19
Differential Revision: https://reviews.llvm.org/D157261
This implication was already done de-facto and there were plenty of users and wrapper functions specifically used to handle the "return-like or RegionBranchTerminatorOpInterface" case. These simply existed due to up until recently missing features in ODS.
With the new capabilities of traits, we can make `ReturnLike` imply `RegionBranchTerminatorOpInterface` and auto generate proper definitions for its methods.
Various occurrences and wrapper methods used for `isa<RegionBranchTerminatorOpInterface>() || hasTrait<ReturnLike>()` have all been removed.
Differential Revision: https://reviews.llvm.org/D157402
Earlier, in the sparse backward dataflow analysis, data from the results
of an op implementing `RegionBranchOpInterface` was considered to flow
into the operands of every op that did not implement the
`RegionBranchTerminatorOpInterface` but was return-like and present
in a region of the former. It was thus also expected that the number of
results of the former be equal to the number of operands in the latter.
This understanding of dataflow is incorrect and thus this expectation is
also not justified. This commit fixes this incorrect understanding.
This commit ensures that these return-like ops are handled just like the
ops implementing the `RegionBranchTerminatorOpInterface`, which means
that, if this op has a region `A` whose successors are regions `B`, `C`,
and `D`, then data flows from the arguments (successor inputs) of `B`,
`C`, and `D` to the corresponding successor operands of this op.
This fix is also propagated to liveness analysis that earlier relied on
this incorrect implementation of the sparse backward dataflow analysis
framework and corrects some incorrect assumptions made in it.
Also cleaned up some unnecessary comments from the test file.
Issue: https://github.com/llvm/llvm-project/issues/64139.
Signed-off-by: Srishti Srivastava <srishtisrivastava.ai@gmail.com>
Reviewed By: jcai19, matthiaskramm, Mogball
Differential Revision: https://reviews.llvm.org/D156376
This commit adds a utility to implement liveness analysis using the
sparse backward data-flow analysis framework. Theoretically, liveness
analysis assigns liveness to each (value, program point) pair in the
program and it is thus a dense analysis. However, since values are
immutable in MLIR, a sparse analysis, which will assign liveness to
each value in the program, suffices here.
Liveness analysis has many applications. It can be used to avoid the
computation of extraneous operations that have no effect on the memory
or the final output of a program. It can also be used to optimize
register allocation. Both of these applications help achieve one very
important goal: reducing runtime.
A value is considered "live" iff it:
(1) has memory effects OR
(2) is returned by a public function OR
(3) is used to compute a value of type (1) or (2).
It is also to be noted that a value could be of multiple types (1/2/3) at
the same time.
A value "has memory effects" iff it:
(1.a) is an operand of an op with memory effects OR
(1.b) is a non-forwarded branch operand and a block where its op could
take the control has an op with memory effects.
A value `A` is said to be "used to compute" value `B` iff `B` cannot be
computed in the absence of `A`. Thus, in this implementation, we say that
value `A` is used to compute value `B` iff:
(3.a) `B` is a result of an op with operand `A` OR
(3.b) `A` is used to compute some value `C` and `C` is used to compute
`B`.
---
It is important to note that there already exists an MLIR liveness
utility here: llvm-project/mlir/include/mlir/Analysis/Liveness.h. So,
what is the need for this new liveness analysis utility being added by
this commit? That need is explained as follows:-
The similarities between these two utilities is that both use the
fixpoint iteration method to converge to the final result of liveness.
And, both have the same theoretical understanding of liveness as well.
However, the main difference between (a) the existing utility and (b)
the added utility is the "scope of the analysis". (a) is restricted to
analysing each block independently while (b) analyses blocks together,
i.e., it looks at how the control flows from one block to the other,
how a caller calls a callee, etc. The restriction in the former implies
that some potentially non-live values could be marked live and thus the
full potential of liveness analysis will not be realised.
This can be understood using the example below:
```
1 func.func private @private_dead_return_value_removal_0() -> (i32, i32) {
2 %0 = arith.constant 0 : i32
3 %1 = arith.addi %0, %0 : i32
4 return %0, %1 : i32, i32
5 }
6 func.func @public_dead_return_value_removal_0() -> (i32) {
7 %0:2 = func.call @private_dead_return_value_removal_0() : () -> (i32, i32)
8 return %0#0 : i32
9 }
```
Here, if we just restrict our analysis to a per-block basis like (a), we
will say that the %1 on line 3 is live because it is computed and then
returned outside its block by the function. But, if we perform a
backward data-flow analysis like (b) does, we will say that %0#1 of line
7 is not live because it isn't returned by the public function and thus,
%1 of line 3 is also not live. So, while (a) will be unable to suggest
any IR optimizations, (b) can enable this IR to convert to:-
```
1 func.func private @private_dead_return_value_removal_0() -> i32 {
2 %0 = arith.constant 0 : i32
3 return %0 : i32
4 }
5 func.func @public_dead_return_value_removal_0() -> i32 {
6 %0 = call @private_dead_return_value_removal_0() : () -> i32
7 return %0 : i32
8 }
```
One operation was removed and one unnecessary return value of the
function was removed and the function signature was modified. This is an
optimization that (b) can enable but (a) cannot. Such optimizations can
help remove a lot of extraneous computations that are currently being
done.
Signed-off-by: Srishti Srivastava <srishtisrivastava.ai@gmail.com>
Reviewed By: matthiaskramm, jcai19
Differential Revision: https://reviews.llvm.org/D153779
