llvm-reduce currently has a file `ReduceDPValues`, which really is
concerned with DbgRecords. Therefore, we rename the file and its
function accordingly.
Patch 1 of 3 to add llvm.dbg.label support to the RemoveDIs project. The
patch stack adds a new base class
-> 1. Add DbgRecord base class for DPValue and the not-yet-added
DPLabel class.
2. Add the DPLabel class.
3. Enable dbg.label conversion and add support to passes.
Patches 1 and 2 are NFC.
In the near future we also will rename DPValue to DbgVariableRecord and
DPLabel to DbgLabelRecord, at which point we'll overhaul the function
names too. The name DPLabel keeps things consistent for now.
LLVM will shortly be able to represent variable locations without
encoding information into intrinsics -- they'll be stored as DPValue
objects instead. We'll still need to be able to llvm-reduce these
variable location assignments just like we can with intrinsics today,
thus, here's an llvm-reduce pass that enumerates and reduces the DPValue
objects.
The test for this is paradoxically written with dbg.value intrinsics:
this is because we're changing all the core parts of LLVM to support
this first, with the textual IR format coming last. Until that arrives,
testing the llvm-reduce'ing of DPValues needs the added test using
intrinsics. We should be able to drop the variable assignment using
%alsoloaded using this method. As with the other llvm-reduce tests, I've
got one set of check lines for making the reduction happen as desired,
and the other set to check the final output.
With the current state of mir support, this is going to generate
a large number of verifier errors. Running the use and def
reductions first helps to mitigate the impact of this.
Reduce by calling the resolver function at the use site, and inserting
an indirect call. Try to delete if there are no uses left over.
We should also probably try to do something about constantexpr uses;
perhaps treat them like aliases.
Main thing I was unsure about was to whether try to delete the now
dead landing blocks, or leave that for the unreachable block reduction.
Personality function is not reduced, but that should be a separate
reduction on the function.
Fixes#58815
The current reduction tries all or nothing elimination of named
metadata. I noticed in one case where one of the module flags was
necessary, but it left the rest. Reduce the individual operands of
named metadata nodes that are known to behave like lists. Be
conservative since some named metadata may have more specific verifier
requirements for the operands.
Previously this would produce many invalid reductions with
"Instruction does not dominate uses" verifier errors.
This fixes issues in cases where the incoming IR
has unreachable blocks, and the resulting reduction
introduced new reachable blocks.
Have basic-blocks skip functions that have unreachable
blocks, Introduce a separate reduction which only
deletes unreachable blocks. Cleanup any newly unreachable
blocks after trimming out the requested deletions.
Includes a variety of meta-reduced tests for llvm-reduce
itself with -abort-on-invalid-reduction that were failing
on different iterations of this patch.
Bugpoint's implementation is much simpler (but currently I don't
understand how it avoids disconnecting interesting blocks from the CFG).
Copy this technique from bugpoint. Before trying to blindly
delete blocks, try to fold branch conditions. This intuitively
makes more sense for a faster reduction, since you can find
dead paths in the function to prune out before trying to bisect
blocks in source order.
Seems to provide some speedup on my multi-hour reduction samples.
This does have the potential to produce testcases with unreachable
blocks. This is already a problem with the existing block
reduction pass. I'm struggling dealing with invalid reductions
in these cases, so in the future this should probably start
deleting those. However, I do sometimes try to reduce failures
in code that becomes unreachable, so I'm not totally sure
what to do here.
Make load and store non-atomic. Make the others monotonic.
We could probably try to incrementally relax the orderings; not sure
how useful that would be.
I often run llvm-reduce on IR that contains debug info, this prevents an
extra step of `opt -passes=strip` I do every time and will result in a
lot less invalid reductions around debug metadata.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D136208
Try some dumb strength reductions to "simpler" opcodes.
Make some opcode substitutions I typically try to get smaller
MIR out of codegen. This is a bit target specific and I have a
lot of increasingly target specific modifications I try
during manual reduction.
Verify all the requested passes exist before trying to run any.
For long reductions, it was really annoying for it to get halfway through
and then I come back later to an incomplete reduction.
Also add a new skip-delta-passes flag. Most of the time I want to opt out
of specific reductions, rather than run a select few.
This new pass for llvm-reduce attempts to reduce DebugInfo metadata.
The process used is:
1. Scan every MD node, keeping track of nodes already visited.
2. Look for DebugInfo nodes, then record any operands that are lists.
3. Bisect though all the elements of the collected lists.
Differential Revision: https://reviews.llvm.org/D132077
so that we can reduce away incidental parts of the CFG in cases where
the full simplifyCFG pass makes the test case uninteresting
Differential Revision: https://reviews.llvm.org/D131920
Try to insert an implicit_def to replace the instruction's value,
replacing the original instruction's def with a dead register. If all
defs are delete the instruction entirely.
This is pretty similar to the instruction reduction, but leaves the
new defs in the same place as the original instruction. This could
possibly replace it. I'm not sure if we should directly delete the
instructions here, or leave dead ones behind.
This could also further work to replace physical register defs.
I have a register allocator failure that only reproduces with IPRA
enabled, and requires the specific regmask if I want to only run the
one relevant pass. The printed custom regmask is enormous and I would
like to reduce it.
This reduces each individual bit in the mask, but it would probably be
better to start at register units and clear all aliasing fields at a
time. This would require stricter verification that all aliasing bits
are set in regmasks (although I would prefer to switch regmasks to use
register units in the first place).
I'm a bit confused by what's actually stored for the allocation
hints. The MIR parser only handles the "simple" case where there's a
single hint. I don't really understand the assertion in
clearSimpleHint, or under what circumstances there are multiple hint
registers.
Many MIR reductions benefit from or require increasing the instruction
count. For example, unlike in the IR, you may need to insert a new
instruction to represent an undef. The current instruction reduction
pass works around this by sticking implicit defs on whatever
instruction happens to be first in the entry block block.
Other strategies I've applied manually include breaking instructions
with multiple defs into separate instructions, or breaking large
register defs into multiple subregister defs.
Make up a simple scoring system based on what I generally try to get
rid of first when manually reducing. Counts implicit defs as free
since reduction passes will be introducing them, although they
probably should count for something. It also might make more sense to
have a comparison the two functions, rather than having to compute a
contextless number. This isn't particularly well tested since overall
the MIR support isn't in a place where it is useful on the kinds of
testcases I want to throw at it.
Previously the options category given to cl::HideUnrelatedOptions was
local to llvm-reduce.cpp and as a result only options declared in that
file were visible in the -help options listing. This was a bit
unfortunate since there were several useful options declared in other
files. This patch addresses that.
Differential Revision: https://reviews.llvm.org/D118682
Add a new "operands-skip" pass whose goal is to remove instructions in the middle of dependency chains. For instance:
```
%baseptr = alloca i32
%arrayidx = getelementptr i32, i32* %baseptr, i32 %idxprom
store i32 42, i32* %arrayidx
```
might be reducible to
```
%baseptr = alloca i32
%arrayidx = getelementptr ... ; now dead, together with the computation of %idxprom
store i32 42, i32* %baseptr
```
Other passes would either replace `%baseptr` with undef (operands, instructions) or move it to become a function argument (operands-to-args), both of which might fail the interestingness check.
In principle the implementation allows operand replacement with any value or instruction in the function that passes the filter constraints (same type, dominance, "more reduced"), but is limited in this patch to values that are directly or indirectly used to compute the current operand value, motivated by the example above. Additionally, function arguments are added to the candidate set which helps reducing the number of relevant arguments mitigating a concern of too many arguments mentioned in https://reviews.llvm.org/D110274#3025013.
Possible future extensions:
* Instead of requiring the same type, bitcast/trunc/zext could be automatically inserted for some more flexibility.
* If undef is added to the candidate set, "operands-skip"is able to produce any reduction that "operands" can do. Additional candidates might be zero and one, where the "reductive power" classification can prefer one over the other. If undefined behaviour should not be introduced, undef can be removed from the candidate set.
Recommit after resolving conflict with D112651 and reusing
shouldReduceOperand from D113532.
Reviewed By: aeubanks
Differential Revision: https://reviews.llvm.org/D111818
This reverts commit fa4210a9a0729eba04593b7df7b701e2b243de39.
It causes compile failures, presumably because conflicting with another
patch landed after I checked locally.
Add a new "operands-skip" pass whose goal is to remove instructions in the middle of dependency chains. For instance:
```
%baseptr = alloca i32
%arrayidx = getelementptr i32, i32* %baseptr, i32 %idxprom
store i32 42, i32* %arrayidx
```
might be reducible to
```
%baseptr = alloca i32
%arrayidx = getelementptr ... ; now dead, together with the computation of %idxprom
store i32 42, i32* %baseptr
```
Other passes would either replace `%baseptr` with undef (operands, instructions) or move it to become a function argument (operands-to-args), both of which might fail the interestingness check.
In principle the implementation allows operand replacement with any value or instruction in the function that passes the filter constraints (same type, dominance, "more reduced"), but is limited in this patch to values that are directly or indirectly used to compute the current operand value, motivated by the example above. Additionally, function arguments are added to the candidate set which helps reducing the number of relevant arguments mitigating a concern of too many arguments mentioned in https://reviews.llvm.org/D110274#3025013.
Possible future extensions:
* Instead of requiring the same type, bitcast/trunc/zext could be automatically inserted for some more flexibility.
* If undef is added to the candidate set, "operands-skip"is able to produce any reduction that "operands" can do. Additional candidates might be zero and one, where the "reductive power" classification can prefer one over the other. If undefined behaviour should not be introduced, undef can be removed from the candidate set.
Reviewed By: aeubanks
Differential Revision: https://reviews.llvm.org/D111818
(Second try. Need to link against CodeGen and MC libs.)
The llvm-reduce tool has been extended to operate on MIR (import, clone and
export). Current limitation is that only a single machine function is
supported. A single reducer pass that operates on machine instructions (while
on SSA-form) has been added. Additional MIR specific reducer passes can be
added later as needed.
Differential Revision: https://reviews.llvm.org/D110527
The llvm-reduce tool has been extended to operate on MIR (import, clone and
export). Current limitation is that only a single machine function is
supported. A single reducer pass that operates on machine instructions (while
on SSA-form) has been added. Additional MIR specific reducer passes can be
added later as needed.
Differential Revision: https://reviews.llvm.org/D110527
Having non-undef constants in a final llvm-reduce output is nicer than
having undefs.
This splits the existing reduce-operands pass into three, one which does
the same as the current pass of reducing to undef, and two more to
reduce to the constant 1 and the constant 0. Do not reduce to undef if
the operand is a ConstantData, and do not reduce 0s to 1s.
Reducing GEP operands very frequently causes invalid IR (since types may
not match up if we index differently into a struct), so don't touch GEPs.
Reviewed By: Meinersbur
Differential Revision: https://reviews.llvm.org/D111765
Instead of setting operands to undef as the "operands" pass does,
convert the operands to a function argument. This avoids having to
introduce undef values into the IR which have some unpredictability
during optimizations.
For instance,
define void @func() {
entry:
%val = add i32 32, 21
store i32 %val, i32* null
ret void
}
is reduced to
define void @func(i32 %val) {
entry:
%val1 = add i32 32, 21
store i32 %val, i32* null
ret void
}
(note that the instruction %val is renamed to %val1 when printing
the IR to avoid ambiguity; ideally %val1 would be removed by dce or the
instruction reduction pass)
Any call to @func is replaced with a call to the function with the
new signature and filled with undef. This is not ideal for IPA passes,
but those out-of-scope for now.
Reviewed By: aeubanks
Differential Revision: https://reviews.llvm.org/D111503
This removes the data layout, target triple, source filename, and module
identifier when possible.
Reviewed By: swamulism
Differential Revision: https://reviews.llvm.org/D108568
