Reviewing the code with Arnold revealed a corner case where forward propagating
a copy into multiple operands of the same instruction wasn't properly detected.
I don't think this case was possible given the language rules, but nonetheless
it is valid SIL and needs to be handled.
This is confusing because in some cases we optimize that situation correctly,
and in other cases we try to assert that it doesn't happen. So, I simplified the
code to bailout anywhere that we see multiple operands of the same value. This
isn't an expected situation that needs to be optimized.
Copy forwarding was designed with some assumptions about symmetry of
operations. If copy_value/destroy_value are expanded somewhere for a given
value, then they should be expanded everywhere. The pass took a conservative
approach to SIL patterns that were guaranteed safe, and bailed out on unknown
patterns. However, due to some over-aggressive code factoring, the assumption
wasn't being checked in one corner case.
This redesign makes a clear distinction between the requirements for forward
vs. backward propagation.
Fixes <rdar://35646292> Swift CI: resilience bot seg faults in
stdlib/RangeReplaceable.swift.gyb
We can just !SILFunction::hasQualifiedOwnership(). Plus as Andy pointed out,
even ignoring the functional aspects, having APIs with names this close can
create confusion.
We don't have a way to represent a cast in SIL where the destination
type introduces new archetypes that depend on the input value, sort
of like opening an existential.
Previously we would only rule out the case where the class itself
was generic, but a class can also be nested inside another generic
context.
Fixes <rdar://problem/35478838>.
Adds a combined API to output both debug message and optimization remarks.
The previously added test partial_specialization_debug.sil ensures that it's an
NFC for debug output.
For now these are underscored attributes, i.e. compiler internal attributes:
@_optimize(speed)
@_optimize(size)
@_optimize(none)
Those attributes override the command-line specified optimization mode for a specific function.
The @_optimize(none) attribute is equivalent to the already existing @_semantics("optimize.sil.never") attribute
This commit is mostly refactoring.
*) Introduce a new OptimizationMode enum and use that in SILOptions and IRGenOptions
*) Allow the optimization mode also be specified for specific SILFunctions. This is not used in this commit yet and thus still a NFC.
Also, fixes a minor bug: we didn’t run mandatory IRGen passes for functions with @_semantics("optimize.sil.never")
This brings the capability from clang to save remarks in an external YAML files.
YAML files can be viewed with tools like the opt-viewer.
Saving the remarks is activated with the new option -save-optimization-record.
Similarly to -emit-tbd, I've only added support for single-compile mode for now.
In this case the default filename is determined by
getOutputFilenameFromPathArgOrAsTopLevel, i.e. unless explicitly specified
with -save-optimization-record-path, the file is placed in the directory of the
main output file as <modulename>.opt.yaml.
I am doing this to ensure that parts of the optimizer that have not yet been
updated for such operations do not see said instructions. This will ensure no
surprise perf regressions from this work.
rdar://31521023
This allows reporting successful and unsuccessful optimizations similar to
clang/llvm.
This first patch adds support for the
options -Rpass=<pass-name-regex> -Rpass-missed=<pass-name-regex>. These allow
reporting successful/unsuccessful optimization on the compiler output for passes
specified by the regex. I've also added one missed and one passed remark type
to the inliner to test the infrastructure.
Clang also has the option of collecting these records in an external YAML data
file. This will be added in a later patch.
A few notes:
* The goal is to use this facility for both user-lever "performance" warnings
and expert-level performance analysis. There will probably be a flag in the
future differentiating the verbosity.
* The intent is match clang/llvm as much as it makes sense. On the other hand I
did make some changes. Unlike in llvm, the emitter is not a pass which
simplifies things. Also the remark class hierarchy is greatly simplified since
we don't derive from DiagnosticInfo. We also don't derive from Diagnostic to
support the streaming API for arbitrary named-value pairs.
* Currently function names are printed mangled which should be fixed.
Support for @noescape SILFunctionTypes.
These are the underlying SIL changes necessary to implement the new
closure capture ABI.
Note: This includes a change to function name mangling that
primarily affects reabstraction thunks.
The new ABI will allow stack allocation of non-escaping closures as a
simple optimization.
The new ABI, and the stack allocation optimization, also require
closure context to be @guaranteed. That will be implemented as the
next step.
Many SIL passes pattern match partial_apply sequences. These all
needed to be fixed to handle the convert_function that SILGen now
emits. The conversion is now needed whenever a function declaration,
which has an escaping type, is passed into a @NoEscape argument.
In addition to supporting new SIL patterns, some optimizations like
inlining and SIL combine are now stronger which could perturb some
benchmark results.
These underlying SIL changes should be merged now to avoid conflicting
with other work. Minor benchmark discrepancies can be investigated as part of
the stack-allocation work.
* Add a noescape attribute to SILFunctionType.
And set this attribute correctly when lowering formal function types to SILFunctionTypes based on @escaping.
This will allow stack allocation of closures, and unblock a related ABI change.
* Flip the polarity on @noescape on SILFunctionType and clarify that
we don't default it.
* Emit withoutActuallyEscaping using a convert_function instruction.
It might be better to use a specialized instruction here, but I'll leave that up to Andy.
Andy: And I'll leave that to Arnold who is implementing SIL support for guaranteed ownership of thick function types.
* Fix SILGen and SIL Parsing.
* Fix the LoadableByAddress pass.
* Fix ClosureSpecializer.
* Fix performance inliner constant propagation.
* Fix the PartialApplyCombiner.
* Adjust SILFunctionType for thunks.
* Add mangling for @noescape/@escaping.
* Fix test cases for @noescape attribute, mangling, convert_function, etc.
* Fix exclusivity test cases.
* Fix AccessEnforcement.
* Fix SILCombine of convert_function -> apply.
* Fix ObjC bridging thunks.
* Various MandatoryInlining fixes.
* Fix SILCombine optimizeApplyOfConvertFunction.
* Fix more test cases after merging (again).
* Fix ClosureSpecializer. Hande convert_function cloning.
Be conservative when combining convert_function. Most of our code doesn't know
how to deal with function type mismatches yet.
* Fix MandatoryInlining.
Be conservative with function conversion. The inliner does not yet know how to
cast arguments or convert between throwing forms.
* Fix PartialApplyCombiner.
Except GenericEnvironment.h, because you can't meaningfully use a
GenericEnvironment without its signature. Lots less depends on
GenericSignature.h now. NFC
This replaces the '[volatile]' flag. Now, class_method and
super_method are only used for vtable dispatch.
The witness_method instruction is still overloaded for use
with both ObjC protocol requirements and Swift protocol
requirements; the next step is to make it only mean the
latter, also using objc_method for ObjC protocol calls.
introduce a common superclass, SILNode.
This is in preparation for allowing instructions to have multiple
results. It is also a somewhat more elegant representation for
instructions that have zero results. Instructions that are known
to have exactly one result inherit from a class, SingleValueInstruction,
that subclasses both ValueBase and SILInstruction. Some care must be
taken when working with SILNode pointers and testing for equality;
please see the comment on SILNode for more information.
A number of SIL passes needed to be updated in order to handle this
new distinction between SIL values and SIL instructions.
Note that the SIL parser is now stricter about not trying to assign
a result value from an instruction (like 'return' or 'strong_retain')
that does not produce any.
So far we did this only for loop headers, but there are cases where we need it for all blocks to prevent infinite loops in SimplifyCFG.
rdar://problem/34576609
The reason to do this is:
1. The check in SILInliner if we can inline can be done without triggering
side-effects.
2. This enables us to know if inlining will succeed before attempting to inline.
This enables for arguments to be adjusted with new SILInstructions and the like
before inlining occurs. I use this in a forthcoming patch that updates mandatory
inlining for ownership.
rdar://31521023
