The PassManager should transform all functions in bottom up order.
This is necessary because when optimizations like inlining looks at the
callee function bodies to compute profitability, the callee functions
should have already undergone optimizations to get better profitability
estimates.
The PassManager builds its function worklist based on bottom up order
on initialization. However, newly created SILFunctions due to
specialization etc, are simply appended to the function worklist. This
can cause us to make bad inlining decisions due to inaccurate
profitability estimates. This change now updates the function worklist such
that, all the callees of the newly added SILFunction are proccessed
before it by the PassManager.
Fixes rdar://52202680
* Fix the mid-level pass pipeline.
Module passes need to be in a separate pipeline, otherwise the
pipeline restart mechanism will be broken.
This makes GlobalOpt and serialization run earlier in the
pipeline. There's no explicit reason for them to be run later, in the
middle of a function pass pipeline.
Also, pipeline boundaries, like serialization and module passes should
be explicit at the the top level function that creates the pass
pipelines.
* SILOptimizer: Add enforcement of function-pass pipelines.
Don't allow module passes to be inserted within a function pass
pipeline. This silently breaks the function pipeline both interfering
with analysis and the normal pipeline restart mechanism.
* Add misssing pass in addFunctionPasses
Co-authored-by: Andrew Trick <atrick@apple.com>
-sil-verify-all flag will verify analyses before and after a pass to
confirm correct invalidations. But if an analysis was never
constructed or invalidated as per current pass order,
it may never detect insufficient invalidations.
-sil-verify-force-analysis will force construct an analysis so that we
can better check for insufficient invalidations.
It is also terribly slow compared to -sil-verify-all.
A request is intended to be a pure function of its inputs. That function could, in theory, fail. In practice, there were basically no requests taking advantage of this ability - the few that were using it to explicitly detect cycles can just return reasonable defaults instead of forwarding the error on up the stack.
This is because cycles are checked by *the Evaluator*, and are unwound by the Evaluator.
Therefore, restore the idea that the evaluate functions are themselves pure, but keep the idea that *evaluation* of those requests may fail. This model enables the best of both worlds: we not only keep the evaluator flexible enough to handle future use cases like cancellation and diagnostic invalidation, but also request-based dependencies using the values computed at the evaluation points. These aforementioned use cases would use the llvm::Expected interface and the regular evaluation-point interface respectively.
Introduce evaluator::SideEffect, the type of a request that performs
some operation solely to execute its side effects. Thankfully, there are
precious few requests that need to use this type in practice, but it's
good to call them out explicitly so we can get around to making them
behave much more functionally in the future.
Add ExecuteSILPipelineRequest which executes a
pipeline plan on a given SIL (and possibly IRGen)
module. This serves as a top-level request for
the SILOptimizer that we'll be able to hang
dependencies off.
Rather than registering individual IRGen passes
when we want to execute them, store function
pointers to all the pass constructors on the
ASTContext. This will make it easier to requestify
the execution of pass pipelines.
We've been running doxygen with the autobrief option for a couple of
years now. This makes the \brief markers into our comments
redundant. Since they are a visual distraction and we don't want to
encourage more \brief markers in new code either, this patch removes
them all.
Patch produced by
for i in $(git grep -l '\\brief'); do perl -pi -e 's/\\brief //g' $i & done
I believe that these were in SILInstruction for historic reasons. This is a
separate API on top of SILInstruction so it makes sense to pull it out into its
own header.
This helps speed up triaging failures caught by -sil-verify-all since this
allows one to trigger the -sil-verify-all verification around specific passes in
the pipeline rather than after every pass run. Was useful for me when tracking
down missing pass manager notification.
This is a verification routine that is only invoked in PassManager
destructors. I am going to use this to ensure that the
PassManagerVerifierAnalysis only runs at such points (even when sil-verify-all
is enabled) since it is too expensive to run otherwise.
NOTE: The default implementation of verifyFull in this commit is a no-op. I
wanted to have verify() be the default implementation of verifyFull(), but I do
not have more time to invest in this and it seems to catch /real/ bugs, albeit
bugs unrelated to pass manager notification verification. Instead I am going to
file an SR for someone to look at it since I need to move on from this work back
to semantic SIL. At least we will not have notification failure issues anymore
and thus a large correctness issue in the compiler has been fixed. Forward
progress!
rdar://42301529
* rename "Name" to "Description" in the pass definition, because it's not really the pass name, but the description of a pass
* remove the getName() from Transforms (which actually returned the description of a pass)
* in debug printing, print the pass ID and not the pass description. It makes it easier to correlate the debug output to the actual pass implementation.
* remove the iteration numbering in the pass manager, because we only run a single iteration anyway.
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")
In the crash output 'While running pass #x SILFunctionTransform "T" on SILFunction "@F".' we printed the previous function which was handled by the pass instead of the current one.
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.
This patch implements collection and dumping of statistics about SILModules, SILFunctions and memory consumption during the execution of SIL optimization pipelines.
The following statistics can be collected:
* For SILFunctions: the number of SIL basic blocks, the number of SIL instructions, the number of SIL instructions of a specific kind, duration of a pass
* For SILModules: the number of SIL basic blocks, the number of SIL instructions, the number of SIL instructions of a specific kind, the number of SILFunctions, the amount of memory used by the compiler, duration of a pass
By default, any collection of statistics is disabled to avoid affecting compile times.
One can enable the collection of statistics and dumping of these statistics for the whole SILModule and/or for SILFunctions.
To reduce the amount of produced data, one can set thresholds in such a way that changes in the statistics are only reported if the delta between the old and the new values are at least X%. The deltas are computed as using the following formula:
Delta = (NewValue - OldValue) / OldValue
Thresholds provide a simple way to perform a simple filtering of the collected statistics during the compilation. But if there is a need for a more complex analysis of collected data (e.g. aggregation by a pipeline stage or by the type of a transformation), it is often better to dump as much data as possible into a file using e.g. -sil-stats-dump-all -sil-stats-modules -sil-stats-functions and then e.g. use the helper scripts to store the collected data into a database and then perform complex queries on it. Many kinds of analysis can be then formulated pretty easily as SQL queries.
A pass has an ID (C++ identifier), Tag (shell identifier),
and Name (human identifier).
Command line options that identify passes should obviously be compatibile with
with the pass' command line identifier. This is also what the user is used to
typing for the -debug-only option.
At some point, pass definitions were heavily macro-ized. Pass
descriptive names were added in two places. This is not only redundant
but a source of confusion. You could waste a lot of time grepping for
the wrong string. I removed all the getName() overrides which, at
around 90 passes, was a fairly significant amount of code bloat.
Any pass that we want to be able to invoke by name from a tool
(sil-opt) or pipeline plan *should* have unique type name, enum value,
commend-line string, and name string. I removed a comment about the
various inliner passes that contradicted that.
Side note: We should be consistent with the policy that a pass is
identified by its type. We have a couple passes, LICM and CSE, which
currently violate that convention.
Previously it was part of swiftBasic.
The demangler library does not depend on llvm (except some header-only utilities like StringRef). Putting it into its own library makes sure that no llvm stuff will be linked into clients which use the demangler library.
This change also contains other refactoring, like moving demangler code into different files. This makes it easier to remove the old demangler from the runtime library when we switch to the new symbol mangling.
Also in this commit: remove some unused API functions from the demangler Context.
fixes rdar://problem/30503344
This pipeline is run as part of IRGen and has access to the IRGenModule.
Passes that run as part of this pipeline can query for the IRGenModule.
We will use it for the AllocStackHoisting pass. It wants to know if a type is of
non-fixed size.
To break the cyclic dependency between IRGen -> SILOptimizer -> IRGen that would
arise from the SILPassManager having to know about the createIRGENPASS()
function IRGen passes instead of exposing this function dynamically have to add
themselves to the pass manager.
Changes:
* Terminate all namespaces with the correct closing comment.
* Make sure argument names in comments match the corresponding parameter name.
* Remove redundant get() calls on smart pointers.
* Prefer using "override" or "final" instead of "virtual". Remove "virtual" where appropriate.
We also either remove or make private the addPass* functions on SILPassManager,
so the only way to execute passes via SILPassManager is by creating a
SILPassPipelinePlan. This beyond adding uniformity ensures that we always
resetAndRemoveTransformations properly after a pipeline is run.
When SILVerifyAll is enabled, individual functions are verified after
function passes are run upon them. This means that any functions created
by a function pass will not be verified after the pass runs. Thus
specialization errors that cause the verifier to trip will be
misattributed to the first pass that makes a change to the specialized
function. This is very misleading and increases triage time.
This change eliminates that problem by ensuring that when SILVerifyAll is
enabled, we always verify newly specialized functions as they are added to the
worklist.
rdar://28706158
An int is already constant and in general in LLVM style we do not create such
const ints. On the other hand, a constexpr I think expresses the intent slightly
better and gives the compiler more freedom anyways.
This adds the typedef and switches uses of NodeType * to NodeRef. This is in
preparation for the eventual NodeRef-ization of the GraphTraits in LLVM. NFC.
