llvm-6502/lib/IR/PassManager.cpp

288 lines
10 KiB
C++
Raw Normal View History

//===- PassManager.cpp - Infrastructure for managing & running IR passes --===//
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
#include "llvm/ADT/STLExtras.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
using namespace llvm;
static cl::opt<bool>
DebugPM("debug-pass-manager", cl::Hidden,
cl::desc("Print pass management debugging information"));
PreservedAnalyses ModulePassManager::run(Module &M, ModuleAnalysisManager *AM) {
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
PreservedAnalyses PA = PreservedAnalyses::all();
if (DebugPM)
dbgs() << "Starting module pass manager run.\n";
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
for (unsigned Idx = 0, Size = Passes.size(); Idx != Size; ++Idx) {
if (DebugPM)
dbgs() << "Running module pass: " << Passes[Idx]->name() << "\n";
[PM] Switch analysis managers to be threaded through the run methods rather than the constructors of passes. This simplifies the APIs of passes significantly and removes an error prone pattern where the *same* manager had to be given to every different layer. With the new API the analysis managers themselves will have to be cross connected with proxy analyses that allow a pass at one layer to query for the analysis manager of another layer. The proxy will both expose a handle to the other layer's manager and it will provide the invalidation hooks to ensure things remain consistent across layers. Finally, the outer-most analysis manager has to be passed to the run method of the outer-most pass manager. The rest of the propagation is automatic. I've used SFINAE again to allow passes to completely disregard the analysis manager if they don't need or want to care. This helps keep simple things simple for users of the new pass manager. Also, the system specifically supports passing a null pointer into the outer-most run method if your pass pipeline neither needs nor wants to deal with analyses. I find this of dubious utility as while some *passes* don't care about analysis, I'm not sure there are any real-world users of the pass manager itself that need to avoid even creating an analysis manager. But it is easy to support, so there we go. Finally I renamed the module proxy for the function analysis manager to the more verbose but less confusing name of FunctionAnalysisManagerModuleProxy. I hate this name, but I have no idea what else to name these things. I'm expecting in the fullness of time to potentially have the complete cross product of types at the proxy layer: {Module,SCC,Function,Loop,Region}AnalysisManager{Module,SCC,Function,Loop,Region}Proxy (except for XAnalysisManagerXProxy which doesn't make any sense) This should make it somewhat easier to do the next phases which is to build the upward proxy and get its invalidation correct, as well as to make the invalidation within the Module -> Function mapping pass be more fine grained so as to invalidate fewer fuction analyses. After all of the proxy analyses are done and the invalidation working, I'll finally be able to start working on the next two fun fronts: how to adapt an existing pass to work in both the legacy pass world and the new one, and building the SCC, Loop, and Region counterparts. Fun times! git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195400 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 00:43:29 +00:00
PreservedAnalyses PassPA = Passes[Idx]->run(M, AM);
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
// If we have an active analysis manager at this level we want to ensure we
// update it as each pass runs and potentially invalidates analyses. We
// also update the preserved set of analyses based on what analyses we have
// already handled the invalidation for here and don't need to invalidate
// when finished.
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
if (AM)
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
PassPA = AM->invalidate(M, std::move(PassPA));
// Finally, we intersect the final preserved analyses to compute the
// aggregate preserved set for this pass manager.
PA.intersect(std::move(PassPA));
M.getContext().yield();
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
}
if (DebugPM)
dbgs() << "Finished module pass manager run.\n";
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
return PA;
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
}
ModuleAnalysisManager::ResultConceptT &
ModuleAnalysisManager::getResultImpl(void *PassID, Module &M) {
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
ModuleAnalysisResultMapT::iterator RI;
bool Inserted;
std::tie(RI, Inserted) = ModuleAnalysisResults.insert(std::make_pair(
PassID, std::unique_ptr<detail::AnalysisResultConcept<Module &>>()));
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
// If we don't have a cached result for this module, look up the pass and run
// it to produce a result, which we then add to the cache.
if (Inserted) {
auto &P = lookupPass(PassID);
if (DebugPM)
dbgs() << "Running module analysis: " << P.name() << "\n";
RI->second = P.run(M, this);
}
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
return *RI->second;
}
ModuleAnalysisManager::ResultConceptT *
ModuleAnalysisManager::getCachedResultImpl(void *PassID, Module &M) const {
ModuleAnalysisResultMapT::const_iterator RI =
ModuleAnalysisResults.find(PassID);
return RI == ModuleAnalysisResults.end() ? nullptr : &*RI->second;
}
void ModuleAnalysisManager::invalidateImpl(void *PassID, Module &M) {
if (DebugPM)
dbgs() << "Invalidating module analysis: " << lookupPass(PassID).name()
<< "\n";
[PM] Split the analysis manager into a function-specific interface and a module-specific interface. This is the first of many steps necessary to generalize the infrastructure such that we can support both a Module-to-Function and Module-to-SCC-to-Function pass manager nestings. After a *lot* of attempts that never worked and didn't even make it to a committable state, it became clear that I had gotten the layering design of analyses flat out wrong. Four days later, I think I have most of the plan for how to correct this, and I'm starting to reshape the code into it. This is just a baby step I'm afraid, but starts separating the fundamentally distinct concepts of function analysis passes and module analysis passes so that in subsequent steps we can effectively layer them, and have a consistent design for the eventual SCC layer. As part of this, I've started some interface changes to make passes more regular. The module pass accepts the module in the run method, and some of the constructor parameters are gone. I'm still working out exactly where constructor parameters vs. method parameters will be used, so I expect this to fluctuate a bit. This actually makes the invalidation less "correct" at this phase, because now function passes don't invalidate module analysis passes, but that was actually somewhat of a misfeature. It will return in a better factored form which can scale to other units of IR. The documentation has gotten less verbose and helpful. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195189 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 04:01:38 +00:00
ModuleAnalysisResults.erase(PassID);
}
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
PreservedAnalyses ModuleAnalysisManager::invalidateImpl(Module &M,
PreservedAnalyses PA) {
// Short circuit for a common case of all analyses being preserved.
if (PA.areAllPreserved())
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
return std::move(PA);
if (DebugPM)
dbgs() << "Invalidating all non-preserved analyses for module: "
<< M.getModuleIdentifier() << "\n";
// FIXME: This is a total hack based on the fact that erasure doesn't
// invalidate iteration for DenseMap.
for (ModuleAnalysisResultMapT::iterator I = ModuleAnalysisResults.begin(),
E = ModuleAnalysisResults.end();
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
I != E; ++I) {
void *PassID = I->first;
// Pass the invalidation down to the pass itself to see if it thinks it is
// necessary. The analysis pass can return false if no action on the part
// of the analysis manager is required for this invalidation event.
if (I->second->invalidate(M, PA)) {
if (DebugPM)
dbgs() << "Invalidating module analysis: "
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
<< lookupPass(PassID).name() << "\n";
ModuleAnalysisResults.erase(I);
}
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
// After handling each pass, we mark it as preserved. Once we've
// invalidated any stale results, the rest of the system is allowed to
// start preserving this analysis again.
PA.preserve(PassID);
}
return std::move(PA);
}
PreservedAnalyses FunctionPassManager::run(Function &F,
FunctionAnalysisManager *AM) {
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
PreservedAnalyses PA = PreservedAnalyses::all();
if (DebugPM)
dbgs() << "Starting function pass manager run.\n";
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
for (unsigned Idx = 0, Size = Passes.size(); Idx != Size; ++Idx) {
if (DebugPM)
dbgs() << "Running function pass: " << Passes[Idx]->name() << "\n";
[PM] Switch analysis managers to be threaded through the run methods rather than the constructors of passes. This simplifies the APIs of passes significantly and removes an error prone pattern where the *same* manager had to be given to every different layer. With the new API the analysis managers themselves will have to be cross connected with proxy analyses that allow a pass at one layer to query for the analysis manager of another layer. The proxy will both expose a handle to the other layer's manager and it will provide the invalidation hooks to ensure things remain consistent across layers. Finally, the outer-most analysis manager has to be passed to the run method of the outer-most pass manager. The rest of the propagation is automatic. I've used SFINAE again to allow passes to completely disregard the analysis manager if they don't need or want to care. This helps keep simple things simple for users of the new pass manager. Also, the system specifically supports passing a null pointer into the outer-most run method if your pass pipeline neither needs nor wants to deal with analyses. I find this of dubious utility as while some *passes* don't care about analysis, I'm not sure there are any real-world users of the pass manager itself that need to avoid even creating an analysis manager. But it is easy to support, so there we go. Finally I renamed the module proxy for the function analysis manager to the more verbose but less confusing name of FunctionAnalysisManagerModuleProxy. I hate this name, but I have no idea what else to name these things. I'm expecting in the fullness of time to potentially have the complete cross product of types at the proxy layer: {Module,SCC,Function,Loop,Region}AnalysisManager{Module,SCC,Function,Loop,Region}Proxy (except for XAnalysisManagerXProxy which doesn't make any sense) This should make it somewhat easier to do the next phases which is to build the upward proxy and get its invalidation correct, as well as to make the invalidation within the Module -> Function mapping pass be more fine grained so as to invalidate fewer fuction analyses. After all of the proxy analyses are done and the invalidation working, I'll finally be able to start working on the next two fun fronts: how to adapt an existing pass to work in both the legacy pass world and the new one, and building the SCC, Loop, and Region counterparts. Fun times! git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195400 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 00:43:29 +00:00
PreservedAnalyses PassPA = Passes[Idx]->run(F, AM);
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
// If we have an active analysis manager at this level we want to ensure we
// update it as each pass runs and potentially invalidates analyses. We
// also update the preserved set of analyses based on what analyses we have
// already handled the invalidation for here and don't need to invalidate
// when finished.
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
if (AM)
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
PassPA = AM->invalidate(F, std::move(PassPA));
// Finally, we intersect the final preserved analyses to compute the
// aggregate preserved set for this pass manager.
PA.intersect(std::move(PassPA));
F.getContext().yield();
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
}
if (DebugPM)
dbgs() << "Finished function pass manager run.\n";
[PM] Add the preservation system to the new pass manager. This adds a new set-like type which represents a set of preserved analysis passes. The set is managed via the opaque PassT::ID() void*s. The expected convenience templates for interacting with specific passes are provided. It also supports a symbolic "all" state which is represented by an invalid pointer in the set. This state is nicely saturating as it comes up often. Finally, it supports intersection which is used when finding the set of preserved passes after N different transforms. The pass API is then changed to return the preserved set rather than a bool. This is much more self-documenting than the previous system. Returning "none" is a conservatively correct solution just like returning "true" from todays passes and not marking any passes as preserved. Passes can also be dynamically preserved or not throughout the run of the pass, and whatever gets returned is the binding state. Finally, preserving "all" the passes is allowed for no-op transforms that simply can't harm such things. Finally, the analysis managers are changed to instead of blindly invalidating all of the analyses, invalidate those which were not preserved. This should rig up all of the basic preservation functionality. This also correctly combines the preservation moving up from one IR-layer to the another and the preservation aggregation across N pass runs. Still to go is incrementally correct invalidation and preservation across IR layers incrementally during N pass runs. That will wait until we have a device for even exposing analyses across IR layers. While the core of this change is obvious, I'm not happy with the current testing, so will improve it to cover at least some of the invalidation that I can test easily in a subsequent commit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195241 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 11:31:50 +00:00
return PA;
[PM] Split the analysis manager into a function-specific interface and a module-specific interface. This is the first of many steps necessary to generalize the infrastructure such that we can support both a Module-to-Function and Module-to-SCC-to-Function pass manager nestings. After a *lot* of attempts that never worked and didn't even make it to a committable state, it became clear that I had gotten the layering design of analyses flat out wrong. Four days later, I think I have most of the plan for how to correct this, and I'm starting to reshape the code into it. This is just a baby step I'm afraid, but starts separating the fundamentally distinct concepts of function analysis passes and module analysis passes so that in subsequent steps we can effectively layer them, and have a consistent design for the eventual SCC layer. As part of this, I've started some interface changes to make passes more regular. The module pass accepts the module in the run method, and some of the constructor parameters are gone. I'm still working out exactly where constructor parameters vs. method parameters will be used, so I expect this to fluctuate a bit. This actually makes the invalidation less "correct" at this phase, because now function passes don't invalidate module analysis passes, but that was actually somewhat of a misfeature. It will return in a better factored form which can scale to other units of IR. The documentation has gotten less verbose and helpful. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195189 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-20 04:01:38 +00:00
}
bool FunctionAnalysisManager::empty() const {
assert(FunctionAnalysisResults.empty() ==
FunctionAnalysisResultLists.empty() &&
"The storage and index of analysis results disagree on how many there "
"are!");
return FunctionAnalysisResults.empty();
}
void FunctionAnalysisManager::clear() {
FunctionAnalysisResults.clear();
FunctionAnalysisResultLists.clear();
}
FunctionAnalysisManager::ResultConceptT &
FunctionAnalysisManager::getResultImpl(void *PassID, Function &F) {
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
FunctionAnalysisResultMapT::iterator RI;
bool Inserted;
std::tie(RI, Inserted) = FunctionAnalysisResults.insert(std::make_pair(
std::make_pair(PassID, &F), FunctionAnalysisResultListT::iterator()));
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
// If we don't have a cached result for this function, look up the pass and
// run it to produce a result, which we then add to the cache.
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
if (Inserted) {
auto &P = lookupPass(PassID);
if (DebugPM)
dbgs() << "Running function analysis: " << P.name() << "\n";
FunctionAnalysisResultListT &ResultList = FunctionAnalysisResultLists[&F];
ResultList.emplace_back(PassID, P.run(F, this));
RI->second = std::prev(ResultList.end());
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
}
return *RI->second->second;
}
FunctionAnalysisManager::ResultConceptT *
FunctionAnalysisManager::getCachedResultImpl(void *PassID, Function &F) const {
FunctionAnalysisResultMapT::const_iterator RI =
FunctionAnalysisResults.find(std::make_pair(PassID, &F));
return RI == FunctionAnalysisResults.end() ? nullptr : &*RI->second->second;
}
void FunctionAnalysisManager::invalidateImpl(void *PassID, Function &F) {
FunctionAnalysisResultMapT::iterator RI =
FunctionAnalysisResults.find(std::make_pair(PassID, &F));
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
if (RI == FunctionAnalysisResults.end())
return;
if (DebugPM)
dbgs() << "Invalidating function analysis: " << lookupPass(PassID).name()
<< "\n";
FunctionAnalysisResultLists[&F].erase(RI->second);
FunctionAnalysisResults.erase(RI);
}
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
PreservedAnalyses
FunctionAnalysisManager::invalidateImpl(Function &F, PreservedAnalyses PA) {
// Short circuit for a common case of all analyses being preserved.
if (PA.areAllPreserved())
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
return std::move(PA);
if (DebugPM)
dbgs() << "Invalidating all non-preserved analyses for function: "
<< F.getName() << "\n";
// Clear all the invalidated results associated specifically with this
// function.
SmallVector<void *, 8> InvalidatedPassIDs;
FunctionAnalysisResultListT &ResultsList = FunctionAnalysisResultLists[&F];
for (FunctionAnalysisResultListT::iterator I = ResultsList.begin(),
E = ResultsList.end();
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
I != E;) {
void *PassID = I->first;
// Pass the invalidation down to the pass itself to see if it thinks it is
// necessary. The analysis pass can return false if no action on the part
// of the analysis manager is required for this invalidation event.
if (I->second->invalidate(F, PA)) {
if (DebugPM)
dbgs() << "Invalidating function analysis: "
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
<< lookupPass(PassID).name() << "\n";
InvalidatedPassIDs.push_back(I->first);
I = ResultsList.erase(I);
} else {
++I;
}
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
// After handling each pass, we mark it as preserved. Once we've
// invalidated any stale results, the rest of the system is allowed to
// start preserving this analysis again.
PA.preserve(PassID);
}
while (!InvalidatedPassIDs.empty())
FunctionAnalysisResults.erase(
std::make_pair(InvalidatedPassIDs.pop_back_val(), &F));
if (ResultsList.empty())
FunctionAnalysisResultLists.erase(&F);
[PM] Fix a pretty nasty bug where the new pass manager would invalidate passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-07 01:58:35 +00:00
return std::move(PA);
Introduce an AnalysisManager which is like a pass manager but with a lot more smarts in it. This is where most of the interesting logic that used to live in the implicit-scheduling-hackery of the old pass manager will live. Like the previous commits, note that this is a very early prototype! I expect substantial changes before this is ready to use. The core of the design is the following: - We have an AnalysisManager which can be used across a series of passes over a module. - The code setting up a pass pipeline registers the analyses available with the manager. - Individual transform passes can check than an analysis manager provides the analyses they require in order to fail-fast. - There is *no* implicit registration or scheduling. - Analysis passes are different from other passes: they produce an analysis result that is cached and made available via the analysis manager. - Cached results are invalidated automatically by the pass managers. - When a transform pass requests an analysis result, either the analysis is run to produce the result or a cached result is provided. There are a few aspects of this design that I *know* will change in subsequent commits: - Currently there is no "preservation" system, that needs to be added. - All of the analysis management should move up to the analysis library. - The analysis management needs to support at least SCC passes. Maybe loop passes. Living in the analysis library will facilitate this. - Need support for analyses which are *both* module and function passes. - Need support for pro-actively running module analyses to have cached results within a function pass manager. - Need a clear design for "immutable" passes. - Need support for requesting cached results when available and not re-running the pass even if that would be necessary. - Need more thorough testing of all of this infrastructure. There are other aspects that I view as open questions I'm hoping to resolve as I iterate a bit on the infrastructure, and especially as I start writing actual passes against this. - Should we have separate management layers for function, module, and SCC analyses? I think "yes", but I'm not yet ready to switch the code. Adding SCC support will likely resolve this definitively. - How should the 'require' functionality work? Should *that* be the only way to request results to ensure that passes always require things? - How should preservation work? - Probably some other things I'm forgetting. =] Look forward to more patches in shorter order now that this is in place. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194538 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-13 01:12:08 +00:00
}
[PM] Switch analysis managers to be threaded through the run methods rather than the constructors of passes. This simplifies the APIs of passes significantly and removes an error prone pattern where the *same* manager had to be given to every different layer. With the new API the analysis managers themselves will have to be cross connected with proxy analyses that allow a pass at one layer to query for the analysis manager of another layer. The proxy will both expose a handle to the other layer's manager and it will provide the invalidation hooks to ensure things remain consistent across layers. Finally, the outer-most analysis manager has to be passed to the run method of the outer-most pass manager. The rest of the propagation is automatic. I've used SFINAE again to allow passes to completely disregard the analysis manager if they don't need or want to care. This helps keep simple things simple for users of the new pass manager. Also, the system specifically supports passing a null pointer into the outer-most run method if your pass pipeline neither needs nor wants to deal with analyses. I find this of dubious utility as while some *passes* don't care about analysis, I'm not sure there are any real-world users of the pass manager itself that need to avoid even creating an analysis manager. But it is easy to support, so there we go. Finally I renamed the module proxy for the function analysis manager to the more verbose but less confusing name of FunctionAnalysisManagerModuleProxy. I hate this name, but I have no idea what else to name these things. I'm expecting in the fullness of time to potentially have the complete cross product of types at the proxy layer: {Module,SCC,Function,Loop,Region}AnalysisManager{Module,SCC,Function,Loop,Region}Proxy (except for XAnalysisManagerXProxy which doesn't make any sense) This should make it somewhat easier to do the next phases which is to build the upward proxy and get its invalidation correct, as well as to make the invalidation within the Module -> Function mapping pass be more fine grained so as to invalidate fewer fuction analyses. After all of the proxy analyses are done and the invalidation working, I'll finally be able to start working on the next two fun fronts: how to adapt an existing pass to work in both the legacy pass world and the new one, and building the SCC, Loop, and Region counterparts. Fun times! git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195400 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 00:43:29 +00:00
char FunctionAnalysisManagerModuleProxy::PassID;
[PM] Switch analysis managers to be threaded through the run methods rather than the constructors of passes. This simplifies the APIs of passes significantly and removes an error prone pattern where the *same* manager had to be given to every different layer. With the new API the analysis managers themselves will have to be cross connected with proxy analyses that allow a pass at one layer to query for the analysis manager of another layer. The proxy will both expose a handle to the other layer's manager and it will provide the invalidation hooks to ensure things remain consistent across layers. Finally, the outer-most analysis manager has to be passed to the run method of the outer-most pass manager. The rest of the propagation is automatic. I've used SFINAE again to allow passes to completely disregard the analysis manager if they don't need or want to care. This helps keep simple things simple for users of the new pass manager. Also, the system specifically supports passing a null pointer into the outer-most run method if your pass pipeline neither needs nor wants to deal with analyses. I find this of dubious utility as while some *passes* don't care about analysis, I'm not sure there are any real-world users of the pass manager itself that need to avoid even creating an analysis manager. But it is easy to support, so there we go. Finally I renamed the module proxy for the function analysis manager to the more verbose but less confusing name of FunctionAnalysisManagerModuleProxy. I hate this name, but I have no idea what else to name these things. I'm expecting in the fullness of time to potentially have the complete cross product of types at the proxy layer: {Module,SCC,Function,Loop,Region}AnalysisManager{Module,SCC,Function,Loop,Region}Proxy (except for XAnalysisManagerXProxy which doesn't make any sense) This should make it somewhat easier to do the next phases which is to build the upward proxy and get its invalidation correct, as well as to make the invalidation within the Module -> Function mapping pass be more fine grained so as to invalidate fewer fuction analyses. After all of the proxy analyses are done and the invalidation working, I'll finally be able to start working on the next two fun fronts: how to adapt an existing pass to work in both the legacy pass world and the new one, and building the SCC, Loop, and Region counterparts. Fun times! git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195400 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 00:43:29 +00:00
FunctionAnalysisManagerModuleProxy::Result
FunctionAnalysisManagerModuleProxy::run(Module &M) {
assert(FAM->empty() && "Function analyses ran prior to the module proxy!");
return Result(*FAM);
}
[PM] Switch analysis managers to be threaded through the run methods rather than the constructors of passes. This simplifies the APIs of passes significantly and removes an error prone pattern where the *same* manager had to be given to every different layer. With the new API the analysis managers themselves will have to be cross connected with proxy analyses that allow a pass at one layer to query for the analysis manager of another layer. The proxy will both expose a handle to the other layer's manager and it will provide the invalidation hooks to ensure things remain consistent across layers. Finally, the outer-most analysis manager has to be passed to the run method of the outer-most pass manager. The rest of the propagation is automatic. I've used SFINAE again to allow passes to completely disregard the analysis manager if they don't need or want to care. This helps keep simple things simple for users of the new pass manager. Also, the system specifically supports passing a null pointer into the outer-most run method if your pass pipeline neither needs nor wants to deal with analyses. I find this of dubious utility as while some *passes* don't care about analysis, I'm not sure there are any real-world users of the pass manager itself that need to avoid even creating an analysis manager. But it is easy to support, so there we go. Finally I renamed the module proxy for the function analysis manager to the more verbose but less confusing name of FunctionAnalysisManagerModuleProxy. I hate this name, but I have no idea what else to name these things. I'm expecting in the fullness of time to potentially have the complete cross product of types at the proxy layer: {Module,SCC,Function,Loop,Region}AnalysisManager{Module,SCC,Function,Loop,Region}Proxy (except for XAnalysisManagerXProxy which doesn't make any sense) This should make it somewhat easier to do the next phases which is to build the upward proxy and get its invalidation correct, as well as to make the invalidation within the Module -> Function mapping pass be more fine grained so as to invalidate fewer fuction analyses. After all of the proxy analyses are done and the invalidation working, I'll finally be able to start working on the next two fun fronts: how to adapt an existing pass to work in both the legacy pass world and the new one, and building the SCC, Loop, and Region counterparts. Fun times! git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195400 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 00:43:29 +00:00
FunctionAnalysisManagerModuleProxy::Result::~Result() {
// Clear out the analysis manager if we're being destroyed -- it means we
// didn't even see an invalidate call when we got invalidated.
FAM->clear();
}
[PM] Switch analysis managers to be threaded through the run methods rather than the constructors of passes. This simplifies the APIs of passes significantly and removes an error prone pattern where the *same* manager had to be given to every different layer. With the new API the analysis managers themselves will have to be cross connected with proxy analyses that allow a pass at one layer to query for the analysis manager of another layer. The proxy will both expose a handle to the other layer's manager and it will provide the invalidation hooks to ensure things remain consistent across layers. Finally, the outer-most analysis manager has to be passed to the run method of the outer-most pass manager. The rest of the propagation is automatic. I've used SFINAE again to allow passes to completely disregard the analysis manager if they don't need or want to care. This helps keep simple things simple for users of the new pass manager. Also, the system specifically supports passing a null pointer into the outer-most run method if your pass pipeline neither needs nor wants to deal with analyses. I find this of dubious utility as while some *passes* don't care about analysis, I'm not sure there are any real-world users of the pass manager itself that need to avoid even creating an analysis manager. But it is easy to support, so there we go. Finally I renamed the module proxy for the function analysis manager to the more verbose but less confusing name of FunctionAnalysisManagerModuleProxy. I hate this name, but I have no idea what else to name these things. I'm expecting in the fullness of time to potentially have the complete cross product of types at the proxy layer: {Module,SCC,Function,Loop,Region}AnalysisManager{Module,SCC,Function,Loop,Region}Proxy (except for XAnalysisManagerXProxy which doesn't make any sense) This should make it somewhat easier to do the next phases which is to build the upward proxy and get its invalidation correct, as well as to make the invalidation within the Module -> Function mapping pass be more fine grained so as to invalidate fewer fuction analyses. After all of the proxy analyses are done and the invalidation working, I'll finally be able to start working on the next two fun fronts: how to adapt an existing pass to work in both the legacy pass world and the new one, and building the SCC, Loop, and Region counterparts. Fun times! git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195400 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-22 00:43:29 +00:00
bool FunctionAnalysisManagerModuleProxy::Result::invalidate(
Module &M, const PreservedAnalyses &PA) {
// If this proxy isn't marked as preserved, then we can't even invalidate
// individual function analyses, there may be an invalid set of Function
// objects in the cache making it impossible to incrementally preserve them.
// Just clear the entire manager.
if (!PA.preserved(ID()))
FAM->clear();
// Return false to indicate that this result is still a valid proxy.
return false;
}
char ModuleAnalysisManagerFunctionProxy::PassID;