2018-04-04 05:31:50 +08:00
|
|
|
//===- CheckerRegistry.cpp - Maintains all available checkers -------------===//
|
2011-08-17 05:24:21 +08:00
|
|
|
//
|
2019-01-19 16:50:56 +08:00
|
|
|
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
|
|
// See https://llvm.org/LICENSE.txt for license information.
|
|
|
|
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
2011-08-17 05:24:21 +08:00
|
|
|
//
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
[analyzer][NFC] Move CheckerRegistry from the Core directory to Frontend
ClangCheckerRegistry is a very non-obvious, poorly documented, weird concept.
It derives from CheckerRegistry, and is placed in lib/StaticAnalyzer/Frontend,
whereas it's base is located in lib/StaticAnalyzer/Core. It was, from what I can
imagine, used to circumvent the problem that the registry functions of the
checkers are located in the clangStaticAnalyzerCheckers library, but that
library depends on clangStaticAnalyzerCore. However, clangStaticAnalyzerFrontend
depends on both of those libraries.
One can make the observation however, that CheckerRegistry has no place in Core,
it isn't used there at all! The only place where it is used is Frontend, which
is where it ultimately belongs.
This move implies that since
include/clang/StaticAnalyzer/Checkers/ClangCheckers.h only contained a single function:
class CheckerRegistry;
void registerBuiltinCheckers(CheckerRegistry ®istry);
it had to re purposed, as CheckerRegistry is no longer available to
clangStaticAnalyzerCheckers. It was renamed to BuiltinCheckerRegistration.h,
which actually describes it a lot better -- it does not contain the registration
functions for checkers, but only those generated by the tblgen files.
Differential Revision: https://reviews.llvm.org/D54436
llvm-svn: 349275
2018-12-16 00:23:51 +08:00
|
|
|
#include "clang/StaticAnalyzer/Frontend/CheckerRegistry.h"
|
2015-07-10 05:43:45 +08:00
|
|
|
#include "clang/Basic/Diagnostic.h"
|
2018-04-04 05:31:50 +08:00
|
|
|
#include "clang/Basic/LLVM.h"
|
2019-05-17 17:51:59 +08:00
|
|
|
#include "clang/Driver/DriverDiagnostic.h"
|
2018-12-16 02:11:49 +08:00
|
|
|
#include "clang/Frontend/FrontendDiagnostic.h"
|
|
|
|
#include "clang/StaticAnalyzer/Checkers/BuiltinCheckerRegistration.h"
|
2015-07-10 05:43:45 +08:00
|
|
|
#include "clang/StaticAnalyzer/Core/AnalyzerOptions.h"
|
2019-04-19 01:32:51 +08:00
|
|
|
#include "clang/StaticAnalyzer/Core/CheckerManager.h"
|
2018-04-04 05:31:50 +08:00
|
|
|
#include "llvm/ADT/STLExtras.h"
|
2011-12-15 09:36:04 +08:00
|
|
|
#include "llvm/ADT/SetVector.h"
|
2018-04-04 05:31:50 +08:00
|
|
|
#include "llvm/ADT/StringMap.h"
|
|
|
|
#include "llvm/ADT/StringRef.h"
|
2018-12-16 02:11:49 +08:00
|
|
|
#include "llvm/Support/DynamicLibrary.h"
|
|
|
|
#include "llvm/Support/Path.h"
|
2012-12-02 01:12:56 +08:00
|
|
|
#include "llvm/Support/raw_ostream.h"
|
2018-04-04 05:31:50 +08:00
|
|
|
#include <algorithm>
|
2011-08-17 05:24:21 +08:00
|
|
|
|
|
|
|
using namespace clang;
|
|
|
|
using namespace ento;
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
using namespace checker_registry;
|
2018-12-16 02:11:49 +08:00
|
|
|
using llvm::sys::DynamicLibrary;
|
|
|
|
|
2020-06-01 03:22:35 +08:00
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Utilities.
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
2019-04-18 23:19:16 +08:00
|
|
|
static bool isCompatibleAPIVersion(const char *VersionString) {
|
|
|
|
// If the version string is null, its not an analyzer plugin.
|
|
|
|
if (!VersionString)
|
2018-12-16 02:11:49 +08:00
|
|
|
return false;
|
|
|
|
|
|
|
|
// For now, none of the static analyzer API is considered stable.
|
|
|
|
// Versions must match exactly.
|
2019-04-18 23:19:16 +08:00
|
|
|
return strcmp(VersionString, CLANG_ANALYZER_API_VERSION_STRING) == 0;
|
2018-12-16 02:11:49 +08:00
|
|
|
}
|
|
|
|
|
2019-01-27 00:35:33 +08:00
|
|
|
static constexpr char PackageSeparator = '.';
|
|
|
|
|
2020-06-01 03:22:35 +08:00
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Methods of CheckerRegistry.
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
2019-01-27 01:27:40 +08:00
|
|
|
CheckerRegistry::CheckerRegistry(
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
CheckerRegistryData &Data, ArrayRef<std::string> Plugins,
|
|
|
|
DiagnosticsEngine &Diags, AnalyzerOptions &AnOpts,
|
2019-04-19 01:32:51 +08:00
|
|
|
ArrayRef<std::function<void(CheckerRegistry &)>> CheckerRegistrationFns)
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
: Data(Data), Diags(Diags), AnOpts(AnOpts) {
|
2019-01-26 22:23:08 +08:00
|
|
|
|
2019-01-27 00:35:33 +08:00
|
|
|
// Register builtin checkers.
|
2018-12-16 02:11:49 +08:00
|
|
|
#define GET_CHECKERS
|
[analyzer] Don't display implementation checkers under -analyzer-checker-help, but do under the new flag -analyzer-checker-help-hidden
During my work on analyzer dependencies, I created a great amount of new
checkers that emitted no diagnostics at all, and were purely modeling some
function or another.
However, the user shouldn't really disable/enable these by hand, hence this
patch, which hides these by default. I intentionally chose not to hide alpha
checkers, because they have a scary enough name, in my opinion, to cause no
surprise when they emit false positives or cause crashes.
The patch introduces the Hidden bit into the TableGen files (you may remember
it before I removed it in D53995), and checkers that are either marked as
hidden, or are in a package that is marked hidden won't be displayed under
-analyzer-checker-help. -analyzer-checker-help-hidden, a new flag meant for
developers only, displays the full list.
Differential Revision: https://reviews.llvm.org/D60925
llvm-svn: 359720
2019-05-02 03:56:47 +08:00
|
|
|
#define CHECKER(FULLNAME, CLASS, HELPTEXT, DOC_URI, IS_HIDDEN) \
|
2019-01-26 22:23:08 +08:00
|
|
|
addChecker(register##CLASS, shouldRegister##CLASS, FULLNAME, HELPTEXT, \
|
[analyzer] Don't display implementation checkers under -analyzer-checker-help, but do under the new flag -analyzer-checker-help-hidden
During my work on analyzer dependencies, I created a great amount of new
checkers that emitted no diagnostics at all, and were purely modeling some
function or another.
However, the user shouldn't really disable/enable these by hand, hence this
patch, which hides these by default. I intentionally chose not to hide alpha
checkers, because they have a scary enough name, in my opinion, to cause no
surprise when they emit false positives or cause crashes.
The patch introduces the Hidden bit into the TableGen files (you may remember
it before I removed it in D53995), and checkers that are either marked as
hidden, or are in a package that is marked hidden won't be displayed under
-analyzer-checker-help. -analyzer-checker-help-hidden, a new flag meant for
developers only, displays the full list.
Differential Revision: https://reviews.llvm.org/D60925
llvm-svn: 359720
2019-05-02 03:56:47 +08:00
|
|
|
DOC_URI, IS_HIDDEN);
|
2019-04-18 23:19:16 +08:00
|
|
|
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
#define GET_PACKAGES
|
|
|
|
#define PACKAGE(FULLNAME) addPackage(FULLNAME);
|
|
|
|
|
2018-12-16 02:11:49 +08:00
|
|
|
#include "clang/StaticAnalyzer/Checkers/Checkers.inc"
|
|
|
|
#undef CHECKER
|
|
|
|
#undef GET_CHECKERS
|
2019-04-18 23:19:16 +08:00
|
|
|
#undef PACKAGE
|
|
|
|
#undef GET_PACKAGES
|
2018-12-16 02:11:49 +08:00
|
|
|
|
2019-01-27 00:35:33 +08:00
|
|
|
// Register checkers from plugins.
|
2019-04-18 23:19:16 +08:00
|
|
|
for (const std::string &Plugin : Plugins) {
|
2018-12-16 02:11:49 +08:00
|
|
|
// Get access to the plugin.
|
2019-04-18 23:19:16 +08:00
|
|
|
std::string ErrorMsg;
|
|
|
|
DynamicLibrary Lib =
|
|
|
|
DynamicLibrary::getPermanentLibrary(Plugin.c_str(), &ErrorMsg);
|
|
|
|
if (!Lib.isValid()) {
|
|
|
|
Diags.Report(diag::err_fe_unable_to_load_plugin) << Plugin << ErrorMsg;
|
2018-12-16 02:11:49 +08:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2019-04-18 23:19:16 +08:00
|
|
|
// See if its compatible with this build of clang.
|
|
|
|
const char *PluginAPIVersion = static_cast<const char *>(
|
|
|
|
Lib.getAddressOfSymbol("clang_analyzerAPIVersionString"));
|
|
|
|
|
|
|
|
if (!isCompatibleAPIVersion(PluginAPIVersion)) {
|
2018-12-16 02:11:49 +08:00
|
|
|
Diags.Report(diag::warn_incompatible_analyzer_plugin_api)
|
2019-04-18 23:19:16 +08:00
|
|
|
<< llvm::sys::path::filename(Plugin);
|
2018-12-16 02:11:49 +08:00
|
|
|
Diags.Report(diag::note_incompatible_analyzer_plugin_api)
|
2019-04-19 01:32:51 +08:00
|
|
|
<< CLANG_ANALYZER_API_VERSION_STRING << PluginAPIVersion;
|
2018-12-16 02:11:49 +08:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
using RegisterPluginCheckerFn = void (*)(CheckerRegistry &);
|
2018-12-16 02:11:49 +08:00
|
|
|
// Register its checkers.
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
RegisterPluginCheckerFn RegisterPluginCheckers =
|
|
|
|
reinterpret_cast<RegisterPluginCheckerFn>(
|
2019-04-19 01:32:51 +08:00
|
|
|
Lib.getAddressOfSymbol("clang_registerCheckers"));
|
2019-04-18 23:19:16 +08:00
|
|
|
if (RegisterPluginCheckers)
|
|
|
|
RegisterPluginCheckers(*this);
|
2018-12-16 02:11:49 +08:00
|
|
|
}
|
2011-08-17 05:24:21 +08:00
|
|
|
|
2019-01-27 01:27:40 +08:00
|
|
|
// Register statically linked checkers, that aren't generated from the tblgen
|
2019-04-19 01:32:51 +08:00
|
|
|
// file, but rather passed their registry function as a parameter in
|
|
|
|
// checkerRegistrationFns.
|
2019-01-27 01:27:40 +08:00
|
|
|
|
2019-04-18 23:19:16 +08:00
|
|
|
for (const auto &Fn : CheckerRegistrationFns)
|
2019-01-27 01:27:40 +08:00
|
|
|
Fn(*this);
|
|
|
|
|
2019-01-27 00:35:33 +08:00
|
|
|
// Sort checkers for efficient collection.
|
|
|
|
// FIXME: Alphabetical sort puts 'experimental' in the middle.
|
|
|
|
// Would it be better to name it '~experimental' or something else
|
|
|
|
// that's ASCIIbetically last?
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
llvm::sort(Data.Packages, checker_registry::PackageNameLT{});
|
|
|
|
llvm::sort(Data.Checkers, checker_registry::CheckerNameLT{});
|
2011-08-17 05:24:21 +08:00
|
|
|
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
#define GET_CHECKER_DEPENDENCIES
|
|
|
|
|
|
|
|
#define CHECKER_DEPENDENCY(FULLNAME, DEPENDENCY) \
|
|
|
|
addDependency(FULLNAME, DEPENDENCY);
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
#define GET_CHECKER_WEAK_DEPENDENCIES
|
|
|
|
|
|
|
|
#define CHECKER_WEAK_DEPENDENCY(FULLNAME, DEPENDENCY) \
|
|
|
|
addWeakDependency(FULLNAME, DEPENDENCY);
|
|
|
|
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
#define GET_CHECKER_OPTIONS
|
2020-02-28 22:07:50 +08:00
|
|
|
#define CHECKER_OPTION(TYPE, FULLNAME, CMDFLAG, DESC, DEFAULT_VAL, \
|
|
|
|
DEVELOPMENT_STATUS, IS_HIDDEN) \
|
|
|
|
addCheckerOption(TYPE, FULLNAME, CMDFLAG, DEFAULT_VAL, DESC, \
|
|
|
|
DEVELOPMENT_STATUS, IS_HIDDEN);
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
|
|
|
|
#define GET_PACKAGE_OPTIONS
|
2020-02-28 22:07:50 +08:00
|
|
|
#define PACKAGE_OPTION(TYPE, FULLNAME, CMDFLAG, DESC, DEFAULT_VAL, \
|
|
|
|
DEVELOPMENT_STATUS, IS_HIDDEN) \
|
|
|
|
addPackageOption(TYPE, FULLNAME, CMDFLAG, DEFAULT_VAL, DESC, \
|
|
|
|
DEVELOPMENT_STATUS, IS_HIDDEN);
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
#include "clang/StaticAnalyzer/Checkers/Checkers.inc"
|
|
|
|
#undef CHECKER_DEPENDENCY
|
|
|
|
#undef GET_CHECKER_DEPENDENCIES
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
#undef CHECKER_WEAK_DEPENDENCY
|
|
|
|
#undef GET_CHECKER_WEAK_DEPENDENCIES
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
#undef CHECKER_OPTION
|
|
|
|
#undef GET_CHECKER_OPTIONS
|
|
|
|
#undef PACKAGE_OPTION
|
|
|
|
#undef GET_PACKAGE_OPTIONS
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
resolveDependencies<true>();
|
|
|
|
resolveDependencies<false>();
|
|
|
|
|
2020-06-16 15:44:02 +08:00
|
|
|
#ifndef NDEBUG
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
for (auto &DepPair : Data.Dependencies) {
|
|
|
|
for (auto &WeakDepPair : Data.WeakDependencies) {
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
// Some assertions to enforce that strong dependencies are relations in
|
|
|
|
// between purely modeling checkers, and weak dependencies are about
|
|
|
|
// diagnostics.
|
|
|
|
assert(WeakDepPair != DepPair &&
|
|
|
|
"A checker cannot strong and weak depend on the same checker!");
|
|
|
|
assert(WeakDepPair.first != DepPair.second &&
|
|
|
|
"A strong dependency mustn't have weak dependencies!");
|
|
|
|
assert(WeakDepPair.second != DepPair.second &&
|
|
|
|
"A strong dependency mustn't be a weak dependency as well!");
|
|
|
|
}
|
|
|
|
}
|
2020-06-12 21:42:29 +08:00
|
|
|
#endif
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
resolveCheckerAndPackageOptions();
|
2019-04-19 19:01:35 +08:00
|
|
|
|
2019-01-27 00:35:33 +08:00
|
|
|
// Parse '-analyzer-checker' and '-analyzer-disable-checker' options from the
|
|
|
|
// command line.
|
2019-08-16 09:53:14 +08:00
|
|
|
for (const std::pair<std::string, bool> &Opt : AnOpts.CheckersAndPackages) {
|
2019-04-18 23:19:16 +08:00
|
|
|
CheckerInfoListRange CheckerForCmdLineArg =
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.getMutableCheckersForCmdLineArg(Opt.first);
|
2011-08-17 05:24:21 +08:00
|
|
|
|
2019-04-18 23:19:16 +08:00
|
|
|
if (CheckerForCmdLineArg.begin() == CheckerForCmdLineArg.end()) {
|
2019-08-16 09:53:14 +08:00
|
|
|
Diags.Report(diag::err_unknown_analyzer_checker_or_package) << Opt.first;
|
2019-01-27 00:35:33 +08:00
|
|
|
Diags.Report(diag::note_suggest_disabling_all_checkers);
|
|
|
|
}
|
2011-08-17 05:24:21 +08:00
|
|
|
|
2019-04-18 23:19:16 +08:00
|
|
|
for (CheckerInfo &checker : CheckerForCmdLineArg) {
|
2019-04-19 01:32:51 +08:00
|
|
|
checker.State = Opt.second ? StateFromCmdLine::State_Enabled
|
|
|
|
: StateFromCmdLine::State_Disabled;
|
2019-01-27 00:35:33 +08:00
|
|
|
}
|
|
|
|
}
|
2020-02-28 22:07:50 +08:00
|
|
|
validateCheckerOptions();
|
|
|
|
}
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Dependency resolving.
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
|
|
|
template <typename IsEnabledFn>
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
static bool collectStrongDependencies(const ConstCheckerInfoList &Deps,
|
|
|
|
const CheckerManager &Mgr,
|
|
|
|
CheckerInfoSet &Ret,
|
|
|
|
IsEnabledFn IsEnabled);
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
/// Collects weak dependencies in \p enabledData.Checkers.
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
template <typename IsEnabledFn>
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
static void collectWeakDependencies(const ConstCheckerInfoList &Deps,
|
|
|
|
const CheckerManager &Mgr,
|
|
|
|
CheckerInfoSet &Ret, IsEnabledFn IsEnabled);
|
2020-02-28 22:07:50 +08:00
|
|
|
|
|
|
|
void CheckerRegistry::initializeRegistry(const CheckerManager &Mgr) {
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
// First, we calculate the list of enabled checkers as specified by the
|
|
|
|
// invocation. Weak dependencies will not enable their unspecified strong
|
|
|
|
// depenencies, but its only after resolving strong dependencies for all
|
|
|
|
// checkers when we know whether they will be enabled.
|
|
|
|
CheckerInfoSet Tmp;
|
|
|
|
auto IsEnabledFromCmdLine = [&](const CheckerInfo *Checker) {
|
|
|
|
return !Checker->isDisabled(Mgr);
|
|
|
|
};
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
for (const CheckerInfo &Checker : Data.Checkers) {
|
2020-03-27 21:29:31 +08:00
|
|
|
if (!Checker.isEnabled(Mgr))
|
2020-02-28 22:07:50 +08:00
|
|
|
continue;
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
CheckerInfoSet Deps;
|
|
|
|
if (!collectStrongDependencies(Checker.Dependencies, Mgr, Deps,
|
|
|
|
IsEnabledFromCmdLine)) {
|
2020-02-28 22:07:50 +08:00
|
|
|
// If we failed to enable any of the dependencies, don't enable this
|
|
|
|
// checker.
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
Tmp.insert(Deps.begin(), Deps.end());
|
2020-02-28 22:07:50 +08:00
|
|
|
|
|
|
|
// Enable the checker.
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
Tmp.insert(&Checker);
|
2020-02-28 22:07:50 +08:00
|
|
|
}
|
2011-08-17 05:24:21 +08:00
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
// Calculate enabled checkers with the correct registration order. As this is
|
|
|
|
// done recursively, its arguably cheaper, but for sure less error prone to
|
|
|
|
// recalculate from scratch.
|
|
|
|
auto IsEnabled = [&](const CheckerInfo *Checker) {
|
|
|
|
return llvm::is_contained(Tmp, Checker);
|
|
|
|
};
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
for (const CheckerInfo &Checker : Data.Checkers) {
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
if (!Checker.isEnabled(Mgr))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
CheckerInfoSet Deps;
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
collectWeakDependencies(Checker.WeakDependencies, Mgr, Deps, IsEnabled);
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
if (!collectStrongDependencies(Checker.Dependencies, Mgr, Deps,
|
|
|
|
IsEnabledFromCmdLine)) {
|
|
|
|
// If we failed to enable any of the dependencies, don't enable this
|
|
|
|
// checker.
|
|
|
|
continue;
|
|
|
|
}
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
// Note that set_union also preserves the order of insertion.
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.EnabledCheckers.set_union(Deps);
|
|
|
|
Data.EnabledCheckers.insert(&Checker);
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
}
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
}
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
template <typename IsEnabledFn>
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
static bool collectStrongDependencies(const ConstCheckerInfoList &Deps,
|
|
|
|
const CheckerManager &Mgr,
|
|
|
|
CheckerInfoSet &Ret,
|
|
|
|
IsEnabledFn IsEnabled) {
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
for (const CheckerInfo *Dependency : Deps) {
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
if (!IsEnabled(Dependency))
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
return false;
|
|
|
|
|
|
|
|
// Collect dependencies recursively.
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
if (!collectStrongDependencies(Dependency->Dependencies, Mgr, Ret,
|
|
|
|
IsEnabled))
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
return false;
|
2019-04-18 23:19:16 +08:00
|
|
|
Ret.insert(Dependency);
|
[analyzer] Reimplement dependencies between checkers
Unfortunately, up until now, the fact that certain checkers depended on one
another was known, but how these actually unfolded was hidden deep within the
implementation. For example, many checkers (like RetainCount, Malloc or CString)
modelled a certain functionality, and exposed certain reportable bug types to
the user. For example, while MallocChecker models many many different types of
memory handling, the actual "unix.MallocChecker" checker the user was exposed to
was merely and option to this modeling part.
Other than this being an ugly mess, this issue made resolving the checker naming
issue almost impossible. (The checker naming issue being that if a checker
registered more than one checker within its registry function, both checker
object recieved the same name) Also, if the user explicitly disabled a checker
that was a dependency of another that _was_ explicitly enabled, it implicitly,
without "telling" the user, reenabled it.
Clearly, changing this to a well structured, declarative form, where the
handling of dependencies are done on a higher level is very much preferred.
This patch, among the detailed things later, makes checkers declare their
dependencies within the TableGen file Checkers.td, and exposes the same
functionality to plugins and statically linked non-generated checkers through
CheckerRegistry::addDependency. CheckerRegistry now resolves these dependencies,
makes sure that checkers are added to CheckerManager in the correct order,
and makes sure that if a dependency is disabled, so will be every checker that
depends on it.
In detail:
* Add a new field to the Checker class in CheckerBase.td called Dependencies,
which is a list of Checkers.
* Move unix checkers before cplusplus, as there is no forward declaration in
tblgen :/
* Add the following new checkers:
- StackAddrEscapeBase
- StackAddrEscapeBase
- CStringModeling
- DynamicMemoryModeling (base of the MallocChecker family)
- IteratorModeling (base of the IteratorChecker family)
- ValistBase
- SecuritySyntaxChecker (base of bcmp, bcopy, etc...)
- NSOrCFErrorDerefChecker (base of NSErrorChecker and CFErrorChecker)
- IvarInvalidationModeling (base of IvarInvalidation checker family)
- RetainCountBase (base of RetainCount and OSObjectRetainCount)
* Clear up and registry functions in MallocChecker, happily remove old FIXMEs.
* Add a new addDependency function to CheckerRegistry.
* Neatly format RUN lines in files I looked at while debugging.
Big thanks to Artem Degrachev for all the guidance through this project!
Differential Revision: https://reviews.llvm.org/D54438
llvm-svn: 352287
2019-01-27 04:06:54 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
template <typename IsEnabledFn>
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
static void collectWeakDependencies(const ConstCheckerInfoList &WeakDeps,
|
|
|
|
const CheckerManager &Mgr,
|
|
|
|
CheckerInfoSet &Ret,
|
|
|
|
IsEnabledFn IsEnabled) {
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
for (const CheckerInfo *Dependency : WeakDeps) {
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
// Don't enable this checker if strong dependencies are unsatisfied, but
|
|
|
|
// assume that weak dependencies are transitive.
|
|
|
|
collectWeakDependencies(Dependency->WeakDependencies, Mgr, Ret, IsEnabled);
|
|
|
|
|
|
|
|
if (IsEnabled(Dependency) &&
|
|
|
|
collectStrongDependencies(Dependency->Dependencies, Mgr, Ret,
|
|
|
|
IsEnabled))
|
|
|
|
Ret.insert(Dependency);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <bool IsWeak> void CheckerRegistry::resolveDependencies() {
|
|
|
|
for (const std::pair<StringRef, StringRef> &Entry :
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
(IsWeak ? Data.WeakDependencies : Data.Dependencies)) {
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
auto CheckerIt = binaryFind(Data.Checkers, Entry.first);
|
|
|
|
assert(CheckerIt != Data.Checkers.end() &&
|
|
|
|
CheckerIt->FullName == Entry.first &&
|
2019-04-19 19:01:35 +08:00
|
|
|
"Failed to find the checker while attempting to set up its "
|
|
|
|
"dependencies!");
|
2019-04-18 23:19:16 +08:00
|
|
|
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
auto DependencyIt = binaryFind(Data.Checkers, Entry.second);
|
|
|
|
assert(DependencyIt != Data.Checkers.end() &&
|
2019-04-19 19:01:35 +08:00
|
|
|
DependencyIt->FullName == Entry.second &&
|
|
|
|
"Failed to find the dependency of a checker!");
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
if (IsWeak)
|
|
|
|
CheckerIt->WeakDependencies.emplace_back(&*DependencyIt);
|
|
|
|
else
|
|
|
|
CheckerIt->Dependencies.emplace_back(&*DependencyIt);
|
2019-04-19 19:01:35 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void CheckerRegistry::addDependency(StringRef FullName, StringRef Dependency) {
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.Dependencies.emplace_back(FullName, Dependency);
|
2019-04-18 23:19:16 +08:00
|
|
|
}
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
void CheckerRegistry::addWeakDependency(StringRef FullName,
|
|
|
|
StringRef Dependency) {
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.WeakDependencies.emplace_back(FullName, Dependency);
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Checker option resolving and validating.
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
2019-05-17 17:51:59 +08:00
|
|
|
/// Insert the checker/package option to AnalyzerOptions' config table, and
|
|
|
|
/// validate it, if the user supplied it on the command line.
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
static void insertAndValidate(StringRef FullName, const CmdLineOption &Option,
|
2019-05-17 17:51:59 +08:00
|
|
|
AnalyzerOptions &AnOpts,
|
|
|
|
DiagnosticsEngine &Diags) {
|
|
|
|
|
|
|
|
std::string FullOption = (FullName + ":" + Option.OptionName).str();
|
|
|
|
|
2020-01-29 03:23:46 +08:00
|
|
|
auto It =
|
|
|
|
AnOpts.Config.insert({FullOption, std::string(Option.DefaultValStr)});
|
2019-05-17 17:51:59 +08:00
|
|
|
|
|
|
|
// Insertation was successful -- CmdLineOption's constructor will validate
|
|
|
|
// whether values received from plugins or TableGen files are correct.
|
|
|
|
if (It.second)
|
|
|
|
return;
|
|
|
|
|
|
|
|
// Insertion failed, the user supplied this package/checker option on the
|
[analyzer] Remove the default value arg from getChecker*Option
Since D57922, the config table contains every checker option, and it's default
value, so having it as an argument for getChecker*Option is redundant.
By the time any of the getChecker*Option function is called, we verified the
value in CheckerRegistry (after D57860), so we can confidently assert here, as
any irregularities detected at this point must be a programmer error. However,
in compatibility mode, verification won't happen, so the default value must be
restored.
This implies something else, other than adding removing one more potential point
of failure -- debug.ConfigDumper will always contain valid values for
checker/package options!
Differential Revision: https://reviews.llvm.org/D59195
llvm-svn: 361042
2019-05-17 23:52:13 +08:00
|
|
|
// command line. If the supplied value is invalid, we'll restore the option
|
|
|
|
// to it's default value, and if we're in non-compatibility mode, we'll also
|
|
|
|
// emit an error.
|
2019-05-17 17:51:59 +08:00
|
|
|
|
|
|
|
StringRef SuppliedValue = It.first->getValue();
|
|
|
|
|
|
|
|
if (Option.OptionType == "bool") {
|
|
|
|
if (SuppliedValue != "true" && SuppliedValue != "false") {
|
|
|
|
if (AnOpts.ShouldEmitErrorsOnInvalidConfigValue) {
|
|
|
|
Diags.Report(diag::err_analyzer_checker_option_invalid_input)
|
|
|
|
<< FullOption << "a boolean value";
|
|
|
|
}
|
[analyzer] Remove the default value arg from getChecker*Option
Since D57922, the config table contains every checker option, and it's default
value, so having it as an argument for getChecker*Option is redundant.
By the time any of the getChecker*Option function is called, we verified the
value in CheckerRegistry (after D57860), so we can confidently assert here, as
any irregularities detected at this point must be a programmer error. However,
in compatibility mode, verification won't happen, so the default value must be
restored.
This implies something else, other than adding removing one more potential point
of failure -- debug.ConfigDumper will always contain valid values for
checker/package options!
Differential Revision: https://reviews.llvm.org/D59195
llvm-svn: 361042
2019-05-17 23:52:13 +08:00
|
|
|
|
2020-01-29 03:23:46 +08:00
|
|
|
It.first->setValue(std::string(Option.DefaultValStr));
|
2019-05-17 17:51:59 +08:00
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (Option.OptionType == "int") {
|
|
|
|
int Tmp;
|
|
|
|
bool HasFailed = SuppliedValue.getAsInteger(0, Tmp);
|
|
|
|
if (HasFailed) {
|
|
|
|
if (AnOpts.ShouldEmitErrorsOnInvalidConfigValue) {
|
|
|
|
Diags.Report(diag::err_analyzer_checker_option_invalid_input)
|
|
|
|
<< FullOption << "an integer value";
|
|
|
|
}
|
[analyzer] Remove the default value arg from getChecker*Option
Since D57922, the config table contains every checker option, and it's default
value, so having it as an argument for getChecker*Option is redundant.
By the time any of the getChecker*Option function is called, we verified the
value in CheckerRegistry (after D57860), so we can confidently assert here, as
any irregularities detected at this point must be a programmer error. However,
in compatibility mode, verification won't happen, so the default value must be
restored.
This implies something else, other than adding removing one more potential point
of failure -- debug.ConfigDumper will always contain valid values for
checker/package options!
Differential Revision: https://reviews.llvm.org/D59195
llvm-svn: 361042
2019-05-17 23:52:13 +08:00
|
|
|
|
2020-01-29 03:23:46 +08:00
|
|
|
It.first->setValue(std::string(Option.DefaultValStr));
|
2019-05-17 17:51:59 +08:00
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
template <class T>
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
static void insertOptionToCollection(StringRef FullName, T &Collection,
|
|
|
|
const CmdLineOption &Option,
|
|
|
|
AnalyzerOptions &AnOpts,
|
|
|
|
DiagnosticsEngine &Diags) {
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
auto It = binaryFind(Collection, FullName);
|
|
|
|
assert(It != Collection.end() &&
|
|
|
|
"Failed to find the checker while attempting to add a command line "
|
|
|
|
"option to it!");
|
|
|
|
|
2019-05-17 17:51:59 +08:00
|
|
|
insertAndValidate(FullName, Option, AnOpts, Diags);
|
2019-05-17 17:29:44 +08:00
|
|
|
|
|
|
|
It->CmdLineOptions.emplace_back(Option);
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
void CheckerRegistry::resolveCheckerAndPackageOptions() {
|
|
|
|
for (const std::pair<StringRef, CmdLineOption> &CheckerOptEntry :
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.CheckerOptions) {
|
|
|
|
insertOptionToCollection(CheckerOptEntry.first, Data.Checkers,
|
2019-05-17 17:51:59 +08:00
|
|
|
CheckerOptEntry.second, AnOpts, Diags);
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
for (const std::pair<StringRef, CmdLineOption> &PackageOptEntry :
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.PackageOptions) {
|
|
|
|
insertOptionToCollection(PackageOptEntry.first, Data.Packages,
|
2019-05-17 17:51:59 +08:00
|
|
|
PackageOptEntry.second, AnOpts, Diags);
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void CheckerRegistry::addPackage(StringRef FullName) {
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.Packages.emplace_back(PackageInfo(FullName));
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
void CheckerRegistry::addPackageOption(StringRef OptionType,
|
|
|
|
StringRef PackageFullName,
|
|
|
|
StringRef OptionName,
|
|
|
|
StringRef DefaultValStr,
|
2019-05-24 06:52:09 +08:00
|
|
|
StringRef Description,
|
|
|
|
StringRef DevelopmentStatus,
|
|
|
|
bool IsHidden) {
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.PackageOptions.emplace_back(
|
2019-05-24 06:07:16 +08:00
|
|
|
PackageFullName, CmdLineOption{OptionType, OptionName, DefaultValStr,
|
2019-05-24 06:52:09 +08:00
|
|
|
Description, DevelopmentStatus, IsHidden});
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
}
|
|
|
|
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
void CheckerRegistry::addChecker(RegisterCheckerFn Rfn,
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
ShouldRegisterFunction Sfn, StringRef Name,
|
[analyzer] Don't display implementation checkers under -analyzer-checker-help, but do under the new flag -analyzer-checker-help-hidden
During my work on analyzer dependencies, I created a great amount of new
checkers that emitted no diagnostics at all, and were purely modeling some
function or another.
However, the user shouldn't really disable/enable these by hand, hence this
patch, which hides these by default. I intentionally chose not to hide alpha
checkers, because they have a scary enough name, in my opinion, to cause no
surprise when they emit false positives or cause crashes.
The patch introduces the Hidden bit into the TableGen files (you may remember
it before I removed it in D53995), and checkers that are either marked as
hidden, or are in a package that is marked hidden won't be displayed under
-analyzer-checker-help. -analyzer-checker-help-hidden, a new flag meant for
developers only, displays the full list.
Differential Revision: https://reviews.llvm.org/D60925
llvm-svn: 359720
2019-05-02 03:56:47 +08:00
|
|
|
StringRef Desc, StringRef DocsUri,
|
|
|
|
bool IsHidden) {
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.Checkers.emplace_back(Rfn, Sfn, Name, Desc, DocsUri, IsHidden);
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
|
|
|
|
// Record the presence of the checker in its packages.
|
|
|
|
StringRef PackageName, LeafName;
|
|
|
|
std::tie(PackageName, LeafName) = Name.rsplit(PackageSeparator);
|
|
|
|
while (!LeafName.empty()) {
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.PackageSizes[PackageName] += 1;
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
std::tie(PackageName, LeafName) = PackageName.rsplit(PackageSeparator);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void CheckerRegistry::addCheckerOption(StringRef OptionType,
|
|
|
|
StringRef CheckerFullName,
|
|
|
|
StringRef OptionName,
|
|
|
|
StringRef DefaultValStr,
|
2019-05-24 06:52:09 +08:00
|
|
|
StringRef Description,
|
|
|
|
StringRef DevelopmentStatus,
|
|
|
|
bool IsHidden) {
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
Data.CheckerOptions.emplace_back(
|
2019-05-24 06:07:16 +08:00
|
|
|
CheckerFullName, CmdLineOption{OptionType, OptionName, DefaultValStr,
|
2019-05-24 06:52:09 +08:00
|
|
|
Description, DevelopmentStatus, IsHidden});
|
[analyzer][NFC] Reimplement checker options
TL;DR:
* Add checker and package options to the TableGen files
* Added a new class called CmdLineOption, and both Package and Checker recieved
a list<CmdLineOption> field.
* Added every existing checker and package option to Checkers.td.
* The CheckerRegistry class
* Received some comments to most of it's inline classes
* Received the CmdLineOption and PackageInfo inline classes, a list of
CmdLineOption was added to CheckerInfo and PackageInfo
* Added addCheckerOption and addPackageOption
* Added a new field called Packages, used in addPackageOptions, filled up in
addPackage
Detailed description:
In the last couple months, a lot of effort was put into tightening the
analyzer's command line interface. The main issue is that it's spectacularly
easy to mess up a lenghty enough invocation of the analyzer, and the user was
given no warnings or errors at all in that case.
We can divide the effort of resolving this into several chapters:
* Non-checker analyzer configurations:
Gather every analyzer configuration into a dedicated file. Emit errors for
non-existent configurations or incorrect values. Be able to list these
configurations. Tighten AnalyzerOptions interface to disallow making such
a mistake in the future.
* Fix the "Checker Naming Bug" by reimplementing checker dependencies:
When cplusplus.InnerPointer was enabled, it implicitly registered
unix.Malloc, which implicitly registered some sort of a modeling checker
from the CStringChecker family. This resulted in all of these checker
objects recieving the name "cplusplus.InnerPointer", making AnalyzerOptions
asking for the wrong checker options from the command line:
cplusplus.InnerPointer:Optimisic
istead of
unix.Malloc:Optimistic.
This was resolved by making CheckerRegistry responsible for checker
dependency handling, instead of checkers themselves.
* Checker options: (this patch included!)
Same as the first item, but for checkers.
(+ minor fixes here and there, and everything else that is yet to come)
There were several issues regarding checker options, that non-checker
configurations didn't suffer from: checker plugins are loaded runtime, and they
could add new checkers and new options, meaning that unlike for non-checker
configurations, we can't collect every checker option purely by generating code.
Also, as seen from the "Checker Naming Bug" issue raised above, they are very
rarely used in practice, and all sorts of skeletons fell out of the closet while
working on this project.
They were extremely problematic for users as well, purely because of how long
they were. Consider the following monster of a checker option:
alpha.cplusplus.UninitializedObject:CheckPointeeInitialization=false
While we were able to verify whether the checker itself (the part before the
colon) existed, any errors past that point were unreported, easily resulting
in 7+ hours of analyses going to waste.
This patch, similarly to how dependencies were reimplemented, uses TableGen to
register checker options into Checkers.td, so that Checkers.inc now contains
entries for both checker and package options. Using the preprocessor,
Checkers.inc is converted into code in CheckerRegistry, adding every builtin
(checkers and packages that have an entry in the Checkers.td file) checker and
package option to the registry. The new addPackageOption and addCheckerOption
functions expose the same functionality to statically-linked non-builtin and
plugin checkers and packages as well.
Emitting errors for incorrect user input, being able to list these options, and
some other functionalies will land in later patches.
Differential Revision: https://reviews.llvm.org/D57855
llvm-svn: 358752
2019-04-19 20:32:10 +08:00
|
|
|
}
|
|
|
|
|
2019-04-18 23:19:16 +08:00
|
|
|
void CheckerRegistry::initializeManager(CheckerManager &CheckerMgr) const {
|
2011-08-17 05:24:21 +08:00
|
|
|
// Initialize the CheckerManager with all enabled checkers.
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
for (const auto *Checker : Data.EnabledCheckers) {
|
2019-09-13 03:09:24 +08:00
|
|
|
CheckerMgr.setCurrentCheckerName(CheckerNameRef(Checker->FullName));
|
2019-04-18 23:19:16 +08:00
|
|
|
Checker->Initialize(CheckerMgr);
|
2011-08-17 05:24:21 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
static void isOptionContainedIn(const CmdLineOptionList &OptionList,
|
|
|
|
StringRef SuppliedChecker,
|
|
|
|
StringRef SuppliedOption,
|
|
|
|
const AnalyzerOptions &AnOpts,
|
|
|
|
DiagnosticsEngine &Diags) {
|
2019-05-17 17:51:59 +08:00
|
|
|
|
|
|
|
if (!AnOpts.ShouldEmitErrorsOnInvalidConfigValue)
|
|
|
|
return;
|
|
|
|
|
|
|
|
auto SameOptName = [SuppliedOption](const CmdLineOption &Opt) {
|
|
|
|
return Opt.OptionName == SuppliedOption;
|
|
|
|
};
|
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
const auto *OptionIt = llvm::find_if(OptionList, SameOptName);
|
2019-05-17 17:51:59 +08:00
|
|
|
|
|
|
|
if (OptionIt == OptionList.end()) {
|
|
|
|
Diags.Report(diag::err_analyzer_checker_option_unknown)
|
|
|
|
<< SuppliedChecker << SuppliedOption;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-01-26 23:59:21 +08:00
|
|
|
void CheckerRegistry::validateCheckerOptions() const {
|
2019-04-18 23:19:16 +08:00
|
|
|
for (const auto &Config : AnOpts.Config) {
|
2019-05-17 17:51:59 +08:00
|
|
|
|
2019-08-16 09:53:14 +08:00
|
|
|
StringRef SuppliedCheckerOrPackage;
|
2019-05-17 17:51:59 +08:00
|
|
|
StringRef SuppliedOption;
|
2019-08-16 09:53:14 +08:00
|
|
|
std::tie(SuppliedCheckerOrPackage, SuppliedOption) =
|
|
|
|
Config.getKey().split(':');
|
2019-05-17 17:51:59 +08:00
|
|
|
|
|
|
|
if (SuppliedOption.empty())
|
2015-07-10 05:43:45 +08:00
|
|
|
continue;
|
|
|
|
|
2019-05-17 17:51:59 +08:00
|
|
|
// AnalyzerOptions' config table contains the user input, so an entry could
|
|
|
|
// look like this:
|
|
|
|
//
|
|
|
|
// cor:NoFalsePositives=true
|
|
|
|
//
|
|
|
|
// Since lower_bound would look for the first element *not less* than "cor",
|
|
|
|
// it would return with an iterator to the first checker in the core, so we
|
|
|
|
// we really have to use find here, which uses operator==.
|
2019-08-16 09:53:14 +08:00
|
|
|
auto CheckerIt =
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
llvm::find(Data.Checkers, CheckerInfo(SuppliedCheckerOrPackage));
|
|
|
|
if (CheckerIt != Data.Checkers.end()) {
|
2019-08-16 09:53:14 +08:00
|
|
|
isOptionContainedIn(CheckerIt->CmdLineOptions, SuppliedCheckerOrPackage,
|
2019-05-17 17:51:59 +08:00
|
|
|
SuppliedOption, AnOpts, Diags);
|
|
|
|
continue;
|
2015-07-10 05:43:45 +08:00
|
|
|
}
|
2019-05-17 17:51:59 +08:00
|
|
|
|
[analyzer] Introduce weak dependencies to express *preferred* checker callback evaluation order
Checker dependencies were added D54438 to solve a bug where the checker names
were incorrectly registered, for example, InnerPointerChecker would incorrectly
emit diagnostics under the name MallocChecker, or vice versa [1]. Since the
system over the course of about a year matured, our expectations of what a role
of a dependency and a dependent checker should be crystallized a bit more --
D77474 and its summary, as well as a variety of patches in the stack
demonstrates how we try to keep dependencies to play a purely modeling role. In
fact, D78126 outright forbids diagnostics under a dependency checkers name.
These dependencies ensured the registration order and enabling only when all
dependencies are satisfied. This was a very "strong" contract however, that
doesn't fit the dependency added in D79420. As its summary suggests, this
relation is directly in between diagnostics, not modeling -- we'd prefer a more
specific warning over a general one.
To support this, I added a new dependency kind, weak dependencies. These are not
as strict of a contract, they only express a preference in registration order.
If a weak dependency isn't satisfied, the checker may still be enabled, but if
it is, checker registration, and transitively, checker callback evaluation order
is ensured.
If you are not familiar with the TableGen changes, a rather short description
can be found in the summary of D75360. A lengthier one is in D58065.
[1] https://www.youtube.com/watch?v=eqKeqHRAhQM
Differential Revision: https://reviews.llvm.org/D80905
2020-05-27 18:29:47 +08:00
|
|
|
const auto *PackageIt =
|
[analyzer][NFC] Move the data structures from CheckerRegistry to the Core library
If you were around the analyzer for a while now, you must've seen a lot of
patches that awkwardly puts code from one library to the other:
* D75360 moves the constructors of CheckerManager, which lies in the Core
library, to the Frontend library. Most the patch itself was a struggle along
the library lines.
* D78126 had to be reverted because dependency information would be utilized
in the Core library, but the actual data lied in the frontend.
D78126#inline-751477 touches on this issue as well.
This stems from the often mentioned problem: the Frontend library depends on
Core and Checkers, Checkers depends on Core. The checker registry functions
(`registerMallocChecker`, etc) lie in the Checkers library in order to keep each
checker its own module. What this implies is that checker registration cannot
take place in the Core, but the Core might still want to use the data that
results from it (which checker/package is enabled, dependencies, etc).
D54436 was the patch that initiated this. Back in the days when CheckerRegistry
was super dumb and buggy, it implemented a non-documented solution to this
problem by keeping the data in the Core, and leaving the logic in the Frontend.
At the time when the patch landed, the merger to the Frontend made sense,
because the data hadn't been utilized anywhere, and the whole workaround without
any documentation made little sense to me.
So, lets put the data back where it belongs, in the Core library. This patch
introduces `CheckerRegistryData`, and turns `CheckerRegistry` into a short lived
wrapper around this data that implements the logic of checker registration. The
data is tied to CheckerManager because it is required to parse it.
Side note: I can't help but cringe at the fact how ridiculously awkward the
library lines are. I feel like I'm thinking too much inside the box, but I guess
this is just the price of keeping the checkers so modularized.
Differential Revision: https://reviews.llvm.org/D82585
2020-06-19 03:40:43 +08:00
|
|
|
llvm::find(Data.Packages, PackageInfo(SuppliedCheckerOrPackage));
|
|
|
|
if (PackageIt != Data.Packages.end()) {
|
2019-08-16 09:53:14 +08:00
|
|
|
isOptionContainedIn(PackageIt->CmdLineOptions, SuppliedCheckerOrPackage,
|
2019-05-17 17:51:59 +08:00
|
|
|
SuppliedOption, AnOpts, Diags);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2019-08-16 09:53:14 +08:00
|
|
|
Diags.Report(diag::err_unknown_analyzer_checker_or_package)
|
|
|
|
<< SuppliedCheckerOrPackage;
|
2015-07-10 05:43:45 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|