Summary:public
With this change, all the `infer-deps.txt` files generated by buck for those targets
running with the `#infer` flavor, will be merged into one `infer-deps.txt` located in the
designated output folder.
Reviewed By: jvillard
Differential Revision: D2994397
fb-gh-sync-id: 14d8109
shipit-source-id: 14d8109
Summary:public
Lazy dynamic dispatch handling works as follows:
Assuming a call of the form:
foo(a);
where the static type of `a` is `A`. If during the symbolic execution, the dynamic type of the variable `a` is `B` where `B <: A`, then we create on-demand a copy `foo(B)` of `foo(A)` where all the uses of the typed parameter `a` are replaced with a parameter of type `B`. Especially, if `foo` contains virtual call, say `get` where `a` is the receiver, then the call gets redirected to the overridden method in `B`, which simulates the runtime behavior of Java.
This lazy dynamic dispatch mode is only turn on for the tracing mode for now in order to avoid conflicts with sblackshear's approach for sound dynamic dispatch.
Reviewed By: sblackshear
Differential Revision: D2888922
fb-gh-sync-id: 3250c9e
shipit-source-id: 3250c9e
Summary:public
Remove back-end infrastructure that exists only when on-demand mode is disabled.
This, together with removing a few command-line options, sheds a lot of weight in the back-end.
No changes expected for on-demand mode.
Reviewed By: sblackshear
Differential Revision: D2960242
fb-gh-sync-id: 220d821
shipit-source-id: 220d821
Summary:public
Deprecate the incremental mode.
Several parts of the back-end can be removed.
The options for incremental analysis -i at the python level are now deprecated, and re-routed to --reactive.
The main difference with --reactive is that it does not produce an analysis of the whole project, but is limited to what is reachable via reactive propagation starting from the changed files.
Reviewed By: sblackshear
Differential Revision: D2960078
fb-gh-sync-id: 6e8b46b
shipit-source-id: 6e8b46b
Summary:public
An observer object that registered to a notification center needs to be
unregistered before it is deallocated.
If not, the notification center may send a notification to a gost object.
This diff introduce a checker for this problem.
Reviewed By: dulmarod
Differential Revision: D2949692
fb-gh-sync-id: 1653cec
shipit-source-id: 1653cec
Summary:public
In tracing mode, we translate the runtime checks done by the JVM, so the checks for null happen independently from the what happens before the dereference.
Reviewed By: cristianoc
Differential Revision: D2981515
fb-gh-sync-id: 695de07
shipit-source-id: 695de07
Summary:public
`print(tracer)` has an automatic conversion to string that was missed when
migrating to `utils.stdout(tracer)`.
Reviewed By: peterogithub
Differential Revision: D2982104
fb-gh-sync-id: 4c26727
shipit-source-id: 4c26727
Summary:public
- s/"/'/ in python strings
- kill `utils.error()` in favour of the new, identical `utils.stderr()`
- one more `print(utils.encode())` to `utils.stderr()` conversion
Reviewed By: jeremydubreil
Differential Revision: D2976710
fb-gh-sync-id: 6c0fdfa
shipit-source-id: 6c0fdfa
Summary:public
Revamped Timeout module by storing elapsed wallclock seconds, and the status of symops, in case of recursive calls.
Extended the API with suspend() and resume() to pause and resume the current timeout.
These are used before and after an on-demand call to the analysis functions.
This achieves the effect that each procedure, even though is interrupted, has its own time and symop counters, which are suspended and resumed as required.
Reviewed By: jeremydubreil
Differential Revision: D2976918
fb-gh-sync-id: 0ed1079
shipit-source-id: 0ed1079
Summary:public
This class expects a working `jwlib.CompilerCommand` even when we're not doing
anything Java-related. Split the java-specific functionality into a new child
class in jwlib.py.
Reviewed By: jeremydubreil
Differential Revision: D2965832
fb-gh-sync-id: e895b33
shipit-source-id: e895b33
Summary:public
Improved/simplified framework for fronend checkers.
Now we have a unique hook from cTrans to run checkers on statements and a unique
hook from cFrontent to run checkers on declarations.
So now when adding a checker we don't have to modify cTrans/cFrontend.
Moreover made more sistematic the way checkers are invoked. This simplify the definition
of checkers and the way we use them.
Code is now simpler.
Reviewed By: jvillard
Differential Revision: D2976589
fb-gh-sync-id: fbe22d4
shipit-source-id: fbe22d4
Summary:public
java-specific code such as this belongs in jwlib.py. It will also help the
refactoring in the next diff.
Reviewed By: sblackshear
Differential Revision: D2965814
fb-gh-sync-id: c3adc03
shipit-source-id: c3adc03
Summary:public
The NoAllocation checker should not report on the creation of exceptions
Reviewed By: sblackshear
Differential Revision: D2969719
fb-gh-sync-id: 4a8ffc8
shipit-source-id: 4a8ffc8
Summary:public
Add extra dereference when accessing fields that have T& type. It is similar
to what is done when accessing variables of T& type.
The only difference is that we need to handle constructor initializer list
separately (this is the only place where the field can be initialized)
Reviewed By: ddino
Differential Revision: D2965887
fb-gh-sync-id: 1b8708b
shipit-source-id: 1b8708b
Summary:public
This just simplifies the end-to-end tests for Dividde By Zero to make the debugging easier when this test fails.
Reviewed By: cristianoc
Differential Revision: D2966296
fb-gh-sync-id: 5eaa1b5
shipit-source-id: 5eaa1b5
Summary:public
This will avoid a circular dependency between analyze.py and jwlib.py in an
upcoming refactoring.
Reviewed By: martinoluca
Differential Revision: D2965734
fb-gh-sync-id: 1cb69d4
shipit-source-id: 1cb69d4
Summary:public
This attempts to properly sanitise text input/output in the Python parts of
infer. Do three things:
- encode user input (coming from the command-line or reading files)
- decode infer output
- in both cases, we may be using the wrong encoding, eg: locale says we're in
ascii, but the source code contains utf-8. In many cases, like error
messages, it's safe to ignore these encoding mismatches.
Also, since we `import __future__.unicode_literals`, it's safe to remove `u'`
prefixes on many unicode literals.
Reviewed By: martinoluca
Differential Revision: D2960493
fb-gh-sync-id: 9812d7d
shipit-source-id: 9812d7d
Summary:public
Do same thing we do to CXXDefaultArgExpr
Reviewed By: dulmarod
Differential Revision: D2954128
fb-gh-sync-id: 2c92c16
shipit-source-id: 2c92c16
Summary:public
Bring merlin config in sync with makefile, in particular so that it can
find values and types in Utils, and so that it reports the same warnings.
Reviewed By: jvillard, cristianoc
Differential Revision: D2959425
fb-gh-sync-id: 9e7b443
shipit-source-id: 9e7b443
Summary:Resolve a bug that Infer does not analyze procedures when the analyzer is run on '/'.
This bug is reported by btakeya on facebook/infer#283.
Closes https://github.com/facebook/infer/pull/284
Reviewed By: cristianoc
Differential Revision: D2960328
Pulled By: jvillard
fb-gh-sync-id: 55e8b0c
shipit-source-id: 55e8b0c
Summary:public
When compiling projects with this macro set to 1 (which is default on my mac),
infer couldn't get specs for some calls. They were replaced with
different functions by preprocessor.
Reviewed By: ddino
Differential Revision: D2944618
fb-gh-sync-id: df8b457
shipit-source-id: df8b457
Summary:public
Add command-line argument --reactive to enable reactive propagation mode.
When the mode is active, the files changed during compilation are detected, and the analysis propagates reactively starting from the modified files.
The reactive mode allows to analyze a subset of the files in a project and follow their dependencies, without storing the results of previous analyses (specs files). Captured files are preserved from previous runs of the analysis (for example, when the previous analysis was the initial capture), so the mode can be used repeatedly while changing code.
Reviewed By: jvillard
Differential Revision: D2931697
fb-gh-sync-id: 9d6dda0
shipit-source-id: 9d6dda0
Summary:public
Running clang-format on symbolic link replaced it with another file.
This is second diff out of two to fix that problem.
Reviewed By: jvillard
Differential Revision: D2960206
fb-gh-sync-id: 0f21a8f
shipit-source-id: 0f21a8f
Summary:public
Running clang-format on symbolic link replaced it with another file.
This is first diff out of two to fix that problem.
Reviewed By: jvillard
Differential Revision: D2960204
fb-gh-sync-id: aaa3231
shipit-source-id: aaa3231
Summary:public
Adds test_build target to toplevel Makefile, which compiles
InferAnalyze, InferPrint, InferClang, and InferJava where warnings
listed in OCAML_FATAL_WARNINGS defined in infer/src/Makefile are fatal.
Other builds do not treat warnings as fatal. The test build is
performed in parallel with the tests by scripts/test.sh. This makes
refactoring and debugging easier, but prevents warnings from slipping
into the repo.
Also, make test target robust wrt parallel make.
Reviewed By: jvillard
Differential Revision: D2953085
fb-gh-sync-id: 5c0282a
shipit-source-id: 5c0282a
Summary:public
This is intended as a target to build during development when quick
compilation and no duplicate compilation warnings are desired.
This builds byte code for InferAnalyze, InferPrint, InferJava, and
InferClang. This is one ocamlbuild invocation, so builds (and reports
warnings on) each source file once. Building bytecode is currently
about twice as fast as native code.
The llvm frontend is currently not built by this since it requires
menhir and standard ocamlyacc is required by the backend, and ocamlbuild
can only use one at a time.
Reviewed By: jvillard
Differential Revision: D2953050
fb-gh-sync-id: 2e90fcd
shipit-source-id: 2e90fcd
Summary:public
Create separate specs for C models compiled in C++. It will allow us to tweak behavior/names of certain
functions based on the compilation language (such as adding `std::` namespace in C++).
Reviewed By: jvillard
Differential Revision: D2938992
fb-gh-sync-id: 73902f8
shipit-source-id: 73902f8
Summary:public
Now use -safe-string and the Bytes module and bytes type.
Reviewed By: jvillard
Differential Revision: D2949369
fb-gh-sync-id: 58aa688
shipit-source-id: 58aa688
Summary:public
Names of templated types can be very long. The default 4k characters is not long
enough to handle folly library. Increase it to 16k characters and log when deserialization
fails.
Reviewed By: jberdine
Differential Revision: D2948935
fb-gh-sync-id: a659b23
shipit-source-id: a659b23
Summary:public
I have seen enough comments in this space by people during code review to switch on the analyses the compiler can already do. This diff is an automated renaming of unused identifiers to _, with a few additional changes made when reading the diff of the results for things that stood out as particularly strange. This base-lines all of the existing warnings. I'm not sure this is a good idea, since it might be better for those familiar with each part of the code to look at these warnings and use them as pointers to suspicious code.
Reviewed By: jeremydubreil
Differential Revision: D2938376
fb-gh-sync-id: 6e67817
shipit-source-id: 6e67817
Summary:public
Add to the code to detect violation of the `NoAllocation` annotation. This diff adds the code to detect such issue based on the code of the `PerformanceCritical` checker. In the next diff, I will refine the list of acceptable allocations, like new exceptions, etc, and add the list of corresponding tests.
Reviewed By: sblackshear
Differential Revision: D2938641
fb-gh-sync-id: 9a047dd
shipit-source-id: 9a047dd
Summary:public
Before this diff, the checker was collecting in a bottom-up fashion all possible call trees from `PerforamanceCritical`-annotated methods to `Expensive`-annotated ones. With this diff, we just collect the names of the direct transitively expensive callees and compute the expensive call stacks when reporting errors only.
Reviewed By: sblackshear
Differential Revision: D2938635
fb-gh-sync-id: dcdd13c
shipit-source-id: dcdd13c
Summary:public We model it as the builtin __instanceof which models the instanceof construct of Java.
The behaviour is the same.
Reviewed By: jvillard
Differential Revision: D2938969
fb-gh-sync-id: 2258de3
shipit-source-id: 2258de3
Summary:public
Translate headers every time they are included provided that they are located inside project_root directory.
While this is suboptimal (we might end up translating same header many times), doing it exactly once
is hard due to parallel compilation and template instantiations
Reviewed By: dulmarod
Differential Revision: D2916799
fb-gh-sync-id: 93b72c4
shipit-source-id: 93b72c4
Summary:public
Is seems that automatically inheriting annotations like `PerformanceCritical` or `NoAllocation` is the right thing to do in general. Otherwise, we need to enforce sub-typing rules which in the best case just adds a little bit of documentation, but could miss important issues when the code is not fully annotated. I am simplifying this part to avoid adding boilerplate code for the `NoAllocation` case.
Reviewed By: sblackshear
Differential Revision: D2938627
fb-gh-sync-id: ddb668b
shipit-source-id: ddb668b
Summary:public
Type-based resolution of fields, constructors, etc. can be ambiguous if
types are not principal. Compile with -principal and enable warnings 18
and 19 to check these cases.
Reviewed By: jvillard
Differential Revision: D2938237
fb-gh-sync-id: bb4237b
shipit-source-id: bb4237b
Summary:
public
This model does not seem to bring anything anymore. Useless because ...
Reviewed By: cristianoc
Differential Revision: D2920118
fb-gh-sync-id: 7f708d7
shipit-source-id: 7f708d7
Summary:
public
Avoid problems of overwriting good type information with incomplete information
when type declaration happens after its complete definition.
The solution is that we will only time we *update* type information is
when struct declaration has definition as well (which should happen once)
Reviewed By: cristianoc, sblackshear
Differential Revision: D2921811
fb-gh-sync-id: 16baba3
shipit-source-id: 16baba3
Summary:
public
The inductive list predicate was not firing during abstraction because of a type mismatch between C and Java. In Java, the second parameter of the `Sil.Sizeof` constructor is always `Sil.Subtype.exact` in C but is `Sil.Subtype.subtypes` in Java. This diff fixes the confution by comparing the `Sil` types only instead of the type expressions.
Reviewed By: jberdine
Differential Revision: D2912493
fb-gh-sync-id: 3f712a8
shipit-source-id: 3f712a8
Summary:
public
The analysis of the Buck project was failing because this script converts the aliases into their expanded target names. It turns out that for Buck, the name of the command is `buck` and the name of the alias is also `buck`, which led to conflicts. This is now fixed by only replacing the targets in the part of the command line that is after `build` in `buck build ...`
Reviewed By: sblackshear
Differential Revision: D2914446
fb-gh-sync-id: ddce4ed
shipit-source-id: ddce4ed
Summary:
public
Java synthetic methods used to be inlined when a procedure was being analyzed.
This was done almost everywhere. A missing case was when a cfg is loaded during an existing analysis because on-demand needs access to a procedure.
Intead of trying to maintain an invariant in all access paths, we now inline them systematically before saving the cfg to disk.
A secondary consequence of this is that in debug mode the cfg dotty file will show the inlined cfg, so there's no difference between that view and what happens during analysis.
Reviewed By: jeremydubreil
Differential Revision: D2903366
fb-gh-sync-id: 252604c
Summary:
public
The PerformanceCritical Checker was not running on Buck project due to a conflict between the incremental mode and the ondemand mode.
Reviewed By: cristianoc
Differential Revision: D2891649
fb-gh-sync-id: 0a503d8
Summary:
public
This expression is used to value-initialize non-class types. Per definition of value initialization for non-class types:
1. If it's an array, value-initialize each of its elements
2. Otherwise, zero-initialize it
http://en.cppreference.com/w/cpp/language/value_initialization
I was unable to reproduce (1) in a way that produced CXXScalarValueInitExpr and so this diff
deals with case (2)
Reviewed By: jvillard
Differential Revision: D2901311
fb-gh-sync-id: beeafa2
Summary:
public Add typ information to trans_state.var_exp and use it
in constructor translation.
This information is necessary to call constructor of class Base on object with
different type (which is derived from Base)
Reviewed By: jberdine
Differential Revision: D2890850
fb-gh-sync-id: 8075db1
Summary:
public
Add type of return parameter to the context. It allows for better translation
of returnStmt and will be necessary for easy implementation of constructor init lists
Reviewed By: cristianoc
Differential Revision: D2890838
fb-gh-sync-id: e791c3d
Summary:
public
The "dotty" field in json reports is only used when reporting retain cycles. It
makes sense not to emit it by default (when it's `None`).
Reviewed By: akotulski
Differential Revision: D2891320
fb-gh-sync-id: 54292a9
Summary:
public
- add to json reports the location in the ocaml code of assertions raised
during the analysis
- only add when in debug mode. For this, add a flag to `InferPrint` that is
passed by the toplevel `infer` whenever `infer -g` is used.
Reviewed By: jeremydubreil
Differential Revision: D2891286
fb-gh-sync-id: ad4577c
Summary:
public
xvalues is concept introduced in C++11. While they are not same as lvalues, they have one common trait:
They have identity which means that:
> it's possible to determine whether the expression refers to the same entity as another expression, such as by comparing addresses of the objects or the functions they identify (obtained directly or indirectly);
It means that as far as backend is concerned, they should be treated in same way. Right now there is no concept of "move" in the backend and so we don't have
to differentiate between them.
Reference:
http://en.cppreference.com/w/cpp/language/value_category
Reviewed By: cristianoc
Differential Revision: D2895593
fb-gh-sync-id: 5101e28
Summary:
public
C++ allows for parameters with empty names (unused/default copy constructors). Make backend happy by assigning a non-empty
name to these variables
Reviewed By: jvillard
Differential Revision: D2895550
fb-gh-sync-id: b466397
Summary:
public
The env variables used by the Infer backend were only passed in multicore mode. They are now passed before the generation of clusters, whether in single core mode or in multicore mode.
Reviewed By: cristianoc
Differential Revision: D2891816
fb-gh-sync-id: 25074af
Summary:
public
It turns out that C-like structs in C++ may have methods generated by clang (constructors for example).
If struct has a method, it needs to have Sil.Class type - make all CXXRecordDecls Sil.Class types by default.
Reviewed By: cristianoc
Differential Revision: D2895567
fb-gh-sync-id: 8eb18c3
Summary:
The annotations won't work on Android unless you target JDK 1.7 (otherwise it can't DEX the app or library).
Closes https://github.com/facebook/infer/pull/257
Reviewed By: jeremydubreil
Differential Revision: D2889334
Pulled By: sblackshear
fb-gh-sync-id: c09adaa
Andrzej Kotulski
Sam Blackshear
martinoluca
jrm
Dino Distefano
Cristiano Calcagno
kstreee
Jules Villard
Dulma Rodriguez
Josh Berdine
Peter O'Hearn
Tony Cosentini