Summary:
This diff is doing three things:
1. Finishes work paved in D20115024, and applies it to nullsafe. In that diff, we hardened API for
file level analysis. Here we use this API in nullsafe, so now we can
analyze things on file-level, not only in proc-level like it was before!
2. Introduces a class-level analysis. For Nullsafe purposes, file is not
an interesting granularity, but we want to analyze a lot of things on
file level. Interesting part here is anonymous classes and how we link
them to their corresponding user-defined classes.
3. Introduces a first (yet to be improved) implementation of class-level
analysis. Namely it is "meta-issues" that tell what is going with class
on high level. For now these are two primitive issues, and we will
refine them in follow up diffs. They are disabled by default.
Follow ups include:
1. Refining semantics of meta-issues.
2. Adding other issues that we could not analyze before or analyzed not
user friendly. Most importantly, we will use it to improve reporting for
FIELD NOT INITIALIZED, which is not very user friendly exactly because
of lack of class-level aggregation.
Reviewed By: artempyanykh
Differential Revision: D20417841
fbshipit-source-id: 59ba7d2e3
Summary: The `FN_loop2` was not actually FN because infer analyzes its complexity as degree 1 correctly.
Reviewed By: dulmarod
Differential Revision: D20468367
fbshipit-source-id: 9e4c19415
Summary: The `iterate_over_mycollection_quad_FN` was not actually FN because infer analyzes its complexity as degree 2 correctly. So, this diff removed `_FN` from there.
Reviewed By: ezgicicek
Differential Revision: D20467398
fbshipit-source-id: b10340612
Summary: There has never been a sufficient formal basis for soundness nor completeness of reports on locals. This diff changes the domain to effectively concern only expressions rooted at formals or globals.
Reviewed By: ezgicicek
Differential Revision: D19769201
fbshipit-source-id: 36ae04d8c
Summary: `Object.clone` modeled as pure until the analysis can distinguish returning a fresh object vs. having no side-effects.
Reviewed By: skcho
Differential Revision: D20439998
fbshipit-source-id: 421054cfb
Summary:
`JavaSplitName` is used to represent Java types (in `Procname` in particular). The type itself is a pair of string (an optional package qualifier) and a "type name" (the quotes are there because it may contain array qualifiers).
For example `java.lang.Object[][]` should be represented as
```
{package=Some "java.lang"; typename="Object[][]"}
```
The constructor `make` was misused to construct instead types such as
```
{package=None; typename="java.lang.Object[][]"}`
```
This is evident when we print the return type of a `Procname` non-verbosely (the default), but we still see the package qualifier.
Obviously this is not just a pretty-printing bug, the values were themselves wrong.
The fix is to use the `of_string` constructor which will parse the package and separate it correctly. Another bug (in response to this one) had to be fixed in `Procname.is_vararg` to maintain behaviour in Nullsafe and Quandary.
Reviewed By: mityal
Differential Revision: D20394146
fbshipit-source-id: 4633902eb
Summary:
Impurity domain was tracking all changes to variables (with a list of traces that containing all write/invalid accesses). This results in having long traces with multiple access events for the same variable. For instance,
```
void swap_impure(int[] array, int i, int j) {
int tmp = array[i];
array[i] = array[j]; \\ included in the trace
array[j] = tmp; \\ included in the trace
}
```
here we recorded both array accesses.
This diff changes the domain to include accesses so that we only keep track of a single trace per access. Array accesses are only recorded once.
Note that we want to record all unique accesses, not just the first one, because impurity will be used for hoisting/cost where we will invalidate impure arguments and consider all the rest as not changing.
Reviewed By: jvillard
Differential Revision: D20385745
fbshipit-source-id: d3647dad3
Summary:
D20362149 missed
- to pass the optional argument `include_value_history` to the recursive call in `PulseTrace.add_to_errlog`.
- to set `include_value_history=false` for skipped calls.
This diff fixes these issues.
Reviewed By: skcho
Differential Revision: D20385604
fbshipit-source-id: 176e4d010
Summary:
Make <infer-out>/report.json the default value for this option, as this
is what is used 99% of the time. Clean up test options using this.
Reviewed By: ngorogiannis
Differential Revision: D20362644
fbshipit-source-id: a1bb18757
Summary: Impurity traces are quite big due to recording values histories. Let's simplify the traces by removing pulse's value histories.
Reviewed By: skcho
Differential Revision: D20362149
fbshipit-source-id: 8a2a6115e
Summary: Type is not enough to say a function call of `Provider.get` is expensive or not.
Reviewed By: jvillard
Differential Revision: D20366206
fbshipit-source-id: 83d3e8741
Summary:
This diff uses a type parameter of `Provider.get` to decide whether assigning expensive cost to the
function call or not. For example, if the type is small one like `Provider<Integer>`, it be
evaluated to have a unit cost, otherwise a linear cost.
To get the return type of `Provider.get`, I added a simple analyzer that collects "casted" types
backwards. In Sil, while the function call statement loses the return type, e.g,
```
n$5=_fun_Object Provider.get()(n$3:javax.inject.Provider*);
```
the `n$5`'s value is usually casted to a specific type at some point later.
```
*&$irvar0:java.lang.Object*=n$5
n$8=*&$irvar0:java.lang.Object*
n$9=_fun___cast(n$8:java.lang.Object*,sizeof(t=java.lang.Integer;sub_t=( sub )(cast)):void)
```
So, the analyzer starts from the cast statements backward, collecting the types to cast for each
variables.
Reviewed By: ezgicicek
Differential Revision: D20345268
fbshipit-source-id: 704b42ec1
Summary: This diff adds a model for Java's `Object.clone()` method (similar to existing shallow_copy).
Reviewed By: jvillard
Differential Revision: D20341073
fbshipit-source-id: 30ae40fe7
Summary:
Some (all?) of this is already tested in other tests, but this feature
is important enough (and the implementation is scattered accross the
whole code), so I found it useful to have a small test that ensures the
very basic things are working as expected.
See `NestedFieldAccess.java` that tests far more advances things, but
here we focus only very basic things: conditions, local variable
assignments, and explicit assignments.
Reviewed By: jvillard
Differential Revision: D20339056
fbshipit-source-id: a6cfd0043
Summary: We forgot to take skipped calls into account for state comparison. This diff fixes that.
Reviewed By: skcho
Differential Revision: D20282739
fbshipit-source-id: 7b4d84bb0
Summary:
These were not used (and were actually activated byt the same config
param). They both are in experimental stage that never reached maturity.
Since the team does not have immediate plans to work on ObjC nullability
checker; and since "eradicate" (now known as nullsafe) is the main
solution for Java, removing it is sensible.
Reviewed By: jvillard
Differential Revision: D20279866
fbshipit-source-id: 79e64992b
Summary:
This is the kind of property for which the previous syntax forced one to
use spurious registers.
Reviewed By: ngorogiannis
Differential Revision: D20118863
fbshipit-source-id: b49740d33
Summary:
This diff renames `ZERO_XXX` issues to more appropriately named and descriptive
`XXX_UNREACHABLE_AT_EXIT` and replaces bottom with
unreachable in cost kinds and issues.
Reviewed By: skcho
Differential Revision: D20140301
fbshipit-source-id: eb6076b30
Summary:
1. It is convenient to stick with the policy "ERROR if and only if it is
enforced". Among other, it makes CI integration much easier to implement
(enforcemend, UI and messaging is decided based on severity).
2. Since Nullsafe annotation is an idiomatic way to indicate classes
with enforced nullability checking, we want it to be the only way to
enforce issues.
3. This means we decrease the priority of GraphQL violation issues.
(In practice they were not enforced so we have plenty of violations in
codebase to reflect reality). The proper way dealing with GraphQL will
be detecting such issues as a special issue type and prioritizing fixing
and Nullsafe-ifying corresponding classes.
4. Among other, we downgrade severity of field overannotated to advice
to keep it consistent with condition redundant.
Reviewed By: artempyanykh
Differential Revision: D20141420
fbshipit-source-id: e2f12835a
Summary:
The issue type `ZERO_EXECUTION_TIME` actually corresponds to bottom state but has been mistakenly used to mean
- unreachable nodes (program never reaching exit state)
- having zero cost (e.g. for allocations).
Note that, for execution costs, the latter doesn't make sense since we always incur a unit cost for the start node. Hence, a function with empty body will have unit cost. For allocations or IO however, we only incur costs for specific primitives, so a function with no allocations/IO could have a zero cost. However, there is no point reporting functions with zero cost as a specific issue type. Instead, what we want to track is the former, i.e. functions whose cost becomes 0 due to program never reaching exit state.
This diff aims to split these cases into two by only reporting on the latter and adds traces to bottom/unreachable cost by creating a special category in polynomials.
Next diff will rename `ZERO_XXX` to `XXX_UNREACHABLE_AT_EXIT`.
Reviewed By: skcho
Differential Revision: D20005774
fbshipit-source-id: 46b9abd5a
Summary:
For Mode.Local this is kind of obvious decision.
But this diff does the same for strict mode as well.
See comment in [ExplicitNonnullThirdParty] for the detailed explanation.
Reviewed By: artempyanykh
Differential Revision: D20140056
fbshipit-source-id: 13c66df81
Summary:
In the previos diff we restructured error rendering utils for
TypeOrigin.MethodCall.
In this diff we do the same with TypeOrigin field: lets make the code
consistent.
We also clearly distinct third party from all other possible cases in
this branch.
This changes messaging and reported errors for strict modes (see test cases), and I believe this is a net improvement.
Reviewed By: artempyanykh
Differential Revision: D20139741
fbshipit-source-id: 84f502553
Summary:
> We don't report when the cost is Top as it corresponds to subsequent 'don't know's. Instead, we
> report Top cost only at the top level per function
The previous code just ignored top costed nodes, so it was able to report a non-top cost that was
from another node. For example,
```
void foo() {
linear-cost();
top-cost();
}
```
It reported inconsistent reports: `EXPENSIVE_EXECUTION_TIME` with a linear cost and
`INFINITE_EXECUTION_TIME` at the same time.
This diff fixes it not to report `EXPENSIVE_EXECUTION_TIME` when there is a node with the top cost.
Reviewed By: ezgicicek
Differential Revision: D20139408
fbshipit-source-id: 9fedd4aec
Summary:
In the previous report, it reported the first cost of node that exceeds a threshold. However, this
may hide a bigger cost of node that appears later. This diff changes this to report the biggest
cost of node among the costs exceeding the threshold.
Reviewed By: ezgicicek
Differential Revision: D20116162
fbshipit-source-id: 06199fb46
Summary:
This syntax
- is less confusing (according to several people who are not me);
objectively, there's less magic under the hood
- gives fine control over register number (because condition/action are separated)
- lets one compare values of different arguments of the same call
(e.g., one could have a transition that is taken only if two
arguments of a method call are equal)
Reviewed By: ngorogiannis
Differential Revision: D20005403
fbshipit-source-id: fad8f3b3d
Summary:
The test shows what that TOPL can express, in addition to bugs,
efficiency properties. However, there seems to be an underlying problem
in biabdaction that prevents this particular problem from being caught.
Reviewed By: ngorogiannis
Differential Revision: D20005404
fbshipit-source-id: 466f79050
Summary: The semantics of the `values` function of Java enum class was missing, when it is called outside the class initializer. This diff gets the size of the enum elements from the summary of class initializer function, `<clinit>`.
Reviewed By: ezgicicek
Differential Revision: D20094880
fbshipit-source-id: 7362bba1c
Summary:
Now when typechecking a class `A` marked with `Nullsafe(LOCAL)`,
classes from trusted list are properly recognized and nullability of
method params and return value are refined to `LocallyCheckedNonnull`
in a context of class `A`.
NOTE: refininng nullability when **accessing fields** on trusted classes
is **not implemented yet**, because the whole business of handling fields
in nullsafe is somewhat convoluted. This should not be a huge issue
though, since in Java fields are commonly accessed via getters any
way.
Reviewed By: mityal
Differential Revision: D20056158
fbshipit-source-id: 496433d90
Summary:
This will help making error reporting more actionable.
Often methods that are nullable in general (like View.findViewById) are used as not-nullable due to app-invariants. In such cases suggesting a non-nullable alternative that does an assertion under the hood makes the error report more actionable and provides necessary guidance with respect to coding best practices
Follow up will include adding more methods to models.
If this goes well, we might support it in user-defined area (nullability
repository)
Reviewed By: artempyanykh
Differential Revision: D20001416
fbshipit-source-id: 46f03467c
Summary:
Introduction of `ThirdPartyNonnull` nullability broke nullability
refinement heuristic for enums. This diff fixes it and also adds tests
so that we hopefully avoid such issues in future.
Reviewed By: mityal
Differential Revision: D19975810
fbshipit-source-id: f9245f305
Summary:
We need to be able to differentiate `UncheckedNonnull`s in internal vs
third-party code. Previously, those were under one `UncheckedNonnull`
nullability which led to hacks for optmistic third-party parameter
checks in `eradicateChecks.ml` and lack of third-party enforcement in
`Nullsafe(LOCAL, trust=all)` mode (i.e. we want to trust internal
unchecked code, but don't want to trust unvetted third-party).
Now such values are properly modelled and can be accounted for
regularly within rules.
Also, various whitelists are refactored using
`Nullability.is_considered_nonnull ~nullsafe_mode nullability`.
`ErrorRenderingUtils` became a tad more convoluted, but oh well, one
step at a time.
Reviewed By: mityal
Differential Revision: D19977086
fbshipit-source-id: 8337a47b9
Summary:
Add support for nullsafe mode with `trust=all` and `trust=none` a case
with a specific trust list is not supported yet and needs to be
implemented separately.
Tests introduce one unexpected
`ERADICATE_INCONSISTENT_SUBCLASS_PARAMETER_ANNOTATION` issue which
complains about `this` having incorrect nullability; it is a bug and
needs to be fixed separately.
Reviewed By: mityal
Differential Revision: D19662708
fbshipit-source-id: 3bc1e3952
Summary: In all other cases we have period at the end, which is inconsistent.
Reviewed By: artempyanykh
Differential Revision: D20001065
fbshipit-source-id: 85ec6d751
Summary:
Use a record of package, class name to store (qualified) Java class names. This saves the round trip of concatenating then splitting again, etc, as well as saves some memory in the type environment as now the package paths can be shared across classes of the same package (about 10% in tests).
Also remove some unfortunate APIs.
Reviewed By: jvillard
Differential Revision: D19969325
fbshipit-source-id: f7b7f5a55
Summary: The way `Mangled.t` is used in `JavaClassName` means that it's always a plain string (we never have a "mangled" part). Remove the indirection and extra allocation. Also, simplify the API by throwing away one function that was used just once and wastefully.
Reviewed By: artempyanykh
Differential Revision: D19950672
fbshipit-source-id: b61fcba6e
Summary:
This adds a violation of baos.topl found in github/seata/seata. However,
it is not a bug (see comment in commit).
Reviewed By: ngorogiannis
Differential Revision: D19518641
fbshipit-source-id: e219245ee
Summary:
Since Javalib 3.2, a new feature allows to rewrite
methods that contain (some specific form of) closures. Infer
now uses it. When loading each class we rewrite them and
new classes generated by Javalib to implements closures
(i.e. Java interfaces)<
Reviewed By: ngorogiannis
Differential Revision: D19389227
fbshipit-source-id: 245dd4404
Summary:
We already warn about lack of nullable annotations in `equals()`, and even have a specialized error message for that.
But lack of an annotation is not as severe as direct dereference: the
latter is a plain bug which is also a time bomb: it will lead to an NPE not immediately.
This is widespread enough to be reported separately.
Reviewed By: dulmarod
Differential Revision: D19719598
fbshipit-source-id: a535d43ea
Summary:
Since we fixed a bug in implementation of FalseOnNull (see stack below),
we can finally ship this change.
Side note: this change is essential for the follow up diff (which adds extra check
for user-defined implementations of equal()), without it the follow
up change would introduce a lot of false positives.
Reviewed By: ngorogiannis
Differential Revision: D19771057
fbshipit-source-id: 7d7cf1ef7
Summary:
If we managed to whitelist a function as TrueOnNull, we should teach
nullsafe the nullability of its arguments, otherwise it will ask not to
pass null here.
This fixes a silly FP warning, see the test.
Reviewed By: dulmarod
Differential Revision: D19770341
fbshipit-source-id: 0f861fae1
Summary:
Yay, the previous refactoring finally makes it possible to do some actual
changes to the code in `TypeCheck.ml`!
Changes in this diff:
1. Fixes the bug: TrueOnNull and FalseOnNull were working only for
static methods. Surpsingly nobody noticed that. It is because the first
argument for non-static method was `this`.
2. Behavior change: TrueOnNull/FalseOnNull were not working correctly
where there are several argumens. See the task attached for the example
of the legit usecase. Now the behavior is the following: if there are
several Nullable arguments infer nullability for all of them.
Reviewed By: skcho
Differential Revision: D19770219
fbshipit-source-id: 7dffe42cd
Summary:
This refactoring unblocks the changes in follow up diffs (plus fixes a
bug).
So what was happening?
Each comparison with null leads to CFG being splitted into two branches, one branch
is PRUNE(a == null) and another is PRUNE(a != null).
PRUNE(a != null) is where most of logic happens, it is the place where
we infer non-null nullability for a, and this is a natural place to
leave a check for redundancy.
Before this diff we effectively checked the same thing twice, and used
`true_branch` (only one of 2 instruction will have it set to true) as a symmetry breaker.
This diff removes the `true_branch` checks, but leaves only one call out
of two, hence breaking symmetry in a different way.
## Bug fix
The code around the removed check was (crazily) doing two things at
once: it processed results of (returning booleans!)
TrueOnNull-annotated functions AND
results of (returning Objects!) other functions, using the fact that all
of them are encoded as zero literals (sic!).
Not surprisingly that lead to a bug where we accidentally call the check
for non intended places (arguments of trueOnNull functions), which lead
to really weird FP.
This diff fixes it.
Reviewed By: dulmarod
Differential Revision: D19744604
fbshipit-source-id: fe4e65a8f
Summary: We don't use allocation costs in prod at the moment. There is no plan to do so in the near future. Let's not report them anymore and also save some space in `costs-report.json`.
Reviewed By: skcho
Differential Revision: D19766828
fbshipit-source-id: 06dffa61d
Summary:
This test tests PropagatesNullable and TrueOnNull/FalseOnNull
annotations.
Both tests suites grew big so it is hard to observe them at glance and
make changes.
I could not figure out better name for TrueFalseOnNull.java, it is sort
of silly but I optimized for searchability, "FalseOnNull" will be
directly searched and "TrueOnNull" will be searched in IDEs that are
smart enough.
Reviewed By: skcho
Differential Revision: D19724512
fbshipit-source-id: 703961342
Summary: This diff returns non-symbolic value (top) for unknown external function calls because the symbolic values sometimes make it hard to understand costs.
Reviewed By: ezgicicek
Differential Revision: D18685715
fbshipit-source-id: 1b39c718b
Summary:
Pulse has an extra invalidation mechanism (introduced in D18726203) to prevent something invalid (e.g. `null`) to be passed by reference to an initialisation function. Therefore, it havocs formals passed by reference to skipped functions. However, I don't think this makes sense in Java. So, let's turn it off.
A nice consequence of this is that in impurity analysis, we do not consider functions that call skipped library calls with object arguments as writing to their formals.
Reviewed By: skcho
Differential Revision: D19697110
fbshipit-source-id: 6e3a71f2a
Summary:
- Thread the two types into one instead of having a record where the `path` field doesn't always make sense (`Class` case).
- Improved pretty printing of class objects (java only).
- Move starvation-specific stuff out of `AbstractAddress` (eg `make_java_synchronized`).
- Slight optimisation of `apply_subst` for when a parameter is used without additional accesses inside a method (then, the substitution need not modify the term substituted for the parameter in any way).
Reviewed By: ezgicicek
Differential Revision: D19639922
fbshipit-source-id: 1cebecf5d
Summary:
`String` and `StringBuilder` both implement `CharSequence`. Let's generalize the model for `String` to `CharSequence` wherever possible and add missing models for
- `StringBuilder.append`
- `StringBuilder.toString`
Reviewed By: skcho
Differential Revision: D19558009
fbshipit-source-id: 0dfdb21af
Summary:
Java's String models were broken for
- initializing a String object with a locally defined constant string (which is an `Object*` in SIL).
- initializing a String object with a `char`/`byte` array
This diff fixes them and also adds models for `new String ()`.
Reviewed By: skcho
Differential Revision: D19662180
fbshipit-source-id: 23968d0aa
Summary:
Prevent returning a negative cost bound when calling `substring(begin_index, end_index)` when either is possible
- `begin_index < 0`
- `begin_index > end_index`
Instead, return unit cost since such cases either throw `IndexOutOfBoundsException ` at runtime or correspond to having two symbolic bounds that cannot be semantically compared.
Reviewed By: dulmarod
Differential Revision: D19619410
fbshipit-source-id: cf5e8cb7b
Summary:
The "access path" memory model (equal access paths iff equal object addresses) is suited to when aliasing occurs only at the roots (i.e. variables). When there is intentional aliasing in the middle of an access path, this model will miss the aliasing. For instance if `[x.f] == [y.g]`, then also `[x.f.h] == [y.g.h]`, but the latter access paths are unequal.
In Java, non-static inner classes consistently alias `this.this$0` inside an inner class, which points to the "parent" outer-class object. So if two inner-class objects (belonging to different inner classes) access `this(type:InnerClassA).this$0.f` and `this(type:InnerClassB).this$0.f` the equality will be missed (many other combinations exist). This isn't strictly due to the memory model -- any alias analysis would have to do some class invariant inference to detect this.
For this purpose `AccessPath.inner_class_normalize` exists (it replaces `this.this$0` with `this` of the appropriate type), but this breaks the invariant that we know which formal parameter is at the root (there may not even exist a `this` parameter if the method is static). So this was buggy.
Here we simply recursively remove the synthetic field prefix of the accesses list, while computing forwards the object type. This is only applied when we check aliasing across threads. This will also allow actuals/parameters substitutions (stacked diff) which normalisation was breaking.
Reviewed By: jberdine
Differential Revision: D19601455
fbshipit-source-id: 7e42667b6
Summary:
- Add `Nullsafe` annotation as a general mechanism to specify
type-checking behaviour for nullsafe.
- Document annotation params and provide usage examples with
explanations.
- Add tests to demonstrate the behaviour with different type-checking
modes.
No implementation is added. This diff serves as an RFC to hash out the
details before I dive into code.
Reviewed By: ngorogiannis
Differential Revision: D19578329
fbshipit-source-id: b1a9f6162
Summary: `String.split(regexp)` returns an array that is split by the given regexp. If the regexp doesn't match, the original string is returned. Hence, the resulting array's length must be in `[1, max(1, n_u -1)]` where`n_u` is the upper bound of the string's length.
Reviewed By: dulmarod
Differential Revision: D19578318
fbshipit-source-id: 675af7376
Summary:
In practice, condition redundant is extremely noisy and low-signal
warning (hence it is turned off by default).
This diff does minor tweaks, without the intention to change anything
substantially:
1/ Change severity to advice
2/ Change "is" to "might be"
3/ Describe the reason in case the origin comes from a method.
The short term motivation is to use 3/ for specific use-case: running nullsafe on codebase and
identify most suspicious functions (that are not annotated by often
compared with null).
Reviewed By: artempyanykh
Differential Revision: D19553571
fbshipit-source-id: 2b43ea0af
Summary:
The order by which the scheduler visits odd and even methods here
will determine if there is any report at all. This is a bad test
so remove.
Reviewed By: fgasperij
Differential Revision: D19535537
fbshipit-source-id: 6b64b0de9
Summary:
1. One should use either a writer or a stream to send a response, but not both.
2. A response should be forwarded only if it was commited.
Both properties are extracted from API comments on classes in the servlet API.
Reviewed By: jvillard
Differential Revision: D19514568
fbshipit-source-id: 79f0257ed
Summary:
If data comes from an outer OutputStream, then this outer OutputStream
needs to be flushed before getting the byte array.
Reviewed By: jvillard
Differential Revision: D19514569
fbshipit-source-id: e3e025394
Summary: We were lacking this kind of test where one interface refines the nullability of the other.
Reviewed By: ngorogiannis
Differential Revision: D19514245
fbshipit-source-id: fa3e781f3
Summary:
This is a common enough case to make error message specific.
Also let's ensure it's modelled.
Reviewed By: artempyanykh
Differential Revision: D19431899
fbshipit-source-id: f34459cb3
Summary:
The previous diff changes the message for params case, this one handles
return.
Reviewed By: artempyanykh
Differential Revision: D19430706
fbshipit-source-id: f897f0e56
Summary: Use more informative method names, and add comments explaining the logic behind each test. Correct two cases which are FPs instead of legitimate reports.
Reviewed By: artempyanykh
Differential Revision: D19465227
fbshipit-source-id: 29332e2b9
Summary:
Currently, impurity analysis is oblivious to skipped functions which might e.g. return a non-deterministic value, write to memory or have some other side-effect. This diff fixes that by relying on Pulse's skipped functions to determine impurity. Any unknown function which is not modeled to be pure is assumed to be impure.
This is a heuristic. We could have assumed them to be pure by default as well.
Reviewed By: jvillard
Differential Revision: D19428514
fbshipit-source-id: 82efe04f9
Summary:
As suggested by Ilya, the current message can be improved in a way that
it can contain more clear action. I also added artempyanykh's explanation at the
end of message to provide an additional justification from common sense
perspective.
But most importantly, the previous message was missing a space which is
eye bleeding, how come haven't I noticed this before, I can't stand it
OMG.
Reviewed By: artempyanykh
Differential Revision: D19430271
fbshipit-source-id: dd31f7adb
Summary:
Introduce a new notion of equality for comparing abstract addresses in distinct threads:
```
(** Abstract address for a lock. There are two notions of equality:
- Equality for comparing two addresses within the same thread/process/trace. Under this,
identical globals and identical class objects compare equal. Locks represented by access paths
rooted at method parameters must have equal access paths to compare equal. Paths rooted at
locals are ignored.
- Equality for comparing two addresses in two distinct threads/traces. Globals and class objects
are compared in the same way, but locks represented by access paths rooted at parameters need
only have equal access lists (ie [x.f.g == y.f.g]). This allows demonically aliasing
parameters in *distinct* threads. This relation is used in [may_deadlock]. *)
```
Reviewed By: skcho
Differential Revision: D19347307
fbshipit-source-id: 9f338731b
Summary:
This diff avoids that `array_sizeof` returns bottom value when given Java enum values, which
introduced unreachable code inadvertently.
Reviewed By: ngorogiannis
Differential Revision: D19409077
fbshipit-source-id: 2816fd995
Summary:
The property SkipAfterRemove already had a test, but not for
intra-procedural violations. This adds a test for that case.
Reviewed By: ngorogiannis
Differential Revision: D19330471
fbshipit-source-id: 1dd1c3ad7
Summary: This diff implements this for Field Not Initialized check
Reviewed By: artempyanykh
Differential Revision: D19393989
fbshipit-source-id: cf60e8d53
Summary: add subdirectories so that we can run each java file against its own topl properties
Reviewed By: rgrig
Differential Revision: D19347302
fbshipit-source-id: 562830774
Summary:
This diff does it for nullable dereference and assignment violations
rules which happen under NullsafeStrict case.
Follow up are to make the same for inheritance and field initializer
violations.
Possible follow up includes making error message more specific and
articulare this this is a nullsafe strict mode.
Reviewed By: artempyanykh
Differential Revision: D19392916
fbshipit-source-id: 2554ac7a7
Summary:
Previously, _override resolution_ considered only the number of
arguments. This led to many FPs in nullsafe's _Inconsistent Subclass
Annotation_ check.
Current version also checks that argument types match. However, we
still don't handle type parameters and erasure, so in this sense the
rules are incomplete.
Reviewed By: ngorogiannis, mityal
Differential Revision: D19393201
fbshipit-source-id: a0c75b8dd
Summary:
This diff avoids that null-retuned path's abstract value ruins that of non-null-returned path.
What this diff does is: when joining two abstract states, one is null-return-path and the other is
non-null-return-path (`return obj;`), it keeps the method calls of `obj` from the
non-null-return-path.
While this design is unsound, I think it should work in practice.
Reviewed By: ezgicicek
Differential Revision: D19348313
fbshipit-source-id: cf5d0f3ff
Summary:
Java treats switch on nullables in a non-obvious way (throws an NPE
surprise) so lets have a decidated test exactly for this.
Reviewed By: ngorogiannis
Differential Revision: D19371280
fbshipit-source-id: d9867b6d6
Summary: This diff use actuall call path in the cost results instead of `class name + method name`.
Reviewed By: ngorogiannis
Differential Revision: D19194969
fbshipit-source-id: b72018586
Summary:
Model array length in Java as returning an unknown interval [0, +inf] for now.
Ideally, we can deal with the size in a more precise manner in the future like in InferBo.
Reviewed By: skcho
Differential Revision: D19312123
fbshipit-source-id: 8c51059a4
Summary: Pulse doesn't care about exceptions yet. With Exceptional CFG, java analysis takes a lot of time due to having many disjuncts. Let's use Normal CFG for now.
Reviewed By: jvillard
Differential Revision: D19194479
fbshipit-source-id: f94bb6078
Summary:
In order to improve the impurity analysis, this diff adds models for
- `hasNext()` and - `Object.equals()` modeled as returning a non-deterministic value (havoc_id)
- `next()` modeled as `StdVector.get` with a fresh index
- `iterator` modeled as just returning the underlying list
Reviewed By: jvillard
Differential Revision: D19177392
fbshipit-source-id: 0babb037a
Summary:
This diff updates the relation between iterator (offset) and integer value not only at
assignments (`x += 1`), but also at function calls (`foo()`) that increase integer values by one in
their side effects.
Reviewed By: ezgicicek
Differential Revision: D19163214
fbshipit-source-id: 47e52f939
Summary: This diff extends the domain to express the relation between iterator's offset and integer value.
Reviewed By: ezgicicek
Differential Revision: D19143670
fbshipit-source-id: 6223bc934
Summary:
Old versions of sawja/javalib got the line numbers slightly wrong. The workaround was to do a regexp search in the source file for the right line.
My understanding is that this is no longer necessary. This diff removes it.
Reviewed By: jvillard
Differential Revision: D19033415
fbshipit-source-id: 2da19d66d
Summary:
This is an optimization. We ask the user to tell us which states are nondeterministic, and we
generate code that handle nondeterminism only for those states. It is common for only one state per
TOPL property to be nondeterministic. This speeds up the biabduction-analysis of the monitor by a
factor of ~10. But, using the monitor is only a little faster.
Facebook
Reviewed By: jvillard
Differential Revision: D19160286
fbshipit-source-id: 4dd39769a
Summary:
In the previous code, it removed non-build-called method calls. For example, it was like
```
{non-build-called: {prop1, prop3}}
{build-called: {}}
b.build();
{non-build-called: {}}
{build-called: {prop1, prop3}}
```
However, this behavior introduced a false positive when there is multiple builders that point to the
same abstract object and `build` is called one by one.
This diff changes the semantics to keep the method calls of non-build-called at `build` calls.
Reviewed By: ezgicicek
Differential Revision: D19144525
fbshipit-source-id: e2ace127f
Summary:
This applies some simplifications that were previously
done after footprint (and therefore lost), and some
simplifications that require looking at both pre and
post.
Reviewed By: ngorogiannis
Differential Revision: D19035494
fbshipit-source-id: bad79534a
Summary:
This havocs event data, so that biabduction doesn't try to
track what was the last event processed by the monitor
(which is redundant as long as the state of the monitor
is tracked).
Reviewed By: ngorogiannis
Differential Revision: D19035491
fbshipit-source-id: a1c75daae
Summary:
Don't instrument SIL when we can determine statically that
biabduction symexec would be a no-op.
Reviewed By: ngorogiannis
Differential Revision: D19116849
fbshipit-source-id: 4d25462a3
Summary: It is confusing to report missing props at the the beginning of the method (especially when there are many components created or when the method has many lines). Let's report them at create methods to better contextualize. Also, make the missing prop bold.
Reviewed By: skcho
Differential Revision: D19141135
fbshipit-source-id: a23d2e7c9
Summary:
In addition to
state1 -> state2: pattern
one can now also write
state1 -> state2: pattern if condition
where "condition" is some conjunction of comparisons (==,<,>) that
involve variables bound by "pattern", registers of the automaton, and
constants.
Reviewed By: ngorogiannis
Differential Revision: D19035496
fbshipit-source-id: 6f6e6a9be
Summary:
According to Java semantics, they are always non-null.
Internally they are represented as static fields, so they have
DeclaredNonnull nullability, which means NullsafeStrict mode would
refuse to use them without strictification.
Lets teach nullsafe that these guys are non-nullables.
See also FN in test case.
Reviewed By: ngorogiannis
Differential Revision: D19024547
fbshipit-source-id: 8c120fa50
Summary:
We do not have the create method in the trace which makes it difficult to understand
- inter-procedural issues where create and prop setting are in different methods
- there are multiple create-build chains in a method
Let's add the create to the beginning of the trace. Moreover, let's simplify the prop printing to make traces easier to understand.
Reviewed By: ngorogiannis
Differential Revision: D19020213
fbshipit-source-id: 7f8a5d4ec
Summary: The new domain is much better than the old one. Let's kill the old one (along with old litho tests) and simplify things.
Reviewed By: skcho
Differential Revision: D18959627
fbshipit-source-id: df77ae20e
Summary:
In order to handle the example added:
changed domain of `MethodCalled`
from `CreatedLocation -> (IsBuildCalled X IsChecked X Set(MethodCall))`
to `(CreatedLocation X IsBuildCalled) -> (IsChecked X Set(MethodCall))`
This avoids joining of two method calls where one is build-called and the other is not, e.g.,
```
if(b) {
o.build();
} else {
// no build call
}
```
changed domain of `NewDomain`
from `Created X MethodCalled`
to `(Created X MethodCalled) X (Created X MethodCalled)`
One is for no returned memory and the other is returned memory. This keeps precision some join
points of branches, e.g.,
```
if(b) {
return;
} else {
// no return
}
```
Reviewed By: ezgicicek
Differential Revision: D18909768
fbshipit-source-id: c39d1a1ef
Summary: It is not used anywhere and there are no plans to revive it. Kill it!
Reviewed By: skcho
Differential Revision: D18934719
fbshipit-source-id: b9b069b96
Summary: Guava uses assertions to ensure a future can be gotten without blocking (this means that if the future is not done, the app will crash). This diff teaches the starvation analyser about a number of such assertions, by treating them as assumes (since we don't care about exceptions).
Reviewed By: jvillard
Differential Revision: D18893427
fbshipit-source-id: 4d26a202b
Summary: A future is guaranteed not to block if `isDone()` has returned true first. Add logic for supporting that by remembering the objects that we have called `isDone` on and by making `assume` do the right thing with that knowledge. All this is achieved with the attribute domain.
Reviewed By: ezgicicek
Differential Revision: D18833901
fbshipit-source-id: 7f4ea0cd1
Summary:
When retrieving a value from a container, we previously had an arbitrary hack which would
- In java, give no ownership to the returned object (trying to be sound)
- In C++ give conditional ownership to the current method's first argument (trying to be complete, but doing it badly, as the first argument may not be the `this` object in a static method, or we might be accessing it through another parameter altogether).
Harmonise both by using the existing ownership of the container as ownership value for the returned object (leaning towards completeness).
Reviewed By: jvillard
Differential Revision: D18882800
fbshipit-source-id: f98f8d315
Summary:
This reverts commit 4fd6165d190bab32544f9f040b777565432c15b2.
We don't need to check for reporting each node anymore. It suffices to just check per function.
Reviewed By: skcho
Differential Revision: D18883833
fbshipit-source-id: 2591b3af3
Summary:
Making `MethodCalled` an inverted map from created location to method calls results in not being able to track a builder that is created in two different branches of a conditional with different types. Instead, we can make `MethodCalled` simply a map and also change `Created` to be a map from access paths to a set of created locations.
To deal with the case of setting a prop only in one branch, we need to ensure that whenever we call a create method, we add a binding to `MethodCalled` with an empty list of methods so that its intersection with a non-empty one is empty.
Reviewed By: skcho
Differential Revision: D18883097
fbshipit-source-id: b3464ca20
Summary: As long as the types match, it should be possible to call build on two components that are created at different locations.
Reviewed By: skcho
Differential Revision: D18881740
fbshipit-source-id: 356f9e168
Summary: Add a FN that is detected by the old domain but not the new one
Reviewed By: ngorogiannis
Differential Revision: D18854389
fbshipit-source-id: 9bdc90a6b
Summary: The map from `CreatedLocation` to `MethodCalls` already takes care of the association from create methods to their set props. `MethodCall` comparison should be oblivious the the receiver, otherwise, we risk mistakenly considering two props set at different locations as different.
Reviewed By: skcho
Differential Revision: D18829388
fbshipit-source-id: b5a0d628d
Summary: This diff check and report on every nodes. Problem of the previous design is that it has to report alarms only with the abstract memory of the exit node. However, the new abstract value becomes imprecise at every join points on the path to the exit node, since it is using inverted map, i.e., under-approximation on collecting called methods. As a solution, this diff report on every nodes where `.build` is called with the abstract memory at that node.
Reviewed By: ezgicicek
Differential Revision: D18809449
fbshipit-source-id: 4fd6165d1
Summary: Current method call comparison is too strong. As exemplified with the new test, one can also set the required prop by calling a version which contains the suffixes. The domain should take care of such cases now.
Reviewed By: skcho
Differential Revision: D18808869
fbshipit-source-id: 9f7672e75
Summary: This diff checks litho condition on the new abstract value. This is triggered with `--new-litho-domain`, but it is intra-procedural as of now.
Reviewed By: ezgicicek
Differential Revision: D18783203
fbshipit-source-id: 98570104e
Summary:
Also add logic for recognising excessive timeouts. Refactor the code
around timeouts a little.
Reviewed By: artempyanykh
Differential Revision: D18807836
fbshipit-source-id: df5a1b566
Summary:
`Object.wait()` must be called on a locked monitor and it releases the
lock immediately, as far as other threads are concerned
(it also magically re-takes the lock when the monitor is `notified`).
Starvation can only occur if the UI thread is waiting
a lock that is distinct to that being waited on.
The check present was over-approximate in that it was checking that there exists a lock held by the UI thread and the thread issuing the `wait`, but did not make sure that lock was *not* the one waited on.
Amusingly, the e2e test was correct, but the reporting code wasn't.
Reviewed By: dulmarod
Differential Revision: D18782919
fbshipit-source-id: b3b98239e
Summary: Similar to constructor established attributes, we do the same here for static initializers. That is, attributes of static properties are injected into the initial state of every method.
Reviewed By: ezgicicek
Differential Revision: D18763192
fbshipit-source-id: 3879a27c5
Summary: This diff extends the bound domain to express multiplication of bounds in some simple cases.
Reviewed By: ezgicicek
Differential Revision: D18745246
fbshipit-source-id: 4f2dcb42c
Summary:
One standard way to schedule work is by starting a thread. We treat this by
- Treating invocations of `start` on a receiver with the `Runnable _` attribute as scheduling that runnable for parallel execution in the background (as opposed to on the UI thread).
- If `start` is used on an object of a subclass of `Thread` everything already works thanks to the `get_exp_attributes` function which will implicitly ascribe to an expression the attribute `Runnable _` if the expression points to an object with a known `run` method. This will even take care of some degree of dynamic dispatch, yay!
- If `start` is used on a `Thread` object which has been created with a constructor call provided with a `Runnable` argument, we have to appropriately model that constructor call, which is what is done in `do_call`.
Reviewed By: jvillard
Differential Revision: D18726676
fbshipit-source-id: 0bd83c28e
Summary:
A current blind spot is when object construction stores specific executors / runnables to object fields, which are then never mutated and accessed from normal methods. IOW the attributes established in the constructor are necessary to report properly inside a normal method (assuming these attributes are not invalidated by method code).
To achieve this, first we retain a subset of the final state attributes in the summary (only those that affect instance variables, in constructor methods). Then, when we analyse a non-constructor method:
- we analyse all constructors
- remove all attributes from the attribute map whose key is not an expression of the form `this.x. ...`
- re-localise all remaining keys so that they appear as rooted in the `this` local variable of the current procedure
- join (intersect) all such attribute maps
- use the result in place of initial state as far as the attribute map is concerned for the analysis of the current procedure, which can now start.
This means we can catch idioms that use side-effectful initialisation for configuring certain object fields like executors or runnables.
Reviewed By: jvillard
Differential Revision: D18707890
fbshipit-source-id: 42ac6108f
Summary: Another way to schedule work in android is by posting it to a `Handler`. A handler can be constructed out of the main looper, which forces it to schedule work on the UI thread. To model all this, we add syntactic models for getting the main looper and for creating handlers, and dataflow attributes for tracking that an expression is a looper/the main looper, or a handler constructed out of a looper.
Reviewed By: skcho
Differential Revision: D18706768
fbshipit-source-id: 7c46e6913
Summary:
This gets rid of false positives when something invalid (eg null) is
passed by reference to an initialisation function. Havoc'ing what the
contents of the pointer to results in being optimistic about said
contents in the future.
Also surprisingly gets rid of some FNs (which means it can also
introduce FPs) in the `std::atomic` tests because a path condition
becomes feasible with havoc'ing.
There's a slight refinement possible where we don't havoc pointers to
const but that's more involved and left as future work.
Reviewed By: skcho
Differential Revision: D18726203
fbshipit-source-id: 264b5daeb
Summary:
This diff fixes the model of substring.
Problem: The cost model of the substring function was to return `size of string - start index` as a
cost. However, sometimes this was a negative number, because of state abstractions on paths, array
elements, call contexts, etc, which caused an exception inadvertently.
This diff changes the model to return just `size of string`, when it cannot say `size of string` is
bigger than `start index`.
Reviewed By: ezgicicek
Differential Revision: D18707954
fbshipit-source-id: 63f27e461
Summary:
Instead of trying to figure out what runnable is directly passed to an executor,
use attributes to track the flow of a runnable. This has several advantages:
- Can track runnables across function return values.
- Can somewhat overcome the information loss under dynamic dispatch.
- Unifies handling with other attributes.
Reviewed By: skcho
Differential Revision: D18672676
fbshipit-source-id: a06a0e6ff
Summary:
- Unify treatment of modelled and annotated executors by making things go through attributes.
- Add a return attribute to summaries, so that we can track flows of thread guards/executors/future stuff through returned values.
- Dispatch modeled functions to model summaries.
This will help in following diffs where runnables will also go through attributes.
Reviewed By: skcho
Differential Revision: D18660185
fbshipit-source-id: e26b1083e
Summary: When we see a call to schedule some work on an executor and we don't have evidence that it is on some specific thread (UI/BG), instead of dropping the work, assign it `UnknownThread` and treat it as running on the background by default.
Reviewed By: jvillard
Differential Revision: D18615649
fbshipit-source-id: e8bad64b6
Mitya Lyubarskiy
Sungkeun Cho
Ezgi Çiçek
Nikos Gorogiannis
Jules Villard
Radu Grigore
Artem Pianykh
David Pichardie