Summary: Variables captured by reference without initialization are missing dereference in their type inside lambda's body. This causes the translation to miss one dereference. To fix this, we want to add missing reference to the type. However, we first need to make sure that lambdas body is translated after the translation of captured variables.
Reviewed By: jvillard
Differential Revision: D23564099
fbshipit-source-id: 6a2ae053d
Summary:
Added a function to check if a method is a cpp lambda call operator that will be used in later diffs
#skipdeadcode
Reviewed By: jvillard
Differential Revision: D23564089
fbshipit-source-id: 144c3d735
Summary:
We were missing assignment to captured variables with initializers.
Consider the following example:
```
S* update_inside_lambda_capture_and_init(S* s) {
S* object = nullptr;
auto f = [& o = object](S* s) { o = s; };
f(s);
return object;
}
```
which was translated to
```
VARIABLE_DECLARED(o:S*&);
*&o:S*&=&object
*&f =(_fun...lambda..._operator(),([by ref]&o &o:S*&))
```
However, we want to capture `o` (which is an address of `object`), rather `&o` in closure.
After the diff
```
VARIABLE_DECLARED(o:S*&);
*&o:S*&=&object
n$7=*&o:S*&
*&f =(_fun...lambda..._operator(),([by ref]n$7 &o:S*&))
```
Reviewed By: jvillard
Differential Revision: D23567346
fbshipit-source-id: 20f77acc2
Summary:
Following the previous diff, the reason we mistakingly introduced new block as complexity increases is because when we are computing the hash of block, we return the string `block` to compute the hash for all block.
This diff fixes this case by returning the non-verbose name of the block when returning hashable name.
Reviewed By: ezgicicek
Differential Revision: D23729710
fbshipit-source-id: a345d3045
Summary:
When inspecting silent introduced issues, we found that an introduced issue is about a complexity increase of a block that is only created in the current diff. Based on the trace view, we find out that this is caused by infer mistakingly consider another block that exists in the previous diff as the same block that is newly created in the current diff.
This diff adds a test case that reproduces this case, and this will be fixed in the next diff.
facebook
Trace view: https://www.internalfb.com/intern/traceview/?id=109896337
Reviewed By: ezgicicek
Differential Revision: D23681550
fbshipit-source-id: 78341268b
Summary:
If the issue is related to the use of an unvetted third party method
that can not be trusted, we record this fact.
Reviewed By: ngorogiannis
Differential Revision: D23705626
fbshipit-source-id: 851328fe5
Summary:
It was inconsistent before (we recorded it only for meta issues).
Now we always record it, which will simplify further diffs.
In the next diffs, we are going to add more fields inside `nullsafe_extra`.
Reviewed By: ngorogiannis
Differential Revision: D23705592
fbshipit-source-id: 8bbb0e7c8
Summary:
This diff introduces `NullsafeIssue.t`: information about the issue
ready to be reported (and put to `result.json`).
Its notion was already implicitly used in a lot of code.
With this change, the achitechure becomes the following:
Firstly, `TypeErr.err_instance` represents issues at the moment of registration
during the typecheck. At this moment we don't always want to report
them, but it is important to store even non-reportable ones (we use it to calculate mode promotions).
Secondly, given the current `NullsafeMode.t`, we can either report the
issue or not. This logic lives in
`TypeErr.make_nullsafe_issue_if_reportable_lazy` that normally redirects
this decision to Rules (e.g. AssignmentRule or DereferenceRule).
Thirdly, if we want to report the issue, it is time to actually figure
out what to report (e.g. the precise error message, or additional
nullsafe specific `.json` params - see next diffs adding them).
Note that the exact logic of deciding if we want to report / how to
report / what should the message or .json param be is in practice
coupled (otherwise we'd have weird bugs when we decided the issue is
reportable, but don't have a good user facing reason).
In practice such logic for complix issues leaves in Rules.
C.f. `DereferenceRule` and `AssignmentRule` code.
The tricky part is that those rules actually share some common code
responsible for reporting, e.g. when it comes to processing third
parties (so the decision making & reporting is unified). See
`ErrorRenderingUtils.mk_nullsafe_issue_for_untrusted_values` which is
called both from `AssignmentRule` and `DereferenceRule`.
`NullsafeIssue.t` glues together those shared parts of code, and make
dependencies explicit.
Check out the next diff when we add more capabilities to
`NullsafeIssue.t` (e.g. ability to store dependent third party methods).
Without this refactoring, implementing this feature would be rather
tricky.
Reviewed By: skcho
Differential Revision: D23705587
fbshipit-source-id: b5246062a
Summary:
Even though it's unused in the implementation this argument is supposed
to be `Typ.t` so spell it out.
Reviewed By: skcho
Differential Revision: D23729433
fbshipit-source-id: d05548f42
Summary:
This can be useful to make pulse forget about tricky parts of the code.
Treat "skipped" procedures as unknown so heuristics for mutating the
return value and parameters passed by reference are applied.
Reviewed By: ezgicicek
Differential Revision: D23729410
fbshipit-source-id: d7a4924a8
Summary: This diff selects an autorelease trace that has a bigger polynomial.
Reviewed By: ezgicicek
Differential Revision: D23731155
fbshipit-source-id: 243591583
Summary:
This diff reports paths under the xcode isysroot as relative in tests.
This was a problem when another machine that has a different isysroot
directory is running the test.
Reviewed By: ezgicicek
Differential Revision: D23729222
fbshipit-source-id: 4e9681f65
Summary:
ObjC++ models are a copy (symlink) into the ObjC ones, albeit with different compile time flags. However, the resulting procnames are identical, and we are left with only one copy of models anyway. This means that only one version (chosen arbitrarily by the build order) is used for analysis.
This diff deletes the ObjC++ version.
Reviewed By: jvillard
Differential Revision: D23704266
fbshipit-source-id: 1dc94251f
Summary:
Handle the case of
- computing the range of `[lb, ub]` where `lb` contains length path and `ub` is constant. E.g. `[a.length, 2]` would give `3` range
- simplifying `c1 +/- min(c2, a.length)` to `c1 +/- min(c2, 0)`
since the length of a collection/array is always nonnegative.
This also removes some existing FPs in tests results.
Facebook
Context: we stumbled upon this issue in ObjC results TV109717121 which is corresponding to the added test case.
Reviewed By: skcho
Differential Revision: D23648807
fbshipit-source-id: 1e89ea246
Summary: This diff adds trace field for autoreleasepool size. Unlike to the other checkers, eg inferbo and operation cost, the autoreleasepool size checker should have traces for constants.
Reviewed By: ezgicicek
Differential Revision: D23678084
fbshipit-source-id: 35e6cf5f5
Summary:
Variables are represented as a subtype of terms. The implementation of
the operations on variables is largely independent of this. This diff
generalizes the implementation over the concrete representation, and
moves it to a separate module.
Reviewed By: ngorogiannis
Differential Revision: D23660290
fbshipit-source-id: 90d8c3ca4
Summary:
Increase the fidelity of the status in the analysis report slightly by
including the number of analysis steps taken.
Reviewed By: jvillard
Differential Revision: D23648493
fbshipit-source-id: 8050b7829
Summary:
Just to avoid confusion over directories named 'bin' being expected to
contain binaries, rather than the sources of executables.
Reviewed By: jvillard
Differential Revision: D23636372
fbshipit-source-id: 8593380e6
Summary:
Add support to ppx_trace to rewrite `[%trace]` into a call to the
corresponding `Trace` function:
```lang=ocaml
val trace :
?call:(pf -> unit)
-> ?retn:(pf -> 'a -> unit)
-> ?rais:(pf -> exn -> Printexc.raw_backtrace -> unit)
-> string
-> string
-> (unit -> 'a)
-> 'a
(** [trace ~call ~retn ~rais mod_name fun_name k] either simply invokes
[k ()], when not enabled, or else increases the indentation level and
emits the [call] message, then invokes [k ()], then decreases the
indentation level and either emits the [retn] or [rais] message,
depending on whether [k ()] returned normally or exceptionally. *)
```
The main motivation over the existing `Trace.call` and `Trace.retn` is
that by packaging them together, unhandled exceptions can be treated
better.
Reviewed By: jvillard
Differential Revision: D23636200
fbshipit-source-id: f61c267fd
Summary:
Construct the identity function explicitly instead of relying on the
`Stdlib.Fun.id` name, which can be problematic depending on compiler
version and which modules have been `open`ed.
Reviewed By: jvillard
Differential Revision: D23636204
fbshipit-source-id: 2fb644c18
Summary:
Version 0.16 of ppxlib breaks merlin in combination with ppx_trace,
due to https://github.com/ocaml-ppx/ppxlib/issues/175.
Reviewed By: jvillard
Differential Revision: D23636202
fbshipit-source-id: 37a5f36b7
Summary:
The expansion of the `[%debug]` extension point is simple and fits
into the pattern supported by `Ppxlib.Extension`. This diff simplifies
the mapper implementing the rest of the trace expansion by moving the
expansion of debug into a separate
`Ppxlib.Context_free.Rule.extension` rule.
Reviewed By: jvillard
Differential Revision: D23636201
fbshipit-source-id: 847c258fd
Summary:
Use `Ppxlib.Ast_traverse.map` instead of
`Ppxlib.Selected_ast.Ast.Ast_mapper` which is included from
`Migrate_parsetree` since `Ppxlib.Selected_ast.Ast` reexports one of
the `Migrate_parsetree.Versions` modules. This change is needed to be
compatible with (ocaml-migrate-parsetree 2.0 and) ppxlib 0.16 since it
no longer re-exports the `Ast_mapper` module from
`Migrate_parsetree`. This ppxlib change is one of the headline
simplification enablers noted in the (announcement of
ocaml-migrate-parsetree
2.0)[https://discuss.ocaml.org/t/ocaml-migrate-parsetree-2-0-0/5991].
Reviewed By: jvillard
Differential Revision: D23636203
fbshipit-source-id: 71e24b46b
Summary:
When equal_or_separate returns Unknown, it is common to sort the args,
which is wasteful since computing equal_or_separate already had to
test if the args are equal, which is most if not all of the work of
comparing them.
Reviewed By: jvillard
Differential Revision: D23636205
fbshipit-source-id: 5b2bcdd8f
Summary:
For complexity issues from O(m) to O(n), we only include the trace of the current complexity O(n). However, this makes it difficult to understand what the original complexity O(m) was. Especially in fixed issues where n=1, we only get a constant cost with no trace attached, so it is difficult to see how the symbol m disappeared.
This diff includes the traces for the previous cost in the cost issues.
Reviewed By: skcho
Differential Revision: D23680360
fbshipit-source-id: 3f2b21b20
Summary:
The best-fit allocator with overhead 120 is slightly faster, and
significantly reduces peak memory usages in the workers.
Reviewed By: jvillard
Differential Revision: D23678794
fbshipit-source-id: ccdfced74
Summary:
When the switch changes to a new version of ocaml, opam may need updating to
see the new compiler version.
Reviewed By: jberdine
Differential Revision: D23678520
fbshipit-source-id: de1205de6
Summary:
The `attr_kind` column has now exactly two possible values: defined and undefined (since D22187238 (61ae2d1e1b)). This is also what should be stored in the field `is_defined` in the procedure attributes. Finally, the cfg can be `NULL` or not. This diff makes all three of these agree, in order to allow for the removal of the `attr_kind` column up the stack, since it will be equivalent to `cfg is NOT NULL`.
The changes in the tests indicate improved correctness: previously, specialising a modelled callee would result in a dummy entry in the procedures table, because all clang procedures were given a procdesc, so the specialisation would specialise the empty procdesc. Now, the model is not shadowed by the specialised dummy procdesc, and is specialised itself (and that's why it also ends up in the capture DB, which is slightly counter-intuitive).
Reviewed By: jvillard
Differential Revision: D23536931
fbshipit-source-id: 983216cb6
Summary: So that it can be meaningfully used for diffing across infer versions, say.
Reviewed By: martintrojer
Differential Revision: D23648560
fbshipit-source-id: 98d634e37
Summary:
Since we don't yet support `.sig` files for fields, hardcoding this in Ocaml
for now.
Reviewed By: jvillard
Differential Revision: D23627390
fbshipit-source-id: 8df78068d
Summary: Polynomial category zero corresponds to unreachable but "zero" is a misnomer and rather confusing. Let's fix it.
Reviewed By: skcho
Differential Revision: D23597735
fbshipit-source-id: f0c96ed26
Summary: Need to use relative paths in order to make the hash consistent across machines
Reviewed By: ngorogiannis
Differential Revision: D23623726
fbshipit-source-id: 44f750658
Summary:
Eliminate the need to serialise procnames when sending work from the restart scheduler to the workers, by sending the proc_uid instead. This is (much) shorter than the byte representation of the proc_name and it's the primary DB key of the procedures table, so it can be used by the worker to obtain the full procname.
Also, reduce GC churn by using folds in the scheduler startup instead of copying lists over and over.
Reviewed By: jberdine
Differential Revision: D23566131
fbshipit-source-id: 1472aa990
Summary:
Script changes to make the clang build work without the FCP submodule.
A llvm+clang+libraries src folder is downloaded and put together in the clang/src/download folder. This is the used for the clang build.
Reviewed By: ngorogiannis
Differential Revision: D23315610
fbshipit-source-id: 0cbce23e2
Summary: dont keep dotfiles in the facebook-clang-plugins folder, rather merge into the main ones
Reviewed By: ngorogiannis
Differential Revision: D23599305
fbshipit-source-id: ced837bc4
Summary: Remove FCP submodule, preparing for folding it into the infer repo
Reviewed By: ngorogiannis
Differential Revision: D23315609
fbshipit-source-id: 6bf0ecd77
Summary:
- freshen up /docs/next/absint-framework to give sensible advice, and
delete outdated bits that are now in the API docs so they remain fresh
- delete SimpleChecker.ml as it's just a source of bitrot
- delete the "adding checkers" page as it's completely outdated and
subsumed by the "AI framework" page + the labs.
Reviewed By: jberdine
Differential Revision: D23597271
fbshipit-source-id: 78b541746
Summary:
We defined cost's plus as "zero+unreachable = unreachable" for the operation cost. The meaning of
the zero cost is that no statment is analyzed yet, and the unreachable(bottom) cost means a program
point is analyzed as unreachable. We thought "zero+unreachable" happens in very specific cases we
need to check, or due to an analyzer bug. For debugging purpose, we defined it to return more
specific unreachable cost.
However, in allocation/autoreleasepool costs, the zero cost is not very special value. If there is
no alloc/autorelease calls, they can have the zero cost. "zero+unreachable = unreachable" doesn't
make sense there.
This diff changes the plus as "zero+unreachable = zero" for the non-operation costs.
Reviewed By: ezgicicek
Differential Revision: D23578869
fbshipit-source-id: 5392eca1c
Daiva Naudziuniene
Sungkeun Cho
Qianyi Shu
Mitya Lyubarskiy
Jules Villard
Nikos Gorogiannis
Ezgi Çiçek
Josh Berdine
Martin Trojer