Summary:
- add the variable being declared so we can report it back in the trace in addition to its location
- distinguish between local vars and formals
Reviewed By: skcho
Differential Revision: D17930348
fbshipit-source-id: a5b863e64
Summary:
When we make the decision to go into a branch "v = N" where some
abstract value is compared to a constant, remember the corresponding
equality. This allows to prune simple infeasible paths
intra-procedurally.
Further work is needed to make this useful interprocedurally, for
instance either or both of these ideas could be explored:
- abduce v=N in the precondition and do not apply summaries when the
equalities in the pre are not satisfied
- prune post-conditions that lead to unsat states where a value has to
be equal to several different constants
Reviewed By: skcho
Differential Revision: D17906166
fbshipit-source-id: 5cc84abc2
Summary:
When we know "x = 3" and we have a condition "x != 3" we know we can
prune the corresponding path.
Reviewed By: skcho
Differential Revision: D17665472
fbshipit-source-id: 988958ea6
Summary:
Unfortunately it is very hard to predict when
`Typ.Procname.describe` will add `()` after the function name, so we
cannot make sure it is always there.
Right now we report clowny stuff like "error while calling `foo()()`",
which this change fixes.
Reviewed By: ezgicicek
Differential Revision: D17665470
fbshipit-source-id: ef290d9c0
Summary:
Pulse didn't treat local variables going out of scope as invalidating the corresponding address in memory. This diff fixes that by
- marking all local variables that exits the scope with the attribute `AddressOfStackVariable`
- before we write the summary for the proc, we make sure to invalidate all such addresses local to the procedure as `Invalid.` If such an address is read, then we would raise a use-after-lifetime issue.
Reviewed By: jvillard
Differential Revision: D16458355
fbshipit-source-id: 3686524cb
Summary:
A common gotcha is the new test. Model the minimum amount of
`std::basic_string` to catch it.
Reviewed By: mbouaziz, ngorogiannis
Differential Revision: D16121090
fbshipit-source-id: 66f06cb43
Summary:
Be more flexible in what type of function calls are allowed in `ViaCall ...` actions to be able to include models.
Also get rid of `here here` in traces /o\
As a side-effect, get more precise (=qualified) procedure names in
traces (but not in messages so as not to be too verbose).
Reviewed By: mbouaziz, ngorogiannis
Differential Revision: D16121092
fbshipit-source-id: fb51b02f8
Summary:
So it turns out we need to translate even more cases. Pulse had a FP
before that this fixes.
Reviewed By: ezgicicek
Differential Revision: D16073629
fbshipit-source-id: c03460b5a
Summary:
This is needed to test some functionality in the next diff. Only one
test changes (no longer a FN), which is now documented. Also, stop
including the "header models" meant for biabduction!
Maybe one day we'll need to have several test modes for different C++
versions. Seems overkill for now, so let's wait until we see some actual
issues (eg FPs) that manifest in one version but not the other.
Reviewed By: mbouaziz
Differential Revision: D16073630
fbshipit-source-id: 1cfdfc933
Summary:
Sometimes the post of a function call has attributes on addresses that
were mentioned in the pre but are no longer reachable in the post. We
don't want to forget these, see added test.
Reviewed By: mbouaziz
Differential Revision: D16050050
fbshipit-source-id: 1ce522b97
Summary:
The previous code would call the destructor for the C++ temporary
*before* the prune nodes, which then try to dereference it. Wrong.
Quick fix: don't destroy temporaries in conditionals.
Reviewed By: mbouaziz
Differential Revision: D16030735
fbshipit-source-id: e11abad58
Summary:
We were skipping some instructions before and that was a problem for
pulse. See added pulse test.
Reviewed By: mbouaziz
Differential Revision: D16030150
fbshipit-source-id: 9c62e6213
Summary: Not sure if anyone uses this but there, now it's modelled.
Reviewed By: mbouaziz
Differential Revision: D16008162
fbshipit-source-id: f4795dcba
Summary:
Prevent false positives about variables captured by value gone out of
scope.
Reviewed By: ezgicicek
Differential Revision: D16008165
fbshipit-source-id: d70e47db4
Summary: We know how to do interprocedural calls so let's use that!
Reviewed By: mbouaziz
Differential Revision: D16008164
fbshipit-source-id: 4c34bf704
Summary:
`function::operator=` is called whenever we assign a literal lambda to a
variable, so it's pretty useful to be able to report anything on
lambdas.
Reviewed By: mbouaziz
Differential Revision: D16008163
fbshipit-source-id: a9d07668d
Summary:
Printing `Exp.Const (Cfun proc_name)` adds `_fun_` in front of the
procedure name, eg `_fun_foo` instead of `foo`. This showed up in pulse
traces.
Reviewed By: mbouaziz
Differential Revision: D16004606
fbshipit-source-id: 72ac6866f
Summary:
Fixes a false positive where the address of a C++ temporary is bound to
a static const reference variable then returned. The fix doesn't try to
establish that the variable is a const reference so could lead to false
negatives but that can be addressed later.
Reviewed By: ezgicicek
Differential Revision: D16004538
fbshipit-source-id: e403dbefe
Summary:
[apologies for the unreviewable diff...]
Get rid of HIL expressions in pulse. This finishes the HIL -> SIL
migration. The first step made pulse start from SIL instructions but
would translate most accesses to HIL to re-use most of the existing
pulse code. This diff gets rid of the intermediate translation of SIL
expressions to HIL expressions.
Big changes:
1. `PulseOperations` mostly rewritten, driven by using `Exp.t` instead of `HilExp.AccessExpression.t` for everything.
2. Stop trying to reverse-engineer what addresses mean in terms of
access paths from program variables. Rely on the trace pointing at
the right places in the code to be enough. This is because it wasn't
that useful (and could even be misleading when wrong) but could be
prohibitively expensive in degenerate cases (eg nodes with tens of
thousands of successive array accesses...)
3. `PulseAbductiveDomain.apply_post` now returns the computed return
value instead of recording it itself.
4. Change of vocabulary: `materialize` -> `eval`, `crumb` -> `event`
5. Function calls arguments are now evaluated prior to doing anything
else, which saves everything else from having to (remember to) do
that. In particular, this changes how models look quite a bit.
Reviewed By: mbouaziz
Differential Revision: D15986373
fbshipit-source-id: 1d79935de
Summary: Inject destructor calls to destroy a temporary when its lifetime ends.
Reviewed By: mbouaziz
Differential Revision: D15674209
fbshipit-source-id: 0f783a906
Summary:
Now that HIL doesn't help us anymore we need to reconstruct its mapping
"SIL logical var -> program access path". We already have everything we
need in pulse: it suffices to walk the current memory graph starting
from program variables until we find the value of the temporary we are
interested in.
This diff also builds some type machinery to make sure all accesses are
explained.
Reviewed By: mbouaziz
Differential Revision: D15824959
fbshipit-source-id: 722c81b39
Summary:
It turns out HIL gets in the way of a precise heap analysis. For
instance, instead of:
```
n$0 = *&x.f
_ = delete(&x)
*&y = n$0
```
HIL tries hard to forget about intermediate variables and shows instead
```
_ = delete(&x)
*&y = *&x.f
```
Oops, that's a use-after-delete, whereas the original code was safe.
While it's easy to write SIL programs that are completely unsound for
HIL, they are not generated very often from the frontends. In fact, the
problem became apparent only when making the clang frontend translate
C++ temporaries destructors, which produces the situation above
routinely.
This diff makes the minimal amount of change to make Pulse build and
produce equivalent results (minus HIL bugs) starting from SIL instead of
HIL. The reporting sucks for now because we need to translate SIL
temporaries back into program access paths. This is done in the next
diff.
Reviewed By: mbouaziz
Differential Revision: D15824961
fbshipit-source-id: 8e4e2a3ed
Summary:
This one isn't caught because we don't destruct temporaries that are
bound to a const reference. According to the C++ standard these should
get destroyed when the const reference gets destroyed but instead we
just don't destroy them for now.
Reviewed By: mbouaziz
Differential Revision: D15760209
fbshipit-source-id: 32c935ec0
Summary:
In a next diff temporaries will get destructed at the end of their
lifetimes and that naive model would be causing false positives.
The flipside is that we lose all reports on closures for now, will need
to model them separately later.
Reviewed By: mbouaziz
Differential Revision: D15695943
fbshipit-source-id: c2c482c02
Summary:
This started as an attempt to understand how to modify the frontend to
inject destructors for C++ temporaries (see next diffs).
This diff rewrites the existing logic for computing the list of
variables that should be destroyed at the end of each statement, either
because it's the end of their syntactic scope or because control flow
branches outside of their syntactic scope.
The frontend translates a function from the last instructions to the
first, but scope computation needs to be done in the other direction, so
it's done in a separate pass *before* the main translation happens. That
first pass creates a map from statements in the AST to the list of
variables that should be destroyed at the end of these statements. This
is still the case now.
Before, that map would be computed in a bit of a weird way: scopes are
naturally a stack but instead of that the structure maintained was a
flat list + a counter to know where the current scope ended in that
list.
In this diff, redo the computation maintaining a stack of scopes
instead, which is a bit cleaner. Also treat more instructions as
introducing a new scope, eg if, for, ...
Reviewed By: mbouaziz
Differential Revision: D15674208
fbshipit-source-id: c92429e82
Summary:
I rewrote the test so it doesn't need any C++ headers so that:
- it's easier to see what's going on
- it's easier to debug: the whole AST is now somewhat readable vs before
the headers made it impossibly long
Reviewed By: ezgicicek
Differential Revision: D15674213
fbshipit-source-id: d98941983
Summary:
This messes with the deduplication heuristic when templated function
names show up in the error messages, since the heuristic demands that
the error messages are the same.
Reviewed By: mbouaziz
Differential Revision: D15374333
fbshipit-source-id: 70232d254
Summary:
Improve the error messages, change is more or less documented in the
code.
Reviewed By: mbouaziz
Differential Revision: D15374334
fbshipit-source-id: f1dd54180
Summary:
Before: the trace would explain how a value was invalidated and
accessed, but not how the value that was invalidated had been
constructed.
Now: `PulseTrace.t` records breadcrumbs of how the value was constructed
in addition to the interproc "action" trace leading to the invalidation
or access action.
Concretely:
```
void bad(X &x) {
X *y = x;
X *z = x;
delete y;
access(z);
}
```
will produce the trace:
Invalidation part:
y = x
delete y
Access part:
z = x
access(z)
access to z->f inside of access(z)
Before this diff the "Access part" would be missing the "z = x" part of
the trace, so it might be confusing why `z` has anything to do with `y`.
However, such "breadcrumbs" are not recorded in the inter-procedural
part, only the sequence of calls is. This is a trade-off for simplicity,
maybe it's enough for developers maybe it isn't, we'll find out later.
Reviewed By: jberdine
Differential Revision: D15354438
fbshipit-source-id: 8d0aed717
Summary:
Feedback from peterogithub:
- mention which access path is being invalidated and accessed in the message
- mention the line at which it was invalidated (the line at which it's accessed is already the line at which we report)
- traces for stack variable/C++ temporary address escapes
- delete double implementation of the same functionality in
`PulseTrace`: `location_of_action_start` is the same as
`outer_location_of_action`...
Reviewed By: jberdine
Differential Revision: D14800294
fbshipit-source-id: 3d9ab9b3d
Summary:
Similarly to function parameters (and the return value), we need to
apply the pre/post of a function call to the globals mentioned in its
summary.
- tigthen summaries further to remember only abducible variables in the
post (as well as in the pre)
- take globals into account when applying pre/post pairs
Reviewed By: jberdine
Differential Revision: D14780800
fbshipit-source-id: fc0d180bb
Summary:
The heuristic to detect variables going out of scope was to detect any
access expression passed as argument to an injected destructor call.
However destructor calls are also injected in destructor bodies to
destruct each field of an object, so the heuristic would detect fields
going out of scope, which, erm, doesn't make sense. Limit the heuristic
to local program variables.
Reviewed By: jberdine
Differential Revision: D14771454
fbshipit-source-id: ffa3c9fe3
Summary:
Only throw values to the pre if they can be followed from "abducible"
variables: formals of the current method and globals.
Because figuring out if a `Pvar.t` is a formal of the current procedure
is actually a giant pain, hack something not too bad instead:
pre-register all formals at the start of the analysis of the
procedure. Then the only other variables we care about in the
precondition are globals, which we can detect easily.
This is mostly an optimisation (summaries won't include irrelevant
"abduced" facts about the procedure's local variables anymore), but it
also fixes a bug where we would sometimes overwrite things in the pre. I
think that's why the tests improved.
Reviewed By: ngorogiannis
Differential Revision: D14753493
fbshipit-source-id: 08e73637f
Summary:
This mostly doesn't make sense. The only thing this would have been good
for was to give the most accurate result on access paths such as
`*(&(x.f))`, but these are normalised anyway (into `x.f`) so we actually
never see these. That said there might be some use to some similar logic
in the future, but in the meantime let's delete the current feature as
it wasn't thought through.
Reviewed By: ezgicicek
Differential Revision: D14753492
fbshipit-source-id: 597cec027
Summary:
The previous message formatting had regressed and produced non-sensical messages.
More importantly, remove template parameters from error messages to
trigger the heuristic in `InferPrint` that deduplicates errors that are
on the same line with the same error type and message. Without this we
get hundreds of reports that correspond to as many instantiations of the
same code.
Reviewed By: ngorogiannis
Differential Revision: D14747979
fbshipit-source-id: 3c4aad2b1
Summary:
We see the magic function `__variable_initialization` at the point where
the variable is declared, eg `int x = foo()`. It's safe to reset `&x` at
that point. This circumvents an issue that pops up in some rare cases
where the ternary conditional operator `?:` and variable initialization
conspire to produce weird frontend results.
Some test becomes a FN again, but I think it was being reported for the
wrong reasons; will investigate more later.
Reviewed By: ngorogiannis
Differential Revision: D14747980
fbshipit-source-id: e75d6e30f
Summary:
This ensures that each attribute type can only be present once per
address. Makes ~80x time improvement on pathological cases such as
Duff's device.
This introduces a new kind of Set in `PrettyPrintable`.
Reviewed By: mbouaziz
Differential Revision: D14645091
fbshipit-source-id: c7f9b760c
Summary:
Detect when a variable goes out of scope. When that's the case, mark its
address *and* its contents as invalid.
Give subsequent uses a USE_AFTER_LIFETIME error type instead of
USE_AFTER_DESTRUCTOR.
Reviewed By: jberdine
Differential Revision: D14387147
fbshipit-source-id: a2c530fda
Summary:
Instead of emitting an ad-hoc builtin on variable declaration emit a new
metadata instruction. This allows us to remove the code matching on that
ad-hoc builtin that had to be inserted in several checkers.
Inferbo & pulse used that information meaningfully and had to undergo
some minor changes to cope with the new metada instruction.
Reviewed By: ezgicicek
Differential Revision: D14833100
fbshipit-source-id: 9b3009d22
Summary:
Bundle all non-semantic-bearing instructions into a `Metadata _`
instruction in SIL.
- On a documentation level this makes clearer the distinction between
instructions that encode the semantics of the program and those that are
just hints for the various backend analysis.
- This makes it easier to add more of these auxiliary instructions in
the future. For example, the next diff introduces a new `Skip` auxiliary
instruction to replace the hacky `ExitScope([], Location.dummy)`.
- It also makes it easier to surface all current and future such
auxiliary instructions to HIL as the datatype for these syntactic hints
can be shared between SIL and HIL. This diff brings `Nullify` and
`Abstract` to HIL for free.
Reviewed By: ngorogiannis
Differential Revision: D14827674
fbshipit-source-id: f68fe2110
Summary:
Some of these tests were wrong, eg `~lambda()` calls `lambda()` then...
takes the bitwise complement or something? The intent was to call the
destructor.
Add interprocedural tests for later.
Reviewed By: jberdine
Differential Revision: D14324762
fbshipit-source-id: 40d2c32f5
Summary:
Previously we would say that `lhs <= rhs` (or `lhs |- rhs`) when a
mapping existed between the abstract addresses of `lhs` and `rhs` such
that `mapping(lhs)` was a supergraph of `rhs`. In particular,
we had that `x |-> x' * x' |-> x'' |- x |-> x'`. This is not entirely
great, in particular once we get pairs of state representing footprint +
current state. I'm not sure I have an extremely compelling argument why
though, except that it's not the usual way we do implication in SL, but
there wasn't a compelling argument for the previous state of affairs
either.
This changes `|-` to be true only when `mapping(lhs) = rhs` (modulo only
considering the addresses reachable from the stack variables).
Reviewed By: jberdine
Differential Revision: D14568272
fbshipit-source-id: 1bb83950e
Summary: This helps convergence when `<=` is based on physical equality for example, and widening is implemented as `widen ~prev ~next = join prev next`.
Reviewed By: skcho
Differential Revision: D14568270
fbshipit-source-id: ded5ed296
Summary:
When joining two lists of disjuncts we try to ensure there isn't a state
that under-approximates another already in the list. This helps reduce
the number of disjuncts that are generated by conditionals and loops.
Before we would always just add more disjuncts unless they were
physically equal but now we do a subgraph computation to assess
under-approximation.
We only do this half-heartedly for now however, only taking into
consideration the "new" disjuncts vs the "old" ones. It probably makes
sense to do a full quadratic search to minimise the number of disjuncts
from time to time but this isn't done here.
Reviewed By: mbouaziz
Differential Revision: D14258482
fbshipit-source-id: c2dad4889
Summary:
The disjunctive domain shouldn't really be a set in the first place as
comparing abstract states for equality is expensive to do naively
(walking the whole maps representing the abstract heap). Moreover in
practice these sets have a small max size (currently 50 for pulse, the
only client), so switching them to plain lists makes sense.
Reviewed By: mbouaziz
Differential Revision: D14258489
fbshipit-source-id: c512169eb
Summary:
It's useful to keep the size of states down, especially when humans are
trying to read it. It will also help keep the size of summaries down in
the inter-procedural pulse.
Reviewed By: mbouaziz
Differential Revision: D14258486
fbshipit-source-id: 45ebcac67
Summary:
Record per-location traces. Actually, that doesn't quite make sense as a
location can be accessed in many ways, so associate a trace to each
*edge* in the memory graph. For instance, when doing `x->f = *y`, we
want to take the history of the `<val of y> --*--> ..` edge, add "assigned
at location blah" to it and store this extended history to the edge
`<val of x> --f--> ..`.
Use this machinery to print nicer traces in `infer explore` and better
error messages too (include the last assignment, like biabduction
messages).
Reviewed By: da319
Differential Revision: D13518668
fbshipit-source-id: 0a62fb55f
Summary: Adding a test case for use after destructor for temporaries. At the moment pulse does not find it as frontend does not inject destructors for temporaries.
Reviewed By: jvillard
Differential Revision: D13506229
fbshipit-source-id: 31b9466f7
Summary:
When a C++ temporary goes out of scope, tag its address in the heap with
a new attribute `AddressOfCppTemporary` so that we can later check that
we don't return it.
Reviewed By: da319
Differential Revision: D13466898
fbshipit-source-id: 8808338b4
Summary:
When assign to the special `return` variable, check that the result is
not the address of a local variable, otherwise report.
Reviewed By: ngorogiannis
Differential Revision: D13466896
fbshipit-source-id: 465da7f13
Summary: Naming it `FP_` was a mistake in the original commit that copied the tests over as pulse has never reported on that method.
Reviewed By: da319
Differential Revision: D13465324
fbshipit-source-id: f8b24ebda
Summary:
It's ok to take an address of a field / array access of an invalid object.
This diff calculates the inner most dereference for an access expression starting with `&` and does not report on the dereference even if the address is invalid.
Reviewed By: jvillard
Differential Revision: D13450758
fbshipit-source-id: 18c038701
Summary:
When we create Dereference edge, we also create TakeAddress back edge. This causes false positives for stack variables. When we write to a stack variable and then take its address, the resulting address is the one from the back edge of the written value. See example `push_back_value_ok`. To solve this issue, this diff changes stack to denote a map from address of variables rather than from variables.
We still have issue for fields, see example, FP_push_back_value_field_ok. To solve this, we probably need to remove back edges.
Reviewed By: jvillard
Differential Revision: D13432415
fbshipit-source-id: 9254a1a6d
Summary:
When a lambda gets created, record the abstract addresses it captures, then
complain if we see some of them be invalidated before it is called.
Add a notion of "allocator" for reporting better messages. The messages are
still a bit sucky, will need to improve them more generally at some point.
```
jul lambda ~ infer 1 infer -g --pulse-only -- clang -std=c++11 -c infer/tests/codetoanalyze/cpp/pulse/closures.cpp
Logs in /home/jul/infer.fb/infer-out/logs
Capturing in make/cc mode...
Found 1 source file to analyze in /home/jul/infer.fb/infer-out
Found 2 issues
infer/tests/codetoanalyze/cpp/pulse/closures.cpp:21: error: USE_AFTER_DESTRUCTOR
`&(f)` accesses address `s` captured by `&(f)` as `s` invalidated by destructor call `S_~S(s)` at line 20, column 3 past its lifetime (debug: 5).
19. f = [&s] { return s.f; };
20. } // destructor for s called here
21. > return f(); // s used here
22. }
23.
infer/tests/codetoanalyze/cpp/pulse/closures.cpp:30: error: USE_AFTER_DESTRUCTOR
`&(f)` accesses address `s` captured by `&(f)` as `s` invalidated by destructor call `S_~S(s)` at line 29, column 3 past its lifetime (debug: 8).
28. f = [&] { return s.f; };
29. }
30. > return f();
31. }
32.
Summary of the reports
USE_AFTER_DESTRUCTOR: 2
```
Reviewed By: da319
Differential Revision: D13400074
fbshipit-source-id: 3c68ff4ea
Summary: Model more `std::vector` functions that can potentially invalidate references to vector's elements (https://en.cppreference.com/w/cpp/container/vector).
Reviewed By: mbouaziz
Differential Revision: D13399161
fbshipit-source-id: 95cf2cae6
Summary:
Some code calls `this->~Obj()` then proceeds to use fields in the current
object, which previously we would report as invalid uses. Assume people know
what they are doing and ignore destructor calls to `this`.
Reviewed By: mbouaziz
Differential Revision: D13401145
fbshipit-source-id: f6b0fb6ec
Summary: Similarly as `std::vector::push_back`, `std::vector::reserve` can invalidate the references to elements if the new size is bigger than the existing one. More info on `std::vector::reserve`: https://en.cppreference.com/w/cpp/container/vector/reserve
Reviewed By: jvillard
Differential Revision: D13340324
fbshipit-source-id: bf99b6923
Summary: Instead of variable having the value of a single location on stack, we now allow variables to have multiple locations. Consequently, we also allow a memory location to point to a set of locations in the heap. We enforce a limit on a maximum number of locations in a set (currently 5).
Reviewed By: jvillard
Differential Revision: D13190876
fbshipit-source-id: 5cb5ba9a6
Summary: Recent improvements in join fixed `FP_allocate_in_branch_ok` because the variable was not read after the join.
Reviewed By: mbouaziz
Differential Revision: D13233441
fbshipit-source-id: 89b701e12
Summary:
It's useful for checkers to know when variables go out of scope to
perform garbage collection in their domains, especially for complex
domains with non-trivial joins. This makes the analyses more precise at
little cost.
This could have been added as a custom function call to a builtin, but I
decided against it because this instruction doesn't have the semantics
of any function call. It's better for each checker to explicitly not
deal with the custom instruction instead.
Reviewed By: jberdine
Differential Revision: D13102951
fbshipit-source-id: 33be22fab
Summary:
Update clang plugin which now gives names to variables captured by lambdas that were empty before.
update-submodule: facebook-clang-plugins
Reviewed By: jvillard
Differential Revision: D12979015
fbshipit-source-id: 0b092fb24
Summary:
It turns out keeping attributes (such as invalidation facts) separate
from the memory is a bad idea and leads to loss of precision and false
positives, as seen in the new test (which previously generated a
report).
Allow me to illustrate on this example, which is a stylised version of
the issue in the added test: previously we'd have:
```
state1 = { x = 1; invalids={} }
state2 = { x = 2; invalids ={1} }
join(state1, state2) = { x = {1, 2}; invalids={{1, 2}} }
```
So even though none of the states said that `x` pointed to an invalid
location, the join state says it does because `1` and `2` have been
glommed together. The fact `x=1` from `state1` and the fact "1 is
invalid" from `state2` conspire together and `x` is now invalid even
though it shouldn't.
Instead, if we record attributes as part of the memory we get that `x`
is still valid after the join:
```
state1 = { x = (1, {}) }
state2 = { x = (2, {}) }
join(state1, state2) = { x = ({1, 2}, {}) }
```
Reviewed By: mbouaziz
Differential Revision: D12958130
fbshipit-source-id: 53dc81cc7
Summary:
I hear that this scheduler is better. I want the best scheduler
possible. Also pulse's join is a bit complex so it might matter one day.
whydididothis
Reviewed By: mbouaziz
Differential Revision: D12958131
fbshipit-source-id: 3bd77ccba
Summary: For a general case of `operator=` we want to create a fresh location for the first parameter as `operator=` behaves as copy assignment.
Reviewed By: jvillard
Differential Revision: D12940635
fbshipit-source-id: 89c6e530d
Summary:
Whenever `vec.reserve(n)` is called, remember that the vector is
"reserved". When doing `vec.push_back(x)` on a reserved vector, assume
enough size has been reserved in advance and do not invalidate the
underlying array.
This gets rid of false positives.
Reviewed By: mbouaziz
Differential Revision: D12939837
fbshipit-source-id: ce6354fc5
Summary:
Instead of keeping at most one invalidation fact for each address, keep
a set of them and call them "attributes". Keeping a set of invalidation
facts is redundant since we always only want the smallest one, but
makes the implementation simpler, especially once we add more kinds of
attributes (used for modelling, see next diffs).
Reviewed By: mbouaziz
Differential Revision: D12939839
fbshipit-source-id: 4a54c2132
Summary:
Copied on the ownership checker logic: return the initial value of the
domain as return. This can probably be improved.
Reviewed By: mbouaziz
Differential Revision: D12888102
fbshipit-source-id: 9e2dac7fc
Summary:
When initialising a variable via semi-exotic means, the frontend loses
the information that the variable was initialised. For instance, it
translates:
```
struct Foo { int i; };
...
Foo s = {42};
```
as:
```
s.i := 42
```
This can be confusing for backends that need to know that `s` actually
got initialised, eg pulse.
The solution implemented here is to insert of dummy call to
`__variable_initiazition`:
```
__variable_initialization(&s);
s.i := 42;
```
Then checkers can recognise that this builtin function does what its
name says.
Reviewed By: mbouaziz
Differential Revision: D12887122
fbshipit-source-id: 6e7214438
Summary:
Now that arrays are dealt with separately (see previous diff), we can
turn the join back into an over-approximation as far as invalid
locations are concerned.
Reviewed By: skcho
Differential Revision: D12881989
fbshipit-source-id: fd85e49c0
Summary:
This prevents the join from wrongly assuming that we haven't seen a
variable on one side of the join.
Reviewed By: skcho
Differential Revision: D12881987
fbshipit-source-id: 42a776adb
Summary: For `operator=(lhs, rhs)` we want to model it as an assignment if rhs is materialized temporary created in the constructor.
Reviewed By: jvillard
Differential Revision: D10462510
fbshipit-source-id: 998341e69
Summary: Do not create a new location for placement new argument if it already exists.
Reviewed By: jvillard
Differential Revision: D12839942
fbshipit-source-id: 758b67a82
Summary:
Get rid of `USE_AFTER_LIFETIME`. This could be useful to deploy pulse
alongside the ownership checker too.
Reviewed By: da319
Differential Revision: D12857477
fbshipit-source-id: 8e2a2a37c
Summary:
Keep `USE_AFTER_LIFETIME` for unclassified errors (for now it contains
vector invalidation too because I can't think of a good name for
them, and maybe it makes sense to wait until we have more types of them
to decide on a name).
Reviewed By: da319
Differential Revision: D12825060
fbshipit-source-id: bd75ef698
Summary:
Getting this right will be long and complex so for now the easiest is to
underreport and only consider as invalid the addresses we know to be invalid on
both sides of a join. In fact the condition for an address to be invalid after
a join is more complex than this: it is invalid only if *all* the addresses in
its equivalence class as discovered by the join are invalid.
Reviewed By: skcho
Differential Revision: D12823925
fbshipit-source-id: 2ca109356
Summary: Similarly as for destructors, we provide an address of an object as a first parameter to constructors. When constructor is called we want to create a fresh location for a new object.
Reviewed By: jvillard
Differential Revision: D10868433
fbshipit-source-id: b60f32953
Summary: We provide an address of an object as a parameter to destructor. When destructor is called the object itself is invalidated, but not the address.
Reviewed By: jvillard
Differential Revision: D12824032
fbshipit-source-id: 516eebcf8
Summary:
The time has come to keep track of which tests pass and which are FP/FN
for pulse.
Reviewed By: mbouaziz
Differential Revision: D10854064
fbshipit-source-id: 60938e48f
Summary:
Turns out once a vector array became invalid it stayed that way, instead
of the vector getting a new valid internal array.
Reviewed By: skcho
Differential Revision: D10853532
fbshipit-source-id: f6f22407f
Summary:
Now the domain can reason about `&` and `*` too. When recording `&`
between two locations also record a back-edge `*`, and vice-versa.
Reviewed By: mbouaziz
Differential Revision: D10509335
fbshipit-source-id: 8091b6ec0
Summary: This is more flexible and allows us to give more details when reporting.
Reviewed By: mbouaziz
Differential Revision: D10509336
fbshipit-source-id: 79c3ac1c8
Summary:
Invalidating addresses for destructors to catch use after destructor errors.
To pass ownership tests for use after destructor errors, we still need to:
(1) fix pointer arithmetic false positives
(2) add model for placement new to fix false positives
(3) add model for operator= to fix false positives
(4) support inter-procedural analysis for destructor_order_bad test
Reviewed By: jvillard
Differential Revision: D10450912
fbshipit-source-id: 2d9b1ee68
Summary:
Instead of the non-sensical piecewise join we had until now write
a proper one. Hopefully the comments explain what it does. Main one:
```
(* high-level idea: maintain some union-find data structure to identify locations in one heap
with locations in the other heap. Build the initial join state as follows:
- equate all locations that correspond to identical variables in both stacks, eg joining
stacks {x=1} and {x=2} adds "1=2" to the unification.
- add all addresses reachable from stack variables to the join state heap
This gives us an abstract state that is the union of both abstract states, but more states
can still be made equal. For instance, if 1 points to 3 in the first heap and 2 points to 4
in the second heap and we deduced "1 = 2" from the stacks already (as in the example just
above) then we can deduce "3 = 4". Proceed in this fashion until no more equalities are
discovered, and return the abstract state where a canonical representative has been chosen
consistently for each equivalence class (this is what the union-find data structure gives
us). *)
```
Reviewed By: mbouaziz
Differential Revision: D10483978
fbshipit-source-id: f6ffd7528
Summary:
`Location` was clashing with the `Location` module, so use `Address`
instead.
When invalidating an address, remember the "actor" of its invalidation,
i.e. the access expression leading to the address and the source
location of the corresponding instruction.
When checking accesses, also pass the actor responsible for the access,
so that when we raise an error we know:
1. when and why a location was invalidated
2. when and why we tried to read it after that
Reviewed By: mbouaziz
Differential Revision: D10446282
fbshipit-source-id: 3ca4fb3d4
Summary:
Model `x[y]` and `x.push_back(i)` to catch the classic bug of "take
reference inside vector, invalidate, then use again".
Reviewed By: da319
Differential Revision: D10445824
fbshipit-source-id: 21ffd9677
Summary:
Do the intersection of the heap and stack domains, and the union of the
invalid location sets. This forgets invalid locations that appear only
in one heap, unfortunately. We can start with this and improve later.
Reviewed By: mbouaziz
Differential Revision: D10445825
fbshipit-source-id: cc24460af
Summary:
New analysis in foetal form to detect invalid use of C++ objects after their
lifetime has ended. For now it has:
- A domain consisting of a graph of abstract locations representing the heap, a map from program variables to abstract locations representing the stack, and a set of locations known to be invalid (their lifetime has ended)
- The heap graph is unfolded lazily when we resolve accesses to the heap down to an abstract location. When we traverse a memory location we check that it's not known to be invalid.
- A simple transfer function reads and updates the stack and heap in a rudimentary way for now
- C++ `delete` is modeled as adding the location that its argument resolves to to the set of invalid locations
- Also, the domain has a really crappy join and widening for now (see comments in the code)
With this we already pass most of the "use after delete" tests from the
Ownership checker. The ones we don't pass are only because we are missing
models.
Reviewed By: mbouaziz
Differential Revision: D10383249
fbshipit-source-id: f414664cb