Summary:
bigmacro_bender
There are 3 ways pulse tracks history. This is at least one too many. So
far, we have:
1. "histories": a humble list of "events" like "assigned here", "returned from call", ...
2. "interproc actions": a structured nesting of calls with a final "action", eg "f calls g calls h which does blah"
3. "traces", which combine one history with one interproc action
This diff gets rid of interproc actions and makes histories include
"nested" callee histories too. This allows pulse to track and display
how a value got assigned across function calls.
Traces are now more powerful and interleave histories and interproc
actions. This allows pulse to track how a value is fed into an action,
for instance performed in callee, which itself creates some more
(potentially now interprocedural) history before going to the next step
of the action (either another call or the action itself).
This gives much better traces, and some examples are added to showcase
this.
There are a lot of changes when applying summaries to keep track of
histories more accurately than was done before, but also a few
simplifications that give additional evidence that this is the right
concept.
Reviewed By: skcho
Differential Revision: D17908942
fbshipit-source-id: 3b62eaf78
Summary:
Java method annotations are ambiguous in that there is no difference between
annotating the return value of a method, and annotating the method itself.
The disambiguation is done entirely based on the meaning of the annotation.
Here, while `UiThread`/`MainThread` are genuine method/class annotations
and not return annotations, the reverse is true for `ForUiThread`/`ForNonUiThread`.
This means that these latter annotations do not determine the thread status of
the method they are attached to.
Here we fix that misunderstanding.
Reviewed By: jvillard
Differential Revision: D17960994
fbshipit-source-id: 5aecfb124
Summary: As per title. These test pass already because the previous thread domain was sufficient to express them. This won't necessarily be true when the whole-program analysis version comes around, because we may decide to not report on the `Threaded` elements (see domain).
Reviewed By: dulmarod
Differential Revision: D17930653
fbshipit-source-id: 2174f6b22
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:
Eventually thread status will be stored inside every critical pair so as to allow path sensitivity. That means that the status can no longer be a whole trace, as this will quickly become intractable, because each domain element would have to maintain its own trace as well as its own thread-status trace.
This is not great, as we lose information here, but I don't see any other way around it that is not super complicated/costly (sharing will be limited when moving from callee to caller).
Other diffs up the stack will clean up infrastructure no longer used meaningfully (ie models and domains).
Reviewed By: mityal
Differential Revision: D17908908
fbshipit-source-id: 3bf353e33
Summary:
Starvation is currently path insensitive. Two special cases of sensitivity cover a large range of useful cases:
- sensitivity on whether the current thread is a UI/background thread;
- sensitivity on whether a lock can be acquired (without blocking) or not.
We add a few tests capturing some of the false positives and negatives of the current analysis.
Reviewed By: mityal
Differential Revision: D17907492
fbshipit-source-id: fbce896ac
Summary:
This diff adopts an array length evaluation function that is conservative. It is useful when our
domain cannot express length result precisely.
For example, suppose there is an array pointer `arr_locs` that may point to two arrays `a` and `b`,
and their lengths are `a.length` and `b.length` (symbols), respectively. Using the usual
evaluation, our current domain cannot express `a.length join b.length` (join of two symbolic
values), so it returns top.
In this case, we can use the conservative function intead. It evaluates the length as `[0,
a.length.ub + b.length.ub]`, since we know every array length is positive. The result is not
precise, but better than top.
Reviewed By: ezgicicek
Differential Revision: D17908859
fbshipit-source-id: 7c0b1591b
Summary:
Instead of a string argument named `~str` pass `Formal | Global` and let
`add_to_errlog` figure out how to print it.
Reviewed By: ezgicicek
Differential Revision: D17907657
fbshipit-source-id: ed09aab72
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:
Let's add basic Java support to impurity checker. Since impurity checker relies on pulse, we need to add Java with Pulse callback as well. Pulse doesn't officially support Java yet, but we can enable it for impurity checker for now.
Many Java primitives/operations are not yet modeled (such as creation of new objects, support for collections etc.). Still, it is good to run impurity checker on the existing tests of the purity checker. Also, it is nice to see that we can identify most of the impure functions correctly in the purity dir. There are a lot of FNs though.
Reviewed By: skcho
Differential Revision: D17906237
fbshipit-source-id: 15308d285
Summary:
This diff introduces inequality for the iterator alias target, as we
did for the size target before.
Reviewed By: ezgicicek
Differential Revision: D17879208
fbshipit-source-id: cc2f6a723
Summary: If we have no pulse summary (most likely caused by pulse finding a legit issue with the code), let's consider the function as impure.
Reviewed By: jvillard
Differential Revision: D17906016
fbshipit-source-id: 671d3e0ba
Summary:
This diff revises the semantics of hasNext model to add the lengths of
arrays, rather than join them to top.
Reviewed By: ezgicicek
Differential Revision: D17882388
fbshipit-source-id: f5edaedb3
Summary:
[androidx.collection.SimpleArrayMap](https://developer.android.com/reference/androidx/collection/SimpleArrayMap.html) also has `keySet` and `entrySet` methods which make them eligible for inefficient keyset checker. Let's add it.
Title
Reviewed By: skcho
Differential Revision: D17831594
fbshipit-source-id: 32e831e18
Summary:
The current usage has several issues reducing code maintainability and
readability:
1. Null_field_access was misleading: it was used for checking accesing
to arrays as well!
2. But actually, when checking access to array via `length`, we sometimes
pretended it is a field access (hence very tricky code in rendering the
error).
3. "Call receiver consistency" is unclear name, was not obvious that it is all about
calling a method in an object.
Let's also consolidate code.
Reviewed By: artempyanykh
Differential Revision: D17789618
fbshipit-source-id: 9b0f58c9c
Summary: Before, we didn't track litho framework callees on client code which was wrong. Now, we replace this with the following: If the callee is `build()` itself or doesn't contain a `build()` in its summary, then we want to track it in the domain. The former makes sense since we always want to track `build()` methods. The latter also makes sense since such a method could be a setter for a prop (as in the case of `prop1` in `buildPropLithoOK` which we were missing before due to the imprecise heuristic that prevented picking up callees in litho).
Reviewed By: ngorogiannis
Differential Revision: D17810704
fbshipit-source-id: 87d88e921
Summary: As a heuristic, litho library calls on non-litho callers are not tracked. This is very imprecise and results in FPs and FNs as exemplified by newly added tests. Instead, we should check to see if the summary contains a `build()` method as will be done in the next diff. This diff adds these tests and refactors the test code.
Reviewed By: skcho
Differential Revision: D17809536
fbshipit-source-id: 6dff1868c
Summary:
Improve the trace by incorporating the callees and their locations in the call chain (i.e. chain of methods starting from `build()` call)
- extend the domain to contain the callee location
- replace the test results with the new traces
This makes our job much easier to debug FPs in a big codebase.
Reviewed By: skcho
Differential Revision: D17788996
fbshipit-source-id: 31938b5fe
Summary: `litho` checker contained two checkers: required-props and graphQL field accesses. Although they use the same domain, their reporting conditions and analysis details are different. However, they were bundled into the same analysis by adding disjunctions to `exec_instr` to handle both cases. Let's separate them into two different checkers, keeping a modular transfer function and analyzer that is reused by these two checkers.
Reviewed By: skcho
Differential Revision: D17788834
fbshipit-source-id: 47d77063b
Summary:
At some point it was thought that we can assume that any annotation starting with "On" means the method is on the UI thread.
That's too imprecise and has led to false positives and negatives. Restrict to a well-known safe set.
Reviewed By: ezgicicek
Differential Revision: D17769376
fbshipit-source-id: 0f8fee059
Summary:
This diff tries to narrowing the fixpoint of outermost loops, so that over-approximated widened values do not flow to the following code.
Problem: There are two phases for finding a fixpoint, widening and narrowing. First, it finds a fixpoint with widening, in function level. After that, it finds a fixpoint with narrowing. A problem is that sometimes an overly-approximated, imprecise, values by widening are flowed to the following loops. They are hard to narrow in the narrowing phase because there is a cycle preventing it.
To mitigate the problem, it tries to do narrowing, in loop level, right after it found a fixpoint of a loop. Thus, it narrows before the widened values are flowed to the following loops. In order to guarantee the termination of the analysis, this eager narrowing is applied only to the outermost loops.
Reviewed By: ezgicicek
Differential Revision: D17740265
fbshipit-source-id: e2d454036
Summary:
This diff extends the alias domain to analyze loop with list comprehensions form in Java precisely.
```
list2 = new List();
for (Element e : list1) {
list2.add(e);
}
```
1. `IteratorOffset` is a relation between a iterator offset and a length of another array. For example, in the above example, after n-times of iterations, the offset of the iterator (if it exists) and the length of `list2` are the same as `n`.
2. `IteratorHasNext` is a relation between iterator and its `hasNext` result.
3. At the conditional nodes, it prunes the alias list length of `list2` by that of `list1`.
* if `hasNext(list1's iterator)` is true, `list2`'s length is pruned by `< list1's length`
* if `hasNext(list1's iterator)` is false, `list2`'s length is pruned by `= list1's length`
Reviewed By: ezgicicek
Differential Revision: D17667128
fbshipit-source-id: 41fb23a45
Summary:
The old domain keeps two sets:
- `events` are things (including lock acquisitions) which eventually happen during the execution of a procedure.
- `order` are pairs of `(lock, event)` such that there is a trace through the procedure which at some point acquires `lock` and before releasing it performs `event`.
A deadlock would be reported if for two procedures, `(lock1,lock2)` is in `order` of procedure 1 and `(lock2,lock1)` is in `order` of procedure 2. This condition/domain allowed for the false positive fixed in the tests, as well as was unwieldy, because it required translating between the two sets.
The new domain has only one set of "critical pairs" `(locks, event)` such that there is a trace where `event` occurs, and *right before it occurs* the locks held are exactly `locks` (no over/under approximation). This allows keeping all information in one set, simplifies the procedure call handling and eliminates the known false positive.
Reviewed By: mityal
Differential Revision: D17686944
fbshipit-source-id: 3c68bb957
Summary: Holding a master lock and then acquiring two other locks inside can generate a false positive as shown.
Reviewed By: mityal
Differential Revision: D17710076
fbshipit-source-id: 5bc910ba2
Summary:
Previously deduplication was always on which is not great for testing.
Also split tests so that we can still test deduplication separately.
Reviewed By: mityal
Differential Revision: D17686877
fbshipit-source-id: 280d91473
Summary:
Ideally the analyser should equate locks `this.x.f` and `a.x.f` in different methods if they can alias.
The heuristic removed here was rarely used and is in the way of a re-write of the analysis.
It was also badly implemented, as this should ideally be the comparison relation of `Lock`.
Reviewed By: mityal
Differential Revision: D17602827
fbshipit-source-id: 4f4576c1a
Summary:
This diff generates a symbolic value when a function returns only
exceptions. Previously, the exception expression is evaluated to top,
thus it was propagated to other functions, which made those costs as
top. For preventing that situation, this diff changed:
* exception expressions are evaluated to bottom, and
* if callee's return value is bottom, it generates a symbolic value
for it.
Reviewed By: ezgicicek
Differential Revision: D17500386
fbshipit-source-id: 0fdcc710d
Summary: This diff introduces an inequality for the size alias targets, in order to get preciser array lengths after loops. The alias domain in inferbo was able to express strict equality between alias source and its targets, e.g. x=size(array). Now, for the size alias target, it can express less than or equal relations, e.g. x>=size(array).
Reviewed By: ezgicicek
Differential Revision: D17606222
fbshipit-source-id: 2557d3bd0
Summary: Component.Builder has its own non-required props that are inherited by the MyComponent.Builder. Add tests where these common props are set in the chain of calls.
Reviewed By: Katalune
Differential Revision: D17710294
fbshipit-source-id: f3c5ef28c
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: `Prop(varArg = myProp) List <?> myPropList` can also be set via `myPropList()` or `myProp()`. Add support for picking up the `varArg` and checking this form of required props.
Reviewed By: ngorogiannis
Differential Revision: D17571997
fbshipit-source-id: 7956cb972
Summary:
Turns out, we did not have such a test in place.
Known issue: we report over-annotated warnings for each fields N times,
one per constructor, which is wrong.
Reviewed By: artempyanykh
Differential Revision: D17574791
fbshipit-source-id: def992691
Summary:
This diff avoids giving the top value to unknown globals in Java,
because they harm precision of the cost checker. Instead, it doesn't
subst the global symbols at function calls.
Reviewed By: ezgicicek
Differential Revision: D17498714
fbshipit-source-id: d1215b3aa
Summary:
This diff adds an eval mode for the substitutions of the cost results, in order to avoid precision
loss by joining two symbols.
The usual join of two different symbolic values, `s1` and `s2`, becomes top due to the limitation of
our domain. On the other hand, in the new eval mode, it returns an upperbound `s1+s2`, because the
cost values only care about the upperbounds.
Reviewed By: ezgicicek
Differential Revision: D17573400
fbshipit-source-id: 2c84743d5
Summary:
This was causing a crash, because when trying to create a procname from a block at that point we don't have the block return type, which is needed for the name. I don't understand why BlockDecl doesn't contain the type, but I looked again and it doesn't (also in clang). So in general we need to pass it from the context, but that's not possible in this case.
Also, one could argue that such a block is not a method from the struct, since it's just a block that is assigned to a field as initialization.
Reviewed By: skcho
Differential Revision: D17575197
fbshipit-source-id: 3974ead3f
Summary:
This continues work for eliminating Annot.Item.t from Nullsafe low-level
code.
The introduced function `from_nullsafe_type` is called when we infer
initial type of the equation based on the function or field formal signature.
Before that, we did it via reading the annotation directly, which
complicates the logic and making introducing Unknown nullability tricky.
## Clarifying the semantics of PropagatesNullable
This diff also clarifies (and changes) the behavior of PropagatesNullable params.
Previously, if the return value of a function that has PropagatesNullable params was
annotated as Nullable, nullsafe was effectively ignoring PropagatesNullable effect.
This is especially bad because one can add Nullable annotation based on the logic "if the function can return `null`, it should be annotated with Nullable`.
In the new design, there is no possibility for such a misuse: the code that
applies the rule "any param is PropagatesNullable hence the return
value is nullable even if not explicitly annotated" lives in NullsafeType.ml, so
this will be automatically taken into account.
Meaning that now we implicitly deduce Nullable annotation for the return value, and providing it explicitly as an alternative that does not change the effect.
In the future, we might consider annotating the return value with `Nullable` explicit.
Reviewed By: jvillard
Differential Revision: D17479157
fbshipit-source-id: 66c2c8777
Summary:
In the cost checker, the range of selected control variables are used to estimate the number of loop iteration. However, sometimes the ranges of control variables are not related to how many times the loop iteration. This diff strengthens the condition for them as:
1. integers from `size` models
2. integers constructed from `+` or `-`
3. integers constructed from `*`
For the last one, the loop iteration is likely to be log scale of the range of the control variable:
```
while (i < c) {
i *= 2;
}
```
We will address this in the future.
Reviewed By: ezgicicek
Differential Revision: D17365796
fbshipit-source-id: c1e709ae8
Summary: Our annotation parameter parsing is too primitive to identify `resType` and before we only assumed that all Prop's can be set by any of the two suffixes: `Attr` and `Res`. After talking to Litho team, there is 3 more additions to these suffixes: `Dip`, `Sip`, and `Px`.
Reviewed By: ngorogiannis
Differential Revision: D17528482
fbshipit-source-id: 8d7f49130
Summary: Before, we were mistakenly checking any annotation that ends with Prop such as TreeProp. This was wrong. Instead, we should only check Prop as adviced by the Litho team.
Reviewed By: ngorogiannis
Differential Revision: D17527769
fbshipit-source-id: b753dd87a
Summary:
Introduce a new experimental checker (`--impurity`) that detects
impurity information, tracking which parameters and global variables
of a function are modified. The checker relies on Pulse to detect how
the state changes: it traverses the pre and post pairs starting from
the parameter/global variable and finds where the pre and post heaps
diverge. At diversion points, we expect to see WrittenTo/Invalid attributes
containing a trace of how the address was modified. We use these to
construct the trace of impurity.
This checker is a complement to the purity checker that exists mainly
for Java (and used for cost and loop-hoisting analyses). The aim of
this new experimental checker is to rely on Pulse's precise
memory treatment and come up with a more precise im(purity)
analysis. To distinguish the two checkers, we introduce a new issue
type `IMPURE_FUNCTION` that reports when a function is impure, rather
than when it is pure (as in the purity checker).
TODO:
- improve the analysis to rely on impurity information of external
library calls. Currently, all library calls are assumed to be nops,
hence pure.
- de-entangle Pulse reporting from analysis.
Reviewed By: skcho
Differential Revision: D17051567
fbshipit-source-id: 5e10afb4f
Summary:
`ModeledRange` represents how many times the interval value can be updated by modeled functions. This
domain is to support the case where there are mismatches between value of a control variable and
actual number of loop iterations. For example,
```
while((c = file_channel.read(buf)) != -1) { ... }
```
the loop will iterates as the file size, but the control variable `c` does not have that value. In
these cases, it assigns a symbolic value of the file size to the modeled range of `c`, then which
is used when calculating the overall cost.
Reviewed By: jvillard
Differential Revision: D17476621
fbshipit-source-id: 9a81376e8
Jules Villard
Nikos Gorogiannis
Sungkeun Cho
Ezgi Çiçek
Mitya Lyubarskiy
Dulma Churchill