Commit Graph

159 Commits

Author SHA1 Message Date
Ted Kremenek c3661decc3 Change ExplodedNode to have its NodeGroups all BumpPtrAllocated, avoiding malloc() traffic when adding successors/predecessors to a node. This was done by introducing BumpVector, which is essentially SmallVector with all memory being BumpPtrAllocated (this can certainly be cleaned up or moved into llvm/ADT).
This change yields a 1.8% speed increase when running the analyzer (with -analyzer-store=region) on a small benchmark file.

llvm-svn: 83439
2009-10-07 00:42:52 +00:00
Ted Kremenek 0ccd06c017 Re-introduce diagnostic caching in BugReporter that was originally added in
r82198 and then reverted. This is an intermediate solution, as diagnostic
caching should not rely on static variables.

llvm-svn: 82301
2009-09-18 22:37:37 +00:00
Ted Kremenek e3d209ff77 Revert most of r82198, which was causing a large number of crashes
when running the analyzer on real projects.  We'll keep the change to
AnalysisManager.cpp in r82198 so that -fobjc-gc analyzes code
correctly in both GC and non-GC modes, although this may emit two
diagnostics for each bug in some cases (a better solution will come
later).

llvm-svn: 82201
2009-09-18 07:31:15 +00:00
Ted Kremenek 82f7f9c080 Introduce caching of diagnostics in BugReporter. This provides extra
pruning of diagnostics that may be emitted multiple times.  This is
accomplished by adding FoldingSet profiling support to PathDiagnostic,
and then having BugReporter record what diagnostics have been issued.

This was motived to a serious bug introduced by moving the
'divide-by-zero' checking outside of GRExprEngine into a separate
'Checker' class.  When analyzing code using the '-fobjc-gc' option, a
given function would be analyzed twice, but the second time various
"internal checks" would be disabled to avoid emitting multiple
diagnostics (e.g., "null dereference") for the same issue.  The
problem is that such checks also effect path pruning and don't just
emit diagnostics.  This resulted in an assertion failure involving a
real divide-by-zero in some analyzed code where we would get an
assertion failure in APInt because the 'DivZero' check was disabled
and didn't prune the logic that resulted in the divide-by-zero in the
analyzer.

The implemented solution is somewhat of a hack, and may not perform
extremely well.  This will need to be cleaned up over time.

As a regression test, 'misc-ps.m' has been modified so that its tests
are run using -fobjc-gc to test this diagnostic pruning behavior.

llvm-svn: 82198
2009-09-18 05:37:41 +00:00
Ted Kremenek 3413f277d0 Fix typo in comment.
llvm-svn: 81825
2009-09-15 03:28:00 +00:00
Ted Kremenek 4112969df1 Fix: <rdar://problem/5905851> do not report a leak when post-dominated by a call
to a noreturn or panic function

llvm-svn: 81803
2009-09-14 22:01:32 +00:00
Ted Kremenek 37a118520c Implement FIXME: free up BugReportEquivClass objects when deleting BugTypes.
llvm-svn: 81783
2009-09-14 20:40:59 +00:00
Ted Kremenek 7020eae076 Introduce "DefinedOrUnknownSVal" into the SVal class hierarchy, providing a way
to statically type various methods in SValuator/GRState as required either a
defined value or a defined-but-possibly-unknown value. This leads to various
logic cleanups in GRExprEngine, and lets the compiler enforce via type checking
our assumptions about what symbolic values are possibly undefined and what are
not.

Along the way, clean up some of the static analyzer diagnostics regarding the uses of uninitialized values.

llvm-svn: 81579
2009-09-11 22:07:28 +00:00
Zhongxing Xu 7e3431b5cf Make AnalysisManager stateless. Now other analyzer components only depends on
local node information.

llvm-svn: 81433
2009-09-10 05:44:00 +00:00
Mike Stump 11289f4280 Remove tabs, and whitespace cleanups.
llvm-svn: 81346
2009-09-09 15:08:12 +00:00
Zhongxing Xu d7674a47f9 Do not call FlushReports() in GRBugReporter's dtor. We already call it in
ActionGRExprEngine().

llvm-svn: 81064
2009-09-05 06:06:49 +00:00
Zhongxing Xu 342950efc5 Remove CodeDecl and CFG from GRExprEngine and GRStateManager.
Now AnalysisManager is the only place we can get CodeDecl.
This leads to an API change: GRState::bindExpr() now takes the CFG argument.

llvm-svn: 79980
2009-08-25 06:51:30 +00:00
Zhongxing Xu 94ec649b33 Remove Decl and CFG from ExplodedGraph. This leads to a series small changes.
llvm-svn: 79973
2009-08-25 03:33:41 +00:00
Zhongxing Xu 068928b2a6 Rename: ProgramPoint::getContext() => ProgramPoint::getLocationContext().
llvm-svn: 79502
2009-08-20 01:23:34 +00:00
Zhongxing Xu 080fa0522b Get the Decl from the current ExplodedNode. Eventually the diagnostic client
and other core analysis logic will be untied to a particular Decl.

llvm-svn: 79420
2009-08-19 12:50:00 +00:00
Zhongxing Xu 7864b9ea0c Remove unused parameter BugReporter due to previous patch.
llvm-svn: 79328
2009-08-18 08:58:41 +00:00
Zhongxing Xu 4c76dbc758 Now we can get the CFG from the ProgramPoint. No need to pass in the
BugReporter.

llvm-svn: 79327
2009-08-18 08:46:04 +00:00
Ted Kremenek 9b2c79de59 Enhance static analyzer diagnostics by introducing a new 'EnhancedBugReporter'
which allows custom checks to register callback creator functions for creating
BugReporterVisitor objects. This allows various checks to include diagnostics
such as 'assuming value is null' with little extra work. Eventually this API
should be refactored to be cleaner and more simple.

llvm-svn: 79302
2009-08-18 01:05:30 +00:00
Zhongxing Xu c90a0c2e81 Core analysis engine template cleanup step 2:
merge ExplodedGraphImpl and ExplodedGraph.

llvm-svn: 78291
2009-08-06 06:28:40 +00:00
Zhongxing Xu 20227f71d7 As GRState seems general enough, it is time to merge some template classes
and their impl base classes. This can greatly simply some code of the core 
analysis engine. This patch merges ExplodedNodeImpl into ExplodedNode.

llvm-svn: 78270
2009-08-06 01:32:16 +00:00
Zhongxing Xu c2394f0ad9 add a fixme
llvm-svn: 77447
2009-07-29 08:13:37 +00:00
Ted Kremenek 4c9f161fd5 Fix helper function GetNextStmt() to look for the first statement that has a
valid SourceLocation.

llvm-svn: 77280
2009-07-28 00:07:15 +00:00
Ted Kremenek bfd28fd596 Refactor 'PostStmt' and 'PreStmt' to subclass a common parent 'StmtPoint'.
Educate GRExprEngine::VisitGraph() about 'PreStmt'.

Mark the constructor of 'PostStmt' to be explicit, preventing implicit
conversions and the selection of the wrong 'generateNode' method in
GRStmtNodeBuilder.

Constify a bunch of arguments, which falls out of the changes to ProgramPoint.

llvm-svn: 76809
2009-07-22 22:35:28 +00:00
Ted Kremenek 6796fbd2e8 Move the source-level CFG from libAST to libAnalysis.
llvm-svn: 76092
2009-07-16 18:13:04 +00:00
Argyrios Kyrtzidis ddcd132a5b Remove the ASTContext parameter from the getBody() methods of Decl and subclasses.
Timings showed no significant difference before and after the commit.

llvm-svn: 74504
2009-06-30 02:35:26 +00:00
Ted Kremenek 1642bdaaa5 Introduce a new concept to the static analyzer: SValuator.
GRTransferFuncs had the conflated role of both constructing SVals (symbolic
expressions) as well as handling checker-specific logic. Now SValuator has the
role of constructing SVals from expressions and GRTransferFuncs just handles
checker-specific logic. The motivation is by separating these two concepts we
will be able to much more easily create richer constraint-generating logic
without coupling it to the main checker transfer function logic.

We now have one implementation of SValuator: SimpleSValuator.

SimpleSValuator is essentially the SVal-related logic that was in GRSimpleVals
(which is removed in this patch). This includes the logic for EvalBinOp,
EvalCast, etc. Because SValuator has a narrower role than the old
GRTransferFuncs, the interfaces are much simpler, and so is the implementation
of SimpleSValuator compared to GRSimpleVals. I also did a line-by-line review of
SVal-related logic in GRSimpleVals and cleaned it up while moving it over to
SimpleSValuator.

As a consequence of removing GRSimpleVals, there is no longer a
'-checker-simple' option. The '-checker-cfref' did everything that option did
but also ran the retain/release checker. Of course a user may not always wish to
run the retain/release checker, nor do we wish core analysis logic buried in the
checker-specific logic. The next step is to refactor the logic in CFRefCount.cpp
to separate out these pieces into the core analysis engine.

llvm-svn: 74229
2009-06-26 00:05:51 +00:00
Ted Kremenek ae01dc73af API cleanup: move more methods from GRStateManager to GRState.
llvm-svn: 73968
2009-06-23 17:55:07 +00:00
Ted Kremenek 095f1a9848 Move clients over from using GRStateManager::BindXXX and friends to
GRState->bindXXX and friends (and constify some arguments along the way).

llvm-svn: 73740
2009-06-18 23:58:37 +00:00
Ted Kremenek a28565ac62 Fix another bug in BugReporter where we wouldn't always select the bug report in a bug equivalence class with the shortest path.
llvm-svn: 71920
2009-05-16 01:11:58 +00:00
Ted Kremenek 73777059f0 BugReporter (extensive diagnostics): Do not include the range of target '}'
llvm-svn: 71832
2009-05-15 02:46:13 +00:00
Ted Kremenek cfe7d02b2c BugReporter (extensive diagnostics): Add control-flow piece to '}' in
loop body when generating 'Looping back to the head of the loop'
diagnostics.

llvm-svn: 71829
2009-05-15 01:50:15 +00:00
Ted Kremenek 5fb7847fbf BugReport::getEndPath() - Only add a Stmt's range to the constructed PathDiagnosticEventPiece if the BugReport contained no explicit ranges.
llvm-svn: 71516
2009-05-11 23:50:59 +00:00
Ted Kremenek 2ee73b86c0 EdgeBuilder: DeclStmts and BinaryOperators are not the enclosing location context when they are used as initialization code for loops.
llvm-svn: 71480
2009-05-11 22:19:32 +00:00
Ted Kremenek c0b879b4a7 EdgeBuilder::cleanUpLocation() should used the PathDiagnosticLocation constructor for a single point, not a range.
llvm-svn: 71477
2009-05-11 21:42:34 +00:00
Ted Kremenek c14b594b14 BugReporter (extensive diagnostics): Add EdgeBuilder::cleanUpLocation for canonicalization locations and use this in both popLocation and rawAddEdge.
llvm-svn: 71470
2009-05-11 19:50:47 +00:00
Ted Kremenek 3978f7972d analyzer:
- Improve -autorelease diagnostics.
- Improve VLA diagnostics.
- Use "short description" for bug when outputting to TextDiagnostics

llvm-svn: 71383
2009-05-10 05:11:21 +00:00
Ted Kremenek 9abe474852 Add preliminary support for enhancing null-pointer dereference diagnostics.
llvm-svn: 71135
2009-05-07 00:45:33 +00:00
Ted Kremenek bb8d546208 Refactor BugReporter interface to have a new 'BugReporterContext' and
'BugReporterVisitor'. This simplifies callbacks from BugReporter to BugReports
(via VisitNode). It also lays the foundation for arbitrary visitor "call backs"
that can be registered to a BugReporterContext as a PathDiagnostic is
constructed. These call backs can help operate as separate "experts" that can
work on constructed pieces of a PathDiagnostic for which they possess special
knowledge.

llvm-svn: 71121
2009-05-06 21:39:49 +00:00
Ted Kremenek a8106efce9 BugReporter (extensive diagnostics): improve location context generation for the
start of 'do' and '@synchronized' statements.

llvm-svn: 71038
2009-05-05 23:13:38 +00:00
Ted Kremenek 98017051ae BugReporter (extensive diagnostics): Fix getEnclosingStmtLocation to reason
about Exprs that are not consumed and fix where the loop iteration diagnostic
goes.

llvm-svn: 71027
2009-05-05 22:19:17 +00:00
Ted Kremenek e396f72ec7 BugReporter (extensive diagnostics): don't mark location contexts that are
control-flow expressions as dead.

llvm-svn: 70887
2009-05-04 18:15:17 +00:00
Ted Kremenek afbff15e64 BugReporter (extensive diagnostics): introduce the notion of a "dead"
location context.  This allows us to postpone the decision of whether
or not a context should add a control-flow piece to the diagnostics
when inspecting its subexpressions.

llvm-svn: 70545
2009-05-01 16:08:09 +00:00
Ted Kremenek 29d0880e50 Remove #if 0'ed code.
llvm-svn: 70542
2009-05-01 15:19:33 +00:00
Ted Kremenek 9f08b842e1 Fix null dereference.
llvm-svn: 70417
2009-04-29 22:05:03 +00:00
Ted Kremenek 47187c6ad2 BugReporter/PathDiagnostics:
- Add an (optional) short description for BugReports for clients that want
  to distinguish between long and short descriptions for bugs
- Make the bug report for VLA less obscene for Plist diagnostics by using
  the short description

llvm-svn: 70415
2009-04-29 21:58:13 +00:00
Ted Kremenek 9bea0dc178 Extensive diagnostics: Do not add a location context for do...while statements.
llvm-svn: 70286
2009-04-28 04:28:12 +00:00
Ted Kremenek 61114bc558 BugReporter (extensive diagnostics): Clean up do...while control-flow edges, and
add "Looping back to the head of the loop" diagnostic for loops.

llvm-svn: 70285
2009-04-28 04:23:15 +00:00
Sebastian Redl a7b98a772c Implement function-try-blocks. However, there's a very subtle bug that I can't track down.
llvm-svn: 70155
2009-04-26 20:35:05 +00:00
Eli Friedman 5cdb8cc267 Minor code cleanup.
llvm-svn: 70144
2009-04-26 19:04:51 +00:00
Chris Lattner f0b64d73a8 split ObjC and C++ Statements out into their own headers.
llvm-svn: 70105
2009-04-26 01:32:48 +00:00