• According to NoQ's suggestion, use assumeZero() instead of isZeroConstant() to determine whether the value is 0.
• Add test memset26_upper_UCHAR_MAX() and memset27_symbol()
• Since void *memset( void *dest, int ch, size_t count ); will converts the value ch to unsigned char, we call evalCast() accordingly.

ping.

• Since there is no perfect way to handle the default binding of non-zero character, remove the default binding of non-zero character. Use bindDefaulrZero() instead of overwriteRegion() to bind the zero character.
• Reuse assume() instead of isZeroConstant() to determine whether it is zero character. The purpose of this is to be able to set the string length when dealing with non-zero symbol character.

In D44934#1088771, @NoQ wrote:

Hmm, ok, it seems that i've just changed the API in D46368, and i should have thought about this use case. Well, at least i have some background understanding of these problems now. Sorry for not keeping eye on this problem.

In D44934#1051002, @NoQ wrote:

Why do you need separate code for null and non-null character? The function's semantics doesn't seem to care.

I guess i can answer myself here:

int32_t x;
memset(&x, 1, sizeof(int32_t));
clang_analyzer_eval(x == 0x1010101); // should be TRUE

I really doubt that we support this case.

So, yeah, zero character is indeed special.

Thank you, Artem! I did not consider this common situation. This patch does not really support this situation, in this patch the value of x will be 1, it's not correct!

• fix typos
• code refactoring, add auxiliary method memsetAux()
• according to a.sidorin's suggestions, remove the useless state splitting.
• make StoreManager::overwriteRegion() pure virtual

Sorry for the long delay, I have just finished my holiday.

ping^2

Since BugReport::addVisitor() has checks for the null Visitor, remove the checks before BugReport->addVisitor().

In D45682#1074407, @george.karpenkov wrote:

I'm new to the taint visitor, but I am quite confused by your change description.

and many checkers rely on it

How can other checkers rely on it if it's private to the taint checker?

Thanks for your review, george! TaintBugVisitor is an utility to add extra information to illustrate where the taint information originated from. There are several checkers use taint information, e.g. ArrayBoundCheckerV2.cpp, in some cases it will report a warning, like warning: Out of bound memory access (index is tainted). If TaintBugVisitor moves to BugReporterVisitors.h, ArrayBoundCheckerV2 can add extra notes like Taint originated here to the report by adding TaintBugVisitor.

Test files for initialization missing? : )

In D45532#1069683, @whisperity wrote:

There is something that came up in my mind:

Consider a construct like this:

class A
{
  A()
  {
    memset(X, 0, 10 * sizeof(int));
  }

  int X[10];
};

I think it's worthy for a test case if this X is considered unitialised at the end of the constructor or not. (And if not, a ticket, or a fix for SA in a subpatch.)

In D45491#1067852, @NoQ wrote:

Yeah, i think this makes sense, thanks! It feels a bit weird that we have to add it as an exception - i wonder if there are other exceptions that we need to make. Widening over the stack memory space should be a whitelist, not a blacklist, because we can easily enumerate all stack variables and see which of them can be modified at all from the loop. But until we have that, this looks like a reasonable workaround.

• Move the CXXThisRegion's check to LoopWidening.cpp
• Use isa<CXXThisRegion>(R) instead of CXXThisRegion::classof(R).

In D45491#1063364, @george.karpenkov wrote:

@MTC what happens for

this.j = 0;
for (int i=0; i<100; i++)
   this.j++;

?

Kindly ping!

Thank you for your reminding, I overlooked this point. However for non-concrete character, the symbol value, if we just invalidate the region, the constraint information of the non-concrete character will be lost. Do we need to consider this?

Fix typo, unsinged -> unsigned

According to @NoQ's suggestion, remove the duplicated code.

Thanks for your review, NoQ!

Thank you for taking the time to pay attention to this problem, @NoQ. The reason for the test regression is that CheckBufferAccess() does not guarantee that CheckNonNull() must be called for the second buffer, see https://github.com/llvm-mirror/clang/blob/master/lib/StaticAnalyzer/Checkers/CStringChecker.cpp#L385.

Add the comments as suggested by @szepet .

In D44557#1042357, @NoQ wrote:

Sorry, one moment, i'm seeing a few regressions after the previous refactoring but i didn't look at them closely yet to provide a reproducer. I'll get back to this.

Just in case: we indeed do not guarantee that SymbolConjured corresponds to a statement; it is, however, not intended, but rather a bug.

Thank you for your explanation and the reasonable example, NoQ.

One small nit for future debugging people: Could you insert a comment line in the test case where you explain what is this all about? E.g what you just have written in the description: "invalidateRegions() will construct the SymbolConjured with null Stmt" or something like this.

Remove the default configuration -analyzer-store=region in the test file.

@NoQ, Very sorry, I've forgotten about this patch, it has now been updated.

Update the taint-generic.c to test both stdin declaration variants.

Thank you for your review, @NoQ!

• If the operand of the ++ operator is of type _Bool, also set to true.
• Add test file _Bool-increment-decement.c.

In D43741#1024949, @alexshap wrote:

i see, but just in case - what about the decrement operator ?

@NoQ Sorry to bother you again. It seems that this patch is useless to analyzer temporarily, if you think so, I will abandon it : ).

rebase

In D42785#995211, @NoQ wrote:

Ew. So it means that checker transitions are currently discarded. Great catch. I guess we don't use this functionality yet, so we can't test it, but the fix should definitely go in.

You are right, that's why I don't know how to add test for this change.

In D42300#982187, @NoQ wrote:

My intuition suggests that this checker shouldn't be path-sensitive; our path-sensitive analysis does very little to help you with this particular checker, and you might end up with a much easier and more reliable checker if you turn it into a simple AST visitor or an AST matcher. Just a heads up.

• Use C++11 range-based for loop to traverse ExplodedNodeSet.
• Define the macro offsetof in system-header-simulator.h.

In D37189#979795, @NoQ wrote:

Oh well, i guess i covered this in my recent patches anyway (esp. r322787/D41406). Sorry, i just fixed everything differently and it became unclear how to integrate your patch into the whole thing.

Thank you for your constant attention to this problem, Artem. I've updated the diff. As you said, this is a complex problem and look forward to your work on this issue.

Use 'return State' instead of 'return nullptr'.

Hi dcoughlin,

Update the llvm_unreachable's description of the BlockEntrance-branch from "Unexpected ProgramPoint" to "Unexpected CFG element at front of block".

1.Use the getAs<> in the if condition.
2.Add an "Unexpected ProgramPoint" assertion to make this patch more complete.

Update diff, use the SourceLocation of the first element of the entered block as the argument of PathDiagnosticLocation.

ping?

ping?

In D39159#910466, @NoQ wrote:

Maybe we could also test both declaration variants, i.e.:

I do not have commit access and hope someone can commit it on my behalf. Thanks a lot!

Split the long "expected" line into multiple lines.

The message about invalidate variable values is temporarily not printed. This work can be done with separate patch.

One of the possible improvements for future work here would be to actually bind the second argument value to the buffer instead of just invalidating it. Like, after memset(buf, 0, sizeof(buf)) the analyzer should know that all values in the buf array are 0. In the analyzer we have the notion of *default bindings* to handle that (see documentation in docs/analyzer/RegionStore.txt for more details).

1. taint-tester.c has some tests about stdin, so I only modified the parts of stdin in Inputs/system-header-simulator.h.
2. The C standard does not specify the implementation of FILE, so I continue to use typedef _FILE FILEto define the FILE type.

(1) Modify the description of the bug report
(2) Update loop-widening-notes.c
(3) PathDiagnosticLocation::create() - Use the location of TerminatorCondition.

Hi peter,

Add the test file and modify the description in the analyzer output.

Update the 'assert' condition and the code comment.

In D37189#854062, @NoQ wrote:

I believe a custom operator new (or new[], regardless) can always return anything it wants. It can return a pointer to a concrete global variable, for example. So the only thing we can assert is that our operator is custom.

In D36708#854060, @NoQ wrote:

We can probably land D37189 first, and then land this patch with a test?

Sorry, remove irrelevant code updates.

In D36708#853442, @zaks.anna wrote:

Hi MTC,

Could you please split out the fix (b) into it's separate patch? We would like to have the patches be as incremental as possible.

Thanks for working on this!

Because the way I submit the diff is not reasonable, the same diff corresponds to multiple purposes. This diff is primarily to address Bug34144.