Re: [PATCH] -fsanitize=vptr instrumentation (take 2)

[Jason Merrill]

On 12/03/2014 07:01 AM, Jakub Jelinek wrote:
One more question.  My current version of the patch adds one ubsan vptr
instrumentation in each of the following functions:
struct S { int s; virtual void foo (); S(); virtual ~S(); };
struct T : S {};
struct U { int u; S s[4]; };
struct V { U v; virtual void bar (); V(); virtual ~V(); };
V v;

int
f1 (V *p)
{
   return p->v.u;
}

int
f2 (V *p)
{
   return p->v.s[2].s;
}

int
f3 ()
{
   return v.v.u;
}

int
f4 ()
{
   return v.v.s[2].s;
}

(in f1 and f3 verifies it for _ZTI1V, in f2 and f4 verifies it for
_ZTI1S).  Should I change it so that we get 2 instrumentations in f2 and f4
and one in f1/f3 (i.e. in f2/f4 check two vptrs, one _ZTI1S and one _ZTI1V),
or do we care just about the outermost one?

I think so; it definitely is possible to clobber the vptr in a normal 
decl and the check would catch that.  Though perhaps we want a separate 
mode that optimizes away the checks when we know the declared type of 
the object.

Note, latest clang has 1 instrumentation in f1 and f4, two in f2 and none
in f3.

That sounds like they mean to avoid checking when there's a declared 
type but array-to-pointer decay is causing them to check the array members.

Jason