libsepol: avoid unsigned integer overflow

Unsigned integer overflow is well-defined and not undefined behavior.
But it is still useful to enable undefined behavior sanitizer checks on
unsigned arithmetic to detect possible issues on counters or variables
with similar purpose.

Use a spaceship operator like comparison instead of subtraction.

Modern compilers will generate a single comparison instruction instead
of actually perform the subtraction.

policydb.c:826:17: runtime error: unsigned integer overflow: 24 - 1699 cannot be represented in type 'unsigned int'

Signed-off-by: Christian Göttsche <cgzones@googlemail.com>
Acked-by: James Carter <jwcart2@gmail.com>
This commit is contained in:
Christian Göttsche 2021-06-08 17:58:55 +02:00 committed by James Carter
parent 42f3d7cceb
commit 1537ea8412

View file

@ -817,11 +817,11 @@ static int filenametr_cmp(hashtab_t h __attribute__ ((unused)),
const filename_trans_key_t *ft2 = (const filename_trans_key_t *)k2;
int v;
v = ft1->ttype - ft2->ttype;
v = (ft1->ttype > ft2->ttype) - (ft1->ttype < ft2->ttype);
if (v)
return v;
v = ft1->tclass - ft2->tclass;
v = (ft1->tclass > ft2->tclass) - (ft1->tclass < ft2->tclass);
if (v)
return v;