At work we have been trying to figure out what this code does, and we feel like it could cause problems. This is a hashmap implementation in C.
static thrift_map_node_t *thrift_map_newnode (thrift_map_base_t *m, union keyDataTypes key, void * value, int vsize)
thrift_map_node_t *node;
int ksize = strlen(key) + 1;
int voffset = ksize + ((sizeof(void*) - ksize) % sizeof(void*));
node = malloc(sizeof(*node) + voffset + vsize);
if (!node) return NULL;
memcpy(node + 1, key, ksize);
node->hash = thrift_hash(key);
node->value = ((char*) (node + 1)) + voffset;
memcpy(node->value, value, vsize);
return node;
The problem we think is here
int voffset = ksize + ((sizeof(void*) - ksize) % sizeof(void*));
We think this could produce a negative value, eg -9 % 4 = -1 in C.
a work around or any help to what this is attempting to do would be nice.
No, it cannot ever produce a negative value if size_t
has rank higher than or equal to int
(and it does except perhaps on some really oddish 16-bit addressing system with 32-bit integers), because then
sizeof (void*) - ksize
will be of type size_t
that is an unsigned type and always positive. Hence if void pointer is 8 bytes wide and ksize is 17, the result of subtraction is (size_t)-9
, or 18446744073709551607 on a 64-bit machine. That modulo 8 is 7, which added to 17 will result in 24, which is a multiple of sizeof (void *)
.
However, this relies on the sizeof (void *)
being a power of 2 , hence it would still be better if you used
sizeof (void*) - (ksize + sizeof (void*)) % sizeof (void *)
for the padding as this avoids the negative numbers altogether
It is true though that would the result subtraction be signed , since C99 the modulo would be always negative: (-3) % 4
is -3
for example.
BTW, the code should not use sizeof (void *)
, but _Alignof (void *)
.
There is still one more dubious thing in the code. Suppose someone can actually feed in a string that is over 2 GiB in length - then a signed overflow occurs on majority of platforms and that is not going to be pretty!
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.