Bound every array a data file can index

Closes Defects items 16 and 17 and Known-and-still-open item 6. All three let
an asset file, or a caller's argument, write past a fixed array.

akgl_sprite_load_json took its frame count straight from the document and wrote
that many entries into a 16-byte frameids -- through a uint32_t * cast of a
uint8_t *, so each write touched four bytes and the overrun reached four bytes
past the array, into the rest of akgl_Sprite and then the next pool slot. The
count is checked first now, each id is read into an int and narrowed
deliberately, and a frame number too large for a uint8_t is refused rather than
truncated into an index for a different tile.

The tilemap loader had the same shape twice: objects[j] with no check against
AKGL_TILEMAP_MAX_OBJECTS_PER_LAYER and tilesets[i] with none against
AKGL_TILEMAP_MAX_TILESETS. akgl_tilemap_load_layers already bounded its own
loop, so the pattern was in the same file. The object one is the reachable
half -- 128 objects is not a large object layer.

akgl_string_initialize zeroed sizeof(akgl_String) starting at `data`, which
begins after the refcount in front of it, so it ran four bytes past the end of
the object and onto the *next* slot's refcount -- the field the allocator reads
to decide whether a slot is free. Same file, same class, fixed with it:
akgl_string_copy accepted a count larger than the buffers, reading past one
pool slot and writing past another, which the header documented as behaviour.

Every case has a test that fails against the old code, with five new fixtures.
Exactly-the-maximum is asserted alongside one-past in each, so the bound cannot
be fixed by making the limit off by one.

25/25 pass, memcheck clean, reindent --check clean.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-08-01 00:24:35 -04:00
parent 6b1cf437d3
commit 230278d303
14 changed files with 3763 additions and 39 deletions

View File

@@ -130,6 +130,111 @@ void reset_string_heap(void)
}
}
/**
* @brief akgl_string_initialize must not write past the buffer it is zeroing.
*
* The `NULL` init path zeroed `sizeof(akgl_String)` bytes starting at `data`.
* `data` begins after the `int refcount` in front of it, so that ran four bytes
* past the end of the object -- straight onto the *next* pool slot's refcount,
* which is the field the allocator uses to decide whether a slot is free.
*
* Claiming two adjacent slots and initializing the first is enough to catch it:
* the second's refcount goes to zero and the pool believes it is free while the
* caller is still holding it.
*/
akerr_ErrorContext *test_akgl_string_initialize_stays_in_bounds(void)
{
PREPARE_ERROR(errctx);
akgl_String *first = NULL;
akgl_String *second = NULL;
ATTEMPT {
CATCH(errctx, akgl_heap_next_string(&first));
CATCH(errctx, akgl_heap_next_string(&second));
TEST_ASSERT(errctx, second == (first + 1),
"the pool did not hand out adjacent slots; this test needs them");
// A sentinel the overrun would land on. akgl_heap_next_string has
// already set it to 1; make it something an accidental write cannot
// coincide with.
second->refcount = 0x5A;
CATCH(errctx, akgl_string_initialize(first, NULL));
TEST_ASSERT(errctx, second->refcount == 0x5A,
"initializing a string wrote past its buffer: the next slot's "
"refcount is %d, expected %d",
second->refcount, 0x5A);
TEST_ASSERT(errctx, first->refcount == 1,
"initializing a string left its own refcount at %d, expected 1",
first->refcount);
TEST_ASSERT(errctx, first->data[0] == '\0',
"initializing a string with NULL did not zero its buffer");
TEST_ASSERT(errctx, first->data[AKGL_MAX_STRING_LENGTH - 1] == '\0',
"initializing a string with NULL did not zero its last byte");
} CLEANUP {
if ( second != NULL ) {
second->refcount = 1;
IGNORE(akgl_heap_release_string(second));
}
if ( first != NULL ) {
IGNORE(akgl_heap_release_string(first));
}
} PROCESS(errctx) {
} FINISH(errctx, true);
SUCCEED_RETURN(errctx);
}
/**
* @brief akgl_string_copy must refuse a count that would leave both buffers.
*
* Both slots are exactly AKGL_MAX_STRING_LENGTH bytes, so a larger count read
* past the end of one and wrote past the end of the other. The header used to
* document that as behaviour.
*/
akerr_ErrorContext *test_akgl_string_copy_bounds_its_count(void)
{
PREPARE_ERROR(errctx);
akgl_String *src = NULL;
akgl_String *dest = NULL;
ATTEMPT {
CATCH(errctx, akgl_heap_next_string(&src));
CATCH(errctx, akgl_heap_next_string(&dest));
CATCH(errctx, akgl_string_initialize(src, "bounded"));
CATCH(errctx, akgl_string_initialize(dest, NULL));
TEST_EXPECT_STATUS(errctx, AKERR_OUTOFBOUNDS,
akgl_string_copy(src, dest, AKGL_MAX_STRING_LENGTH + 1),
"copying one byte more than a pool string holds");
TEST_EXPECT_STATUS(errctx, AKERR_OUTOFBOUNDS,
akgl_string_copy(src, dest, -1),
"copying a negative number of bytes");
// The boundary itself is legal: it is exactly the buffer.
TEST_EXPECT_OK(errctx, akgl_string_copy(src, dest, AKGL_MAX_STRING_LENGTH),
"copying exactly a pool string's length");
TEST_ASSERT(errctx, strcmp((char *)&dest->data, "bounded") == 0,
"a full-length copy did not transfer the contents");
// 0 still means "the whole buffer" rather than "nothing".
CATCH(errctx, akgl_string_initialize(dest, NULL));
TEST_EXPECT_OK(errctx, akgl_string_copy(src, dest, 0),
"copying with a count of zero");
TEST_ASSERT(errctx, strcmp((char *)&dest->data, "bounded") == 0,
"a zero-count copy did not transfer the whole buffer");
} CLEANUP {
if ( dest != NULL ) {
IGNORE(akgl_heap_release_string(dest));
}
if ( src != NULL ) {
IGNORE(akgl_heap_release_string(src));
}
} PROCESS(errctx) {
} FINISH(errctx, true);
SUCCEED_RETURN(errctx);
}
int main(void)
{
@@ -151,6 +256,12 @@ int main(void)
reset_string_heap();
printf("test_akgl_string_initialize....\n");
test_akgl_string_initialize();
reset_string_heap();
printf("test_akgl_string_initialize_stays_in_bounds ...\n");
CATCH(errctx, test_akgl_string_initialize_stays_in_bounds());
reset_string_heap();
printf("test_akgl_string_copy_bounds_its_count ...\n");
CATCH(errctx, test_akgl_string_copy_bounds_its_count());
} CLEANUP {
} PROCESS(errctx) {
} FINISH_NORETURN(errctx);