CDRIVER-5756 Coverity fixes #1867
Conversation
CID: 100057
CID: 100087
CID: 115840, 115848, 115851, 115853
CID: 134019
CID: 138986
CID: 138989
CID: 156373
CID: 157950
| bson_mutex_lock (&topology->tpld_modification_mtx); | ||
| for (size_t i = 0u; i < n_rtt_monitors; i++) { | ||
| server_monitor = mongoc_set_get_item (topology->rtt_monitors, i); | ||
| mongoc_server_monitor_request_shutdown (server_monitor); | ||
| } | ||
| bson_mutex_unlock (&topology->tpld_modification_mtx); |
There was a problem hiding this comment.
I'm not sure if this mutex is relevant here. Do you have a link to the Coverity warning related to these lines?
There was a problem hiding this comment.
I sent you the link in a DM.
| jsn->jprs = (jsonsl_jpr_t *)malloc(sizeof(jsonsl_jpr_t) * njprs); | ||
| if (!jsn->jprs) { | ||
| return; | ||
| } |
There was a problem hiding this comment.
| jsn->jprs = (jsonsl_jpr_t *)malloc(sizeof(jsonsl_jpr_t) * njprs); | |
| if (!jsn->jprs) { | |
| return; | |
| } | |
| jsn->jprs = (jsonsl_jpr_t *) bson_malloc (sizeof (jsonsl_jpr_t) * njprs); |
Suggest using bson_malloc. bson_malloc already checks (and aborts) if allocation fails. I expect that will fix the Coverity warning. Using bson_malloc ensures allocators set with bson_mem_set_vtable are used.
Also replace free(jsn->jprs); with bson_free(jsn->jprs);
| jsn->jpr_root = (size_t*)calloc(1, sizeof(size_t) * njprs * jsn->levels_max); | ||
| if (!jsn->jpr_root) { | ||
| free(jsn->jprs); | ||
| return; | ||
| } |
There was a problem hiding this comment.
| jsn->jpr_root = (size_t*)calloc(1, sizeof(size_t) * njprs * jsn->levels_max); | |
| if (!jsn->jpr_root) { | |
| free(jsn->jprs); | |
| return; | |
| } | |
| jsn->jpr_root = (size_t *) bson_malloc0 (sizeof (size_t) * njprs * jsn->levels_max); |
Similarly, use bson_malloc0 in place of calloc. Suggest updating other malloc / free calls in this file.
There was a problem hiding this comment.
OK. That makes sense. Should I do that as a drive-by in this PR, or a separate PR?
There was a problem hiding this comment.
I think it is OK to include in this PR. Though the other malloc calls check their returns, I think it is a related enough change to include in this PR.
| @@ -262,14 +262,20 @@ static const uint8_t mongoc_b64rmap_invalid = 0xff; | |||
| #if defined(BSON_OS_UNIX) | |||
There was a problem hiding this comment.
I expect this entire #if / #else / #endif block can be removed. Add #include <common-thread-private.h> and instead use the bson_once* and BSON_ONCE* APIs. bson_once already has an assert.
| @@ -622,9 +622,10 @@ _bson_json_read_integer (bson_json_reader_t *reader, uint64_t val, int64_t sign) | |||
|
|
|||
| if (rs == BSON_JSON_REGULAR) { | |||
| BASIC_CB_BAIL_IF_NOT_NORMAL ("integer"); | |||
| BSON_ASSERT (mlib_in_range (int, len)); | |||
There was a problem hiding this comment.
Rather than assert, suggest checking if len is greater than INT_MAX and returning an error. I expect this applies to all types being read. That may avoid a possible crash if the input JSON has a very large (over INT_MAX length) key.
Here is a commit with a test (which is only run when the environment variable MONGOC_TEST_LARGE_ALLOCATIONS=ON is defined) and suggested change.
This PR fixes some small Coverity issues (w/ full Evergreen patch build).
@kevinAlbs I wasn't sure who would be available to review, so please feel free to adjust the reviewers as needed.