crypto/octeontx: fix null pointer dereferencing

The function otx_cpt_get_resource() would be setting the pointer
'instance'. In case of error, 'instance' would be set to NULL, and
returns rte_errno. If rte_errno when 'instance' is set to NULL, it can
lead to NULL pointer dereferencing.

Coverity Issue: 323486, 323489
Fixes: bfe2ae495ee2 ("crypto/octeontx: add PMD skeleton")
Fixes: 0961348fdf52 ("crypto/octeontx: add queue pair functions")

Signed-off-by: Anoob Joseph <anoob.joseph@caviumnetworks.com>
This commit is contained in:
Anoob Joseph 2018-11-12 18:14:12 +00:00 committed by Thomas Monjalon
parent acbaf3cedc
commit accf8ca7b0
2 changed files with 2 additions and 2 deletions

View File

@ -100,8 +100,8 @@ otx_cpt_pci_remove(struct rte_pci_device *pci_dev)
if (rte_eal_process_type() == RTE_PROC_PRIMARY)
rte_free(cryptodev->data->dev_private);
cryptodev->device = NULL;
cryptodev->device->driver = NULL;
cryptodev->device = NULL;
cryptodev->data = NULL;
/* free metapool memory */

View File

@ -216,7 +216,7 @@ otx_cpt_que_pair_setup(struct rte_cryptodev *dev,
}
ret = otx_cpt_get_resource(cptvf, 0, &instance);
if (ret != 0) {
if (ret != 0 || instance == NULL) {
CPT_LOG_ERR("Error getting instance handle from device %s : "
"ret = %d", dev->data->name, ret);
return ret;