On 6/21/2016 4:18 PM, Reshma Pattan wrote:
> getenv can return a NULL pointer if the match for
> SOCKET_PATH_HOME is not found in the environment.
> NULL check is added to return immediately without
> calling mkdir.
>
> Coverity issue 127344: return value check
> Coverity issue 127347: null pointer dereference
>
> Fixes: 278f945402c5 ("pdump: add new library for packet capture")
> Fixes: 278f945402c5 ("pdump: add new library for packet capture")
>
> Signed-off-by: Reshma Pattan <reshma.pattan at intel.com>
...
> /* get socket path (/var/run if root, $HOME otherwise) */
> -static void
> +static int
> pdump_get_socket_path(char *buffer, int bufsz, enum rte_pdump_socktype type)
> {
> const char *dir = NULL;
> @@ -451,8 +451,16 @@ pdump_get_socket_path(char *buffer, int bufsz, enum
> rte_pdump_socktype type)
> else if (type == RTE_PDUMP_SOCKET_CLIENT && client_socket_dir[0] != 0)
> dir = client_socket_dir;
> else {
> - if (getuid() != 0)
> + if (getuid() != 0) {
> dir = getenv(SOCKET_PATH_HOME);
> + if (!dir) {
> + RTE_LOG(ERR, PDUMP,
> + "Failed to get environment variable"
> + "value for %s, %s:%d\n",
> + SOCKET_PATH_HOME, __func__, __LINE__);
> + return -1;
Instead of failing, does it make sense to fallback to a default path?
Is it possible that sometimes end user doesn't really care where socket
created as long as it created and runs smoothly?
