Fixed memory leak in RPC parsing code.

Problem in prs_set_buffer_size() was Realloc returns a NULL when newsize is zero (equivalent to a free()). We were returning a failure here without resetting the buffer_size or the data_p pointer in the prs_struct. And we weren't checking for a failure from prs_set_buffer_size(). So realloc's to zero size were not reflected in the prs_struct: memory leak.
(This used to be commit 590d9ece8449b1feecfe1aa13e61bcd8fea4e5bf)

1 files changed, 3 insertions, 1 deletions

diff --git a/source3/rpc_parse/parse_prs.c b/source3/rpc_parse/parse_prs.c
index dafff63ad9..422b420a3c 100644
--- a/source3/rpc_parse/parse_prs.c
+++ b/source3/rpc_parse/parse_prs.c
@@ -164,9 +164,11 @@ BOOL prs_set_buffer_size(prs_struct *ps, uint32 newsize)
 
 	if (newsize < ps->buffer_size) {
 		char *new_data_p = Realloc(ps->data_p, newsize);
-		if (new_data_p == NULL) {
+		/* if newsize is zero, Realloc acts like free() & returns NULL*/
+		if (new_data_p == NULL && newsize != 0) {
 			DEBUG(0,("prs_set_buffer_size: Realloc failure for size %u.\n",
 				(unsigned int)newsize));
+			DEBUG(0,("prs_set_buffer_size: Reason %s\n",strerror(errno)));
 			return False;
 		}
 		ps->data_p = new_data_p;