RxRPC: Allocate tokens with kzalloc to avoid oops in rxrpc_destroy

With slab poisoning enabled, I see the following oops:

  Unable to handle kernel paging request for data at address 0x6b6b6b6b6b6b6b73
  ...
  NIP [c0000000006bc61c] .rxrpc_destroy+0x44/0x104
  LR [c0000000006bc618] .rxrpc_destroy+0x40/0x104
  Call Trace:
  [c0000000feb2bc00] [c0000000006bc618] .rxrpc_destroy+0x40/0x104 (unreliable)
  [c0000000feb2bc90] [c000000000349b2c] .key_cleanup+0x1a8/0x20c
  [c0000000feb2bd40] [c0000000000a2920] .process_one_work+0x2f4/0x4d0
  [c0000000feb2be00] [c0000000000a2d50] .worker_thread+0x254/0x468
  [c0000000feb2bec0] [c0000000000a868c] .kthread+0xbc/0xc8
  [c0000000feb2bf90] [c000000000020e00] .kernel_thread+0x54/0x70

We aren't initialising token->next, but the code in destroy_context relies
on the list being NULL terminated. Use kzalloc to zero out all the fields.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

net/rxrpc/ar-key.c

@@ -89,11 +89,11 @@ static int rxrpc_instantiate_xdr_rxkad(struct key *key, const __be32 *xdr,
 		return ret;
 
 	plen -= sizeof(*token);
-	token = kmalloc(sizeof(*token), GFP_KERNEL);
+	token = kzalloc(sizeof(*token), GFP_KERNEL);
 	if (!token)
 		return -ENOMEM;
 
-	token->kad = kmalloc(plen, GFP_KERNEL);
+	token->kad = kzalloc(plen, GFP_KERNEL);
 	if (!token->kad) {
 		kfree(token);
 		return -ENOMEM;
@@ -731,10 +731,10 @@ static int rxrpc_instantiate(struct key *key, const void *data, size_t datalen)
 		goto error;
 
 	ret = -ENOMEM;
-	token = kmalloc(sizeof(*token), GFP_KERNEL);
+	token = kzalloc(sizeof(*token), GFP_KERNEL);
 	if (!token)
 		goto error;
-	token->kad = kmalloc(plen, GFP_KERNEL);
+	token->kad = kzalloc(plen, GFP_KERNEL);
 	if (!token->kad)
 		goto error_free;