Merge tag 'optee-rpc-arg-for-v5.19' of...

Merge tag 'optee-rpc-arg-for-v5.19' of https://git.linaro.org/people/jens.wiklander/linux-tee into arm/drivers

OP-TEE RPC argument cache

Adds caching of the OP-TEE argument structure used to pass request to
secure world. This reduces quite a bit of unnecessary alloc/free and
possibly switching back and forth to secure work in order to register
the buffers in some configurations, most notably FF-A.

* tag 'optee-rpc-arg-for-v5.19' of https://git.linaro.org/people/jens.wiklander/linux-tee:
  optee: cache argument shared memory structs
  optee: add FF-A capability OPTEE_FFA_SEC_CAP_ARG_OFFSET
  optee: add OPTEE_SMC_CALL_WITH_RPC_ARG and OPTEE_SMC_CALL_WITH_REGD_ARG
  optee: rename rpc_arg_count to rpc_param_count

Link: https://lore.kernel.org/r/20220504201759.GA180315@jadeSigned-off-by: Arnd Bergmann <arnd@arndb.de>

drivers/tee/optee/core.c

@@ -171,6 +171,7 @@ void optee_remove_common(struct optee *optee)
 	optee_unregister_devices();
 
 	optee_notif_uninit(optee);
+	optee_shm_arg_cache_uninit(optee);
 	teedev_close_context(optee->ctx);
 	/*
 	 * The two devices have to be unregistered before we can free the

drivers/tee/optee/ffa_abi.c

@@ -601,6 +601,7 @@ static int optee_ffa_yielding_call(struct tee_context *ctx,
  * optee_ffa_do_call_with_arg() - Do a FF-A call to enter OP-TEE in secure world
  * @ctx:	calling context
  * @shm:	shared memory holding the message to pass to secure world
+ * @offs:	offset of the message in @shm
  *
  * Does a FF-A call to OP-TEE in secure world and handles eventual resulting
  * Remote Procedure Calls (RPC) from OP-TEE.
@@ -609,24 +610,33 @@ static int optee_ffa_yielding_call(struct tee_context *ctx,
  */
 
 static int optee_ffa_do_call_with_arg(struct tee_context *ctx,
-				      struct tee_shm *shm)
+				      struct tee_shm *shm, u_int offs)
 {
 	struct ffa_send_direct_data data = {
 		.data0 = OPTEE_FFA_YIELDING_CALL_WITH_ARG,
 		.data1 = (u32)shm->sec_world_id,
 		.data2 = (u32)(shm->sec_world_id >> 32),
-		.data3 = shm->offset,
+		.data3 = offs,
 	};
 	struct optee_msg_arg *arg;
 	unsigned int rpc_arg_offs;
 	struct optee_msg_arg *rpc_arg;
 
-	arg = tee_shm_get_va(shm, 0);
+	/*
+	 * The shared memory object has to start on a page when passed as
+	 * an argument struct. This is also what the shm pool allocator
+	 * returns, but check this before calling secure world to catch
+	 * eventual errors early in case something changes.
+	 */
+	if (shm->offset)
+		return -EINVAL;
+
+	arg = tee_shm_get_va(shm, offs);
 	if (IS_ERR(arg))
 		return PTR_ERR(arg);
 
 	rpc_arg_offs = OPTEE_MSG_GET_ARG_SIZE(arg->num_params);
-	rpc_arg = tee_shm_get_va(shm, rpc_arg_offs);
+	rpc_arg = tee_shm_get_va(shm, offs + rpc_arg_offs);
 	if (IS_ERR(rpc_arg))
 		return PTR_ERR(rpc_arg);
 
@@ -678,7 +688,8 @@ static bool optee_ffa_api_is_compatbile(struct ffa_device *ffa_dev,
 
 static bool optee_ffa_exchange_caps(struct ffa_device *ffa_dev,
 				    const struct ffa_dev_ops *ops,
-				    unsigned int *rpc_arg_count)
+				    u32 *sec_caps,
+				    unsigned int *rpc_param_count)
 {
 	struct ffa_send_direct_data data = { OPTEE_FFA_EXCHANGE_CAPABILITIES };
 	int rc;
@@ -693,7 +704,8 @@ static bool optee_ffa_exchange_caps(struct ffa_device *ffa_dev,
 		return false;
 	}
 
-	*rpc_arg_count = (u8)data.data1;
+	*rpc_param_count = (u8)data.data1;
+	*sec_caps = data.data2;
 
 	return true;
 }
@@ -772,11 +784,13 @@ static void optee_ffa_remove(struct ffa_device *ffa_dev)
 static int optee_ffa_probe(struct ffa_device *ffa_dev)
 {
 	const struct ffa_dev_ops *ffa_ops;
-	unsigned int rpc_arg_count;
+	unsigned int rpc_param_count;
 	struct tee_shm_pool *pool;
 	struct tee_device *teedev;
 	struct tee_context *ctx;
+	u32 arg_cache_flags = 0;
 	struct optee *optee;
+	u32 sec_caps;
 	int rc;
 
 	ffa_ops = ffa_dev_ops_get(ffa_dev);
@@ -788,8 +802,11 @@ static int optee_ffa_probe(struct ffa_device *ffa_dev)
 	if (!optee_ffa_api_is_compatbile(ffa_dev, ffa_ops))
 		return -EINVAL;
 
-	if (!optee_ffa_exchange_caps(ffa_dev, ffa_ops, &rpc_arg_count))
+	if (!optee_ffa_exchange_caps(ffa_dev, ffa_ops, &sec_caps,
+				     &rpc_param_count))
 		return -EINVAL;
+	if (sec_caps & OPTEE_FFA_SEC_CAP_ARG_OFFSET)
+		arg_cache_flags |= OPTEE_SHM_ARG_SHARED;
 
 	optee = kzalloc(sizeof(*optee), GFP_KERNEL);
 	if (!optee)
@@ -805,7 +822,7 @@ static int optee_ffa_probe(struct ffa_device *ffa_dev)
 	optee->ops = &optee_ffa_ops;
 	optee->ffa.ffa_dev = ffa_dev;
 	optee->ffa.ffa_ops = ffa_ops;
-	optee->rpc_arg_count = rpc_arg_count;
+	optee->rpc_param_count = rpc_param_count;
 
 	teedev = tee_device_alloc(&optee_ffa_clnt_desc, NULL, optee->pool,
 				  optee);
@@ -838,6 +855,7 @@ static int optee_ffa_probe(struct ffa_device *ffa_dev)
 	mutex_init(&optee->call_queue.mutex);
 	INIT_LIST_HEAD(&optee->call_queue.waiters);
 	optee_supp_init(&optee->supp);
+	optee_shm_arg_cache_init(optee, arg_cache_flags);
 	ffa_dev_set_drvdata(ffa_dev, optee);
 	ctx = teedev_open(optee->teedev);
 	if (IS_ERR(ctx)) {

drivers/tee/optee/optee_ffa.h

@@ -81,8 +81,16 @@
  *                   as the second MSG arg struct for
  *                   OPTEE_FFA_YIELDING_CALL_WITH_ARG.
  *        Bit[31:8]: Reserved (MBZ)
- * w5-w7: Note used (MBZ)
+ * w5:	  Bitfield of secure world capabilities OPTEE_FFA_SEC_CAP_* below,
+ *	  unused bits MBZ.
+ * w6-w7: Not used (MBZ)
+ */
+/*
+ * Secure world supports giving an offset into the argument shared memory
+ * object, see also OPTEE_FFA_YIELDING_CALL_WITH_ARG
  */
+#define OPTEE_FFA_SEC_CAP_ARG_OFFSET	BIT(0)
+
 #define OPTEE_FFA_EXCHANGE_CAPABILITIES OPTEE_FFA_BLOCKING_CALL(2)
 
 /*
@@ -112,6 +120,8 @@
  *	  OPTEE_MSG_GET_ARG_SIZE(num_params) follows a struct optee_msg_arg
  *	  for RPC, this struct has reserved space for the number of RPC
  *	  parameters as returned by OPTEE_FFA_EXCHANGE_CAPABILITIES.
+ *	  MBZ unless the bit OPTEE_FFA_SEC_CAP_ARG_OFFSET is received with
+ *	  OPTEE_FFA_EXCHANGE_CAPABILITIES.
  * w7:    Not used (MBZ)
  * Resume from RPC. Register usage:
  * w3:    Service ID, OPTEE_FFA_YIELDING_CALL_RESUME

drivers/tee/optee/optee_private.h

@@ -59,6 +59,16 @@ struct optee_notif {
 	u_long *bitmap;
 };
 
+#define OPTEE_SHM_ARG_ALLOC_PRIV	BIT(0)
+#define OPTEE_SHM_ARG_SHARED		BIT(1)
+struct optee_shm_arg_entry;
+struct optee_shm_arg_cache {
+	u32 flags;
+	/* Serializes access to this struct */
+	struct mutex mutex;
+	struct list_head shm_args;
+};
+
 /**
  * struct optee_supp - supplicant synchronization struct
  * @ctx			the context of current connected supplicant.
@@ -121,7 +131,7 @@ struct optee;
  */
 struct optee_ops {
 	int (*do_call_with_arg)(struct tee_context *ctx,
-				struct tee_shm *shm_arg);
+				struct tee_shm *shm_arg, u_int offs);
 	int (*to_msg_param)(struct optee *optee,
 			    struct optee_msg_param *msg_params,
 			    size_t num_params, const struct tee_param *params);
@@ -143,7 +153,7 @@ struct optee_ops {
  * @notif:		notification synchronization struct
  * @supp:		supplicant synchronization struct for RPC to supplicant
  * @pool:		shared memory pool
- * @rpc_arg_count:	If > 0 number of RPC parameters to make room for
+ * @rpc_param_count:	If > 0 number of RPC parameters to make room for
  * @scan_bus_done	flag if device registation was already done.
  * @scan_bus_wq		workqueue to scan optee bus and register optee drivers
  * @scan_bus_work	workq to scan optee bus and register optee drivers
@@ -157,11 +167,12 @@ struct optee {
 		struct optee_smc smc;
 		struct optee_ffa ffa;
 	};
+	struct optee_shm_arg_cache shm_arg_cache;
 	struct optee_call_queue call_queue;
 	struct optee_notif notif;
 	struct optee_supp supp;
 	struct tee_shm_pool *pool;
-	unsigned int rpc_arg_count;
+	unsigned int rpc_param_count;
 	bool   scan_bus_done;
 	struct workqueue_struct *scan_bus_wq;
 	struct work_struct scan_bus_work;
@@ -273,8 +284,18 @@ void optee_cq_wait_for_completion(struct optee_call_queue *cq,
 void optee_cq_wait_final(struct optee_call_queue *cq,
 			 struct optee_call_waiter *w);
 int optee_check_mem_type(unsigned long start, size_t num_pages);
-struct tee_shm *optee_get_msg_arg(struct tee_context *ctx, size_t num_params,
-				  struct optee_msg_arg **msg_arg);
+
+void optee_shm_arg_cache_init(struct optee *optee, u32 flags);
+void optee_shm_arg_cache_uninit(struct optee *optee);
+struct optee_msg_arg *optee_get_msg_arg(struct tee_context *ctx,
+					size_t num_params,
+					struct optee_shm_arg_entry **entry,
+					struct tee_shm **shm_ret,
+					u_int *offs);
+void optee_free_msg_arg(struct tee_context *ctx,
+			struct optee_shm_arg_entry *entry, u_int offs);
+size_t optee_msg_arg_size(size_t rpc_param_count);
+
 
 struct tee_shm *optee_rpc_cmd_alloc_suppl(struct tee_context *ctx, size_t sz);
 void optee_rpc_cmd_free_suppl(struct tee_context *ctx, struct tee_shm *shm);

drivers/tee/optee/optee_smc.h

@@ -107,14 +107,22 @@ struct optee_smc_call_get_os_revision_result {
 /*
  * Call with struct optee_msg_arg as argument
  *
- * When calling this function normal world has a few responsibilities:
+ * When called with OPTEE_SMC_CALL_WITH_RPC_ARG or
+ * OPTEE_SMC_CALL_WITH_REGD_ARG in a0 there is one RPC struct optee_msg_arg
+ * following after the first struct optee_msg_arg. The RPC struct
+ * optee_msg_arg has reserved space for the number of RPC parameters as
+ * returned by OPTEE_SMC_EXCHANGE_CAPABILITIES.
+ *
+ * When calling these functions, normal world has a few responsibilities:
  * 1. It must be able to handle eventual RPCs
  * 2. Non-secure interrupts should not be masked
  * 3. If asynchronous notifications has been negotiated successfully, then
- *    asynchronous notifications should be unmasked during this call.
+ *    the interrupt for asynchronous notifications should be unmasked
+ *    during this call.
  *
- * Call register usage:
- * a0	SMC Function ID, OPTEE_SMC*CALL_WITH_ARG
+ * Call register usage, OPTEE_SMC_CALL_WITH_ARG and
+ * OPTEE_SMC_CALL_WITH_RPC_ARG:
+ * a0	SMC Function ID, OPTEE_SMC_CALL_WITH_ARG or OPTEE_SMC_CALL_WITH_RPC_ARG
  * a1	Upper 32 bits of a 64-bit physical pointer to a struct optee_msg_arg
  * a2	Lower 32 bits of a 64-bit physical pointer to a struct optee_msg_arg
  * a3	Cache settings, not used if physical pointer is in a predefined shared
@@ -122,6 +130,15 @@ struct optee_smc_call_get_os_revision_result {
  * a4-6	Not used
  * a7	Hypervisor Client ID register
  *
+ * Call register usage, OPTEE_SMC_CALL_WITH_REGD_ARG:
+ * a0	SMC Function ID, OPTEE_SMC_CALL_WITH_REGD_ARG
+ * a1	Upper 32 bits of a 64-bit shared memory cookie
+ * a2	Lower 32 bits of a 64-bit shared memory cookie
+ * a3	Offset of the struct optee_msg_arg in the shared memory with the
+ *	supplied cookie
+ * a4-6	Not used
+ * a7	Hypervisor Client ID register
+ *
  * Normal return register usage:
  * a0	Return value, OPTEE_SMC_RETURN_*
  * a1-3	Not used
@@ -154,6 +171,10 @@ struct optee_smc_call_get_os_revision_result {
 #define OPTEE_SMC_FUNCID_CALL_WITH_ARG OPTEE_MSG_FUNCID_CALL_WITH_ARG
 #define OPTEE_SMC_CALL_WITH_ARG \
 	OPTEE_SMC_STD_CALL_VAL(OPTEE_SMC_FUNCID_CALL_WITH_ARG)
+#define OPTEE_SMC_CALL_WITH_RPC_ARG \
+	OPTEE_SMC_STD_CALL_VAL(OPTEE_SMC_FUNCID_CALL_WITH_RPC_ARG)
+#define OPTEE_SMC_CALL_WITH_REGD_ARG \
+	OPTEE_SMC_STD_CALL_VAL(OPTEE_SMC_FUNCID_CALL_WITH_REGD_ARG)
 
 /*
  * Get Shared Memory Config
@@ -202,7 +223,11 @@ struct optee_smc_get_shm_config_result {
  * a0	OPTEE_SMC_RETURN_OK
  * a1	bitfield of secure world capabilities OPTEE_SMC_SEC_CAP_*
  * a2	The maximum secure world notification number
- * a3-7	Preserved
+ * a3	Bit[7:0]: Number of parameters needed for RPC to be supplied
+ *		  as the second MSG arg struct for
+ *		  OPTEE_SMC_CALL_WITH_ARG
+ *	Bit[31:8]: Reserved (MBZ)
+ * a4-7	Preserved
  *
  * Error return register usage:
  * a0	OPTEE_SMC_RETURN_ENOTAVAIL, can't use the capabilities from normal world
@@ -227,6 +252,8 @@ struct optee_smc_get_shm_config_result {
 #define OPTEE_SMC_SEC_CAP_MEMREF_NULL		BIT(4)
 /* Secure world supports asynchronous notification of normal world */
 #define OPTEE_SMC_SEC_CAP_ASYNC_NOTIF		BIT(5)
+/* Secure world supports pre-allocating RPC arg struct */
+#define OPTEE_SMC_SEC_CAP_RPC_ARG		BIT(6)
 
 #define OPTEE_SMC_FUNCID_EXCHANGE_CAPABILITIES	9
 #define OPTEE_SMC_EXCHANGE_CAPABILITIES \
@@ -236,7 +263,7 @@ struct optee_smc_exchange_capabilities_result {
 	unsigned long status;
 	unsigned long capabilities;
 	unsigned long max_notif_value;
-	unsigned long reserved0;
+	unsigned long data;
 };
 
 /*
@@ -358,6 +385,9 @@ struct optee_smc_disable_shm_cache_result {
  * should be called until all pended values have been retrieved. When a
  * value is retrieved, it's cleared from the record in secure world.
  *
+ * It is expected that this function is called from an interrupt handler
+ * in normal world.
+ *
  * Call requests usage:
  * a0	SMC Function ID, OPTEE_SMC_GET_ASYNC_NOTIF_VALUE
  * a1-6	Not used
@@ -390,6 +420,12 @@ struct optee_smc_disable_shm_cache_result {
 #define OPTEE_SMC_GET_ASYNC_NOTIF_VALUE \
 	OPTEE_SMC_FAST_CALL_VAL(OPTEE_SMC_FUNCID_GET_ASYNC_NOTIF_VALUE)
 
+/* See OPTEE_SMC_CALL_WITH_RPC_ARG above */
+#define OPTEE_SMC_FUNCID_CALL_WITH_RPC_ARG	18
+
+/* See OPTEE_SMC_CALL_WITH_REGD_ARG above */
+#define OPTEE_SMC_FUNCID_CALL_WITH_REGD_ARG	19
+
 /*
  * Resume from RPC (for example after processing a foreign interrupt)
  *