Re: [PATCH v2] optee: add timeout value to optee_notif_wait() to support timeout - OP-TEE

6 May 2024

On Mon, May 6, 2024 at 10:19 AM Gavin Liu (劉哲廷) Gavin.Liu@mediatek.com wrote:
...
Hi Jens,
Thanks for the reviews.
On Mon, 2024-05-06 at 09:48 +0200, Jens Wiklander wrote:
...
External email : Please do not click links or open attachments until
you have verified the sender or the content.
 On Thu, May 2, 2024 at 10:56 AM gavin.liu gavin.liu@mediatek.com
wrote:
...
From: Gavin Liu gavin.liu@mediatek.com
Add timeout value to support self waking when timeout to avoid
waiting
...
indefinitely.
Signed-off-by: Gavin Liu gavin.liu@mediatek.com
Change in v2:
Change commit message.
Add description for value[0].c in optee_rpc_cmd.h.


drivers/tee/optee/notif.c         |  9 +++++++--
 drivers/tee/optee/optee_private.h |  2 +-
 drivers/tee/optee/optee_rpc_cmd.h |  1 +
 drivers/tee/optee/rpc.c           | 10 ++++++++--
 4 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/drivers/tee/optee/notif.c b/drivers/tee/optee/notif.c
index 05212842b0a5..d5e5c0645609 100644
--- a/drivers/tee/optee/notif.c
+++ b/drivers/tee/optee/notif.c
@@ -29,7 +29,7 @@ static bool have_key(struct optee *optee, u_int
key)
...
    return false;

}
-int optee_notif_wait(struct optee *optee, u_int key)
+int optee_notif_wait(struct optee *optee, u_int key, u32 timeout)
 {
        unsigned long flags;
        struct notif_entry *entry;
@@ -70,7 +70,12 @@ int optee_notif_wait(struct optee *optee, u_int
key)
...
     * Unlock temporarily and wait for completion.
     */
    spin_unlock_irqrestore(&optee->notif.lock, flags);


  wait_for_completion(&entry->c);




  if (timeout != 0) {


          if (!wait_for_completion_timeout(&entry->c,



timeout))
...

                  rc = -ETIMEDOUT;


  } else {


          wait_for_completion(&entry->c);


  }
  spin_lock_irqsave(&optee->notif.lock, flags);

  list_del(&entry->link);



diff --git a/drivers/tee/optee/optee_private.h
b/drivers/tee/optee/optee_private.h
...
index 7a5243c78b55..da990c4016ec 100644
--- a/drivers/tee/optee/optee_private.h
+++ b/drivers/tee/optee/optee_private.h
@@ -252,7 +252,7 @@ struct optee_call_ctx {
int optee_notif_init(struct optee *optee, u_int max_key);
 void optee_notif_uninit(struct optee *optee);
-int optee_notif_wait(struct optee *optee, u_int key);
+int optee_notif_wait(struct optee *optee, u_int key, u32 timeout);
 int optee_notif_send(struct optee *optee, u_int key);
u32 optee_supp_thrd_req(struct tee_context *ctx, u32 func, size_t
num_params,
...
diff --git a/drivers/tee/optee/optee_rpc_cmd.h
b/drivers/tee/optee/optee_rpc_cmd.h
...
index f3f06e0994a7..99342aa66263 100644
--- a/drivers/tee/optee/optee_rpc_cmd.h
+++ b/drivers/tee/optee/optee_rpc_cmd.h
@@ -41,6 +41,7 @@

Waiting on notification
[in]    value[0].a      OPTEE_RPC_NOTIFICATION_WAIT
[in]    value[0].b      notification value



[in]    value[0].c      timeout in millisecond or 0 if no



timeout
milliseconds
Ok, I'll change it.
...
...


Sending a synchronous notification
[in]    value[0].a      OPTEE_RPC_NOTIFICATION_SEND

diff --git a/drivers/tee/optee/rpc.c b/drivers/tee/optee/rpc.c
index e69bc6380683..14e6246aaf05 100644
--- a/drivers/tee/optee/rpc.c
+++ b/drivers/tee/optee/rpc.c
@@ -130,6 +130,8 @@ static void
handle_rpc_func_cmd_i2c_transfer(struct tee_context *ctx,
...
static void handle_rpc_func_cmd_wq(struct optee *optee,
                                   struct optee_msg_arg *arg)
 {

  int rc = 0;


  if (arg->num_params != 1)
          goto bad;




@@ -139,7 +141,8 @@ static void handle_rpc_func_cmd_wq(struct optee
*optee,
...
    switch (arg->params[0].u.value.a) {
    case OPTEE_RPC_NOTIFICATION_WAIT:


          if (optee_notif_wait(optee, arg-



params[0].u.value.b))

          rc = optee_notif_wait(optee, arg-



params[0].u.value.b, arg->params[0].u.value.c);

          if (rc)
                  goto bad;
          break;
  case OPTEE_RPC_NOTIFICATION_SEND:



@@ -153,7 +156,10 @@ static void handle_rpc_func_cmd_wq(struct
optee *optee,
...
    arg->ret = TEEC_SUCCESS;
    return;

bad:

  arg->ret = TEEC_ERROR_BAD_PARAMETERS;




  if (rc == -ETIMEDOUT)


          arg->ret = TEEC_ERROR_BUSY;



TEEC_ERROR_BUSY is confusing. How about TEE_ERROR_TIMEOUT? We
normally
only use TEEC_XXX error codes, but GP doesn't define a
TEEC_ERROR_TIMEOUT so it's better to use the GP-defined
TEE_ERROR_TIMEOUT.
Cheers,
Jens
Agreed, we also use TEE_ERROR_TIMEOUT in the corresponding patch for
optee-os. But, the "TEE_ERROR_TIMEOUT" is currently undefined in Linux
Kernel. Do you have a suggestion for which header would be better to
place it in?
Please put it in drivers/tee/optee/optee_private.h below
TEEC_ERROR_SHORT_BUFFER with a comment that it's from the GP TEE
Internal Core API Specification.
Cheers,
Jens

    