uvm_sequence_base

The constructor for uvm_sequence_base.

Returns 1 on items and 0 on sequences.  As this object is a sequence, is_item will always return 0.

Returns the sequence state as an enumerated value.  Can use to wait on the sequence reaching or changing from one or more states.

wait(get_sequence_state() & (UVM_STOPPED|UVM_FINISHED));

Waits until the sequence reaches one of the given state.  If the sequence is already in one of the state, this method returns immediately.

wait_for_sequence_state(UVM_STOPPED|UVM_FINISHED);

Returns the integral recording transaction handle for this sequence.  Can be used to associate sub-sequences and sequence items as child transactions when calling uvm_component::begin_child_tr.

Executes this sequence, returning when the sequence has completed.

The sequencer argument specifies the sequencer on which to run this sequence.  The sequencer must be compatible with the sequence.

If parent_sequence is null, then this sequence is a root parent, otherwise it is a child of parent_sequence.  The parent_sequence’s pre_do, mid_do, and post_do methods will be called during the execution of this sequence.

By default, the priority of a sequence is the priority of its parent sequence.  If it is a root sequence, its default priority is 100.  A different priority may be specified by this_priority.  Higher numbers indicate higher priority.

If call_pre_post is set to 1 (default), then the pre_body and post_body tasks will be called before and after the sequence body is called.

This task is a user-definable callback that is called before the optional execution of pre_body.  This method should not be called directly by the user.

This task is a user-definable callback that is called before the execution of body only when the sequence is started with start.  If start is called with call_pre_post set to 0, pre_body is not called.  This method should not be called directly by the user.

This task is a user-definable callback task that is called on the parent sequence, if any sequence has issued a wait_for_grant() call and after the sequencer has selected this sequence, and before the item is randomized.

Although pre_do is a task, consuming simulation cycles may result in unexpected behavior on the driver.

This method should not be called directly by the user.

This function is a user-definable callback function that is called after the sequence item has been randomized, and just before the item is sent to the driver.  This method should not be called directly by the user.

This is the user-defined task where the main sequence code resides.  This method should not be called directly by the user.

This function is a user-definable callback function that is called after the driver has indicated that it has completed the item, using either this item_done or put methods.  This method should not be called directly by the user.

This task is a user-definable callback task that is called after the execution of body only when the sequence is started with start.  If start is called with call_pre_post set to 0, post_body is not called.  This task is a user-definable callback task that is called after the execution of the body, unless the sequence is started with call_pre_post=0.  This method should not be called directly by the user.

This task is a user-definable callback that is called after the optional execution of post_body.  This method should not be called directly by the user.

Returns the ‘starting phase’.

If non-null, the starting phase specifies the phase in which this sequence was started.  The starting phase is set automatically when this sequence is started as the default sequence on a sequencer.  See uvm_sequencer_base::start_phase_sequence for more information.

Internally, the uvm_sequence_base uses an uvm_get_to_lock_dap to protect the starting phase value from being modified after the reference has been read.  Once the sequence has ended its execution (either via natural termination, or being killed), then the starting phase value can be modified again.

Sets the ‘starting phase’.

Internally, the uvm_sequence_base uses a uvm_get_to_lock_dap to protect the starting phase value from being modified after the reference has been read.  Once the sequence has ended its execution (either via natural termination, or being killed), then the starting phase value can be modified again.

Sets the ‘automatically object to starting phase’ bit.

The most common interaction with the starting phase within a sequence is to simply raise the phase’s objection prior to executing the sequence, and drop the objection after ending the sequence (either naturally, or via a call to kill).  In order to simplify this interaction for the user, the UVM provides the ability to perform this functionality automatically.

For example

function my_sequence::new(string name="unnamed");
  super.new(name);
  set_automatic_phase_objection(1);
endfunction : new

From a timeline point of view, the automatic phase objection looks like:

start() is executed
  --! Objection is raised !--
  pre_start() is executed
  pre_body() is optionally executed
  body() is executed
  post_body() is optionally executed
  post_start() is executed
  --! Objection is dropped !--
start() unblocks

This functionality can also be enabled in sequences which were not written with UVM Run-Time Phasing in mind:

my_legacy_seq_type seq = new("seq");
seq.set_automatic_phase_objection(1);
seq.start(my_sequencer);

Internally, the uvm_sequence_base uses a uvm_get_to_lock_dap to protect the automatic_phase_objection value from being modified after the reference has been read.  Once the sequence has ended its execution (either via natural termination, or being killed), then the automatic_phase_objection value can be modified again.

NEVER set the automatic phase objection bit to 1 if your sequence runs with a forever loop inside of the body, as the objection will never get dropped!

Returns (and locks) the value of the ‘automatically object to starting phase’ bit.

If 1, then the sequence will automatically raise an objection to the starting phase (if the starting phase is not null) immediately prior to pre_start being called.  The objection will be dropped after post_start has executed, or kill has been called.

The priority of a sequence may be changed at any point in time.  When the priority of a sequence is changed, the new priority will be used by the sequencer the next time that it arbitrates between sequences.

The default priority value for a sequence is 100.  Higher values result in higher priorities.

This function returns the current priority of the sequence.

The default is_relevant implementation returns 1, indicating that the sequence is always relevant.

Users may choose to override with their own virtual function to indicate to the sequencer that the sequence is not currently relevant after a request has been made.

When the sequencer arbitrates, it will call is_relevant on each requesting, unblocked sequence to see if it is relevant.  If a 0 is returned, then the sequence will not be chosen.

If all requesting sequences are not relevant, then the sequencer will call wait_for_relevant on all sequences and re-arbitrate upon its return.

Any sequence that implements is_relevant must also implement wait_for_relevant so that the sequencer has a way to wait for a sequence to become relevant.

This method is called by the sequencer when all available sequences are not relevant.  When wait_for_relevant returns the sequencer attempt to re-arbitrate.

Returning from this call does not guarantee a sequence is relevant, although that would be the ideal.  The method provide some delay to prevent an infinite loop.

If a sequence defines is_relevant so that it is not always relevant (by default, a sequence is always relevant), then the sequence must also supply a wait_for_relevant method.

Requests a lock on the specified sequencer.  If sequencer is null, the lock will be requested on the current default sequencer.

A lock request will be arbitrated the same as any other request.  A lock is granted after all earlier requests are completed and no other locks or grabs are blocking this sequence.

The lock call will return when the lock has been granted.

Requests a lock on the specified sequencer.  If no argument is supplied, the lock will be requested on the current default sequencer.

A grab request is put in front of the arbitration queue.  It will be arbitrated before any other requests.  A grab is granted when no other grabs or locks are blocking this sequence.

The grab call will return when the grab has been granted.

Removes any locks or grabs obtained by this sequence on the specified sequencer.  If sequencer is null, then the unlock will be done on the current default sequencer.

Removes any locks or grabs obtained by this sequence on the specified sequencer.  If sequencer is null, then the unlock will be done on the current default sequencer.

Returns a bit indicating whether this sequence is currently prevented from running due to another lock or grab.  A 1 is returned if the sequence is currently blocked.  A 0 is returned if no lock or grab prevents this sequence from executing.  Note that even if a sequence is not blocked, it is possible for another sequence to issue a lock or grab before this sequence can issue a request.

Returns 1 if this sequence has a lock, 0 otherwise.

Note that even if this sequence has a lock, a child sequence may also have a lock, in which case the sequence is still blocked from issuing operations on the sequencer.

This function will kill the sequence, and cause all current locks and requests in the sequence’s default sequencer to be removed.  The sequence state will change to UVM_STOPPED, and the post_body() and post_start() callback methods will not be executed.

If a sequence has issued locks, grabs, or requests on sequencers other than the default sequencer, then care must be taken to unregister the sequence with the other sequencer(s) using the sequencer unregister_sequence() method.

This function is a user hook that is called whenever a sequence is terminated by using either sequence.kill() or sequencer.stop_sequences() (which effectively calls sequence.kill()).

Create_item will create and initialize a sequence_item or sequence using the factory.  The sequence_item or sequence will be initialized to communicate with the specified sequencer.

start_item and finish_item together will initiate operation of a sequence item.  If the item has not already been initialized using create_item, then it will be initialized here to use the default sequencer specified by m_sequencer.  Randomization may be done between start_item and finish_item to ensure late generation

finish_item, together with start_item together will initiate operation of a sequence_item.  Finish_item must be called after start_item with no delays or delta-cycles.  Randomization, or other functions may be called between the start_item and finish_item calls.

This task issues a request to the current sequencer.  If item_priority is not specified, then the current sequence priority will be used by the arbiter.  If a lock_request is made, then the sequencer will issue a lock immediately before granting the sequence.  (Note that the lock may be granted without the sequence being granted if is_relevant is not asserted).

When this method returns, the sequencer has granted the sequence, and the sequence must call send_request without inserting any simulation delay other than delta cycles.  The driver is currently waiting for the next item to be sent via the send_request call.

The send_request function may only be called after a wait_for_grant call.  This call will send the request item to the sequencer, which will forward it to the driver.  If the rerandomize bit is set, the item will be randomized before being sent to the driver.

A sequence may optionally call wait_for_item_done.  This task will block until the driver calls item_done or put.  If no transaction_id parameter is specified, then the call will return the next time that the driver calls item_done or put.  If a specific transaction_id is specified, then the call will return when the driver indicates completion of that specific item.

Note that if a specific transaction_id has been specified, and the driver has already issued an item_done or put for that transaction, then the call will hang, having missed the earlier notification.

When called with enable set to 1, responses will be sent to the response handler.  Otherwise, responses must be retrieved using get_response.

By default, responses from the driver are retrieved in the sequence by calling get_response.

An alternative method is for the sequencer to call the response_handler function with each response.

Returns the state of the use_response_handler bit.

When the use_response_handler bit is set to 1, this virtual task is called by the sequencer for each response that arrives for this sequence.

By default, if the response_queue overflows, an error is reported.  The response_queue will overflow if more responses are sent to this sequence from the driver than get_response calls are made.  Setting value to 0 disables these errors, while setting it to 1 enables them.

When this bit is 0 (default value), error reports are generated when the response queue overflows.  When this bit is 1, no such error reports are generated.

The default maximum depth of the response queue is 8.  These method is used to examine or change the maximum depth of the response queue.

Setting the response_queue_depth to -1 indicates an arbitrarily deep response queue.  No checking is done.

Returns the current depth setting for the response queue.

Empties the response queue for this sequence.