context.c

Include dependency graph for context.c:

Defines

IMPL_EXECUTE_LOOP

DEFAULT_STACK_SIZE 8

BYTES_PER_TERM (TERM_BITS / 8)

Functions

static struct Monitor *context_monitors_handle_terminate(Context *ctx)

static void destroy_extended_registers(Context *ctx, unsigned int live)

Context *context_new(GlobalContext *glb)

Creates a new context.

Allocates a new Context struct and initialize it. The newly created context is also inserted into the processes table, however it is not scheduled, allowing for further initialization.

Parameters:

• glb – The global context of this virtual machine instance.

Returns:

created context.

void context_destroy(Context *ctx)

Destroys a context.

Frees context resources and memory and removes it from the processes table. This should be called from the scheduler only. To actually delete a context that was created with context_new, use scheduler_terminate.

Parameters:

• c – the context that will be destroyed.

void context_process_kill_signal(Context *ctx, struct TermSignal *signal)

Process a kill signal, setting the exit reason and changing the killed flag.

Parameters:

• ctx – the context being executed

• signal – the kill message

void context_process_process_info_request_signal(Context *ctx, struct BuiltInAtomRequestSignal *signal)

Process a process info request signal.

Parameters:

• ctx – the context being executed

• signal – the kill message

bool context_process_signal_trap_answer(Context *ctx, struct TermSignal *signal)

Process a trap answer signal.

Parameters:

• ctx – the context being executed

• signal – the answer message

Returns:

true if successful, false in case of memory error

void context_process_flush_monitor_signal(Context *ctx, uint64_t ref_ticks, bool info)

Process a flush monitor signal.

Parameters:

• ctx – the context being executed

• ref_ticks – the monitor reference

• info – whether to return FALSE_ATOM if no message was flushed.

void context_update_flags(Context *ctx, int mask, int value)

Set or clear a flag on another context.

atomically update flags <- (flags & mask) | value

Parameters:

• ctx – the context to set/clear flag on.

• mask – the mask to apply on flags

• value – the value to set

size_t context_message_queue_len(Context *ctx)

Returns number of messages in the process’s mailbox.

Parameters:

• ctx – a valid context.

Returns:

the number of messages in the process’s mailbox

size_t context_size(Context *ctx)

Returns total amount of size (in byes) occupied by the process.

Parameters:

• ctx – a valid context.

Returns:

total amount of size (in byes) occupied by the process

bool context_get_process_info(Context *ctx, term *out, size_t *term_size, term atom_key, Heap *heap)

Get process information.

Parameters:

• ctx – the context being executed

• out – the answer term. Can be NULL if only the size matters.

• term_size – the size of the answer term, in words.

• atom_key – the key representing the info to get

• heap – the heap to allocate the answer to

Returns:

true if successful, false in case of an error in which case *out is filled with an exception atom if it was not NULL

struct Monitor *monitor_link_new(term link_pid)

Half-link process to another process.

Parameters:

• monitor_pid – process to link to

Returns:

the allocated monitor or NULL if allocation failed

struct Monitor *monitor_new(term monitor_pid, uint64_t ref_ticks, bool is_monitoring)

Create a monitor on a process.

Parameters:

• monitor_pid – monitoring process

• ref_ticks – reference of the monitor

• is_monitoring – if ctx is the monitoring process

Returns:

the allocated monitor or NULL if allocation failed

struct Monitor *monitor_resource_monitor_new(void *resource, uint64_t ref_ticks)

Create a resource monitor.

Parameters:

• resource – resource object

• ref_ticks – reference associated with the monitor

• process_id – process being monitored

Returns:

the allocated resource monitor or NULL if allocation failed

void context_add_monitor(Context *ctx, struct Monitor *new_monitor)

Add a monitor on a process.

Called within the process only. This function is called from MonitorSignal. Monitor is not added if it already exists. Monitors are identified by a reference, but links have no reference and a link can only exist once.

Parameters:

• ctx – the context being executed

• new_monitor – monitor object to add

void context_unlink(Context *ctx, term link_pid)

Half-unlink process to another process.

Called within the process only. For the other end of the link, an UnlinkSignal is sent that calls this function.

Parameters:

• ctx – the context being executed

• monitor_pid – process to unlink from

Returns:

0 on success

void context_demonitor(Context *ctx, uint64_t ref_ticks)

Destroy a monitor on a process.

Called within the process only. This function is called from DemonitorSignal.

Parameters:

• ctx – the context being executed

• ref_ticks – reference of the monitor to remove

Returns:

0 on success

term context_get_monitor_pid(Context *ctx, uint64_t ref_ticks, bool *is_monitoring)

Get target of a monitor.

Parameters:

• ctx – the context being executed

• ref_ticks – reference of the monitor to remove

• is_monitoring – whether ctx is the monitoring process.

Returns:

pid of monitoring process, self() if process is monitoring (and not monitored) or term_invalid() if no monitor could be found.