Struct TraceEventQueue 

pub struct TraceEventQueue {
    pub async_id_read: Id,
    pub async_id_write: Id,
    pub cpu_id: u32,
    /* private fields */
}

Stores all trace events.

Fields

async_id_read: Id

Async ID for read track grouping.

async_id_write: Id

Async ID for write track grouping.

cpu_id: u32

CPU ID for trace tracks.

Implementations

impl<'a> TraceEventQueue

pub fn new(inspect_node: &Node, cpu_id: u32) -> Result<Self, Errno>

pub fn read_track_name(&self) -> Cow<'static, str>

pub fn write_track_name(&self) -> Cow<'static, str>

pub fn from(kernel: &Kernel) -> Arc<Self>

Uses the TraceEventQueue from the kernel or initializes a new one if not present.

pub fn is_enabled(&self) -> bool

pub fn enable(&self) -> Result<(), Errno>

pub fn disable(&self) -> Result<(), Errno>

Disables the event queue and resets it to empty. The number of dropped pages are recorded for reading via tracefs.

pub fn read(&self, buf: &mut dyn OutputBuffer) -> Result<usize, Errno>

Reads a page worth of events. Currently only reads pages that are full.

From https://docs.kernel.org/trace/ring-buffer-design.html, when memory is mapped, a reader page can be swapped with the header page to avoid copying memory.

pub fn push_event( &self, event: TraceEvent, data: &[u8], ) -> Result<Duration<BootTimeline>, Errno>

Write event into ring_buffer. If event does not fit in the current page, move on to the next.

Should eventually allow for a writer to preempt another writer. See https://docs.kernel.org/trace/ring-buffer-design.html. Returns the delta duration between this event and the previous event written.