Struct TrainingConfig 

#[non_exhaustive]
pub struct TrainingConfig {
    pub ffe_levels: FfeLevels,
    pub flt_ready_polls: u32,
    pub ltp_polls: u32,
    pub frl_start_polls: u32,
    pub no_timeout_poll_cap: u32,
    pub max_retrains: u32,
    pub exit_to_tmds_on_error: bool,
}

Per-attempt training configuration.

Construct via TrainingConfig::default and override fields as needed. Poll limits are exact counts: N means exactly N polls before the state gives up. The defaults assume one poll every 2 ms.

One train call makes at most `flt_ready_polls + (n_rates + max_retrains) × ltp_polls

• (max_retrains + 1) × frl_start_pollspolls, each limit replaced byno_timeout_poll_capwhile the sink setsFLT_no_timeout: 850 to 1350 polls with the defaults for one to six rates, 4500 to 7000 under FLT_no_timeout`.

Fields (Non-exhaustive)

Non-exhaustive structs could have additional fields added in future. Therefore, non-exhaustive structs cannot be constructed in external crates using the traditional Struct { .. } syntax; cannot be matched against without a wildcard ..; and struct update syntax will not work.

ffe_levels: FfeLevels

Highest TxFFE level the source supports, written to Config_1 (limited per rate). Default 3, as the AMD and Intel drivers advertise: the sink can then request up to three TxFFE raises per lane. Set it to 0 for a PHY that cannot apply TxFFE, so the source does not advertise levels it cannot deliver; requests to raise a level are then not acted on.

flt_ready_polls: u32

Poll limit for FLT_ready in LTS:2. Default 50 (100 ms at 2 ms per poll).

ltp_polls: u32

Poll limit for LTS:3. Default 100 (200 ms at 2 ms per poll).

frl_start_polls: u32

Poll limit for FRL_start in LTS:P. Default 100 (200 ms at 2 ms per poll, the FRL_start wait of the AMD and Intel drivers).

no_timeout_poll_cap: u32

The LTS:2, LTS:3 and LTS:P poll limit while the sink sets FLT_no_timeout. When it runs out the attempt ends with TrainingOutcome::NoTimeoutHold, leaving the link as it is. Default 500 (1 s at 2 ms per poll, the AMD driver’s cap).

max_retrains: u32

How many times one train call returns from LTS:P to LTS:3 when the sink requests retraining (FLT_update). The next request after that ends the attempt with FallbackReason::RetrainsExhausted; 0 falls back on the first one. Default 3, the number of retries the AMD driver allows a failed procedure (it reruns the whole procedure rather than LTS:3).

Unlike the poll limits, this bound holds under FLT_no_timeout as well: it is what guarantees that the LTS:P ↔ LTS:3 cycle ends, so exhausting it falls back rather than ending in TrainingOutcome::NoTimeoutHold. A compliance test that needs more retrains raises it.

exit_to_tmds_on_error: bool

Whether an SCDC or PHY error is followed by LTS:L, returning both ends to TMDS as a fallback does. Default true. Turn it off to leave the sink and PHY as the error left them, for example to inspect the sink’s registers; TrainingError then reports TmdsExit::Skipped. While the sink sets FLT_no_timeout an error never leads to LTS:L (TmdsExit::HeldUnderTest).

Trait Implementations

impl Clone for TrainingConfig

fn clone(&self) -> TrainingConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for TrainingConfig

impl Debug for TrainingConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for TrainingConfig

fn default() -> TrainingConfig

Returns the “default value” for a type.

impl Eq for TrainingConfig

impl PartialEq for TrainingConfig

fn eq(&self, other: &TrainingConfig) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for TrainingConfig