refactor(pf-encode): extract the range-family RFI recovery policy (WP7.2)
The two direct-NVENC backends carried hand-copied twins of the same loss-recovery decision: range validity, covering-range dedup, DPB window, clamp — ~30 duplicated lines each. The decision now lives once as nvenc_core::plan_range_recovery (the range half of WP7.2; the slot half is enc/rfi.rs), pure and unit-tested; each backend keeps its session gate, its unsafe per-timestamp driver loop, and its state stores. The step order is load-bearing and now pinned by tests: the covering dedup runs with the UNCLAMPED last and BEFORE the DPB window (a covered re-ask never touches the driver even when the range has since aged out of the DPB), the boundary at next_ts - RFI_DPB is inclusive, and the Invalidate carries the CLAMPED last — which is also what the caller records in last_rfi_range, exactly as the inline code stored it. A driver failure mid-loop still returns false with NO range recorded and no anchor armed. Decline deliberately clears nothing (neither twin touched pending_anchor on decline — same shape as Vulkan's non-clear, opposite of AMF/QSV; do not harmonize). The exact-cover → Covered test records EXISTING behavior including that a covered range survives a forced IDR with zero driver calls — a recorded fact, not an endorsement. RFI_DPB's import leaves both twins: its only per-backend use was the arithmetic that moved. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
@@ -61,9 +61,9 @@
|
||||
#![deny(clippy::undocumented_unsafe_blocks)]
|
||||
|
||||
use super::nvenc_core::{
|
||||
apply_low_latency_config, build_init_params, cached_ceiling, codec_guid, resolve_slices,
|
||||
resolve_split_mode, resolve_subframe, store_ceiling, CeilingKey, LowLatencyConfig, NvStatusExt,
|
||||
RFI_DPB,
|
||||
apply_low_latency_config, build_init_params, cached_ceiling, codec_guid, plan_range_recovery,
|
||||
resolve_slices, resolve_split_mode, resolve_subframe, store_ceiling, CeilingKey,
|
||||
LowLatencyConfig, NvStatusExt, RangePlan,
|
||||
};
|
||||
use super::nvenc_status;
|
||||
use super::{AuChunk, ChromaFormat, Codec, EncodedFrame, Encoder, EncoderCaps};
|
||||
@@ -1847,48 +1847,46 @@ impl Encoder for NvencCudaEncoder {
|
||||
}
|
||||
|
||||
fn invalidate_ref_frames(&mut self, first: i64, last: i64) -> bool {
|
||||
if self.encoder.is_null() || !self.rfi_supported || first < 0 || first > last {
|
||||
// Range validity, covering-range dedup, DPB window and clamp all live in
|
||||
// `nvenc_core::plan_range_recovery` — one policy for both direct-NVENC backends; only the
|
||||
// session gate and the driver loop are this backend's.
|
||||
if self.encoder.is_null() || !self.rfi_supported {
|
||||
return false;
|
||||
}
|
||||
// Already invalidated a covering range for this loss event — re-arm the anchor (the previous
|
||||
// anchor AU may itself have been lost) but skip the driver calls.
|
||||
if let Some((pf, pl)) = self.last_rfi_range {
|
||||
if first >= pf && last <= pl {
|
||||
match plan_range_recovery(first, last, self.frame_idx, self.last_rfi_range) {
|
||||
// Already invalidated a covering range for this loss event — re-arm the anchor (the
|
||||
// previous anchor AU may itself have been lost) but skip the driver calls.
|
||||
RangePlan::Covered => {
|
||||
self.pending_anchor = true;
|
||||
return true;
|
||||
true
|
||||
}
|
||||
}
|
||||
// The DPB holds `[frame_idx - RFI_DPB, frame_idx - 1]`; a lost frame older than that can't be
|
||||
// invalidated, so the only correct recovery is an IDR.
|
||||
let oldest_in_dpb = self.frame_idx - RFI_DPB as i64;
|
||||
if first < oldest_in_dpb {
|
||||
return false;
|
||||
}
|
||||
let last = last.min(self.frame_idx - 1);
|
||||
if first > last {
|
||||
return false;
|
||||
}
|
||||
// Each input's `inputTimeStamp` is the WIRE frame index (pinned by `submit_indexed`), so the
|
||||
// client's lost-frame range maps 1:1 onto the timestamps NVENC invalidates here.
|
||||
// SAFETY: `invalidate_ref_frames` is a function pointer from the runtime table; `self.encoder`
|
||||
// was checked non-null and is the live session; this runs on the encode thread (no concurrent
|
||||
// NVENC use). Each `ts` is clamped to `[oldest_in_dpb, frame_idx - 1]`, naming a frame still
|
||||
// in the DPB; the call passes only that `u64` (no struct).
|
||||
unsafe {
|
||||
for ts in first..=last {
|
||||
if (api().invalidate_ref_frames)(self.encoder, ts as u64)
|
||||
.nv_ok()
|
||||
.is_err()
|
||||
{
|
||||
return false;
|
||||
RangePlan::Decline => false,
|
||||
RangePlan::Invalidate { first, last } => {
|
||||
// Each input's `inputTimeStamp` is the WIRE frame index (pinned by
|
||||
// `submit_indexed`), so the client's lost-frame range maps 1:1 onto the timestamps
|
||||
// NVENC invalidates here.
|
||||
// SAFETY: `invalidate_ref_frames` is a function pointer from the runtime table;
|
||||
// `self.encoder` was checked non-null and is the live session; this runs on the
|
||||
// encode thread (no concurrent NVENC use). The plan clamped each `ts` to
|
||||
// `[oldest_in_dpb, frame_idx - 1]`, naming a frame still in the DPB; the call
|
||||
// passes only that `u64` (no struct).
|
||||
unsafe {
|
||||
for ts in first..=last {
|
||||
if (api().invalidate_ref_frames)(self.encoder, ts as u64)
|
||||
.nv_ok()
|
||||
.is_err()
|
||||
{
|
||||
return false;
|
||||
}
|
||||
}
|
||||
}
|
||||
self.last_rfi_range = Some((first, last));
|
||||
// The next submitted frame is the clean re-anchor — arm the tag so its AU ships
|
||||
// with `recovery_anchor` and the client lifts its post-loss freeze on it.
|
||||
self.pending_anchor = true;
|
||||
true
|
||||
}
|
||||
}
|
||||
self.last_rfi_range = Some((first, last));
|
||||
// The next submitted frame is the clean re-anchor — arm the tag so its AU ships with
|
||||
// `recovery_anchor` and the client lifts its post-loss freeze on it.
|
||||
self.pending_anchor = true;
|
||||
true
|
||||
}
|
||||
|
||||
fn poll(&mut self) -> Result<Option<EncodedFrame>> {
|
||||
|
||||
@@ -294,10 +294,158 @@ mod tests {
|
||||
|
||||
/// Reference-frame DPB depth when RFI is supported (Apollo uses 5). A deeper DPB lets an invalidated
|
||||
/// reference fall back to an older still-valid frame instead of a full IDR; `numRefL0 = 1` keeps each
|
||||
/// P-frame single-reference for low latency. Also the window the backends' `invalidate_ref_frames`
|
||||
/// paths check against (`frame_idx - RFI_DPB` = the oldest frame still in the DPB).
|
||||
/// P-frame single-reference for low latency. Also the window [`plan_range_recovery`] checks against
|
||||
/// (`next_ts - RFI_DPB` = the oldest frame still in the DPB).
|
||||
pub(super) const RFI_DPB: u32 = 5;
|
||||
|
||||
/// One loss event's recovery decision for the timestamp-range RFI both direct-NVENC backends run
|
||||
/// (the range half of WP7.2's policy extraction; the slot half — AMF/QSV/Vulkan — is
|
||||
/// `crate::rfi`). The mechanism (the per-timestamp `nvEncInvalidateRefFrames` loop, the
|
||||
/// `last_rfi_range`/`pending_anchor` stores, the null-handle/`rfi_supported` gate) stays in each
|
||||
/// backend.
|
||||
pub(super) enum RangePlan {
|
||||
/// The last successful invalidation already covers this range — no new driver calls, no IDR.
|
||||
/// The caller must still RE-ARM its recovery anchor: the client re-asking means the previous
|
||||
/// anchor AU may itself have been lost, and the next frame is just as clean a re-anchor.
|
||||
Covered,
|
||||
/// Invalidate `first..=last` (the CLAMPED range — this is also what the caller must record in
|
||||
/// `last_rfi_range` on success, exactly as the inline code stored the post-clamp values).
|
||||
Invalidate { first: i64, last: i64 },
|
||||
/// Recovery without an IDR is impossible (nonsense range, loss older than the DPB, or a range
|
||||
/// entirely in the future) — the caller returns `false` and its (coalesced) keyframe path
|
||||
/// recovers. Deliberately NOT paired with any state clearing: neither twin touches
|
||||
/// `pending_anchor` on decline (matching Vulkan's decline, opposite of AMF/QSV's
|
||||
/// `pending_force` clear — see `crate::rfi`'s module doc before "harmonizing").
|
||||
Decline,
|
||||
}
|
||||
|
||||
/// The range-RFI policy, extracted verbatim from the two backends' `invalidate_ref_frames` (they
|
||||
/// were hand-copied twins). Step order is load-bearing and pinned by tests:
|
||||
///
|
||||
/// 1. nonsense range (`first < 0 || first > last`) → [`RangePlan::Decline`];
|
||||
/// 2. covering-range dedup — checked with the UNCLAMPED `last`, BEFORE the DPB window, so a
|
||||
/// covered re-ask never touches the driver even when the range has since left the DPB;
|
||||
/// 3. DPB window: `first < next_ts - RFI_DPB` → Decline (a lost frame older than the DPB cannot
|
||||
/// be invalidated; the only correct recovery is an IDR);
|
||||
/// 4. clamp `last` to `next_ts - 1` (never invalidate a timestamp never assigned); an inverted
|
||||
/// range after the clamp (loss entirely in the future — a prediction desync) → Decline.
|
||||
///
|
||||
/// `next_ts` is the backend's `frame_idx`: the NEXT timestamp to assign, which `submit_indexed`
|
||||
/// pins to the wire frame index — so the client's lost-frame range maps 1:1 onto the timestamps
|
||||
/// the driver invalidates, across every rebuild/reset. Note `teardown()` clears `last_rfi_range`
|
||||
/// but NOT `frame_idx`, so a post-reset call legitimately sees a stale-high `next_ts` with a
|
||||
/// `None` range — the same view the inline code had.
|
||||
pub(super) fn plan_range_recovery(
|
||||
first: i64,
|
||||
last: i64,
|
||||
next_ts: i64,
|
||||
last_rfi_range: Option<(i64, i64)>,
|
||||
) -> RangePlan {
|
||||
if first < 0 || first > last {
|
||||
return RangePlan::Decline;
|
||||
}
|
||||
if let Some((pf, pl)) = last_rfi_range {
|
||||
if first >= pf && last <= pl {
|
||||
return RangePlan::Covered;
|
||||
}
|
||||
}
|
||||
let oldest_in_dpb = next_ts - RFI_DPB as i64;
|
||||
if first < oldest_in_dpb {
|
||||
return RangePlan::Decline;
|
||||
}
|
||||
let last = last.min(next_ts - 1);
|
||||
if first > last {
|
||||
return RangePlan::Decline;
|
||||
}
|
||||
RangePlan::Invalidate { first, last }
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod range_policy_tests {
|
||||
use super::{plan_range_recovery, RangePlan, RFI_DPB};
|
||||
|
||||
/// Convenience: the plan with no prior invalidation recorded.
|
||||
fn plan(first: i64, last: i64, next_ts: i64) -> RangePlan {
|
||||
plan_range_recovery(first, last, next_ts, None)
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn nonsense_ranges_decline() {
|
||||
assert!(matches!(plan(-1, 5, 100), RangePlan::Decline));
|
||||
assert!(matches!(plan(7, 5, 100), RangePlan::Decline));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn covering_range_dedups_partial_overlap_does_not() {
|
||||
let prior = Some((90i64, 95i64));
|
||||
// Exact cover and sub-range → Covered. This pins EXISTING behavior, including that a
|
||||
// covered range survives a forced IDR with zero driver calls (nothing clears
|
||||
// `last_rfi_range` on a keyframe) — a recorded fact, not an endorsement.
|
||||
assert!(matches!(
|
||||
plan_range_recovery(90, 95, 100, prior),
|
||||
RangePlan::Covered
|
||||
));
|
||||
assert!(matches!(
|
||||
plan_range_recovery(92, 94, 100, prior),
|
||||
RangePlan::Covered
|
||||
));
|
||||
// Partial overlap re-invalidates the FULL new range (next_ts = 98 keeps the window open:
|
||||
// oldest_in_dpb = 93; at next_ts = 100 the same range would age out and Decline instead).
|
||||
assert!(matches!(
|
||||
plan_range_recovery(93, 97, 98, prior),
|
||||
RangePlan::Invalidate {
|
||||
first: 93,
|
||||
last: 97
|
||||
}
|
||||
));
|
||||
}
|
||||
|
||||
/// The covering check runs BEFORE the DPB window: a covered re-ask stays Covered (no driver
|
||||
/// calls needed) even when the range has since aged out of the DPB.
|
||||
#[test]
|
||||
fn covered_is_checked_before_the_dpb_window() {
|
||||
let prior = Some((10i64, 12i64));
|
||||
assert!(matches!(
|
||||
plan_range_recovery(10, 12, 100, prior),
|
||||
RangePlan::Covered
|
||||
));
|
||||
// ...whereas the same range with no prior invalidation is outside the window → Decline.
|
||||
assert!(matches!(plan(10, 12, 100), RangePlan::Decline));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn dpb_window_boundary() {
|
||||
let next_ts = 100i64;
|
||||
let oldest = next_ts - RFI_DPB as i64; // 95: the oldest timestamp still in the DPB
|
||||
assert!(matches!(
|
||||
plan(oldest, oldest, next_ts),
|
||||
RangePlan::Invalidate { .. }
|
||||
));
|
||||
assert!(matches!(
|
||||
plan(oldest - 1, oldest, next_ts),
|
||||
RangePlan::Decline
|
||||
));
|
||||
}
|
||||
|
||||
/// `last` clamps to `next_ts - 1` (the newest encoded frame); the Invalidate carries the
|
||||
/// CLAMPED value — which is also what the caller records in `last_rfi_range`.
|
||||
#[test]
|
||||
fn clamps_to_newest_encoded() {
|
||||
assert!(matches!(
|
||||
plan(98, 150, 100),
|
||||
RangePlan::Invalidate {
|
||||
first: 98,
|
||||
last: 99
|
||||
}
|
||||
));
|
||||
// A range entirely in the future inverts under the clamp → Decline (prediction desync).
|
||||
assert!(matches!(plan(100, 150, 100), RangePlan::Decline));
|
||||
// Fresh session (`frame_idx == 0`): window passes (oldest = -5) but the clamp gives
|
||||
// last = -1 < first → Decline. The inline code behaved identically.
|
||||
assert!(matches!(plan(0, 3, 0), RangePlan::Decline));
|
||||
}
|
||||
}
|
||||
|
||||
/// The per-session knobs both direct-NVENC backends feed [`apply_low_latency_config`]. `Copy` so the
|
||||
/// backend fills it from `self` at the call. The two input-format fields bridge the only real
|
||||
/// divergence between the CUDA and D3D11 paths (which surface formats can carry full chroma / 10-bit
|
||||
|
||||
@@ -37,9 +37,9 @@
|
||||
#![deny(clippy::undocumented_unsafe_blocks)]
|
||||
|
||||
use super::nvenc_core::{
|
||||
apply_low_latency_config, build_init_params, cached_ceiling, codec_guid, resolve_slices,
|
||||
resolve_split_mode, resolve_subframe, store_ceiling, CeilingKey, LowLatencyConfig, NvStatusExt,
|
||||
RFI_DPB,
|
||||
apply_low_latency_config, build_init_params, cached_ceiling, codec_guid, plan_range_recovery,
|
||||
resolve_slices, resolve_split_mode, resolve_subframe, store_ceiling, CeilingKey,
|
||||
LowLatencyConfig, NvStatusExt, RangePlan,
|
||||
};
|
||||
use super::nvenc_status;
|
||||
use super::{ChromaFormat, Codec, EncodedFrame, Encoder, EncoderCaps};
|
||||
@@ -1679,61 +1679,56 @@ impl Encoder for NvencD3d11Encoder {
|
||||
}
|
||||
|
||||
fn invalidate_ref_frames(&mut self, first: i64, last: i64) -> bool {
|
||||
// No live session, the GPU can't invalidate, or a nonsense range → caller forces a full IDR.
|
||||
// (NVENC handles are single-threaded; this runs on the encode thread, like submit/poll.)
|
||||
if self.encoder.is_null() || !self.rfi_supported || first < 0 || first > last {
|
||||
// No live session or the GPU can't invalidate → caller forces a full IDR. (NVENC handles
|
||||
// are single-threaded; this runs on the encode thread, like submit/poll.) Everything else
|
||||
// — range validity, covering-range dedup, the DPB window, the clamp — is
|
||||
// `nvenc_core::plan_range_recovery`, one policy for both direct-NVENC backends.
|
||||
if self.encoder.is_null() || !self.rfi_supported {
|
||||
return false;
|
||||
}
|
||||
// Already invalidated a covering range for this loss event — no new driver calls needed,
|
||||
// no IDR. RE-ARM the anchor though: the client re-asking means the previous recovery
|
||||
// anchor AU may itself have been lost, and the next frame is just as clean a re-anchor
|
||||
// (it too references only valid frames).
|
||||
if let Some((pf, pl)) = self.last_rfi_range {
|
||||
if first >= pf && last <= pl {
|
||||
match plan_range_recovery(first, last, self.frame_idx, self.last_rfi_range) {
|
||||
// Already invalidated a covering range for this loss event — no new driver calls
|
||||
// needed, no IDR. RE-ARM the anchor though: the client re-asking means the previous
|
||||
// recovery anchor AU may itself have been lost, and the next frame is just as clean a
|
||||
// re-anchor (it too references only valid frames).
|
||||
RangePlan::Covered => {
|
||||
self.pending_anchor = true;
|
||||
return true;
|
||||
true
|
||||
}
|
||||
}
|
||||
// `frame_idx` is the NEXT timestamp to assign, so the last encoded frame is `frame_idx - 1`
|
||||
// and the DPB holds `[frame_idx - RFI_DPB, frame_idx - 1]`. A lost frame older than that
|
||||
// can't be invalidated, so the only correct recovery is an IDR.
|
||||
let oldest_in_dpb = self.frame_idx - RFI_DPB as i64;
|
||||
if first < oldest_in_dpb {
|
||||
return false;
|
||||
}
|
||||
// Clamp to frames we've actually encoded (don't invalidate a timestamp we never assigned).
|
||||
let last = last.min(self.frame_idx - 1);
|
||||
if first > last {
|
||||
return false;
|
||||
}
|
||||
// Each input's `inputTimeStamp` is `frame_idx`, which `submit_indexed` pins to the WIRE
|
||||
// frame index the AU carries — so the client's lost-frame range maps 1:1 onto the
|
||||
// timestamps NVENC invalidates here, and stays 1:1 across encoder rebuilds/resets (an
|
||||
// internal counter would desync on the first adaptive-bitrate rebuild and RFI would then
|
||||
// clamp every range into first > last, silently degrading to IDR-only forever).
|
||||
// SAFETY: `invalidate_ref_frames` is a function pointer from the runtime-loaded `EncodeApi` table.
|
||||
// `self.encoder` was checked non-null at the top of this fn and is the live session; this runs
|
||||
// on the encode thread (like submit/poll), so there is no concurrent NVENC use. Each `ts` was
|
||||
// clamped to `[oldest_in_dpb, frame_idx - 1]` above, so it names a frame still in the session's
|
||||
// DPB; the call passes only that `u64` timestamp (no struct), so there is no struct-size or
|
||||
// lifetime concern.
|
||||
unsafe {
|
||||
for ts in first..=last {
|
||||
if (api().invalidate_ref_frames)(self.encoder, ts as u64)
|
||||
.nv_ok()
|
||||
.is_err()
|
||||
{
|
||||
return false; // any failure → fall back to IDR
|
||||
RangePlan::Decline => false,
|
||||
RangePlan::Invalidate { first, last } => {
|
||||
// Each input's `inputTimeStamp` is `frame_idx`, which `submit_indexed` pins to the
|
||||
// WIRE frame index the AU carries — so the client's lost-frame range maps 1:1 onto
|
||||
// the timestamps NVENC invalidates here, and stays 1:1 across encoder
|
||||
// rebuilds/resets (an internal counter would desync on the first adaptive-bitrate
|
||||
// rebuild and RFI would then clamp every range into first > last, silently
|
||||
// degrading to IDR-only forever).
|
||||
// SAFETY: `invalidate_ref_frames` is a function pointer from the runtime-loaded
|
||||
// `EncodeApi` table. `self.encoder` was checked non-null at the top of this fn and
|
||||
// is the live session; this runs on the encode thread (like submit/poll), so there
|
||||
// is no concurrent NVENC use. The plan clamped each `ts` to
|
||||
// `[oldest_in_dpb, frame_idx - 1]`, so it names a frame still in the session's
|
||||
// DPB; the call passes only that `u64` timestamp (no struct), so there is no
|
||||
// struct-size or lifetime concern.
|
||||
unsafe {
|
||||
for ts in first..=last {
|
||||
if (api().invalidate_ref_frames)(self.encoder, ts as u64)
|
||||
.nv_ok()
|
||||
.is_err()
|
||||
{
|
||||
return false; // any failure → fall back to IDR
|
||||
}
|
||||
}
|
||||
}
|
||||
self.last_rfi_range = Some((first, last));
|
||||
// The next submitted frame is the first one encoded after the invalidation — the
|
||||
// clean re-anchor P-frame. Arm the tag so its AU ships with `recovery_anchor` and
|
||||
// the client lifts its post-loss freeze on it (instead of waiting ~1 s for the
|
||||
// cooldown-suppressed IDR fallback).
|
||||
self.pending_anchor = true;
|
||||
true
|
||||
}
|
||||
}
|
||||
self.last_rfi_range = Some((first, last));
|
||||
// The next submitted frame is the first one encoded after the invalidation — the clean
|
||||
// re-anchor P-frame. Arm the tag so its AU ships with `recovery_anchor` and the client
|
||||
// lifts its post-loss freeze on it (instead of waiting ~1 s for the cooldown-suppressed
|
||||
// IDR fallback).
|
||||
self.pending_anchor = true;
|
||||
true
|
||||
}
|
||||
|
||||
fn poll(&mut self) -> Result<Option<EncodedFrame>> {
|
||||
|
||||
Reference in New Issue
Block a user