From 0508df6fbbee2c956bfc2d92d89cb87994101ffc Mon Sep 17 00:00:00 2001 From: hathach Date: Sat, 29 Aug 2026 01:54:42 +0700 Subject: docs: point the pr3851 followup at the merged PR (#3866) commit --- docs/superpowers/followup/pr3851-audiod-short-fifo-read.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/superpowers/followup/pr3851-audiod-short-fifo-read.md b/docs/superpowers/followup/pr3851-audiod-short-fifo-read.md index a9bb86c42..cf7fa7ace 100644 --- a/docs/superpowers/followup/pr3851-audiod-short-fifo-read.md +++ b/docs/superpowers/followup/pr3851-audiod-short-fifo-read.md @@ -12,8 +12,8 @@ audio_test_freertos: "Audio mismatch at sample 8702: expected 8702, got 8696"). previous packet's data — the host hears a replayed fragment, offset by exactly one packet (6 samples at HS 48 kHz mono 16-bit: "expected 8702, got 8696"). - Reproduced whenever the producer lets the FIFO underrun; the example was fixed - to hold the FIFO at threshold (PR #3851, `20ff2b498`), but any application - with >~2 ms producer starvation can still hit the driver bug. + to hold the FIFO at threshold (this PR, #3866), but any application with + >~2 ms producer starvation can still hit the driver bug. - A naive short-write is not a fix: sending fewer bytes than the flow-control size is legal ISO behavior, but a 1-byte pad would byte-misalign the stream — the packet must shrink to the bytes actually read (sample-aligned). -- cgit v1.3.1