PR #22726 opened by u6bkep
URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/22726
Patch URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/22726.patch

The segment split logic in hlsenc compares `pkt->pts - vs->start_pts` without
normalizing time bases. When audio packets arrive before video, `start_pts` gets
stored in the audio stream's time base (e.g. 1/48000), but the comparison runs
against video packets in a different time base (e.g. 1/90000). The 
cross-timebase
subtraction produces a wildly inflated elapsed time, so every keyframe triggers 
a
segment split regardless of `-hls_time`.

The fix normalizes `start_pts` to AV_TIME_BASE_Q on assignment and converts
`pkt->pts` before comparison.

I have a stand alone reproducer script if reviewers want.

This may have been hitting users transcoding Blu-ray remuxes with video 
passthrough +
audio transcode (fmp4 segments, non-zero starting PTS):
- https://github.com/jellyfin/jellyfin/issues/14702
- https://github.com/jellyfin/jellyfin/issues/15656
I noticed it during testing for some tangentially related work I am doing on 
jellyfin.


From 45cb56c52030395ef40e480594e6c0eab9f018df Mon Sep 17 00:00:00 2001
From: Ben Kepner <[email protected]>
Date: Sun, 5 Apr 2026 16:53:06 -0400
Subject: [PATCH] avformat/hlsenc: fix segment duration with mixed stream time
 bases
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

When audio and video streams have different time bases (e.g. video at
1/90000 and audio at 1/48000), vs->start_pts was stored as a raw PTS
from whichever stream's packet arrived first. The segment split
comparison then subtracted this value from the current packet's PTS
without accounting for the time base difference, producing incorrect
elapsed time calculations.

This caused segments to be split at wrong points — either too
frequently (on every keyframe) or not at all, depending on the
relative magnitudes of the time bases.

Fix by normalizing vs->start_pts to AV_TIME_BASE_Q at the point of
assignment and converting pkt->pts to the same base before comparison.
This ensures the segment split decision is always unit-consistent
regardless of which stream's packet is being evaluated.

The bug is most easily triggered by HLS muxing with video passthrough
and audio transcode, where the video retains its container time base
while the audio encoder outputs in its native time base.

Signed-off-by: Ben Kepner <[email protected]>
---
 libavformat/hlsenc.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/libavformat/hlsenc.c b/libavformat/hlsenc.c
index 8eaab37a2c..db2ae2125d 100644
--- a/libavformat/hlsenc.c
+++ b/libavformat/hlsenc.c
@@ -2448,13 +2448,16 @@ static int hls_write_packet(AVFormatContext *s, 
AVPacket *pkt)
     }
 
     if (vs->start_pts == AV_NOPTS_VALUE) {
-        vs->start_pts = pkt->pts;
+        vs->start_pts = av_rescale_q(pkt->pts, st->time_base, AV_TIME_BASE_Q);
         if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
             vs->start_pts_from_audio = 1;
     }
-    if (vs->start_pts_from_audio && st->codecpar->codec_type == 
AVMEDIA_TYPE_VIDEO && vs->start_pts > pkt->pts) {
-        vs->start_pts = pkt->pts;
-        vs->start_pts_from_audio = 0;
+    if (vs->start_pts_from_audio && st->codecpar->codec_type == 
AVMEDIA_TYPE_VIDEO) {
+        int64_t video_start = av_rescale_q(pkt->pts, st->time_base, 
AV_TIME_BASE_Q);
+        if (vs->start_pts > video_start) {
+            vs->start_pts = video_start;
+            vs->start_pts_from_audio = 0;
+        }
     }
 
     if (vs->has_video) {
@@ -2485,8 +2488,8 @@ static int hls_write_packet(AVFormatContext *s, AVPacket 
*pkt)
     }
 
     can_split = can_split && (pkt->pts - vs->end_pts > 0);
-    if (vs->packets_written && can_split && av_compare_ts(pkt->pts - 
vs->start_pts, st->time_base,
-                                                          end_pts, 
AV_TIME_BASE_Q) >= 0) {
+    if (vs->packets_written && can_split && (av_rescale_q(pkt->pts, 
st->time_base, AV_TIME_BASE_Q) - vs->start_pts
+                                                          >= end_pts)) {
         int64_t new_start_pos;
         int byterange_mode = (hls->flags & HLS_SINGLE_FILE) || 
(hls->max_seg_size > 0);
         double cur_duration;
-- 
2.52.0

_______________________________________________
ffmpeg-devel mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to