Skip to content

Bug/red5 dev 2156#438

Merged
mondain merged 2 commits intomainfrom
bug/RED5DEV-2156
May 1, 2026
Merged

Bug/red5 dev 2156#438
mondain merged 2 commits intomainfrom
bug/RED5DEV-2156

Conversation

@mondain
Copy link
Copy Markdown
Member

@mondain mondain commented May 1, 2026

Pull Request

This PR fixes #Enter issue number

Changes proposed in this pull request:

mondain added 2 commits April 30, 2026 18:40
…stamps

When PlayEngine pushes a buffered keyframe (or any prelude event)
captured before the live anchor (videoBaseTs / audioBaseTs / publisher
timestamp offset), the rebase formula eventTime - baseTs + 2 produces
a negative result. The negative int writes to the wire as the equivalent
unsigned-32 value, which RTMP clients render as a far-future timestamp
(e.g. -97 ms -> 0xFFFFFF9F -> ~24.85 days when bit 31 is stripped, or
~49.7 days when read as full unsigned-32).

Clamp negative adjusted timestamps to 1 so buffered events precede the
first live frame at ts=2 without leaking into the unsigned-32 upper
half. Verified via ffplay end-to-end against an SRT/TS publisher: video
start went from 2147483.551s to 0.001s.
@mondain mondain merged commit acb191e into main May 1, 2026
@mondain mondain deleted the bug/RED5DEV-2156 branch May 1, 2026 01:43
@sonarqubecloud
Copy link
Copy Markdown

sonarqubecloud Bot commented May 1, 2026

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant