You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Camera session: an unrequested bulk record (subtype 0x06) drops a thumbnail and keeps the session open for minutes #256
Seen on hardware on 2026-11-15 (S3 main board, TC100 F4_C1_56_06_ED_3B, main with #239 and #252), while testing #239.
What happened
A scan found two new recordings. The thumbnail of the first (dir 100 num 34) was stored. Then:
2035.06 s: GetThumbnail (cmdId 772) for the second one (num 33, a video). The camera starts the record on bulk subtype 0x04, 1 of 21 frames arrives.
2035.15 s: the camera sends a record on bulk subtype 0x06 (ver 9, typ 1, 25 payload bytes, one frame). handle_bulk_frame() has one reassembler for all subtypes and resets it for the new record; the thumbnail is lost.
The remaining thumbnail frames match no open record, are acknowledged as stray frames and dropped (logged on DEBUG only). check_thumbnail_stall() does not fire, because it only looks at a record on subtype 0x04.
Until 2278 s the camera sends 53 such records (typ counting up 1..53, 25 or 26 bytes, every 2 to 40 s). Each one counts as progress and restarts the idle timeout.
2338.6 s: No answer or data from the camera for 60 s; closing the camera session, ESP_ERR_TIMEOUT, 1 item(s) stored (session ended early).
The session kept the camera's Wi-Fi on for about five minutes; the video has no thumbnail in the catalog (flags 0).
Two defects
A record on another subtype destroys the thumbnail being received.
Nothing asks for the thumbnail again, and frames nobody asked for count as progress.
Open
What the camera reports on subtype 0x06 (the log did not show the content).
Why the records start after the second thumbnail request of the session and were not seen in earlier scans.
Activity