Chip STBC TX diversity + txdemo event-loop EINTR fix (#132)#144
Merged
Conversation
WiFiDriverTxDemo could only pick a TX adapter by first-match VID:PID, so two identical RTL8814AU dongles (same VID:PID:serial) couldn't be told apart on the TX side — the RX demo already had this (demo/main.cpp). Needed to TX-select one of several identical adapters, e.g. for the #132 chip-STBC TX-diversity measurement. Same semantics: DEVOURER_USB_BUS + optional DEVOURER_USB_PORT (dotted libusb port path); unset falls back to the VID:PID open loop. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
tests/stbc_sanity.sh: TX canonical beacons at MCS1/STBC from an 8814 (chip encodes Alamouti internally), RX on an 8812 (whose descriptor exposes the received STBC bit, unlike the 8814 RX), and report whether frames arrive with stbc=1. PASS = chip emits decodable STBC; FAIL(stbc=0) = descriptor STBC ignored (would need CSD); INCONCLUSIVE = link too weak. The gate that decides whether the #132 chip-STBC TX-diversity measurement can proceed. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bare `wait` blocked on the still-running TX beacon, so the script never printed its verdict. Capture the RX pid and wait on it alone.
…STBC toggle Two changes from the #132 chip-STBC TX-diversity work: 1. FIX: usb_event_loop() broke out of the loop on any non-TIMEOUT return from libusb_handle_events_timeout_completed, including the transient LIBUSB_ERROR_INTERRUPTED (EINTR — a signal hitting the underlying poll, more likely under concurrent USB load). Exiting that thread permanently stops servicing async TX-completion callbacks, so submitted URBs are never freed and within a few frames every libusb_submit_transfer fails — a "TX works briefly then every send fails" wedge, seen intermittently while an RX ran concurrently. Now INTERRUPTED is ignored like TIMEOUT and other errors are logged (rate- limited) without breaking. This is a provable-by-inspection bug matching the observed symptom; it could not be verified against the intermittent failure because that no longer reproduces. 2. FEATURE: DEVOURER_TX_STBC_TOGGLE=1 alternates the STBC bit every frame (rate from DEVOURER_TX_RATE, must be HT/VHT). The RX reports the received STBC bit per frame, so an alternating TX lets one moving capture compare STBC vs single-stream delivery on the same fading — the transmit-side mobility measurement, tagged for free by the receiver. Verified: RX sees ~50/50 stbc=0/stbc=1, alternating per sequence number. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
josephnef
added
enhancement
Closed
10 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Transmit-side spatial diversity (chip STBC), the TX dual of the #130 RX-MRC
result — plus a real TX bug fix uncovered along the way. Scoped to #132.
The bug fix (worth merging on its own)
txdemo's async-TX completion event loop broke out permanently on anynon-timeout return from
libusb_handle_events_timeout_completed, including thetransient
LIBUSB_ERROR_INTERRUPTED(EINTR). Once that thread exits, TXcompletion callbacks stop firing, submitted URBs are never freed, and within a
few frames every
libusb_submit_transferfails — a "TX works briefly then everysend fails" wedge, seen intermittently under concurrent-RX load. Now EINTR is
ignored like TIMEOUT and other errors are logged (rate-limited) without breaking.
(Provable-by-inspection bug matching the symptom; couldn't be verified against
the intermittent failure because it stopped reproducing.)
The measurement enablers
txdemo:DEVOURER_USB_BUS/PORTtopology selector (mirror of the RX demo —lets you TX-select one of several identical adapters).
txdemo:DEVOURER_TX_STBC_TOGGLE=1alternates the STBC bit every frame; theRX reports the received STBC bit per frame, so one moving capture compares STBC
vs single-stream on the same fading, tagged for free.
tests/stbc_sanity.sh: confirms the chip emits decodable STBC on-air (TX 8814MCS1/STBC→ RX 8812 reads the STBC bit).The result (see #132)
Chip STBC TX works on-air (sanity gate: 6336/6336
stbc=1). At a marginal rate(MCS5), alternating STBC vs single-stream from a moving TX:
STBC's advantage grows under motion (coding gain static, + diversity moving), and
it helps even a single-antenna receiver because the win is TX-side — the
moving-drone FPV case.
Testing
EINTR fix); STBC toggle verified ~50/50 alternating by sequence number.
🤖 Generated with Claude Code