Follow-ups left open by the 8822B DFIR and J3 runtime-thread reviews (#497, #498)

Ouverte
#499 11 commentaires 0 réactions 0 personnes assignées Voir sur GitHub

Les mainteneurs répondent en général sous 1 jour

Personne n'a encore pris cette issue.

Évaluation

Difficulté
5/5
Temps estimé
Plus d'une semaine
Accessibilité débutants
35/100
Type d'issue
Bug
Clarté
Plutôt claire
Activité
Active
Stack technique
c, linux

Piste de recherche

Start with the review findings in issues #497 and #498 and the Jaguar3 notes in src/jaguar3/CLAUDE.md; the payload names no test files or commands. Reproduce the stated transmitter and receiver setups, then investigate each of the four open measurements: runtime-thread duration and chip variants, MCS7 EVM and distance, 2.4 GHz HT40 qualification, and Jaguar1 wide-bandwidth RX. Record results and update the relevant documentation or qualification findings.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Description

enhancement

Things the #497 and #498 reviews measured but did not settle. Bench for all of them: an 8812EU (Jaguar3, 0bda:a81a) transmitting and an 8812BU (8822B cut 3, CF-924AC 0bda:b812) receiving.

1. Does Jaguar3 TX go silent with the runtime thread off, given longer runs or the 8822C? (#498)

src/jaguar3/CLAUDE.md says sustained 5 GHz TX needs the coex runtime thread; without it the coex firmware silences the antenna. On the 8822E I could not reproduce that within 22 s. TX-only, ch36 MCS1/20, 10 000 frames:

rep 1 rep 2
thread on (default) 99.4 % 99.2 %
DEVOURER_J3_RUNTIME=0 99.5 % 99.4 %

Not tested: runs of minutes rather than seconds, the 8822C (8812CU/8822CU), and the upper 5 GHz channels, where pwr_track matters. If the silencing does not reproduce, the CLAUDE.md claim should be narrowed. If it does, the j3_runtime_thread docs should say after how long and on which die.

2. Is MCS7 on this pair really worse at 20 MHz than at 40 MHz?

Same pair, same channel (ch36), same session, 1500 frames per cell:

  • MCS7/20: 34.0 / 63.1 / 54.1 / 43.7 % across four runs, all on byte-identical 20 MHz code paths.
  • MCS7/40: 95.5–98.7 % (after #497).

The 20 MHz spread is far above the usual ±3-point probe noise, and the 20 MHz result is worse than the 40 MHz one, which is backwards. One unexplored lead: in the earlier runs at this distance the rx.txhit RSSI read 102–104, i.e. near-field. At 20 MHz the same TX power is concentrated in half the bandwidth, so this may be receiver saturation, which would show as poor EVM rather than poor RSSI. Check EVM and repeat at more distance (or lower TX power) before believing the number.

3. Re-qualify the 8822B b812 for HT40 on 2.4 GHz

After #497 the b812 is flat up to MCS7 on ch36 HT40 (88.7 / 89.9 / 89.2 / 89.0 / 93.5 / 96.6 % across MCS1/3/4/5/6/7). It was already qualified at 20 MHz. ch6 HT40 is still unmeasured: the 8812EU module is wired with 5 GHz antennas and is not a usable 2.4 GHz transmitter (ch6, 20 MHz: 6M 29.5 %, MCS1 0 %). Repeat the ladder with a transmitter that works on 2.4 GHz.

4. Does Jaguar1 have the same wide-RX hole?

Before #497, rxdemo with DEVOURER_BW=40 delivered nothing on Jaguar2. The 8822B half of that was its RX DFIR being left in the 20 MHz setting (0 % → ~96 % at VHT80). The 8821C path already switched the DFIR by width. Jaguar1 (8812AU/8814AU/8821AU) has not been checked. Run the same A/B: an 8812EU sending HT40/VHT80 on ch36 into an 8812AU at DEVOURER_BW=40/80.

Langage dominant
C
Étoiles
56
Forks
29
Merge moyen
3 h 43 min
PR mergées (30 j)
75

Préparer son environnement

Ce projet ne fournit ni conteneur de développement, ni Dockerfile, ni guide de contribution : l'installation est à votre charge. Commencez par son README, et consultez notre guide de la première contribution pour les étapes générales.

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de OpenIPC/devourer

Toutes les issues de OpenIPC/devourer

Issues similaires

Plus d'issues C

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.