claude-noether
3a38286e6f
N=3 stress reps on ohm with v3 module (srcversion 371C6606B73AF19299228CA), 3 min @ 4 MB/s each, all within fresh-chip uptime window (200/391/582 s). | rep | MB/s | sdio_rx_work | bh_work redispatches | |----:|----:|-:|-:| | 1 | 2.363 | 0 | 0 | | 2 | 2.590 | 0 | 0 | | 3 | 2.102 | 0 | 0 | N=3 mean: 2.352 MB/s · median 2.363 MB/s · min 2.102 MB/s. vs Patch B baseline (1.362 MB/s, run-20260507-patchC-preflight): +73%. vs original Bug #5 floor (75 KB/s rep 3 death): 28× improvement. Plan §4.5 prediction verified: - sdio_rx_work dispatch rate: 86.4/s -> 0/s (function deleted) - bes2600_bh_work redispatches: 0 (preserved invariant) - observed receive @ 4 MB/s: floor lifts toward >= 1 MB/s (exceeded — floor is 2.10 MB/s) Bonus finding: sdio_tx_work dispatch rate dropped from 276.1/s to 0.8/s. The post-tx queue_work(rx_work) call I rewired to self->irq_handler() was actually firing more often than predicted; folding it into bh-wake-up cuts ~99.7% of the workqueue dispatches. No WARN/BUG/oops on any rep — the v2 race that wedged Patch C v1 within 13 s under stress did NOT reproduce on v3. Phase 8 lesson distilled as feedback_mine_upstream_ancestor memory: when patching a fork-from-upstream driver, mine the ancestor's fix history BEFORE writing fixes from scratch. cw1200 mining drove the structural pivot from v2's atomic_t wrapper to v3's no-relay architecture. Without the mine, we'd have shipped v2. Phase 7 receipts checklist met (N=3, fresh-chip, identical instrumentation, predicted delta verified, no-WARN under stress).
BESser — BES2600 driver mainlining for Pine64 PineTab2
Working tree for upstreaming the Bestechnic BES2600WM WiFi/BT combo driver currently shipped out-of-tree by the danctnix kernel for the Pine64 PineTab2.
Why "BESser"
"BES" (the chip vendor) + German besser ("better"). Sibling to:
- Bin — u-boot upstreaming for the CoolPi GenBook (RK3588)
- MegabitChip — DDR init blob RE for the same SoC
- SouthAfrica — OpenPineBuds / 1MORE Fit SE BES2300 firmware RE (chip family relative)
- RockHard — RK3588 mainline kernel work
Hardware target
| Item | Value |
|---|---|
| Device | PineTab2 |
| Host name | ohm (ohm.fritz.box, ohm.vpn = 10.170.16.14) |
| SoC | Rockchip RK3566 (aarch64) |
| WiFi/BT | Bestechnic BES2600WM — WiFi 4 (11n dual-band 1x1) + BT 5.2/5.3 |
| WiFi interface | SDIO 3.0, vendor 0xBE57 device 0x2002 |
| BT interface | UART HCI (separate bes2600_btuart module) |
| OS | Arch Linux ARM, danctnix kernel 6.19.10-danctnix1-1-pinetab2 |
Driver lineage
ST-Ericsson CW1200 (mainline drivers/net/wireless/st/cw1200/)
│ author: Dmitry Tarnyagin <dmitry.tarnyagin@stericsson.com>
▼
Bestechnic BES2600 (closed fork ~2015–2023, same author)
│ adds: BT-over-UART, 5GHz, ePTA coexistence, SDIO ADMA, char-dev API
▼
Released for Chinese Android burner-phone vendor kernels (~2023)
│
▼
arjan-vlek/bes2600 (GitLab, 2023-07) → 26-commit checkpatch cleanup
│
├── TuxThePenguin0/bes2600 (preserves pre-removal Android source)
├── salsa.debian.org/Mobian-team/devices/bes2600-dkms (active, our PR target)
└── codeberg.org/DanctNIX/linux-pinetab2 (in-kernel at drivers/staging/bes2600/)
CW1200-ancestry markers in current source: same author Dmitry Tarnyagin,
CONFIG_BES2600_USE_STE_EXTENSIONS (STE = ST-Ericsson),
CONFIG_BES2600_WSM_DEBUG (WSM = CW1200's host↔FW protocol).
Repos
| Role | Path |
|---|---|
| This umbrella | git.reauktion.de/marfrit/besser — patches/, scripts/, fw-analysis/, notes/ |
| Mobian DKMS fork (PR target) | git.reauktion.de/marfrit/bes2600-dkms — branches per patch; upstream = salsa.debian.org/Mobian-team/devices/bes2600-dkms |
Patch series
| Patch | Status | Description |
|---|---|---|
| c1 | ✅ MERGED-LOCAL, deployed on ohm | bes2600: use request_firmware() for factory.txt read — kills filp_open / kernel_read in the read path; fixes the (NULL device *) read and check /lib/firmware/bes2600_factory.txt error boot spam |
| c2 | pending | PM handshake gate + retry — fixes bes2600_pwr_enter_lp_mode wait pm ind timeout recurring every 5–10s |
| c3 | pending | bes_chardev removal + nl80211 testmode hookup — clears the second upstream blocker (custom char device); secondary effect on bug #3 |
| c4 | pending | Enable CONFIG_BES2600_TESTMODE=y + route patch_wifi_testMode — exposes hidden firmware testmode features |
| c5 | pending | Investigate factory.txt parse fail — uncovered by c1; original bug masked it; parser chokes on data request_firmware now successfully returns |
| c1.1 | pending follow-up | Drop kernel_write from factory.c + bes_chardev.c + bes_fw.c (/lib/firmware/bes2002_fw_write.bin debug file!) |
| c1.2 | pending follow-up | Thread struct device * through factory API → eliminates (NULL device *) log prefix |
Workflow
- Edit on Mobian fork (
bes2600-dkms) — primary upstream PR target - Topic branch per patch:
bes2600/<short-name> git commit --signoffwith full kernel-style message- checkpatch.pl --no-tree --strict before push (must be clean)
git format-patch -1 -o ../besser/patches/to canonicalize- Push branch to
origin(Gitea), open MR upstream when ready
Build + deploy
See scripts/:
build-bes2600-on-ohm.sh— native aarch64 build on ohm (requireslinux-pinetab2-headers)deploy-c1-to-ohm.sh— install patched modules to/lib/modules/<ver>/extra/+ depmodbackup-ohm-kernel.sh— timestamped tar.zst of /boot + /lib/modules + /lib/firmware/bes2600
Firmware RE state (task d)
Firmware blobs (4 .bin + factory.txt) are not committed here (binary,
ships from Bestechnic). On boltzmann:~/src/besser/firmware-from-ohm/ and
on oppenheimer:/work/besser/ (Ghidra host).
What IS committed under fw-analysis/:
*.strings.txt— full sorted-unique strings extraction per blob*.fnnames.txt— function-name pattern extraction (191 candidates frombest2002_fw_sdio.bin)
Build provenance: Dec 26-27 2023 build from Bestechnic Jenkins
(GCC-10-pipeline), source git commit 8125e20be (NOT public).
Hidden features in firmware (per task d): testmode patch (CONFIG_BES2600_TESTMODE
gates it), ePTA coexistence with bypass (gEptaBypass: %d), DPD per-channel
calibration, hal_trace_crash_dump_register, METAL_ID + patch_version
introspection, RF tone generation (nosignal variant), memory dump (nosignal
variant), DAC/IQ calibration v2, fast-PSM tuning hooks.
Adjacent infrastructure
- oppenheimer (CT131 on data Proxmox): Ghidra 11.3.2 host. SSH via
mfritsche@192.168.88.63. Staging at/work/besser/. - boltzmann (Rock 5 ITX+): primary build/dev, ARM64 native, hub-tools preferred path.
- ohm: target hardware; UART-over-USB-C debug rig also available (Tigard FT2232H Channel A @ 1.8V).
Authors / contacts
- Markus Fritsche
<fritsche.markus@gmail.com>(this fork) - Driver upstream: Dmitry Tarnyagin
<dmitry.tarnyagin@stericsson.com>(CW1200 / BES2600 base) - BT companion: Dang Huynh
<dang.huynh@mainlining.org>(bes2600_btuart.c, 2025) - Mobian DKMS: Julian
<mail@julianfairfax.ch>(current packaging maintainer)