Skip to content

ak3918ev300_lite_heimvision-pe3020w: Majestic, with the IR-cut pins - #253

Merged
openipc-ai merged 2 commits into
masterfrom
ak3918ev300-pe3020w-majestic
Nov 17, 2026
Merged

openipc-ai merged 2 commits into
masterfrom
ak3918ev300-pe3020w-majestic

Conversation

@openipc-ai

Copy link
Copy Markdown
Contributor

Adds the Majestic streamer to the PE3020-W image (#234), plus the IR-cut pins it needs.

What's needed for streaming, and where it stands

  • Kernel: anyka: EV300 runs its Allegro encoder through the UIO codec device linux#242 is merged into anyka-ak3918ev300, the branch firmware builds the Anyka kernel from. It enables the Allegro encoder through the UIO codec device, adds an ION carve-out for the encoder's memory and turns on the encoder node.
  • Majestic: the ak3918ev300 binary is published as majestic.ak3918ev300.lite.master.tar.bz2, the file firmware's majestic package downloads.

This PR

Local build

  • uImage 1600/2048 KB; rootfs 5068/5120 KB, so 52 KB headroom.
  • The kernel contains the anyka,ak3918ev300-uio-vencoder node and the PE3020-W model string.
  • The rootfs contains /usr/bin/majestic, S95majestic, the web UI and the customizer lines.

Not tested on hardware yet. The #234 reporter is the next step. I don't know which of the two pins Majestic drives for day and which for night, so they may need swapping.

The PE3020-W image now carries Majestic. It captures from the kernel's ISP2
host and encodes H.264/H.265 and JPEG on the SoC's Allegro encoder; the
kernel side is OpenIPC/linux#242, already on the anyka-ak3918ev300 branch
the firmware builds. The Majestic binary for ak3918ev300 is published as
majestic.ak3918ev300.lite.master.tar.bz2.

The customizer gives Majestic the IR-cut pins the board was measured with
(GPIO 41 day, GPIO 42 night, #234) and selects H.264. No fonts: Majestic's
OSD is not ported to Anyka yet.

Local build: uImage 1600/2048 KB, rootfs 5068/5120 KB (52 KB headroom).
The kernel carries the Allegro encoder node and the PE3020-W model string.
Streaming is not yet tested on hardware.
@qodo-free-for-open-source-projects

Copy link
Copy Markdown

PR Summary by Qodo

Enable Majestic and IR-cut configuration for HeimVision PE3020-W

✨ Enhancement 📝 Documentation 🕐 10-20 Minutes

Grey Divider

AI Description

• Add Majestic to the PE3020-W image so its existing web UI has a streaming backend.
• Configure the measured IR-cut GPIOs and select H.264 as the default video codec.
• Record successful flash boot and clarify that streaming remains untested on hardware.
Diagram

graph TD
  B["Build config"] --> F["PE3020-W image"] --> M["Majestic"] --> K["ISP and encoder"]
  C["Board customizer"] --> M --> I["IR-cut GPIOs"]
Loading
High-Level Assessment

Selecting the existing Majestic package and applying board-specific settings through the existing customizer is the appropriate scope for this image. A separate streamer or kernel implementation would duplicate support already provided elsewhere; the remaining priority is hardware validation, particularly IR-cut direction.

Files changed (3) +10 / -3

Enhancement (1) +5 / -2
customizer.shSet IR-cut pins and default H.264 codec +5/-2

Set IR-cut pins and default H.264 codec

• Configures Majestic's IR-cut pins as GPIO 41 and 42 and sets video0 to H.264. Updates the pin comment to reflect board measurements.

devices/ak3918ev300_lite_heimvision-pe3020w/general/overlay/usr/share/openipc/customizer.sh

Documentation (1) +1 / -1
README.mdClarify PE3020-W boot and streaming status +1/-1

Clarify PE3020-W boot and streaming status

• Records successful flash boot and the measured IR-cut GPIOs. Explicitly notes that Majestic streaming has not yet been tested on hardware.

README.md

Other (1) +4 / -0
ak3918ev300_lite_heimvision-pe3020w_defconfigInclude Majestic in the PE3020-W firmware +4/-0

Include Majestic in the PE3020-W firmware

• Enables the Majestic package alongside the already-selected web UI. Documents its kernel capture and encoder dependency and why fonts are omitted.

devices/ak3918ev300_lite_heimvision-pe3020w/br-ext-chip-anyka/configs/ak3918ev300_lite_heimvision-pe3020w_defconfig

@qodo-free-for-open-source-projects

qodo-free-for-open-source-projects Bot commented Nov 17, 2026 •

Copy link
Copy Markdown

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (2) 🎨 UX issues (0) 🔗 Cross-repo conflicts (0) 📜 Skill insights (0)

Grey Divider


Remediation recommended

1. IR-cut pin roles are stated as fact ✓ Resolved
Description
The customizer comment says the GPIO 41 day / GPIO 42 night assignment was measured on the board,
and it sets .nightMode.irCutPin1 to 41 and .nightMode.irCutPin2 to 42. The PR description says
the author does not know which pin Majestic drives for day and which for night, and that they may
need swapping. The config therefore presents an unverified wiring mapping as confirmed hardware
fact, and a swap would make the IR-cut switch the wrong way at night without any error.

2. Kernel size undercuts flash-boot status 📎 Requirement gap ≡ Correctness
Description
The README status for the PE3020-W on NOR_8M changes from RAM-boot only to flash boot ok, despite
a reported 1600 KB uImage exceeding the 1,572,864-byte stock KERNEL slot. With no qualified
partition-layout migration or hardware flash-boot test shown, the status treats local build results
and the earlier RAM diagnostic as evidence of a complete flash boot.
Code

README.md[45]

+HeimVision PE3020-W      AK3918EV300  JXF37     RTL8188FU_USB    NOR_8M    testing: flash boot ok, ISP capture-ok (OpenIPC kernel), ircut GPIO 41/42; majestic streaming not yet hardware-tested
Evidence
The changed README line claims flash boot ok, while the PR reports a locally built 1600/2048 KB
uImage, says it has not been tested on hardware, and cites no qualified partition-layout migration.
The reported 1600 KB kernel exceeds the stock KERNEL slot, and the prior status documented only RAM
boot; together, these facts support neither flash readiness under rule 2 nor a complete hardware
boot under rule 6.

Respect the stock NOR partition limits
README.md[45-45]

Agent prompt
The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The README claims `flash boot ok` for the PE3020-W on NOR_8M, but the reported 1600 KB uImage exceeds the 1,572,864-byte stock KERNEL slot, no qualified layout migration is cited, and the PR says the image has not been tested on hardware. The prior status documented only RAM boot.
## Fix Focus Areas
- README.md[45-45]
## Recommended Fix
Reword the status to reflect the evidence held, such as a RAM-boot diagnostic and an unvalidated flash boot. Retain a flash-ready claim only if the kernel is shown to fit the stock KERNEL slot or a separately validated layout migration is documented, and a hardware flash-boot test record supports the boot claim.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools


3. Camera capture claimed without F37 proof 📎 Requirement gap ≡ Correctness
Description
The rewritten README status keeps the claim ISP capture-ok (OpenIPC kernel) for the PE3020-W,
which is a native-capture claim. The PR adds no validation of the active F37 sensor, its tuning,
format, or stop/restart behaviour, and its own description says streaming is not yet tested on
hardware, so the capture claim has no board-specific evidence in the change.
Code

README.md[45]

+HeimVision PE3020-W      AK3918EV300  JXF37     RTL8188FU_USB    NOR_8M    testing: flash boot ok, ISP capture-ok (OpenIPC kernel), ircut GPIO 41/42; majestic streaming not yet hardware-tested
Evidence
Rule 4 requires independent confirmation of the active F37 sensor and capture configuration before
native capture is claimed to work. The README line states ISP capture-ok (OpenIPC kernel) with no
such confirmation in the diff.

Identify the active F37 sensor and validate native capture claims
README.md[45-45]

Agent prompt
The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution
The issue below was found during a code review. Follow the provided context and guidance below and implement a solution
Issue description
The README status for the PE3020-W states `ISP capture-ok (OpenIPC kernel)`, a native capture claim. The diff contains no validation of the active F37 sensor, its tuning, format, or stop/restart behaviour, and the PR says streaming is untested on hardware.
Fix Focus Areas
- README.md[45-45]
Recommended Fix
Either attach board-specific capture validation for the F37 sensor and its configuration, or qualify the README wording to say capture has been observed on the OpenIPC kernel without repeatable validation.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools


Grey Divider

Tip of the day
💡 Did you know, you can enable the Remediation agent and Qodo fixes findings in a dedicated fix PR

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

Grey Divider

Qodo Logo

@openipc-ai

Copy link
Copy Markdown
Contributor Author

Qodo review:

  1. IR-cut roles stated as fact: fixed in 4d82931. The measured part stays: on this board GPIO 41 pulses the filter to day and GPIO 42 to night, from the reporter's tests in HeimVision PE3020-W: AK3918 V300/F37 board findings, 1.5 MiB kernel-slot mismatch and bring-up questions #234. The comment now also says it isn't yet confirmed which of Majestic's two pins drives which way, so they may need swapping.
  2. "Flash boot ok" vs the 1.5 MB stock kernel slot: not changed. The image doesn't use the stock layout: it needs the OpenIPC U-Boot and 8 MB layout, whose kernel slot is 2048 KB, and the 1600 KB uImage fits it. The reporter cold-booted the released image from NOR that way on the physical camera (HeimVision PE3020-W: AK3918 V300/F37 board findings, 1.5 MiB kernel-slot mismatch and bring-up questions #234 (comment) replies to that report).
  3. "ISP capture-ok" without F37 proof: not changed. On a RAM boot of the OpenIPC kernel, the reporter's camera probed the F37 (Probing Sensor ID: 0xf37), loaded /etc/sensor/jxf37.conf and captured 1080p frames (HeimVision PE3020-W: AK3918 V300/F37 board findings, 1.5 MiB kernel-slot mismatch and bring-up questions #234, linux#200). The row says streaming is not yet tested on hardware, which is the part still unverified.

@openipc-ai
openipc-ai merged commit c0f1a5a into master Nov 17, 2026
7 checks passed
@openipc-ai
openipc-ai deleted the ak3918ev300-pe3020w-majestic branch November 17, 2026 14:08
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