Add Smitch 360 PTZ (HI3518EV300, JXF23, RTL8188FU, NOR 16MB)

[siddhant01]

Device details

• Model: Smitch 360 PTZ (Indian brand, servers closed)
• Board: T5810hct-m02
• SOC: HI3518EV300
• Flash: EN25QH128 (16MB NOR)
• RAM: 64MB
• WiFi: RTL8188FU (internal USB, GPIO 40 power-up)
• Sensor: JXF23
• Variant: ultimate

Notes

• WiFi power-up via GPIO 40 — wlandev profile rtl8188fu-hi3518ev300-t5810hct-m02
  already merged in firmware repo: add wifi support for t5810hct-m02 board firmware#1373
• QR WiFi provisioning tested and working
• Motors, IR cut, LEDs, reset button not yet mapped (GPIO unknown)

Status

video-yes, wifi-yes, motors-no

[openipc-ai]

Thanks for this, and sorry it sat so long. It is in good shape — reviewed against current master.

Good news first: you no longer need a master.yml matrix entry. Since #120 the build matrix is read off the tree, so a device is registered simply by having devices/<dir>/.../<dir>_defconfig. Your device therefore registers itself, and this repo now builds pull requests — a push here will build hi3518ev300_ultimate_smitch-360ptz-t5810hct-m02 for the first time, narrowed to just that one device. (CLAUDE.md step 6 still describes the old flow; that's on us to fix.)

What checks out

• fw_setenv wlandev rtl8188fu-hi3518ev300-t5810hct-m02 — already upstream in firmware's general/overlay/etc/wireless/usb, complete with a # HI3518EV300 t5810hct-m02 comment. So the Wi-Fi half is landed.
• The exclude list correctly keeps libsns_f23.so, jxf23_i2c_1080p.ini and iq/f23.ini for a JXF23 board, and its name (hi3518ev300_ultimate.list) matches the soc+flavour convention.
• The UBI block matches what both hi3516ev300_ultimate_* devices do, so that's right for the family.

Three small things before merge

1. Drop BR2_PACKAGE_JSONFILTER=y. package: stop selecting jsonfilter, mirroring firmware #2304 #128 removed it from all 97 defconfigs after firmware/#2304, because it drags in libubox for nothing on a device that enables neither uqmi-openipc nor wifibroadcast-ng — it was worth 32KB and had three boards over their rootfs cap.
2. Please drop the README wording change on the "Create firmware with built-in credentials" heading — it's unrelated to the device and just makes the diff harder to read. The device-table row is welcome, keep that.
3. customizer.sh is missing its trailing newline.

Rebase on master with those and CI will tell us whether it builds.

[openipc-ai]

Marking this changes-requested so it shows up as such in the queue. Nothing new — the same three items from my review above are still outstanding:

1. Drop BR2_PACKAGE_JSONFILTER=y from the defconfig (#128 removed it from all 97 defconfigs after firmware/#2304).
2. Drop the unrelated README wording change on the "Create firmware with built-in credentials" heading. The device-table row is welcome, keep that.
3. customizer.sh is missing its trailing newline.

The branch still merges cleanly, so no rebase is needed — push those three and CI will build the device.

[openipc-ai]

Picking this back up rather than letting it age out — the device work was fine,
only the three small items were outstanding, so I have pushed them to your
branch myself:

• dropped BR2_PACKAGE_JSONFILTER=y from the defconfig;
• restored the repack.sh section heading to what master has (your device-table
  row is untouched);
• added the missing trailing newline to customizer.sh.

I also rebased onto current master. When I last looked the branch still merged
cleanly, but the README device table has taken a number of new rows since, and
your row now collided with the SpezVision SVI-252B entry. Both are kept, with
Smitch sorted ahead of SpezVision.

Note this was a force-push, so if you have a local copy of the branch, reset to
the remote rather than merging.

ci-matrix --self-test and the cli path lint both pass, and the matrix
narrows to hi3518ev300_ultimate_smitch-360ptz-t5810hct-m02. I have approved
the build; assuming it comes back green I will merge. Thanks for the profile,
and sorry it sat as long as it did.

[openipc-ai]

Merged. One more fix was needed beyond the three above, worth writing down
because it is a real property of the profile rather than paperwork.

The first build failed late, after generating rootfs.ubifs:

sed: can't read .../general/scripts/ubifs/ubinize_hisilicon.cfg: No such file or directory
make[1]: *** [fs/ubi/ubi.mk:51: .../rootfs.ubi] Error 2

BR2_TARGET_ROOTFS_UBI_CUSTOM_CONFIG_FILE pointed at ubinize_hisilicon.cfg,
which is not in the firmware tree — only ubinize.cfg, ubinize_rockchip.cfg
and ubinize_sigmastar.cfg are. The two devices that use that name each ship
their own private copy under general/scripts/ubifs/ in their device
directory, and this profile referenced the name without carrying the file.

Worth knowing that the UBI block is not dead weight on a NOR board, so deleting
it would have been the wrong fix: CI publishes both a -nor.tgz and a
-nand.tgz for these ultimate profiles, and the nand repack consumes
rootfs.ubi. Dropping it would have silently stopped producing the NAND image.

I switched it to the in-tree ubinize.cfg, which is what both hi3518ev200
ultimate boards use. That suits the nand repack better than the hisilicon
variant, which adds a [kernel] volume while the repack already passes
uImage separately. Your NOR image is unaffected — that path repacks uImage
plus rootfs.squashfs.

Both now build with room to spare: NOR rootfs 6828/8192 KB, NAND rootfs.ubi
13568/16384 KB, kernel 1873/2048 KB.

If the motors get working, a follow-up updating the README status would be
welcome. Thanks for the profile, and again, sorry it waited so long.