You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm not able to help with integration itself, but if you will need anything get from cameras or be a beta tester with firmware I can be that person :).
I still didn't catch network traffic / didn't try to / while upgrading FW, so I don't know process how camera is possible to flash firmware without uboot / uart usage. It's not possible in camera UI itself, but possible by some external tool. But it's work for some of next days for me, if I'll have some time :)
I have a number of the ASECAM Vatilon PB1 cameras. They can be affected by something on the network that causes them to all reboot at once. This can continue for some time, all rebooting regularly. I am willing to assist where I can. I upgraded the firmware using the Windows tool. It didn't help with the reboots but it may be the way to install alternative firmware.
I have a number of the ASECAM Vatilon PB1 cameras. They can be affected by something on the network that causes them to all reboot at once. This can continue for some time, all rebooting regularly. I am willing to assist where I can. I upgraded the firmware using the Windows tool. It didn't help with the reboots but it may be the way to install alternative firmware.
In the camera WebGui is possible to setup autoreboot. Maybe you have it enabled?
I have no issue with reboots of camera when this settings is disabled. But I have it enabled to reboot once per day to cleanup possible FW memleaks, sync time again, etc.
On Fri, 29 Nov 2024, 7:44 am Matej Trakal, ***@***.***> wrote:
I have a number of the ASECAM Vatilon PB1 cameras. They can be affected by
something on the network that causes them to all reboot at once. This can
continue for some time, all rebooting regularly. I am willing to assist
where I can. I upgraded the firmware using the Windows tool. It didn't help
with the reboots but it may be the way to install alternative firmware.
In the camera WebGui is possible to setup autoreboot. Maybe you have it
enabled?
I have no issue with reboots of camera when this settings is disabled. But
I have it enabled to reboot once per day to cleanup possible FW memleaks,
sync time again, etc.
image.png (view on web)
<https://github.com/user-attachments/assets/32aa564c-4c76-4f30-bf07-61cf1c37d30f>
—
Reply to this email directly, view it on GitHub
<#1610 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABTSUVMWCQQBVC2ZOI2M2UD2C6FFJAVCNFSM6AAAAABSE5CH7WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMBWG44DCMRTGY>
.
You are receiving this because you commented.Message ID:
***@***.***>
I upgraded the firmware using the Windows tool. It didn't help with the reboots but it may be the way to install alternative firmware.
Yeah, I already caught FW upgrade packets, but it's a bit useless because it's encrypted. So App and Cam changed the keys and session and encrypted the rest of the communication. So until we know the algorithm and how the key is build, we are lost I think this simple way for upgrade FW, but we have 2 others ⏬.
Discover camera is based on Broadcast on Network L2 (ARP broadcast on same network area):
But we already have ROOT access to telnet, so I don!t think, that we need use standard way how to upgrade firmware, when you can (I supose) re-mount partition as write or use dd to rewrite flash memory. Not sure, but expecting, that with root bash / bussybox it is possible.
And second tested variant - the author of articles on Reddit already used UART and flash his own custom FW (changed default password to telnet with custom one). So flash new FW is not the main problem I think :)
I opened one up and took some pics. Trouble is, after I checked the chip
numbers I didn't get a pic of the camera chip. Only the flash and ethernet.
I've ordered a new one to have a decent look around inside.
On Sat, 30 Nov 2024, 8:03 am Matej Trakal, ***@***.***> wrote:
I upgraded the firmware using the Windows tool. It didn't help with the
reboots but it may be the way to install alternative firmware.
Yeah, I already caught FW upgrade packets, but it's a bit useless because
it's encrypted. So App and Cam changed the keys and session and encrypted
the rest of the communication. So until we know the algorithm and how the
key is build, we are lost I think this simple way for upgrade FW, but we
have 2 others ⏬.
Discover camera is based on Broadcast on Network L2 (ARP broadcast on same
network area):
{
"DeviceType": "IPC",
"ChannelNum": 1,
"DeviceModel": "PB4",
"DeviceModel2": "PB",
"DeviceName": "PB4",
"CloudID": "__CLEARED__",
"SerialNo": "__CLEARED__",
"NetType": 1,
"MAC": "__CLEARED__",
"DHCP": 0,
"Address": "10.0.0.33",
"Submask": "255.255.255.0",
"GateWay": "10.0.0.254",
"DNS": "10.0.0.252",
"TCPPort": 8600,
"RTSPPort": 554,
"HTTPPort": 80,
"Language": "English",
"SoftwareVersion": "V1.13.23-20240524",
"DeviceRunTime": 78,
"QWTEnable": false,
"IPAdaptive": 1,
"LampType": 1,
"DayNightSwitch": 0,
"P2PType": 4,
"P2PPID": "__CLEARED__",
"P2PQRCode": "__CLEARED__",
"LensCount": 1,
"StreamCount": 2,
"StreamNumPerChannel": 2,
"NetTypeSet": 1,
"NetProtocolSet": 143,
"Status": 2,
"Version": 1,
"Scode": "__CLEARED__"
}
After that application sends UDP packet to camera IP address with body:
{
"operation": "login",
"param": {
"user": "JmsRxTtVEW+J5da5aSt3lw==",
"password": "bC+UKGMPDDiynIUE9ViJzw==",
"heartbeat_timeinterval": 1,
"heartbeat_timeoutcounts": 3,
"sessionid": 179511328,
"encryptionkey": "EOtDcH4DFPJZj2CDhLFNAg==",
"roletype": 5
}
}
So we know, that decrypted user = admin and password = 123456.
Response is:
{
"operation": "login_response",
"code": 0,
"msg": "successful ",
"sessionid": 179511328,
"encryptionkey": "HO55v18ZnC9cL6pJe0U9Tg=="
}
And the rest is encryped with new key.
------------------------------
But we already have ROOT access to telnet, so I don!t think, that we need
use standard way how to upgrade firmware, when you can (I supose) re-mount
partition as write or use dd to rewrite flash memory. Not sure, but
expecting, that with root bash / bussybox it is possible.
------------------------------
And second tested variant - the author of articles on Reddit
<https://www.reddit.com/r/hardwarehacking/comments/1cuu1wf/hacking_an_asecam_ip_camera_part_1/>
already used UART and flash his own custom FW (changed default password to
telnet with custom one). So flash new FW is not the main problem I think :)
—
Reply to this email directly, view it on GitHub
<#1610 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABTSUVPWI7IBURFKECQFO7D2DDQEXAVCNFSM6AAAAABSE5CH7WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMBYGY4TMMJRGQ>
.
You are receiving this because you commented.Message ID:
***@***.***>
I have multiple Vatilon cameras, including PB4. They actually work pretty well if you block internet comms to them. I see we can already get to Uboot and have root access to the firmware. I can contribute. Nice cheap cameras these.
i have two of these: https://www.aliexpress.com/item/1005006826151802.html
i have used nmap to scan the ports, and 2360 was not listed as open, but i can indeed telnet to that port.
but the admin123456 doesn't work there (it works through the web UI, though).
any other ideas for login/pw?
it seems with the v1.14 and v1.15 there is a bug that does not let you save the time settings on the webgui on the PB4... I have one camera with v1.13 and that works fine but I used IPC tool and tried to upgrade from v1.14 to v1.15 and neither allows me to save the time settings
hello everyone i have a chinese camera which runs a V1.14.91-20241120, but everytime my nvr reboots (once a week) the camera fails to connect and needs a manual password input
can someone supply me with the V1.12 version
thank you
Just here to add some info as there isn't much on the internet. I have two of these cameras and can help any sort of testing if needed.
I bought some REVODATA fisheye cameras in 2022 that turned out to run these Vatilon boards. I can confirm I also had the same issue darrylb123 had where both of these cameras would just reboot on their own, and I didn't have auto reboot on. These cameras eventually broke and started boot looping, the webUI would show up just for a moment and I couldn't get in.
in case the link goes down: FH8852V200+GC5053 V1.14.91-20241120 PA4 uboot-2016-22
Well these ones eventually broke in the exact same way so I dug further and found the info discussed here: the open telnet port and cracked root password. root/ipc@hs66
I got in and found the /tmp/ipcam binary was segfaulting on boot, which then caused it to restart forever. After reading the /usr/app/run.sh I found you can force a factory reset which I did, but still it was bootlooping. The fix for me was to manually edit the /tmp/etc/config/ipcam.conf and disable ONVIF. I don't know what changed, cause it worked before but if that camera booted with ONVIF it would segfault on startup, and no way to get in the webUI. My camera now boots and I just have to use RTSP
point being, if you have this Vatilon firmware and your camera boot loops, I fixed it by killing ipcam process to keep it up, editing the /tmp/etc/config/ipcam.conf to disable ONVIF. So having a nice working OpenIPC would be great, there are tons of cheap cameras out there that that use these chips =)
@darrylb123 thank you, that log helped a lot. It shows two things.
The memory layout fix (#2872) works on your PB1. The video memory pool is at 0xA1D00000~0xA7DFFFFF, load_fullhan reported no problem, and the video drivers loaded. That's the first time this has run on a real camera.
The missing video has a different cause. majestic tried to load /usr/lib/sensors/libgc4653_mipi.so, and that file doesn't exist for this chip. Fullhan's SDK for the FH8856 V100 ships sensor libraries for other sensors but not for the GC4653 in your camera, so the image never had one.
#2879 adds one, built from source in a new repository, OpenIPC/openfullhan. It also fixes where the image installs the sensor tuning files, so the camera actually finds them.
Test build
This is a test build. The sensor driver has only been run in emulation, never on a camera, so yours would be the first.
It is the normal fh8856v100 image plus the GC4653 library. It's for cameras installed with the fixed 8 MB layout from my steps above, as yours is. On the camera:
If the camera can't reach GitHub, download openipc.fh8856v100-nor-lite.tgz on your PC (don't unpack it). Run python3 -m http.server 8000 in that folder, then use --url=http://YOUR_PC_IP:8000/openipc.fh8856v100-nor-lite.tgz. Your settings and password are kept.
If you'd rather not reflash, the release also has the three files on their own. Copy libgc4653_mipi.so to /usr/lib/sensors/, and gc4653_mipi_attr.hex and gc4653_mipi_night_attr.hex to /etc/sensors/, then reboot.
To go back, sysupgrade -k -r installs the regular latest image.
What to send back
Whether rtsp://<camera-ip>/stream=0 plays, and if it does, a snapshot from http://<camera-ip>/image.jpg. If the picture comes up but the colours look off, that's expected: until someone tunes it, the colour tuning is borrowed from the closest sensor the SDK has, the GC4633.
If there's no video, that log is what I need to fix it, for example if the MIPI speed setting is off. That setting is the one value no source states.
@mtrakal thank you for the PB4 data. plat_id 0x19112201 and FH8856V200 confirm the PB4 is the newer V200 chip, unlike the PB1, so it uses the fh8856v200 image, not this test build. Your /proc/driver/isp output also helped write this driver: the frame size it reports matches the timing the new GC4653 driver uses.
@eribito your IF56N units run the same firmware as the PB1 dump, so this test build applies to them as well, whenever you get to the bench.
@darrylb123 thank you for testing so quickly, and sorry for the wait. Your log shows the new parts working on your camera, up to the very last step:
the GC4653 library loads, and majestic picks it up as gc4653_i2c;
it accepts 2560x1440 at 25 fps;
the tuning file is found in /etc/sensors;
the image processor initialises.
Only the final start call fails: API_ISP_SensorKick failed with -1.
That's my mistake. In this chip's ISP library, that call runs a kick function the sensor library is supposed to provide, and returns -1 when it is missing. I wrote the driver from Fullhan's own V100 example drivers, and neither of them provides it. The library now does: OpenIPC/openfullhan@1564015
If it gets past the start call and still shows no picture, the next suspects are the MIPI speed setting and the timing, and that log is what will tell them apart.
@darrylb123 sorry, that one is my fault again. Your camera is most likely still running test build 1. Both test builds carried the same version stamp, so sysupgrade saw "Same version, nothing to update" and wrote nothing. Your log is identical to the first one, line for line.
You can check which library you have:
sha256sum /usr/lib/sensors/libgc4653_mipi.so
15fd4c88… is test build 1 and d56fb0c1… is test build 2.
The quickest way to get test build 2 is to replace just that one file, since nothing else changed:
Initially asked for a login and password.
ran vlc http:///image.jpg and an image flashed up briefly
ran vlc rtsp:///stream=0 and it asked for login then said it couldn't display the video format.
Changed to H264 but can no longer get anything to work, video or image.
root@openipc-fh8856v100:~# logread | grep -i -E 'sensor|gc4653|mipi|isp|majestic' | head -80
Nov 17 17:44:19 openipc-fh8856v100 kern.info kernel: [ 9.181855] isp - (dev. name: ispp - IRQ #7
Nov 17 17:44:19 openipc-fh8856v100 kern.info kernel: [ 9.181952] isp - (dev. name: ispf - IRQ #8
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: Majestic Lite for Fullhan (fh8852v100), version master+80cb39d, built on 2026-11-16 17:41
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: Loading main configuration: /etc/majestic.yaml
Nov 17 17:44:22 openipc-fh8856v100 user.debug majestic[812]: Detected Synopsys DesignWare Watchdog
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: Watchdog armed: asked 300 s, accepted 300 s, fed every 150.000s, magic close supported
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: no identity: no chip identity on this platform
Nov 17 17:44:22 openipc-fh8856v100 user.debug majestic[812]: Set server path: /var/www
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: HTTP server started on :::80
Nov 17 17:44:22 openipc-fh8856v100 user.warn majestic[812]: HTTPS port :::443 configured without cert/key; skipping TLS server
Nov 17 17:44:22 openipc-fh8856v100 user.debug majestic[812]: Autodetected sensor as 'gc4653_i2c'
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: Using gc4653_i2c sensor
Nov 17 17:44:22 openipc-fh8856v100 user.debug majestic[812]: 2560 1440 25 maxw 2560
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: [dsp] version: V1.2.0(gcd1ec21),build: 2020-03-24
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: [isp] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: [ispcore] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 17:44:22 openipc-fh8856v100 user.debug majestic[812]: trying to load /usr/lib/sensors/libgc4653_mipi.so
Nov 17 17:44:22 openipc-fh8856v100 user.debug majestic[812]: ADV_ISP version: V1.2.0(g4bb12fc)
Nov 17 17:44:22 openipc-fh8856v100 user.debug majestic[812]: Error: FHAdv_Isp_SensorInit failed, ret = -2
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: [sensor] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: [mipi] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: Error: sensor function get_bayer_format is NULL
Nov 17 17:44:22 openipc-fh8856v100 user.warn majestic[812]: mirror 0, flip 0 not set: -1(ffffffff)
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: joined 239.255.255.250 on eth0 (172.25.1.99)
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: onvif discovery sent Hello via eth0 as 172.25.1.99
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: also answering for openipc-fh8856v100.local.
Nov 17 17:44:22 openipc-fh8856v100 user.info majestic[812]: joined 224.0.0.251 on eth0 (172.25.1.99)
Nov 17 17:44:23 openipc-fh8856v100 user.info majestic[812]: announced openipc.local. and openipc-fh8856v100.local. as 172.25.1.99 on eth0
Nov 17 17:44:23 openipc-fh8856v100 user.info majestic[812]: RTSP listener on IPv4 refused (Protocol not available); listening without SO_REUSEPORT
Nov 17 17:44:23 openipc-fh8856v100 user.info majestic[812]: RTSP listener on IPv6 refused (Protocol not available); listening without SO_REUSEPORT
Nov 17 17:44:23 openipc-fh8856v100 user.info majestic[812]: RTSP server started on port 554
Nov 17 17:44:23 openipc-fh8856v100 user.info majestic[812]: H.264 SPS said nothing about picture reordering; publishing it with max_num_reorder_frames=0 (20 -> 23 bytes) so decoders that would otherwise hold a level's whole DPB show the first frame at once
Nov 17 17:44:23 openipc-fh8856v100 user.info majestic[812]: H.264 SPS said nothing about picture reordering; publishing it with max_num_reorder_frames=0 (20 -> 23 bytes) so decoders that would otherwise hold a level's whole DPB show the first frame at once
Nov 17 17:44:23 openipc-fh8856v100 user.debug majestic[812]: DTLS identity: P-256, 289-byte certificate
Nov 17 17:44:23 openipc-fh8856v100 user.info majestic[812]: WebRTC fingerprint sha-256 49:DA:A7:CE:34:7F:65:BA:0B:49:56:C0:B8:DC:E1:EA:C9:80:61:6A:CC:62:ED:67:E9:F0:EE:1E:7A:D6:BE:47
Nov 17 17:44:24 openipc-fh8856v100 user.debug majestic[812]: announced again as 172.25.1.99 on eth0
Nov 17 17:44:24 openipc-fh8856v100 user.debug majestic[812]: onvif discovery sent Probe via eth0
Nov 17 17:44:25 openipc-fh8856v100 user.debug majestic[812]: onvif discovery sent Hello again via eth0
Nov 17 17:44:26 openipc-fh8856v100 user.debug majestic[812]: announced again as 172.25.1.99 on eth0
Nov 17 17:44:29 openipc-fh8856v100 user.debug majestic[812]: onvif discovery sent Hello again via eth0
Nov 17 17:44:30 openipc-fh8856v100 user.debug majestic[812]: announced again as 172.25.1.99 on eth0
Nov 17 17:44:33 openipc-fh8856v100 user.info majestic[812]: Live backlog budget: 5209 KiB of 25868 KiB RAM (board-class bound, memory available 17024 KiB); 1280 KiB per session, so about 4 concurrent live viewers across /ws/video, WebRTC, RTSP-over-TCP, /mjpeg and /video.mp4. Low
Nov 17 17:44:33 openipc-fh8856v100 user.warn majestic[812]: this kernel publishes no MemAvailable, so the budget and the low-memory floor are read from MemFree + Buffers + Cached; both understate what is reclaimable and tighten early
Nov 17 17:44:38 openipc-fh8856v100 kern.err kernel: [ 28.151544] isp_release start
Nov 17 17:44:38 openipc-fh8856v100 kern.err kernel: [ 28.171295] fh81_isp_release finish
Nov 17 17:44:38 openipc-fh8856v100 user.notice majestic: crashed (signal 8); starting it again in 5s
Nov 17 17:44:43 openipc-fh8856v100 user.info majestic[864]: Majestic Lite for Fullhan (fh8852v100), version master+80cb39d, built on 2026-11-16 17:41
Nov 17 17:44:43 openipc-fh8856v100 user.info majestic[864]: Loading main configuration: /etc/majestic.yaml
Nov 17 17:44:43 openipc-fh8856v100 user.warn majestic[864]: the previous majestic crashed with SIGFPE (8); its crash dump is in /etc/crash. Its video pools may still be held and the hardware watchdog may still be armed from it
Nov 17 17:44:43 openipc-fh8856v100 user.debug majestic[864]: Detected Synopsys DesignWare Watchdog
Nov 17 17:44:43 openipc-fh8856v100 user.info majestic[864]: Watchdog armed: asked 300 s, accepted 300 s, fed every 150.000s, magic close supported
Nov 17 17:44:43 openipc-fh8856v100 user.info majestic[864]: no identity: no chip identity on this platform
Nov 17 17:44:43 openipc-fh8856v100 user.debug majestic[864]: Set server path: /var/www
Nov 17 17:44:43 openipc-fh8856v100 user.info majestic[864]: HTTP server started on :::80
Nov 17 17:44:43 openipc-fh8856v100 user.warn majestic[864]: HTTPS port :::443 configured without cert/key; skipping TLS server
Nov 17 17:44:44 openipc-fh8856v100 user.debug majestic[864]: Autodetected sensor as 'gc4653_i2c'
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: Using gc4653_i2c sensor
Nov 17 17:44:44 openipc-fh8856v100 user.debug majestic[864]: 2560 1440 25 maxw 2560
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: [dsp] version: V1.2.0(gcd1ec21),build: 2020-03-24
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: [isp] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: [ispcore] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 17:44:44 openipc-fh8856v100 user.debug majestic[864]: trying to load /usr/lib/sensors/libgc4653_mipi.so
Nov 17 17:44:44 openipc-fh8856v100 user.debug majestic[864]: ADV_ISP version: V1.2.0(g4bb12fc)
Nov 17 17:44:44 openipc-fh8856v100 user.debug majestic[864]: Error: FHAdv_Isp_SensorInit failed, ret = -2
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: [sensor] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: [mipi] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: Error: sensor function get_bayer_format is NULL
Nov 17 17:44:44 openipc-fh8856v100 user.warn majestic[864]: mirror 0, flip 0 not set: -1(ffffffff)
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: joined 239.255.255.250 on eth0 (172.25.1.99)
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: onvif discovery sent Hello via eth0 as 172.25.1.99
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: also answering for openipc-fh8856v100.local.
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: joined 224.0.0.251 on eth0 (172.25.1.99)
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: announced openipc.local. and openipc-fh8856v100.local. as 172.25.1.99 on eth0
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: RTSP listener on IPv4 refused (Protocol not available); listening without SO_REUSEPORT
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: RTSP listener on IPv6 refused (Protocol not available); listening without SO_REUSEPORT
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: RTSP server started on port 554
Nov 17 17:44:44 openipc-fh8856v100 user.debug majestic[864]: DTLS identity: P-256, 289-byte certificate
Nov 17 17:44:44 openipc-fh8856v100 user.info majestic[864]: WebRTC fingerprint sha-256 47:21:D0:A9:5D:24:78:AE:3C:4A:FA:79:06:BF:78:94:CC:E8:3D:08:7A:1E:35:94:48:8D:56:DA:74:DA:66:36
Nov 17 17:44:45 openipc-fh8856v100 user.debug majestic[864]: wait pic_end fail, errno=0
Nov 17 17:44:45 openipc-fh8856v100 user.debug majestic[864]: announced again as 172.25.1.99 on eth0
root@openipc-fh8856v100:~# dmesg | tail -40
[ 28.151443] [vmm] clean vmm zone anonymous
[ 28.151456] [vmm] clean vmm zone anonymous
[ 28.151467] [vmm] clean vmm zone anonymous
[ 28.151479] [vmm] clean vmm zone anonymous
[ 28.151490] [vmm] clean vmm zone anonymous
[ 28.151502] [vmm] clean vmm zone anonymous
[ 28.151514] [vmm] clean vmm zone anonymous
[ 28.151526] [vmm] clean vmm zone anonymous
[ 28.151544] isp_release start
[ 28.171295] fh81_isp_release finish
[ 33.355224] [wdt] set topval: 5
[ 33.400033] [VPU]vpu_parse_cmdline 349: VPU Revc Command: vi_2560_1440
[ 33.400078] [VPU]vpu_parse_cmdline 638: set command success! (vi)
[ 33.400295] [VPU]vpu_parse_cmdline 349: VPU Revc Command: cap_0_2560_1440
[ 33.400335] [VPU]vpu_parse_cmdline 638: set command success! (cap)
[ 33.400512] [VPU]vpu_parse_cmdline 349: VPU Revc Command: cap_1_1280_720
[ 33.400551] [VPU]vpu_parse_cmdline 638: set command success! (cap)
[ 33.401267] [MEDIA]fh_media_open 111: media_process open success.
[ 33.403463] [ENC]pae_enc_isr 3285: [PAE]---PAE RUN!!---
[ 33.422204] [ENC]pae_open 69: pae open success!
[ 33.422813] [HEVC]hevc_open 42: hevc open success!
[ 33.423106] [JPEG]jpeg_open 35: jpeg open success.
[ 33.423309] [VPU]vpu_open 544: vpu open success.
[ 33.423422] [BGM]bgm_open 329: bgm open success.
[ 33.423658] remap_pfn_range start~end: 0x406ea000 0x4070c000, pyh:0xa1d00 , nocached.
[ 33.424789] remap_pfn_range start~end: 0x4136c000 0x42375000, pyh:0xa1d22 , nocached.
[ 33.425761] remap_pfn_range start~end: 0x42446000 0x429ab000, pyh:0xa2d2b , nocached.
[ 33.426367] remap_pfn_range start~end: 0x42a5c000 0x42e77000, pyh:0xa3290 , nocached.
[ 33.426860] remap_pfn_range start~end: 0x408ec000 0x40942000, pyh:0xa36ab , nocached.
[ 33.441724] [MEDIA]check_dst_chan 455: [unlegality chan]:chan 1 0x0 0 0
[ 33.441758] [MEDIA]check_dst_chan 455: [unlegality chan]:chan 5 0x0 0 0
[ 33.442259] remap_pfn_range start~end: 0x42e77000 0x4356c000, pyh:0xa3701 , nocached.
[ 33.442514] [MEDIA]check_dst_chan 455: [unlegality chan]:chan 7 0x0 0 0
[ 33.442679] [MEDIA]check_dst_chan 455: [unlegality chan]:chan 2 0x0 0 0
[ 33.442942] remap_pfn_range start~end: 0x43644000 0x43833000, pyh:0xa3df6 , nocached.
[ 33.443127] [MEDIA]check_dst_chan 455: [unlegality chan]:chan 8 0x0 0 0
[ 33.445543] remap_pfn_range start~end: 0x4386f000 0x44ed6000, pyh:0xa3ff0 , nocached.
[ 33.521925] remap_pfn_range start~end: 0x44ed6000 0x45099000, pyh:0xa5657 , nocached.
[ 33.522129] [MEDIA]check_dst_chan 455: [unlegality chan]:chan 16 0x0 0 0
[ 63.780250] set time :rtc core busy
@darrylb123 that's real progress, thank you. With test build 2 the sensor is streaming. The image that flashed up in VLC was a real frame from the GC4653, and the log shows the encoder starting (H.264 SPS ...). So the sensor setup, the MIPI link and the timing all work on your camera.
What stops it now is that majestic crashes about 16 seconds in (crashed (signal 8), a division by zero). After the automatic restart it gets no frames at all (wait pic_end fail), which is why nothing worked after you switched to H.264. Keeping H.264 is fine.
Two lines just before streaming starts are related: FHAdv_Isp_SensorInit failed, ret = -2 and sensor function get_bayer_format is NULL. They show a vendor call that registers the sensor's colour-order functions failing in majestic's start-up. That affects every sensor on this chip, not just yours, and it's being looked at on the majestic side.
majestic saved a crash dump, and that's what will show where the division by zero happens. Could you please:
Reboot the camera, so it starts clean, and wait about a minute.
If you can, also attach the dump file itself. On the camera, run gzip -c /etc/crash/majestic.dump > /tmp/majestic.dump.gz. Copy it to your PC (for example scp root@<camera-ip>:/tmp/majestic.dump.gz .) and drag it into a comment here. It holds the registers and stack at the moment of the crash, which the readable part doesn't.
If the camera crashes again after the reboot, the dump will be from that crash, which is fine.
@darrylb123 there's a test build 3 with a fix for the FHAdv_Isp_SensorInit failed, ret = -2 line from your log. It contains a test majestic that registers the GC4653 library with the vendor's ISP layer correctly on this chip. That registration failed on every start before, and it's the most likely cause of the crash at 16 seconds. The sensor library is the same as in build 2.
The dashboard displays an image. Image cannot be seen over http and no video.
root@openipc-fh8856v100:~# logread | grep -i -E 'sensor|gc4653|mipi|isp|majestic|picture|crash' | head -80
Nov 17 19:59:04 openipc-fh8856v100 kern.info kernel: [ 9.541124] isp - (dev. name: ispp - IRQ #7
Nov 17 19:59:04 openipc-fh8856v100 kern.info kernel: [ 9.541221] isp - (dev. name: ispf - IRQ #8
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: Majestic Lite for Fullhan (fh8852v100), version fullhan-v100-sensor-info+5705db24, built on 2026-11-17 08:21
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: Loading main configuration: /etc/majestic.yaml
Nov 17 19:59:07 openipc-fh8856v100 user.debug majestic[828]: Detected Synopsys DesignWare Watchdog
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: Watchdog armed: asked 300 s, accepted 300 s, fed every 150.000s, magic close supported
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: no identity: no chip identity on this platform
Nov 17 19:59:07 openipc-fh8856v100 user.debug majestic[828]: Set server path: /var/www
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: HTTP server started on :::80
Nov 17 19:59:07 openipc-fh8856v100 user.warn majestic[828]: HTTPS port :::443 configured without cert/key; skipping TLS server
Nov 17 19:59:07 openipc-fh8856v100 user.debug majestic[828]: Autodetected sensor as 'gc4653_i2c'
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: Using gc4653_i2c sensor
Nov 17 19:59:07 openipc-fh8856v100 user.debug majestic[828]: 2560 1440 25 maxw 2560
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: [dsp] version: V1.2.0(gcd1ec21),build: 2020-03-24
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: [isp] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: [ispcore] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 19:59:07 openipc-fh8856v100 user.debug majestic[828]: trying to load /usr/lib/sensors/libgc4653_mipi.so
Nov 17 19:59:07 openipc-fh8856v100 user.debug majestic[828]: ADV_ISP version: V1.2.0(g4bb12fc)
Nov 17 19:59:07 openipc-fh8856v100 user.debug majestic[828]: probe sensor: gc4653_mipi
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: [sensor] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 19:59:07 openipc-fh8856v100 user.info majestic[828]: [mipi] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: picture upright
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: H.264 SPS said nothing about picture reordering; publishing it with max_num_reorder_frames=0 (20 -> 23 bytes) so decoders that would otherwise hold a level's whole DPB show the first frame at once
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: joined 239.255.255.250 on eth0 (172.25.1.99)
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: onvif discovery sent Hello via eth0 as 172.25.1.99
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: also answering for openipc-fh8856v100.local.
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: joined 224.0.0.251 on eth0 (172.25.1.99)
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: announced openipc.local. and openipc-fh8856v100.local. as 172.25.1.99 on eth0
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: RTSP listener on IPv4 refused (Protocol not available); listening without SO_REUSEPORT
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: RTSP listener on IPv6 refused (Protocol not available); listening without SO_REUSEPORT
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: RTSP server started on port 554
Nov 17 19:59:08 openipc-fh8856v100 user.debug majestic[828]: DTLS identity: P-256, 290-byte certificate
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: WebRTC fingerprint sha-256 F4:DA:38:85:1D:F8:44:72:CC:6B:83:47:13:3D:39:19:66:47:93:86:78:23:B6:1E:24:71:20:E8:1A:25:95:CA
Nov 17 19:59:08 openipc-fh8856v100 user.info majestic[828]: H.264 SPS said nothing about picture reordering; publishing it with max_num_reorder_frames=0 (20 -> 23 bytes) so decoders that would otherwise hold a level's whole DPB show the first frame at once
Nov 17 19:59:09 openipc-fh8856v100 user.debug majestic[828]: announced again as 172.25.1.99 on eth0
Nov 17 19:59:10 openipc-fh8856v100 user.debug majestic[828]: onvif discovery sent Probe via eth0
Nov 17 19:59:10 openipc-fh8856v100 user.debug majestic[828]: onvif discovery sent Hello again via eth0
Nov 17 19:59:11 openipc-fh8856v100 user.debug majestic[828]: announced again as 172.25.1.99 on eth0
Nov 17 19:59:12 openipc-fh8856v100 user.info majestic[828]: session restored from stay-signed-in token for root
Nov 17 19:59:12 openipc-fh8856v100 user.debug majestic[828]: JPEG was sent to client
Nov 17 19:59:12 openipc-fh8856v100 user.debug majestic[828]: JPEG: 356184 bytes
Nov 17 19:59:14 openipc-fh8856v100 user.debug majestic[828]: onvif discovery sent Hello again via eth0
Nov 17 19:59:17 openipc-fh8856v100 kern.err kernel: [ 22.330645] isp_release start
Nov 17 19:59:17 openipc-fh8856v100 kern.err kernel: [ 22.350403] fh81_isp_release finish
Nov 17 19:59:17 openipc-fh8856v100 user.notice majestic: crashed (signal 8); starting it again in 5s
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: Majestic Lite for Fullhan (fh8852v100), version fullhan-v100-sensor-info+5705db24, built on 2026-11-17 08:21
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: Loading main configuration: /etc/majestic.yaml
Nov 17 19:59:22 openipc-fh8856v100 user.warn majestic[906]: the previous majestic crashed with SIGFPE (8); its crash dump is in /etc/crash. Its video pools may still be held and the hardware watchdog may still be armed from it
Nov 17 19:59:22 openipc-fh8856v100 user.debug majestic[906]: Detected Synopsys DesignWare Watchdog
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: Watchdog armed: asked 300 s, accepted 300 s, fed every 150.000s, magic close supported
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: no identity: no chip identity on this platform
Nov 17 19:59:22 openipc-fh8856v100 user.debug majestic[906]: Set server path: /var/www
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: HTTP server started on :::80
Nov 17 19:59:22 openipc-fh8856v100 user.warn majestic[906]: HTTPS port :::443 configured without cert/key; skipping TLS server
Nov 17 19:59:22 openipc-fh8856v100 user.debug majestic[906]: Autodetected sensor as 'gc4653_i2c'
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: Using gc4653_i2c sensor
Nov 17 19:59:22 openipc-fh8856v100 user.debug majestic[906]: 2560 1440 25 maxw 2560
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: [dsp] version: V1.2.0(gcd1ec21),build: 2020-03-24
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: [isp] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: [ispcore] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 19:59:22 openipc-fh8856v100 user.debug majestic[906]: trying to load /usr/lib/sensors/libgc4653_mipi.so
Nov 17 19:59:22 openipc-fh8856v100 user.debug majestic[906]: ADV_ISP version: V1.2.0(g4bb12fc)
Nov 17 19:59:22 openipc-fh8856v100 user.debug majestic[906]: probe sensor: gc4653_mipi
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: [sensor] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: [mipi] version: V1.2.0(gaee6fa1),build: 2020-03-24
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: picture upright
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: joined 239.255.255.250 on eth0 (172.25.1.99)
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: onvif discovery sent Hello via eth0 as 172.25.1.99
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: also answering for openipc-fh8856v100.local.
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: joined 224.0.0.251 on eth0 (172.25.1.99)
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: announced openipc.local. and openipc-fh8856v100.local. as 172.25.1.99 on eth0
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: RTSP listener on IPv4 refused (Protocol not available); listening without SO_REUSEPORT
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: RTSP listener on IPv6 refused (Protocol not available); listening without SO_REUSEPORT
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: RTSP server started on port 554
Nov 17 19:59:22 openipc-fh8856v100 user.debug majestic[906]: DTLS identity: P-256, 289-byte certificate
Nov 17 19:59:22 openipc-fh8856v100 user.info majestic[906]: WebRTC fingerprint sha-256 24:78:56:4F:2E:E1:0E:21:FD:7C:1B:56:13:7A:5B:31:99:27:D2:FA:AB:2F:21:61:32:89:0F:0F:54:15:29:5E
Nov 17 19:59:23 openipc-fh8856v100 user.debug majestic[906]: wait pic_end fail, errno=0
Nov 17 19:59:23 openipc-fh8856v100 user.debug majestic[906]: announced again as 172.25.1.99 on eth0
Nov 17 19:59:24 openipc-fh8856v100 user.debug majestic[906]: wait pic_end fail, errno=0
Nov 17 19:59:24 openipc-fh8856v100 user.debug majestic[906]: onvif discovery sent Hello again via eth0
root@openipc-fh8856v100:~# strings /etc/crash/majestic.dump | grep -E '^(signal|code|addr|thread|version)='
signal=8
code=-6
addr=0x33c
thread=majestic
version=fullhan-v100-sensor-info+5705db24, 2026-11-17 08:21
root@openipc-fh8856v100:~# strings /etc/crash/majestic.dump | grep -E '^[0-9]{2}:[0-9]{2}:[0-9]{2} ' | tail -40
19:59:07 INFO <majestic> [main] main@1397: Majestic Lite for Fullhan (fh8852v100), version fullhan-v100-sensor-info+5705db24, built on 2026-11-17 08:21
19:59:07 INFO <majestic> [main] main@1430: Loading main configuration: /etc/majestic.yaml
19:59:07 DEBUG <majestic> [watchdog] watchdog_start@341: Detected Synopsys DesignWare Watchdog
19:59:07 INFO <majestic> [watchdog] watchdog_start@376: Watchdog armed: asked 300 s, accepted 300 s, fed every 150.000s, magic close supported
19:59:07 INFO <majestic> [licence] licence_evaluate@166: no identity: no chip identity on this platform
19:59:07 DEBUG <majestic> [httpd] start_web_server@812: Set server path: /var/www
19:59:07 INFO <majestic> [httpd] new_http_server@778: HTTP server started on :::80
19:59:07 WARN <majestic> [httpd] new_http_server@550: HTTPS port :::443 configured without cert/key; skipping TLS server
19:59:07 DEBUG <majestic> [sensor_lookup] sensor_hint@139: Autodetected sensor as 'gc4653_i2c'
19:59:07 INFO <majestic> [sdk] start_sdk@2438: Using gc4653_i2c sensor
19:59:07 DEBUG <majestic> [sdk] start_sdk@2503: 2560 1440 25 maxw 2560
19:59:07 INFO <majestic> [puts] @0: [dsp] version: V1.2.0(gcd1ec21),build: 2020-03-24
19:59:07 INFO <majestic> [puts] @0: [isp] version: V1.2.0(gaee6fa1),build: 2020-03-24
19:59:07 INFO <majestic> [puts] @0: [ispcore] version: V1.2.0(gaee6fa1),build: 2020-03-24
19:59:07 DEBUG <majestic> [sdk] try_to_load@311: trying to load /usr/lib/sensors/libgc4653_mipi.so
19:59:07 DEBUG <majestic> [log] printf@381: ADV_ISP version: V1.2.0(g4bb12fc)
19:59:07 DEBUG <majestic> [log] printf@381: probe sensor: gc4653_mipi
19:59:07 INFO <majestic> [puts] @0: [sensor] version: V1.2.0(gaee6fa1),build: 2020-03-24
19:59:07 INFO <majestic> [puts] @0: [mipi] version: V1.2.0(gaee6fa1),build: 2020-03-24
19:59:08 INFO <majestic> [sdk] sdk_apply_image_orientation@3043: picture upright
19:59:08 INFO <majestic> [commonsdk] publish_sps@219: H.264 SPS said nothing about picture reordering; publishing it with max_num_reorder_frames=0 (20 -> 23 bytes) so decoders that would otherwise hold a level's whole DPB show the first frame at once
19:59:08 INFO <majestic> [network] mcast_follow@859: joined 239.255.255.250 on eth0 (172.25.1.99)
19:59:08 INFO <majestic> [discovery] send_on@565: onvif discovery sent Hello via eth0 as 172.25.1.99
19:59:08 INFO <majestic> [mdns] init_owned_names@154: also answering for openipc-fh8856v100.local.
19:59:08 INFO <majestic> [network] mcast_follow@859: joined 224.0.0.251 on eth0 (172.25.1.99)
19:59:08 INFO <majestic> [mdns] announce@705: announced openipc.local. and openipc-fh8856v100.local. as 172.25.1.99 on eth0
19:59:08 INFO <majestic> [server] rtsp_listen@553: RTSP listener on IPv4 refused (Protocol not available); listening without SO_REUSEPORT
19:59:08 INFO <majestic> [server] rtsp_listen@553: RTSP listener on IPv6 refused (Protocol not available); listening without SO_REUSEPORT
19:59:08 INFO <majestic> [rtsp] rtsp_init@28: RTSP server started on port 554
19:59:08 DEBUG <majestic> [dtls] rtc_identity_new@257: DTLS identity: P-256, 290-byte certificate
19:59:08 INFO <majestic> [rtc] start_rtc@104: WebRTC fingerprint sha-256 F4:DA:38:85:1D:F8:44:72:CC:6B:83:47:13:3D:39:19:66:47:93:86:78:23:B6:1E:24:71:20:E8:1A:25:95:CA
19:59:08 INFO <majestic> [commonsdk] publish_sps@219: H.264 SPS said nothing about picture reordering; publishing it with max_num_reorder_frames=0 (20 -> 23 bytes) so decoders that would otherwise hold a level's whole DPB show the first frame at once
19:59:09 DEBUG <majestic> [mdns] announce@710: announced again as 172.25.1.99 on eth0
19:59:10 DEBUG <majestic> [discovery] send_on@559: onvif discovery sent Probe via eth0
19:59:10 DEBUG <majestic> [discovery] send_on@559: onvif discovery sent Hello again via eth0
19:59:11 DEBUG <majestic> [mdns] announce@710: announced again as 172.25.1.99 on eth0
19:59:12 INFO <majestic> [httpd] check_remember@3617: session restored from stay-signed-in token for root
19:59:12 DEBUG <majestic> [on_jpg] send_regular_jpeg@22: JPEG was sent to client
19:59:12 DEBUG <majestic> [sdk] capture_jpeg@2281: JPEG: 356184 bytes
@darrylb123 thanks, the log and dump answer the first two questions: the sensor now registers properly (no -2, no get_bayer_format is NULL) and the picture comes up upright. The crash is new information. I decoded your dump, and the divide by zero happens inside Fullhan's own ISP library, in its white-balance step.
The cause looks like the tuning file. There's no Fullhan tuning for the GC4653, so I borrowed the GC4633's. One byte in it tells the ISP library that the sensor driver supplies the white-balance gains. GC4633's driver does that; ours doesn't. So the ISP is left with zero gains and divides by one about 10 seconds in. Every other 2560x1440 tuning in Fullhan's SDK has that byte cleared, and test build 4 clears it too:
Or try only the corrected tuning on build 3, without reflashing
Only two small files change, and /etc is writable:
cd /etc/sensors
forfin gc4653_mipi_attr.hex gc4653_mipi_night_attr.hex;do
curl -L -o $f https://github.com/OpenIPC/openfullhan/releases/download/test-fh8856v100-gc4653-4/$fdone
sha256sum gc4653_mipi*_attr.hex # both should print b0ff6c82857b1668...
reboot
Reboot rather than restart majestic. After a crash the video hardware stays stuck until a reboot, which is the wait pic_end fail in your second run.
After it boots, please let it run for a couple of minutes and post the same output as last time:
logread | grep -i -E 'sensor|gc4653|mipi|isp|majestic|picture|crash'| head -80
Two more things that would help:
A snapshot from http://<camera-ip>/image.jpg. It works even when the video doesn't, and a colour cast or a missing colour would point at the next thing to fix.
Whether rtsp://<camera-ip>/stream=0 plays.
If it still crashes, the crash dump file from the web page again, as you did this time.
Activity
darrylb123 commentedon Nov 26, 2024
I have a number of the ASECAM Vatilon PB1 cameras. They can be affected by something on the network that causes them to all reboot at once. This can continue for some time, all rebooting regularly. I am willing to assist where I can. I upgraded the firmware using the Windows tool. It didn't help with the reboots but it may be the way to install alternative firmware.
mtrakal commentedon Nov 28, 2024
In the camera WebGui is possible to setup autoreboot. Maybe you have it enabled?
I have no issue with reboots of camera when this settings is disabled. But I have it enabled to reboot once per day to cleanup possible FW memleaks, sync time again, etc.
darrylb123 commentedon Nov 28, 2024
mtrakal commentedon Nov 29, 2024
Yeah, I already caught FW upgrade packets, but it's a bit useless because it's encrypted. So App and Cam changed the keys and session and encrypted the rest of the communication. So until we know the algorithm and how the key is build, we are lost I think this simple way for upgrade FW, but we have 2 others ⏬.
Discover camera is based on Broadcast on Network L2 (ARP broadcast on same network area):
{ "DeviceType": "IPC", "ChannelNum": 1, "DeviceModel": "PB4", "DeviceModel2": "PB", "DeviceName": "PB4", "CloudID": "__CLEARED__", "SerialNo": "__CLEARED__", "NetType": 1, "MAC": "__CLEARED__", "DHCP": 0, "Address": "10.0.0.33", "Submask": "255.255.255.0", "GateWay": "10.0.0.254", "DNS": "10.0.0.252", "TCPPort": 8600, "RTSPPort": 554, "HTTPPort": 80, "Language": "English", "SoftwareVersion": "V1.13.23-20240524", "DeviceRunTime": 78, "QWTEnable": false, "IPAdaptive": 1, "LampType": 1, "DayNightSwitch": 0, "P2PType": 4, "P2PPID": "__CLEARED__", "P2PQRCode": "__CLEARED__", "LensCount": 1, "StreamCount": 2, "StreamNumPerChannel": 2, "NetTypeSet": 1, "NetProtocolSet": 143, "Status": 2, "Version": 1, "Scode": "__CLEARED__" }After that application sends UDP packet to camera IP address with body:
{ "operation": "login", "param": { "user": "JmsRxTtVEW+J5da5aSt3lw==", "password": "bC+UKGMPDDiynIUE9ViJzw==", "heartbeat_timeinterval": 1, "heartbeat_timeoutcounts": 3, "sessionid": 179511328, "encryptionkey": "EOtDcH4DFPJZj2CDhLFNAg==", "roletype": 5 } }So we know, that decrypted
user = adminandpassword = 123456.Response is:
{ "operation": "login_response", "code": 0, "msg": "successful ", "sessionid": 179511328, "encryptionkey": "HO55v18ZnC9cL6pJe0U9Tg==" }And the rest is encryped with new key.
But we already have ROOT access to telnet, so I don!t think, that we need use standard way how to upgrade firmware, when you can (I supose) re-mount partition as write or use dd to rewrite flash memory. Not sure, but expecting, that with root bash / bussybox it is possible.
And second tested variant - the author of articles on Reddit already used UART and flash his own custom FW (changed default password to telnet with custom one). So flash new FW is not the main problem I think :)
darrylb123 commentedon Nov 29, 2024
L4ky commentedon Dec 28, 2024
Hi, i have a few Simicam cameras from Aliexpress.
They are Vatilon PB4 boards and various firmware version:
On Vatilon website there is 1.15.11 firmware version but still didn't tried it.
I have root access via telnet on port 2360.
Let me know if i can contribute in some way.
kellerman commentedon Feb 22, 2025
I have multiple Vatilon cameras, including PB4. They actually work pretty well if you block internet comms to them. I see we can already get to Uboot and have root access to the firmware. I can contribute. Nice cheap cameras these.
attila-lendvai commentedon Feb 23, 2025
i also have two ASECAM cameras that say they are PB1 in their web UI.
i'm relatively experienced in lowlevel hw hacking, i can also help, but i'm very new to openipc.
attila-lendvai commentedon Feb 24, 2025
i have two of these: https://www.aliexpress.com/item/1005006826151802.html
i have used
nmapto scan the ports, and 2360 was not listed as open, but i can indeed telnet to that port.but the
admin123456doesn't work there (it works through the web UI, though).any other ideas for login/pw?
L4ky commentedon Feb 25, 2025
Try root/ipc@hs66
attila-lendvai commentedon Feb 25, 2025
it worked! :) thank you!
crankmosh commentedon May 26, 2025
it seems with the v1.14 and v1.15 there is a bug that does not let you save the time settings on the webgui on the PB4... I have one camera with v1.13 and that works fine but I used IPC tool and tried to upgrade from v1.14 to v1.15 and neither allows me to save the time settings
oglabogla commentedon Oct 7, 2025
hello everyone i have a chinese camera which runs a V1.14.91-20241120, but everytime my nvr reboots (once a week) the camera fails to connect and needs a manual password input
can someone supply me with the V1.12 version
thank you
ETE-Design commentedon Nov 8, 2025
@mtrakal Any progress on this? Also have an Vatilon PB4 with the fh8856v200 so would be nice to flash OpenIPC on it :-)
tylerwight commentedon Nov 18, 2025
Just here to add some info as there isn't much on the internet. I have two of these cameras and can help any sort of testing if needed.
I bought some REVODATA fisheye cameras in 2022 that turned out to run these Vatilon boards. I can confirm I also had the same issue darrylb123 had where both of these cameras would just reboot on their own, and I didn't have auto reboot on. These cameras eventually broke and started boot looping, the webUI would show up just for a moment and I couldn't get in.
I didn't know much about them at the time, opened them up and bought new SOCs/mobos for them, so these cameras now run this module:
https://www.aliexpress.us/item/3256804887434971.html
in case the link goes down: FH8852V200+GC5053 V1.14.91-20241120 PA4 uboot-2016-22
Well these ones eventually broke in the exact same way so I dug further and found the info discussed here: the open telnet port and cracked root password. root/ipc@hs66
I got in and found the /tmp/ipcam binary was segfaulting on boot, which then caused it to restart forever. After reading the /usr/app/run.sh I found you can force a factory reset which I did, but still it was bootlooping. The fix for me was to manually edit the /tmp/etc/config/ipcam.conf and disable ONVIF. I don't know what changed, cause it worked before but if that camera booted with ONVIF it would segfault on startup, and no way to get in the webUI. My camera now boots and I just have to use RTSP
point being, if you have this Vatilon firmware and your camera boot loops, I fixed it by killing ipcam process to keep it up, editing the /tmp/etc/config/ipcam.conf to disable ONVIF. So having a nice working OpenIPC would be great, there are tons of cheap cameras out there that that use these chips =)
17 remaining items
darrylb123 commentedon Nov 17, 2026
fullhan-opensdk: build the V100 GC4653 sensor library from OpenIPC/op…
openipc-ai commentedon Nov 17, 2026
@darrylb123 thank you, that log helped a lot. It shows two things.
The memory layout fix (#2872) works on your PB1. The video memory pool is at
0xA1D00000~0xA7DFFFFF,load_fullhanreported no problem, and the video drivers loaded. That's the first time this has run on a real camera.The missing video has a different cause. majestic tried to load
/usr/lib/sensors/libgc4653_mipi.so, and that file doesn't exist for this chip. Fullhan's SDK for the FH8856 V100 ships sensor libraries for other sensors but not for the GC4653 in your camera, so the image never had one.#2879 adds one, built from source in a new repository, OpenIPC/openfullhan. It also fixes where the image installs the sensor tuning files, so the camera actually finds them.
Test build
This is a test build. The sensor driver has only been run in emulation, never on a camera, so yours would be the first.
https://github.com/OpenIPC/openfullhan/releases/tag/test-fh8856v100-gc4653-1
It is the normal
fh8856v100image plus the GC4653 library. It's for cameras installed with the fixed 8 MB layout from my steps above, as yours is. On the camera:If the camera can't reach GitHub, download
openipc.fh8856v100-nor-lite.tgzon your PC (don't unpack it). Runpython3 -m http.server 8000in that folder, then use--url=http://YOUR_PC_IP:8000/openipc.fh8856v100-nor-lite.tgz. Your settings and password are kept.If you'd rather not reflash, the release also has the three files on their own. Copy
libgc4653_mipi.soto/usr/lib/sensors/, andgc4653_mipi_attr.hexandgc4653_mipi_night_attr.hexto/etc/sensors/, then reboot.To go back,
sysupgrade -k -rinstalls the regular latest image.What to send back
rtsp://<camera-ip>/stream=0plays, and if it does, a snapshot fromhttp://<camera-ip>/image.jpg. If the picture comes up but the colours look off, that's expected: until someone tunes it, the colour tuning is borrowed from the closest sensor the SDK has, the GC4633.If there's no video, that log is what I need to fix it, for example if the MIPI speed setting is off. That setting is the one value no source states.
@mtrakal thank you for the PB4 data.
plat_id 0x19112201andFH8856V200confirm the PB4 is the newer V200 chip, unlike the PB1, so it uses thefh8856v200image, not this test build. Your/proc/driver/ispoutput also helped write this driver: the frame size it reports matches the timing the new GC4653 driver uses.@eribito your IF56N units run the same firmware as the PB1 dump, so this test build applies to them as well, whenever you get to the bench.
darrylb123 commentedon Nov 17, 2026
No video
fh8852v100/gc4653: set kick, which API_ISP_SensorKick needs
openipc-ai commentedon Nov 17, 2026
@darrylb123 thank you for testing so quickly, and sorry for the wait. Your log shows the new parts working on your camera, up to the very last step:
gc4653_i2c;/etc/sensors;Only the final start call fails:
API_ISP_SensorKick failed with -1.That's my mistake. In this chip's ISP library, that call runs a
kickfunction the sensor library is supposed to provide, and returns -1 when it is missing. I wrote the driver from Fullhan's own V100 example drivers, and neither of them provides it. The library now does: OpenIPC/openfullhan@1564015Test build 2
https://github.com/OpenIPC/openfullhan/releases/tag/test-fh8856v100-gc4653-2
Or copy just the new
libgc4653_mipi.sofrom the release to/usr/lib/sensors/and reboot. The tuning files haven't changed.Then the same as before: does
rtsp://<camera-ip>/stream=0play (and a snapshot fromhttp://<camera-ip>/image.jpg), and the output of:If it gets past the start call and still shows no picture, the next suspects are the MIPI speed setting and the timing, and that log is what will tell them apart.
darrylb123 commentedon Nov 17, 2026
No video or image
openipc-ai commentedon Nov 17, 2026
@darrylb123 sorry, that one is my fault again. Your camera is most likely still running test build 1. Both test builds carried the same version stamp, so
sysupgradesaw "Same version, nothing to update" and wrote nothing. Your log is identical to the first one, line for line.You can check which library you have:
15fd4c88…is test build 1 andd56fb0c1…is test build 2.The quickest way to get test build 2 is to replace just that one file, since nothing else changed:
curl -L -o /usr/lib/sensors/libgc4653_mipi.so https://github.com/OpenIPC/openfullhan/releases/download/test-fh8856v100-gc4653-2/libgc4653_mipi.so sha256sum /usr/lib/sensors/libgc4653_mipi.so # should print d56fb0c11eaa12f3... rebootI've also re-uploaded the full test build 2 image with its own version stamp, so the
sysupgradecommand from my last message now works too.After the reboot, the same two outputs as before would tell me where it stops next, if it does:
Plus whether
rtsp://<camera-ip>/stream=0orhttp://<camera-ip>/image.jpgshows anything.darrylb123 commentedon Nov 17, 2026
Initially asked for a login and password.
ran vlc http:///image.jpg and an image flashed up briefly
ran vlc rtsp:///stream=0 and it asked for login then said it couldn't display the video format.
Changed to H264 but can no longer get anything to work, video or image.
openipc-ai commentedon Nov 17, 2026
@darrylb123 that's real progress, thank you. With test build 2 the sensor is streaming. The image that flashed up in VLC was a real frame from the GC4653, and the log shows the encoder starting (
H.264 SPS ...). So the sensor setup, the MIPI link and the timing all work on your camera.What stops it now is that majestic crashes about 16 seconds in (
crashed (signal 8), a division by zero). After the automatic restart it gets no frames at all (wait pic_end fail), which is why nothing worked after you switched to H.264. Keeping H.264 is fine.Two lines just before streaming starts are related:
FHAdv_Isp_SensorInit failed, ret = -2andsensor function get_bayer_format is NULL. They show a vendor call that registers the sensor's colour-order functions failing in majestic's start-up. That affects every sensor on this chip, not just yours, and it's being looked at on the majestic side.majestic saved a crash dump, and that's what will show where the division by zero happens. Could you please:
gzip -c /etc/crash/majestic.dump > /tmp/majestic.dump.gz. Copy it to your PC (for examplescp root@<camera-ip>:/tmp/majestic.dump.gz .) and drag it into a comment here. It holds the registers and stack at the moment of the crash, which the readable part doesn't.If the camera crashes again after the reboot, the dump will be from that crash, which is fine.
openipc-ai commentedon Nov 17, 2026
@darrylb123 there's a test build 3 with a fix for the
FHAdv_Isp_SensorInit failed, ret = -2line from your log. It contains a test majestic that registers the GC4653 library with the vendor's ISP layer correctly on this chip. That registration failed on every start before, and it's the most likely cause of the crash at 16 seconds. The sensor library is the same as in build 2.https://github.com/OpenIPC/openfullhan/releases/tag/test-fh8856v100-gc4653-3
This image has its own version stamp, so
sysupgradewill actually install it this time:Or try only the new majestic, without reflashing
It's 1.6 MB, too big for the camera's writable space, so this runs it from RAM until the next reboot:
curl -L -o /tmp/majestic https://github.com/OpenIPC/openfullhan/releases/download/test-fh8856v100-gc4653-3/majestic sha256sum /tmp/majestic # should print 641cd99ddb17a253... chmod +x /tmp/majestic /etc/init.d/S95majestic stop mount --bind /tmp/majestic /usr/bin/majestic /etc/init.d/S95majestic startIf that gives
wait pic_end failagain, the video drivers are still stuck from the last crash, and the full image above is the better test.After it boots, please let it run for at least a minute and post:
I'm looking for three things:
FHAdv_Isp_SensorInitwithout-2, and noget_bayer_format is NULL.picture uprightline instead ofmirror 0, flip 0 not set.rtsp://<camera-ip>/stream=0play (H.264 is fine), and doeshttp://<camera-ip>/image.jpgshow a picture?If it still crashes, the readable part of the new crash dump would help as before:
darrylb123 commentedon Nov 17, 2026
The dashboard displays an image. Image cannot be seen over http and no video.
darrylb123 commentedon Nov 17, 2026
Crash dump from web browser dump page
majestic_172.25.1.99_2026-11-17_20-11-26.dump.gz
fh8852v100/gc4653: leave white balance to the ISP, as the other 2560x…
openipc-ai commentedon Nov 17, 2026
@darrylb123 thanks, the log and dump answer the first two questions: the sensor now registers properly (no
-2, noget_bayer_format is NULL) and the picture comes up upright. The crash is new information. I decoded your dump, and the divide by zero happens inside Fullhan's own ISP library, in its white-balance step.The cause looks like the tuning file. There's no Fullhan tuning for the GC4653, so I borrowed the GC4633's. One byte in it tells the ISP library that the sensor driver supplies the white-balance gains. GC4633's driver does that; ours doesn't. So the ISP is left with zero gains and divides by one about 10 seconds in. Every other 2560x1440 tuning in Fullhan's SDK has that byte cleared, and test build 4 clears it too:
https://github.com/OpenIPC/openfullhan/releases/tag/test-fh8856v100-gc4653-4
Or try only the corrected tuning on build 3, without reflashing
Only two small files change, and
/etcis writable:Reboot rather than restart majestic. After a crash the video hardware stays stuck until a reboot, which is the
wait pic_end failin your second run.After it boots, please let it run for a couple of minutes and post the same output as last time:
Two more things that would help:
http://<camera-ip>/image.jpg. It works even when the video doesn't, and a colour cast or a missing colour would point at the next thing to fix.rtsp://<camera-ip>/stream=0plays.If it still crashes, the crash dump file from the web page again, as you did this time.