AMBE Server stops working

[dl1esz]

AMBE Server on Raspberry Pi3 stops working
aktuall raspian AMBE device: AMBE3000 Hardware on Arduino Duo Software from DVMEGA
AMBEServer -d -p2461 compiled for 230400 Baud. Power option on Raspi ok no rttheling voltage ok.
When stops AMBEServer (maybe after hours, simulated by hig loads of cpu on rpi) following Error Messaege is seen

Serial payload exceeds buffer size:65332
processSerial failed to read packet

a new AMBEServer start brings it to work for a while.

If i use a DV30 USB AMBE Stick (460800 Baud) this issue is not given only by using AMBE3000 shild with arduino (230400Baud)
On Windows BlueDV Software connected to USB the Arduinoshild with AMBE3000 work fine without stoping.
Different RPI ( Pi zeroW, RPI2b) frsh installation and each fresh compiled comes to the same result

[lwkoon]

I am also having some issue on the AMBEserver service is running but actually dying, running -x for now (debugging mode), the only way to solve is to restart the AMBEserver process.

a script of checking AMBE process won't help as the PID is still runing on linux (im running raspbian on rpi 3b+)

[pa7lim]

The traffic to the AMBEServer is UDP so not reliable. If the packet is corrupted and the AMBEServer is receiving an AMBE packet length of e.g. 322 but the AMBE header is 65132 ( e.g. due bad internet) ( see crash report), the AMBEserver will crash.
The only way to fix this is by making a good check/parsing of the AMBE packet before sending it to the AMBE3000. This also prevents the lock up of the AMBE3000 due a bad internet connection.

[lwkoon]

packet

I am not good at C language but i tot some checking of length payload check is there?

if (bytesRead != dv3k_packet_size(packet)) {
fprintf(stderr, "AMBEserver: invalid length when reading from the socket, %zd=%zd", bytesRead, dv3k_packet_size(packet));
return 1;
}

[dl1esz]

Hi Eddie and David. I saw this line also in the code and what David describes is also a good idea. But what is the different between the DV30 and the AMBE3000 Shield communication ,why i ask this, the DV3000 shield with the Arduino 2560 give me the impression that always after a day ambeserver stucks i can simulate this with high Load on the rpi and bring the ambeserver with this high load to break down, but if i connect the DV30 USB Stick instead the Arduino/AMBE3000 shield everything works well on same rpi, only different Baudrate ( the DV30 USB Stick runs native on 460800Baud, the AMBE3000/Arduino have only 230400 Baud). So if this issue depends on loss of UDP Packets  (was also my first thought) it must affect also the DV30 sollution. I am not shure if a change the data cache payload will helps here? A implementation of a renudancy check (checksum between packed TRX and Packet recieved) will also overflow my expierience in C#. @david: your implementation AMBE Server on Windows APP, will have also this issue reported from users with AMBE 3000Shield with Arduino 2560/CH340 Controller, maybe you have also in Win APP AMBESERVER adapted the same  AMBEServer C code? In the meantime i wrote a small script and let all 5minutes (by cron) check if the demon service of ambeserver its allive, ive its brocken script will restart ambeserver, but thats ugly ;-)) Am 22.05.19 um 16:02 schrieb Eddie:

…

I am not good at C language but i tot some checking of length payload check is there? if (bytesRead != dv3k_packet_size(packet)) { fprintf(stderr, "AMBEserver: invalid length when reading from the socket, %zd=%zd", bytesRead, dv3k_packet_size(packet)); return 1; } — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#5?email_source=notifications&email_token=AMDEK6KTZPWJ7Y6252XY6H3PWVG57A5CNFSM4HNVQVX2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODV7EWMI#issuecomment-494816049>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AMDEK6JFNLO5Z5A7VOSCCR3PWVG57ANCNFSM4HNVQVXQ>.

[marrold]

Hi all, apologies I seem to have missed this thread. I am not so active with DMR now, my AMBE chips are in storage and I'm not very good at C!

@lwkoon can you create a fork with your suggested change and then test it to see if it resolves the issue?

Thanks

[lwkoon]

After glance through with my limited knowledge of C++ from the code:-

the return of the checking of processSocket() should be return 0 instead of 1 after checking
maybe i try to put return 0 of function packet.start_byte != DV3K_START_BYTE and bytesRead != dv3k_packet_size(packet) and recompile and see how it goes

int processSocket(int sockFd, int serialFd)
{
socklen_t addrLen = sizeof(struct sockaddr_in);
ssize_t bytesRead;
struct dv3k_packet packet;

bytesRead = recvfrom(sockFd, &packet, sizeof(packet), 0, (struct sockaddr *) &sa, &addrLen);
if (bytesRead < 0) {
    fprintf(stderr, "AMBEserver: error when reading from the socket: %s\n", strerror(errno));
    return 0;
}

if (packet.start_byte != DV3K_START_BYTE) {
    fprintf(stderr, "AMBEserver: invalid start byte when reading from the socket, 0x%02X", packet.start_byte);
    return 0;
}

if (bytesRead != dv3k_packet_size(packet)) {
    fprintf(stderr, "AMBEserver: invalid length when reading from the socket, %zd=%zd", bytesRead, dv3k_packet_size(packet));
    return 0;
}

if (debug)
    dump("Socket data", &packet);

if (write(serialFd, &packet, dv3k_packet_size(packet)) != dv3k_packet_size(packet)) {
    fprintf(stderr, "AMBEserver: error when writing to the serial port: %s\n", strerror(errno));
    return 0;
}

return 1;

}

[lwkoon]

solved:
edit /boot/config.txt

Add
init_uart_clock=50000000