search

g4klx / MMDVMHost

The host program for the MMDVM
GNU General Public License v2.0
379 stars 274 forks source link

Lots of "MMDVM RX buffer has overflowed" #327

Open m1geo opened 7 years ago

m1geo commented 7 years ago

Hi folks. I have an issue running

M: 2017-08-13 23:24:37.636 MMDVMHost-20170719 is starting M: 2017-08-13 23:24:37.636 Built 00:21:58 Aug 14 2017 (GitID #3749b8d)

and

I: 2017-08-13 23:24:39.650 MMDVM protocol version: 1, description: MMDVM 20170501 TCXO (D-Star/DMR/System Fusion/P25/RSSI/CW Id) GitID #04ca97e

together. I get pages and pages of

E: 2017-08-13 23:27:39.695 MMDVM RX buffer has overflowed E: 2017-08-13 23:27:40.449 MMDVM RX buffer has overflowed E: 2017-08-13 23:27:41.205 MMDVM RX buffer has overflowed

I have rolled the MMDVM modem back to:

I: 2017-08-13 23:34:29.405 MMDVM protocol version: 1, description: MMDVM 20170501 TCXO (D-Star/DMR/System Fusion/P25/RSSI/CW Id) GitID #669acb9

which was known good and the problem persists (although not as bad; I get a few chunks of the error [typically 3-4 seconds in 500ms chunks {see attached}] following any D-STAR activity, and then it stops). Thus I assume it to be related to MMDVMHost #3749b8d.

Error is repeatable, but on a deployed repeater. This was fine before updating to the latest two commits.

George M1GEO.

screenshot_2017-08-14_00-49-27

g0wfv commented 7 years ago

George,

Either you haven't done a 'make clean' before recompiling, or you have an older firmware ...

I: 2017-08-13 23:24:39.650 MMDVM protocol version: 1, description: MMDVM 20170501 TCXO (D-Star/DMR/System Fusion/P25/RSSI/CW Id) GitID

04ca97e

The highlighted date should change with a good compile to latest code.

What platform is this on (DUE, STM, Teensy etc) and how do you compile (CLI, IDE etc)

Tony

m1geo commented 7 years ago

In which case, "make -f Makefile.Arduino clean" doesn't work correctly. I certainly did do a make clean with the Arduino Makefile.

I re-did the modem build from scratch, make clean and make -f Makefile.Arduino clean followed by make -f Makefile.Arduino {complie,upload} and flashed the modem again. The flash succeeded and verified OK, but, I still get the following:

I: 2017-08-14 11:02:44.958 MMDVM protocol version: 1, description: MMDVM 2017050 1 TCXO (D-Star/DMR/System Fusion/P25/RSSI/CW Id) GitID #04ca97e

George.

m1geo commented 7 years ago

@g0wfv Looking at what yous aid, the version date of MMDVM hasn't been bumped for 4 months: https://github.com/g4klx/MMDVM/blob/master/SerialPort.cpp

phl0 commented 7 years ago

Ok maybe the Makefile.Arduino has a problem. I will investigate.

m1geo commented 7 years ago

@phl0 I don't think it has. Looking at the source file, the version hasn't been touched in 4 months. The git-hash is correct, so that part must have been recompiled. However, I'm not sure the Makefile.Arduino file does a make clean correctly? It doesn't appear to remove object files for me, but maybe I'm mad...

phl0 commented 7 years ago

Oh so the Makefile does not have a problem then?

m1geo commented 7 years ago

You know, I think this is more an MMDVMHost issue; it errors constantly from startup until something D-STAR says/does something. And then, it's fine for a while (unsure if connected with hang times or not). Latest origin master on modem and host.

image

kc2vrj commented 7 years ago

I have the same issues since my update this week

g0wfv commented 7 years ago

@m1geo I do apologise! Looks like @g4klx didn't bump the firmware version at the same time as all the other components. MMDVM hasn't been bumped since 1 May (at a quick glance) whereas everything else got bumped on 19 Jul.

Not sure if this was intentional, Jon, but it's easier to identify compatible software/firmware if versions match?

As for the original problem, George, any movement towards a solution? (My make clean comment is most likely a red herring taking into account the above revelation re version numbers!)

m1geo commented 7 years ago

No problem @g0wfv - I did double/treble check I wasn't doing anything stupid, and I'm fairly certain I'm not (famous last words!). I still all works, I can find nothing different to previous builds, but, it spews this error out regularly for a while, then stops, then starts again. The log files have become orders of magnitude larger! ;) Needs someone who better understands the system to have a nose. I'm a bit out of touch and snowed under at work! :(

g4klx commented 7 years ago

I didn't bump the version on the firmware as I didn't want to make it an official release. Its still work in progress regarding the boxcar and the D-Star correlator. The problem is that the Due doesn't have enough horsepower for the new features so I need to look at optimizing the processing. I'd go with the master branch and go back in time until you find a version that doesn't exhibit these faults. By all means use the latest host, although there are some significant YSF upgrades coming to that fairly soon.

m1geo commented 7 years ago

Thanks, @g4klx - this was/is on the master branch, but the latest commits - I wasn't sure what it was. At least now we know there's an issue. Frustratingly I have an Arduino on KH, but, I could easily upgrade that to STM32 board in the near future, if that's the way forward. For now, I'll revert back to an earlier modem. I guess it's safe (protocol wise) to run the latest version and an earlier modem?

g4klx commented 7 years ago

I'm still on the Arduino on my development system. Don't go back too far with the firmware, but you shouldn't need to back too far to ensure that things are OK. I suggest start a few weeks ago as there was a brief period when DMR was broken.

m1geo commented 7 years ago

Thanks Jon. I'm currently running the latest once and ignoring errors - it seems to be working fine..?

iw3row commented 6 years ago

I'm experiencing the same issue, the log is full of "MMDVM RX buffer has overflowed" (once per second or so). The message seems to almost disappear during the DMR signal reception. I'm also having random DMR decoding issues, not sure if this can be in some way related.

immagine

Latest versions of MMDVM and Host installed. I've the same issue on ZUM board with Nucleo F446RE and STM32 DVM board. I think on the Nucleo that mount a powerful STM32 processor the message rate is lower, may this help?

UPDATE: keeping the same hardware, Ive changed the sd card on raspberry with latest version of pi-star and the problem disappeared, so it seems to be a problem related to a specific installation or version.

JoshuaCurry commented 4 years ago

I had this issue, and turns out in the default config, MMDVMHost is set to use /dev/ttyAMA0 for [TFT Serial], and I was also using that for my MMDVM Board (raspberrry Pi hat). I commented out the lines for TFT Serial and solved the issue for me.