"Segmentation Fault" Error with DSD 1.4.1 Ubunto

[fdnybuff18]

Getting a Segmentation Fault error after the program monitoring a channel with TRBO Data/Signal for about 5 minutes? Program then stops running. I noticed this only on TRBO Channels that are running constant data.

Any thoughts?

 

[n4yek]

If you come up or get a solution, please post it. I asked this same question a few months ago with no solution.

 

[mtindor]

If you come up or get a solution, please post it. I asked this same question a few months ago with no solution.

I _think_ that when I ran it using a few switches, it didn't happen -- but if I ran it without any switches it did happen. I'll test again today and let you guys know what specific switches I am using when it does _not_ crash.

Mike

 

[mtindor]

I _think_ that when I ran it using a few switches, it didn't happen -- but if I ran it without any switches it did happen. I'll test again today and let you guys know what specific switches I am using when it does _not_ crash.

Mike

I was wrong. Switches or no switches, it still bails. It only does that on control channels spewing constant info. I never have it crap out on a voice transmission. Just long term constant CC info.

Mike

 

[troffasky]

You will probably need to run it with a debugger and post the output here when it segfaults, unless you can work out exactly what input is causing it to segfault.

 

[mtindor]

You can set up a batch file to restart it by creating a batch file in your dsd working directory:

* dmr.bat is the name of the batch file I run
* my working directory is c;\dsd -f yours is something else change cd\dsd
* the switches are specific to TRBO (with -xr because I need to switch invertedness on my particular audio card)
* I renamed my dsd.exe [from woodpecker, with filter] to dsd_filter.exe

echo off
set /a i=0

:loop
if %i%==1000 GOTO END
echo This is iteration %i%
cd\dsd
dsd_filter -ff -mg -xr -i /dev/dsp -o /dev/dsp
set /a i=%i%+1
GOTO loop

:end
echo That's It!

- this will restart it 1000 times (modify 1000 to set number of times you want it to restart)
- doesn't matter what you echo - you don't even need echo
- after all, you'll never see what's echoed during restarts most of the time

Mike

 

[slicerwizard]

The error is fairly obvious; the DSD authors allocate a 1,000,000 element array and point dibit_buf at its first entry; they then use the pointer dibit_buf_p to walk through the array, storing dibit values as they're generated by getDibit() or getFrameSync(); this is done so that they can step backwards to the start of a DMR frame when they find a DMR sync pattern (which is always located in the middle of a DMR frame)

Under most circumstances, frame decoding (of any supported protocol) stops when a channel drops (RF carrier goes away) and this leads to the noCarrier() function being called, where dibit_buf_p is reset to point back near the start of the array.

When monitoring DMR channels that don't go away (e.g. control channels), dibit_buf_p eventually walks off the end of the 1,000,000 allocated storage elements and starts clobbering whatever lays in its path. Before long, you get a segfault.

 

[fdnybuff18]

You can set up a batch file to restart it by creating a batch file in your dsd working directory:

* dmr.bat is the name of the batch file I run
* my working directory is c;\dsd -f yours is something else change cd\dsd
* the switches are specific to TRBO (with -xr because I need to switch invertedness on my particular audio card)
* I renamed my dsd.exe [from woodpecker, with filter] to dsd_filter.exe

echo off
set /a i=0

:loop
if %i%==1000 GOTO END
echo This is iteration %i%
cd\dsd
dsd_filter -ff -mg -xr -i /dev/dsp -o /dev/dsp
set /a i=%i%+1
GOTO loop

:end
echo That's It!

- this will restart it 1000 times (modify 1000 to set number of times you want it to restart)
- doesn't matter what you echo - you don't even need echo
- after all, you'll never see what's echoed during restarts most of the time

Mike

Mike,

This worked awesome.....

Thank you

 

[mtindor]

Mike,

This worked awesome.....

Thank you

You're welcome. I'm glad it was of use to someone else In case anyone wondered, I just stole an example windows batch file loop that I found on the internet and modified things so that it restart my DSD.

Mike

 

[n4yek]

Fixed my problem, thanks for the code.

 

[krokus]

The error is fairly obvious; the DSD authors allocate a 1,000,000 element array and point dibit_buf at its first entry; they then use the pointer dibit_buf_p to walk through the array, storing dibit values as they're generated by getDibit() or getFrameSync(); this is done so that they can step backwards to the start of a DMR frame when they find a DMR sync pattern (which is always located in the middle of a DMR frame)

Under most circumstances, frame decoding (of any supported protocol) stops when a channel drops (RF carrier goes away) and this leads to the noCarrier() function being called, where dibit_buf_p is reset to point back near the start of the array.

When monitoring DMR channels that don't go away (e.g. control channels), dibit_buf_p eventually walks off the end of the 1,000,000 allocated storage elements and starts clobbering whatever lays in its path. Before long, you get a segfault.

They are using an array, instead of a bi-directional linklist?