[pskmail] Re: FLDIGI Hangs
- From: Remi Chateauneu <remi.chateauneu@xxxxxxxxx>
- To: pskmail@xxxxxxxxxxxxx
- Date: Thu, 22 Mar 2012 11:36:50 +0000
Could you maybe try to catch it ?
http://sunsite.ualberta.ca/Documentation/Gnu/gdb-4.18/html_chapter/gdb_6.html
gdb> handle SIGPIPE stop
So we could see which thread is complaining.
Thanks
R
On Thu, Mar 22, 2012 at 10:26 AM, Rein Couperus <rein@xxxxxxxxxxxx> wrote:
> Here is the first example of wrong error handling in fldigi using the ARQ
> interface:
>
> Program received signal SIGPIPE, Broken pipe.
> 0x0012d422 in __kernel_vsyscall ()
> (gdb)
>
> When the error would be handled properly, the display would not freeze...
>
> BTW, when restarting the program, gdb says:
> The program being debugged has been started already!
> Do you want to start from the beginning? (y or n) y
>
> I guess this is a general problem related to all error handling in the ARQ
> interface...
> It also happens with errors in the socket and sound interfaces.
>
> Rein PA0R
>
>
>
>
>
> 1. If fldigi hangs, the display is not updated anymore (see attached
> picture).
> 2. I have run it under gdb before... the event log shows incoming ARQ
> messages.
> (e.g. mdde switching every minute).
>
> 3. Yes, you normally see outgoing messages in the output window.
> 4. Either sound errors or rigctl socket errors.
> 5. PI4TUE uses soft PTT via hamlib.
> 6. PI4TUE uses the USB soundcard in the IC7200.
> 7. Fldigi uses the USB sound driver of ubuntu 10.04 LTS.
>
> PI4TUE is now running under gdb.
>
> Rein PA0R
>
>
>
> "It only happens when using the ARQ interface for pskmail."
>
> Ah, ok, I never used this part. I have to investigate a bit the code.
> Extra questions please:
>
> "It is always related to PTT (fldigi starts sending)."
>
> That is: jPSKMail sends a message through the ARQ interface. And the
> problem happens sometimes only ?
>
> "The thread running X stops, most other threads (e.g. the ARQ thread)
> still work."
>
> (1) The display is frozen ?
> (2) How do you know the other threads are running ? Do they answer ?
> (3) When jPSKMail can succesfully send through the ARQ interface, is there
> anything visible on the X display, apart from the normal TX red spot ? And
> is the red spot visible ?
>
> (4) "I will start PI4TUE under gdb, "
>
> User PI4TUE mentionned in
> https://docs.google.com/spreadsheet/ccc?key=0ApD_rVrTY8hNdGc4alVqelRMY3BxR1RfeDdvUTJXOXc#gid=0
>
> "E: trx_trx_receive loop: Sound error: Connection timed out."
>
> Any other detail about this specific problem ? (This is called precisely
> when transmitting).
>
> Aprt from running gdb, can someone do it under valgrind ? (Which detects
> uninitialised variables).
>
> Something interesting in the log window ?
>
> Thanks
>
> R
>
>
> On Wed, Mar 21, 2012 at 4:37 PM, Rein Couperus <rein@xxxxxxxxxxxx> wrote:
>
>> It only happens when using the ARQ interface for pskmail.
>> It also happens with hamlib, I never tried rigcat.
>> It is mainly happening with the pskmail server. The server does NOT use
>> the xmlrpc interface.
>> It is always related to PTT (fldigi starts sending).
>> The thread running X stops, most other threads (e.g. the ARQ thread)
>> still work.
>> I will start PI4TUE under gdb, to show what happens.
>> As far as I can remember it just says thread 'xxx' stopped.
>>
>> Rein PA0R
>>
>>
>>
>>
>> I do not use rigcat, just hamlib. Maybe this is the reason why the very
>> rare problems I had had nothing to do with yours.
>>
>> Some questions please.
>>
>> (1) Could someone please try to do the same tests without rigcat ?
>>
>> (2) I am about to commit a new version which can create a core dump when
>> receiving a signal (And with the correct environment variable). When
>> available, it would be nice if some used it and tried to reproduce the
>> problem.
>>
>> Cheers
>>
>> R
>>
>> On Wed, Mar 21, 2012 at 2:51 PM, Pär Crusefalk <per@xxxxxxxxxxxx> wrote:
>>
>>> Hi Eric,
>>>
>>> I have no solution for you, just wanted to let you know that you are not
>>> alone in experiencing this. I tried to debug and find the cause a while
>>> ago and came as far as pointing the finger at PTT handling in rigcat
>>> (which is used even if rigcat is not used). I also tried to gather
>>> information about other affected stations to see if we had something in
>>> common that triggers the issue. Can't say I found any smoking gun there
>>> but you are more than welcome to add your data to our finds:
>>> http://goo.gl/6vNPT
>>>
>>> The bottom line is that fldigi has had this issue a long time and
>>> nothing is being done about it. The way ahead for us is likely that we
>>> get rid of fldigi and move to a java based modem.
>>>
>>> 73 de Per. sm0rwo
>>>
>>>
>>>
>>>
>>>
>>> Eric Davenport skrev 2012-03-21 15:40:
>>> > Good Morning,
>>> >
>>> > I have been working with fldigi+jpskmail for some time.
>>> >
>>> > I have been trying to find an energy efficient way of running a server
>>> > so I have been working on virtual machines and standalone computers for
>>> > a while.
>>> >
>>> > I first tried to install it on Ubuntu 10 on a VM Server I gave up when
>>> I
>>> > could not get the Signalink USB to transfer any information. Same
>>> > situation using VirtualBox.
>>> >
>>> > So now I am trying a Laptop running Ubunto 10.04 fldigi 3.21.40 from
>>> > source and using fltk 1.3.
>>> >
>>> > So to test it I am using jPSKMAIL 1.5.7
>>> >
>>> > I am running fldigi minimized but that does not seem to be the problem.
>>> >
>>> > This works for 2 or 3 hours and fldigi freezes.
>>> >
>>> > Nothing in the Logs anywhere it just stops.
>>> >
>>> > Maybe this is related to running jPSKMAIL and would not happen with the
>>> > server.
>>> >
>>> > When I ran server some years ago I do not recall any problems and it
>>> ran
>>> > on a much slower machine at the time for weeks on end.
>>> >
>>> > I have seen references by Rein and others to freezing and wonder if
>>> > their is something I need to change that I do not remember.
>>> >
>>> > Maybe I should just get the server up and running and see if that
>>> freezes.
>>> >
>>> > Any suggestions are appreciated.
>>> >
>>> > Eric
>>> >
>>>
>>>
>>> --
>>> Mobile: +46703784299<#13639f149980f9cc_13636b936a6b730c_136361ee36ac5f36_>
>>>
>>
>>
>
Other related posts:
