[mgetty]: Problems with hanging sendfax (Solaris 2.5)

• Messages sorted by:[ date ][ thread ][ subject ][ author ]

Hi Christian,

Christian Baur wrote:
> I have a problem with sendfax from the mgetty distribution (0.99-Aug96)
> and I didn't find a solution yet (although I already browsed through the
> mail-archive of this list).

It's a quite unusual problem.

> I am using sendfax under Solaris 2.5 with a ELSA Microlink 28.8 TQV
> (fax-device ttyb). [..]

Could be a "Solaris 2.5 has broken serial drivers" problem, but maybe
not. Doesn't feel like a typical solaris problem.

[..]
> Here is a typical sequence from the logfile:
> 
> 09/05 21:15:01 tyb  fax_wait_for(CONNECT)
> 09/05 21:17:01 tyb  Warning: got alarm signal!
> 09/05 21:17:01 tyb  mdm_read_byte: read returned -1: Interrupted system
> call
> 09/05 21:17:01 tyb  mdm_get_line: cannot read byte, return: Interrupted
> system call
> 09/05 21:17:01 tyb  AT+FDT -> some error (-6), abort fax send!
> 09/05 21:17:01 ##### failed transmitting f1.g3: phone="4293003",
> +FHS:-6, dev=ttyb, time=120s, acct=""
> 09/05 21:17:01 tyb  fax_send: 'AT+FCLASS=0'

I can't really say what could be causing this, but I'd guess:

 - hardware flow control mixed up (bad modem cable, bad plug, bad modem
   firmware...), modem sees/thinks "RTS low" and doesn't send data to the 
   host --> timeout.  Host sees "CTS low" and can't send out the final 
   AT+FCLASS=0 command.

   Please check this with a RS232 line tester.


 - SunOS 4.x and Solaris 2.4 are known to behave badly if RTS/CTS flow
   control is activated but the DCD line drops (can't receive any
   characters from the port any more). So, what I could imagine here
   is that the modem fails to properly start the pre-page handshake
   initiated by AT+FDT, lowers DCD, and *then* sends an error string
   to the Sun, which can't be read.

   Please try #define FAXSEND_FLOW FLOW_SOFT in policy.h, and recompile.
   While in the general case hardware flow control is superiour, for
   fax sending software flow control usually works fine as well, and
   will work around this Sun problem (if it's this).

   Why this bug would cause the AT+FCLASS=0 to get stuck as well, I
   don't know. Maybe the ELSA lowers CTS as well when DCD goes low,
   didn't investigate this enough yet.


Please try this, and if it doesn't help, I'll try to think up a new
"guess".

gert
-- 
Gert Doering
Mobile communications ... right now writing from *AWAY* :-)) 
phone me: +49 8251 4470   or mail me:  gert@greenie.muc.de