Re: flexfax: faxq hangs...

[Jay Davis <Jay_Davis@compuserve.com>]

Well, wherever you are, I'm there too. This is the first time I'm trying
to run this software and I was sort-of hoping I wouldn't have to debug
it before getting one fax out. Just not my week.

Here is the info on my setup:
Red Hat Linux 6.0
Hylafax 4.1beta2 compiled with gcc version egcs-2.91.66
USR Sportster Vi 28.8 (doesn't matter since it never gets that far).

I've tested the modem and serial ports out and they seem fine. I started
the modem with a simple faxmodem command to reduce the number
of things involved in the problem. I checked port.h and CONFIG_OPENFIFO
is set to O_RDWR. The build when smoothly except for having to fix the TIFF
version check in the configure script. I send a text file with the sendfax program
to a local number. Here's what I get:

The main faxq process seems to keep running properly.
The child faxq process, runs fine while creating the faxable
     images, but then continues to run at 90-95% cpu
     indefinately (at least an hour).
The files in the docq directory are as follows:
-rw-rw----   1 uucp     60002       10889 Dec 23 23:35 doc1.cover
-rw-------   1 uucp     wheel       12718 Dec 23 23:35 doc1.cover;30
-rw-rw----   1 uucp     60002       23992 Dec 23 23:35 doc1.ps.1
-rw-------   1 uucp     wheel      166695 Dec 23 23:36 doc1.ps;30
-rw-------   1 uucp     60002           1 Dec 23 23:35 seqf
I checked the doc1.cover;30 and doc1.ps;30 files with viewfax and they
     look OK. I had full logging on the the message log showed ps2fax running
     properly on both files. It also showed the modem being ready and the job
     being submitted.
The sendq directory, as expected, just has:
-rw-rw----   1 uucp     60002         685 Dec 23 23:35 q1
-rw-------   1 uucp     60002           1 Dec 23 23:35 seqf
The queue status shows the job as being process.

So everthing looks fine except that the child faxq process is stuck. It has the following
files open:
lrwx------   1 root     root           64 Dec 23 23:41 0 -> /dev/null
lrwx------   1 root     root           64 Dec 23 23:41 1 -> /dev/null
lr-x------   1 root     root           64 Dec 23 23:41 10 -> /var/spool/fax/docq/doc1.cover;30
lrwx------   1 root     root           64 Dec 23 23:41 2 -> /dev/null
lr-x------   1 root     root           64 Dec 23 23:41 3 -> pipe:[793]
lr-x------   1 root     root           64 Dec 23 23:41 4 -> pipe:[796]
lrwx------   1 root     root           64 Dec 23 23:41 5 -> /dev/null
lrwx------   1 root     root           64 Dec 23 23:41 6 -> /var/spool/fax/FIFO
lrwx------   1 root     root           64 Dec 23 23:41 7 -> socket:[850]
l-wx------   1 root     root           64 Dec 23 23:41 8 -> /var/spool/fax/client/723 (deleted)
lrwx------   1 root     root           64 Dec 23 23:41 9 -> /var/spool/fax/sendq/q1

If you notice, in this case, it also has a deleted file open in the client directory. Now, since
I have never used the program before and I just started looking at the source, I have
no idea what it is supposed to be doing next. It doesn't appear as if any attempt
to use the modem is made. What happens between creating the fax images and accessing
the modem?

So, any ideas? Anyone?

Jay Davis
 
 

Tom Lear wrote:

 
Okay here's the deal, I can get to the modems with other programs, and
faxgetty reports valid states, but when I send a fax faxq spawns a child
that grabs all the cpu.

An strace shows this faxq to not be making any system calls.  I need to
kill this process and the faxgetty before I can kill the job.

Please CC me as I'm not on the list...

Hylafax version: 4.0.2
Linux 2.2.9 #7 SMP Sat Oct 2 17:56:48 PDT 1999 i686 unknown

                                                        - Tom