Re: [hylafax-users] MultiTech MT5634ZPX-PCI-V92 and Wedging

["Mike McMullen" <mlm@xxxxxxxxxxxxxxxxxx>]

> Mike,
> We have three of these running in production in one box with no issues 
> with wedging.  As I mentioned, they are the -PCI not the -U versions, but 
> that seems to be a small difference.
>
> I thought I'd give you some info to compare to:
>
> Here is what they look like in lspci -vv (note that mine get mapped into
> high IRQs.  The fact that yours are on 5 and 9 bother me.  I know that 5
> is often used for printers, and I thought that 2 and 9 were used to shift
> registers, but this is very old data in the depths of a buggy mind.
>
>
> If you want other examples, just ask.
>
> Bill

<PCI Output Removed>

Hi Bill,

My server is old. It was built around ~2000. I brought the machine down and
went into the BIOS. The PCI configurations were set for AUTO so I picked
IRQ 12 and 14 which seemed to be free. My choices only went up to 15.

I rebooted the system but when I run lspci the output shows the modems are
set still set to 5 & 9 like they were originally setup from AUTO.

Looking at setserial. the man page states the following:

"The selection of an alternative IRQ line is difficult, since most of them are already used.  
The following table lists the "standard MS-DOS" assignments of available IRQ lines:

             IRQ 3: COM2
             IRQ 4: COM1
             IRQ 5: LPT2
             IRQ 7: LPT1

Most  people  find that IRQ 5 is a good choice, assuming that there is only one parallel 
port active in the computer.  Another good choice is IRQ 2 (aka IRQ 9); although this 
IRQ is sometimes used by network cards, and very rarely  VGA  cards will be 
configured to use IRQ 2 as a vertical retrace interrupt."

So I guess using IRQ 5 & 9 seems to be what is recommended.

I've also reconfigured the config files NOT to use the symlinked device names
"modem" and "modem1" and am using ttyS14 & ttyS15. Test faxes today
seem  to go with no problem but I'll know more when we our normal fax load
starts Monday or maybe Tuesday with the holiday.

The setserial man page also says this about 16550A UARTs which are what the
internal ZPXs are using. The man page was written in 2000 so I don't know if this
is relevant anymore:

"Some internal modems are billed as having a "16550A UART with a 1k buffer".  This is a lie.  
They do  not have  really  have  a 16550A compatible UART; instead what they have is a 
16450 compatible UART with a 1k receive buffer to prevent receiver overruns.  This is 
important, because they  do  not  have  a  transmit FIFO.   Hence,  they  are  not 
compatible with a 16550A UART, and the autoconfiguration process will correctly identify 
them as 16450s.  If you attempt to override this using the uart parameter, you will  see
dropped  characters  during  file transmissions.  These UARTs usually have other problems: 
the skip_test parameter also often must be specified."

My system autoconfigures them as 16550As so I am assuming that is correct.

I've googled "LSR safety check engaged" and learned more than I ever cared to
about the topic but none of it seems relevant here. I get this error only after
a fax reception has failed and Hylafax is trying to reset the modem. At that point
it appears the modem is hung. As I mentioned in a message very early in this
thread, the modem after the failure hangs and I keep hearing tones come out of
the box like it was trying to retrain. Only power cycling the system AND unplugging
the phone lines bring it back.

Any insight appreciated,

Mike




.

____________________ HylaFAX(tm) Users Mailing List _______________________
 To subscribe/unsubscribe, click http://lists.hylafax.org/cgi-bin/lsg2.cgi
On UNIX: mail -s unsubscribe hylafax-users-request@xxxxxxxxxxx < /dev/null
 *To learn about commercial HylaFAX(tm) support, mail sales@xxxxxxxxx*