[Asterisk-Users] Issue with TE405P and Adaptec U160 SCSI

[Steven R. Ringwald]

I am  upgrading from an X100P to a TE405P (T1/E1/QuadSpan) card. The 
asterisk server is an IBM xSeries 300 running Fedora Core 1.

uname -a reports:
Linux hermes 2.4.22-1.2199.nptl #1 Wed Aug 4 12:21:48 EDT 2004 i686

The system has an Adaptec U160 SCSI card in it.

/proc/pci:
    SCSI storage controller: Adaptec AIC-7892A U160/m (rev 2).
      IRQ 11.
      Master Capable.  Latency=32.  Min Gnt=40.Max Lat=25.
      I/O at 0xdc00 [0xdcff].
      Non-prefetchable 64 bit memory at 0xdd0c3000 [0xdd0c3fff].


When I put the TE405P card into the machine, I get the following 
messages on the system console, and the machine eventually freezes 
(usually before booting up to a point where I can look and see what the 
interactions are).

/var/log/messages:
Sep 16 16:54:15 hermes kernel: scsi0: PCI error Interrupt at seqaddr = 0x104
Sep 16 16:54:15 hermes kernel: scsi0: Data Parity Error has been 
reported via PERR#
Sep 16 16:54:15 hermes kernel: scsi0: PCI error Interrupt at seqaddr = 0x17d
Sep 16 16:54:15 hermes kernel: scsi0: Data Parity Error has been 
reported via PERR#
Sep 16 16:54:15 hermes kernel: scsi0: PCI error Interrupt at seqaddr = 0x83
Sep 16 16:54:15 hermes kernel: scsi0: Data Parity Error has been 
reported via PERR#
Sep 16 16:54:15 hermes kernel: scsi0: PCI error Interrupt at seqaddr = 0x17d
Sep 16 16:54:15 hermes kernel: scsi0: Data Parity Error has been 
reported via PERR#
Sep 16 16:54:15 hermes kernel: scsi0: PCI error Interrupt at seqaddr = 0x82
Sep 16 16:54:15 hermes kernel: scsi0: Data Parity Error has been 
reported via PERR#

With the TE card out of the machine, and just the X100P in, these parity 
errors do not appear, and the system comes up and runs just fine.

Any ideas what the problem might be, as we would really like to use this 
card, and it seems odd that we would have this kind of interaction in a 
"plug-and-play" system... Tomorrow I plan on bringing the card up an a 
different machine, and seeing if I can replicate the problem elsewhere.

Thanks in advance,
Steve RIngwald