Menu

#3 Kernel freeze with multiport serial card

open
nobody
None
5
2001-01-16
2001-01-16
Anonymous
No

Discussion

  • Nobody/Anonymous

    Logged In: NO
    Browser: Mozilla/5.0 (X11; U; Linux 2.2.16-3 i686; en-US; m18) Gecko/20001107 Netscape6/6.0

    Kernel freeze with multiport serial card

    Environment tested
    ------------------

    Serial driver:
    serial-5.05

    Operating system:
    RedHat 6.2 on kernel 2.2.16
    RedHat 6.2 on kernel 2.4.0

    PC-Hardware:
    Siemens-Fujitsu Pentium III ~500 MHz
    Unisys Pentium II ~266 MHz

    Serial Card:
    Exsys EX-4065A 8 serial RS-232 PCI Card 4 x 16C550
    Chips with 16-Byte buffer

    Tools:
    setserial-2.17

    Problem description
    -------------------

    We have all 8 serial lines, of the serial card,
    connected to console ports
    of remote VME crates, using 10m serial extension cables
    (TP Typ 5).

    <Serial Card>[]<fan-out>[]-------- TP Typ 5 (9m)
    ------[]<VME Crate>
    | []-------- TP Typ 5 (9m)
    ------[]<VME Crate>
    | []-------- TP Typ 5 (9m)
    ------[]<VME Crate>
    | []-------- TP Typ 5 (9m)
    ------[]<VME Crate>
    |
    []<fan-out>[]-------- TP Typ 5 (9m)
    ------[]<VME Crate>
    []-------- TP Typ 5 (9m)
    ------[]<VME Crate>
    []-------- TP Typ 5 (9m)
    ------[]<VME Crate>
    []-------- TP Typ 5 (9m)
    ------[]<VME Crate>

    Things work fine if all the cables and extensions are
    connected to the remote
    devices.
    However, if we disconnect one of the extension, at the
    VME site, but leave the
    same cable still connected to the fan-out cable, at the
    serial card on the PC,
    the kernel will deadly freeze when executing 'setserial
    -a /dev/ttySn'.

    <Serial Card>[]<fan-out>[]-------- TP Typ 5 (9m)
    ------[ ]<VME Crate>

    If the cable extension is also disconnected, on the
    fan-out cable, on the PC
    site, executing 'setserial -a /dev/ttySn' works fine.

    <Serial Card>[]<fan-out>[ ]-------- TP Typ 5 (9m)
    ------[]<VME Crate>

    This happens on ports 2-8 on the EX-4065A 8 serial
    card, but not on port 1.

    The kernel doesn't write any panic to the console or to
    a log file.

     
  • Ryan Belcher

    Ryan Belcher - 2001-05-14

    Logged In: YES
    user_id=91184

    This suspisciously sounds like a problem with unterminated
    signals. If you ever have to disconnect one of the cables
    from your device(s), attach a loopback device up to the
    hanging line. Due to the cable length, density, and nature
    of the signals, crosstalk is a very perplexing and real
    threat to normal ops. Make sure you terminate both the
    data, and modem control signals to ensure no extraneous
    interrupt occur. The reason it only causes a problem on
    ports 2-8, is because on port 1, more than likely, is the
    ISP (Interrupt Status Port) within the SPR (Scratch Pad
    Register) on the primary UART which in turn is *supposed*
    to inform the driver as to which port is actually
    generating the interrupt. Another possible problem which
    could possibly be adding to this problem and causing a real
    mess of a troubleshoot, is from what I noticed (or lack
    there of) on your post, is that you didn't mention
    the "^fourport autoconfig" in any of the setserial
    information (which needs to be set for multiport stuff); as
    well as the ISP masking information something similar
    to "portX 0xYY maskZZ 0xAA matchZZ 0xAA" which (in most
    cases) resides on the first UART-Port of the card. One
    last thing, you need to make sure that your kernel config
    has multiport boards and a few other things included in
    there. Hope it helps.

     
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.