Re: Fwd: realtime performance of adl_pci9118

Can anyone give me some hints about it? thanks

regards,

Wicky
  ----- Original Message ----- 
  From: wicky Zhang 
  To: Calin A. Culianu 
  Cc: comedi_at_comedi.org 
  Sent: Tuesday, November 01, 2005 7:39 PM
  Subject: Re: Fwd: realtime performance of adl_pci9118


  hi, Calin and Michal

  thank you for your instruction, but i can't resolve the provlem up to now.

  1: for the TRIG_WAKE_EOS, i try to find where to set or clear the bit, then i 

  find the following lines in the file kcomedilib_main.c:

   if(async->cb_mask & COMEDI_CB_EOS)
    cmd->flags |= TRIG_WAKE_EOS;

  in my user code, there is nothing about it. i am confused how to clear this bit.

  2: i found that my card uses bus mastering, but then how?

  3: my system now will often apear the following messages

    "unexpected FPU trap from hard PID="

  i think this may due to my CPU which is an AMD xp2500 working at xp3200, the cpuinfo is:

  processor : 0
  vendor_id : AuthenticAMD
  cpu family : 6
  model  : 10
  model name : AMD Athlon(tm) XP 3200+
  stepping : 0
  cpu MHz  : 2205.002
  cache size : 512 KB
  fdiv_bug : no
  hlt_bug  : no
  f00f_bug : no 
  coma_bug : no
  fpu  : yes
  fpu_exception : yes
  cpuid level : 1
  wp  : yes
  flags  : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
  bogomips : 4404.01
   
  thank you

  regards,

  wicky

   
  On 10/16/05, Calin A. Culianu <calin_at_ajvar.org> wrote: 


    On Thu, 13 Oct 2005, wicky Zhang wrote:

    > ---------- Forwarded message ----------
    > From: wicky Zhang < wicky.zhang_at_gmail.com>
    > Date: Oct 13, 2005 9:22 PM
    > Subject: Re: realtime performance of adl_pci9118
    > To: Michal Dobes < dobes_at_tesnet.cz>
    > Cc: rtai_at_rtai.org
    >
    >
    >
    > do you use cmd->flags|=TRIG_WAKE_EOS ?
    >> If yes, then after every sample is generated interrupt and 50 
    >> kinterrupts/s
    >> will kill any system.
    >
    >  I have not do that operation, but would you tell me how to check this
    > setting
    >  or reset it?

    Just make sure cmd->flags doesn't contain the TRIG_WAKE_EOS bit. 
    If you didn't set it, it should be off by itself (you *did* memset() your
    comedi_cmd struct to 0, right?  Either that or it is a global var,
    right?).

    If you want to be paranoid (something I always encourage), read the 
    sources to your driver to make sure it doesn't default to have the board
    interrupt after each scan -- if it does that (and the driver supports
    not-interrupting-after-each-scan) that's a bug.

    Also, as someone else suggested, try and make sure the board is doing DMA 
    transfers (by figuring out if the driver is using DMA).

    -Calin