Re: device driver race conditions problems (kernel 2.4)

From: Norm Dresner (ndrez_at_att.net)
Date: 08/28/04 

Date: Sat, 28 Aug 2004 13:15:19 GMT

The simplest way to manage multiple, identical cards that have to be
serviced regularly (you say that you have a 2 ms interrupt from your
"firmware") is to have the single driver poll each of the cards to see if
they need servicing during that interval. You can't get any race conditions
from multiple interrupts this way.

Yes, I know that you can't easily get a 2 ms interrupt in software from a
normal linux kernel, but you can do it very easily with the freeware version
of real-time linux. If you in fact have a hardware interrupt at the 2 ms
interval that you've mentioned, then you can use this in normal linux
forever as a timing mechanism.

Norm

Relevant Pages

  • Re: Fault-Tolerant List Construction
    ... Here's another attempt that avoids the multiple value problem by using ... avoids the NIL problem of my other attempt and short-ciruits ... producing the interrupt does or does not produce a value. ...
    (comp.lang.lisp)
  • Re: ATA in WinCE5.0 and Samsung HD -- Not a repost.
    ... > the code relies upon the registry setting here for the DRQ data block ... > READ MULTIPLE and WRITE MULTIPLE is used, wherein, the DRQ block size is ... > number of sectors set using the SET MULTIPLE command. ... > supported by the disk and then waits for an interrupt at the end of the ...
    (microsoft.public.windowsce.platbuilder)
  • Re: Error in DebugVersion of wince5.0
    ... You can have multiple sources trigger that ... have a GPIO driver for which i need to trigger GPIO Interrupts ... interrupt the service calls interruptdone to free the interrupt ...
    (microsoft.public.windowsce.platbuilder)