Re: SRQ support using asyn

[Eric Norum <norume@aps.anl.gov>]

I looked through my copy of the VXI-11 specification and agree with  
your analysis.   The only mention of the 'lid' value returned by a  
create_link operation is in Rule B.6.3, "The value of lid SHALL be  
unique for all currently active links within a network instrumentation  
server".     There is nothing saying that the value must be non-zero.
Please send your changes as a patch file to me or Mark Rivers.

Thank you.

On Jan 18, 2009, at 8:12 PM, Takashi.Asakawa@jp.yokogawa.com wrote:

> Dear all,
>
>  We have a question about a SRQ support which uses asyn(R4-10).
> We are developing a device support for a waveform digitizer which  
> employs
> the vxi11 protocol. It basically works properly except for the SRQ  
> function.
> What happens is as follows:
>
>  When we receive SRQ from the device, the following message comes up
> and the readout of the status byte fails:
>
> 2009/01/15 17:21:35.042 inst0 vxiCreateDeviceLink error VXI:
> out of resources L0 vxiCreateDevLink failed for addr 0
> 2009/01/15 17:21:35.042 L0 addr 0 asynGpib:srqPoll serialPoll error
>
>  We chased this error within the asyn code and found that it comes
> from the function of vxiSerialPll() in asyn/vxi11/drvVxi11.c:
> ---------------------------
> static asynStatus vxiSerialPoll(void *drvPvt, int addr,
>    double timeout,int *statusByte)
> {
>  ...
>    if(!pdevLink->lid) {
>        Device_Link lid;
>        if(!vxiCreateDevLink(pvxiPort,addr,&lid)) {
>            printf("%s vxiCreateDevLink failed for addr %d\n",
>                pvxiPort->portName,addr);
>            return asynError;
>        }
>        pdevLink->lid = lid;
>    }
>  ...
> }
> ----------------------------
>  This function judges whether the connection has been made or not by  
> checking
> the value of "lid"; "non-zero" means "connected", and "zero" means  
> "not connected".
>  In the case of the device we are struggling with, Device_Link can  
> be "zero"
> even if the connection succeeds. (I heard that this behavior of the  
> device
> does not violate the vxi11 specifications.)
> And the problem happens when Device_Link is zero.
>
>  We therefore modified the above code as shown below:
> ----------------------------
>    if(!pdevLink->conneced) {         /* Modified */
>        Device_Link lid;
>        if(!vxiCreateDevLink(pvxiPort,addr,&lid)) {
>            printf("%s vxiCreateDevLink failed for addr %d\n",
>                pvxiPort->portName,addr);
>            return asynError;
>        }
>        pdevLink->lid = lid;
>        pdevLink->conneced = TRUE;  /* Added */
>    }
> ----------------------------
>  By this modification, everything seems to work properly now.
> What we'd like to know is whether this modification is reasonable or  
> not;
> we wonder if the original code employs "lid" instead of "connected"
> because of some important reason we don't know.
>
>  Does anybody kindly tell us about this, please?
>
> Many thanks in advance.
>
> Best regards,
>
> Takashi Asakawa
> Yokogawa Electric Corporation, Japan
--
Eric Norum <norume@aps.anl.gov>
Advanced Photon Source
Argonne National Laboratory
(630) 252-4793