Is issueing multiple IBWAITs the best way to incorporate max time?

[C. Kennedy]

My application is given a max-time parameter that tells how long it should wait to recieve data from my instrument after requesting data. The max-time can be anywhere from 2 seconds to 480 seconds.
Currently I send an IBWAIT (TIMO|ERR|SRQI). If SRQI then I send IBRSP to the instrument. I don't feel comfortable changing the GPIB time-out time. Is it possible to put the IBWAIT command in a 'while' loop and compare time from when issued and when finished, and issue another IBWAIT if TIMO was returned on the first IBWAIT? Or, is there a better way to do this?
Thank you for any help.

[DiegoF]

Hi,

I don't see any problems. A timeout error (or any error) does not close the GPIB reference, so you can call other driver functions, such as ibwait.

As a reference, if you are waiting for a service request for a particular device (using device-level unit descriptors), you should use the RQS mask for ibwait. When a service request occurs, the driver performs an Automatic serial poll, to determine which device generated the RQS. The driver will wake up only the ibwait call for that particular device, if you are waiting for several devices to request service. The SRQI mask is to detect if the SRQ line was asserted (any device requesting service).

DiegoF
National Instruments.

[Julien Brutus]

If I have 5 devices on my GPIB bus, can I call ibwait in each of the 5 threads I have created to read the response of each device ?
I am afraid that the CPU will be tied up?
Don't U think?
Thanks

[LoganS]

You should be able to call IBWAIT on each device as long as when one thread within a process calls ibwait on a given descriptor, another thread within the same process is not already performing an ibwait using that same descriptor (see NI-488.2 Help under "EWIP"). I haven't been able to find any documentation about the CPU getting tied up with 5 calls to IBWAIT at once.

[Julien Brutus]

I am actually programming under linux RedHat 8.0 with C langage.
The truth is my gpib release is 0.8.5 and I wonder if this release is thread safe
Is there any example of synchronous blocking read in C under linux?
Thank U

[Shawn B.]

Julien,

No the 0.8.6 driver is not multithread safe. If you need this functionality you can try the latest beta driver available at www.ni.com/beta.

Shawn B.
National Instruments

[Julien Brutus]

Hi Scott,

I was in contact weeks ago about a pb with gpibintctrl under linux RH 8.0 with beta 2.3.
I have downloaded 2.5 as U suggested it, and install it in place of 2.3

But the pb is the same, the buffer returned by ibrd under gpibintctrl cannot be greater than 16 bytes !!!
What is the pb?

Example :

: ibdev 0 9 0 13 0 0x0a

ud0: ibwrt "*IDN?\r\n"
[0100] ( cmpl )
count: 7

ud0: ibrd 16
[0100] ( cmpl )
count: 16
48 45 57 4c 45 54 54 2d H E W L E T T -
50 41 43 4b 41 52 44 2c P A C K A R D ,


ud0: ibrd 17
[2100] ( end cmpl )
count: 15
00 00 00 00 00 00 00 00 . . . . . . . .
00 00 00 00 00 00 00 . . . . . . .

[NI-488.2 master]

Julien,

Did you restart your machine following the installation? We did experience a very similar issue with DMA that required us to reboot the machine before using the driver.

Craig A.
National Instruments Engineer

[Julien Brutus]

Yes, I do it many times and I use many GPIB devices for the same behaviour.