Re: Multiple asyn ports inside asynPortDriver class

[Mark Rivers via Tech-talk <tech-talk at aps.anl.gov>]

Hi Abdalla,

• I have another question, right now I am relying on asyn’s address as the register’s address on the device but I think the driver could be designed in a better way, can you give pointers on how to parse asyn parameter strings and create corresponding asyn parameters?

I recently wrote a driver for the SIS3153, which is a USB to VME interface card.  It allows access to VME bus registers from a Linux machine over USB with no CPU card in the VME crate.

https://github.com/epics-modules/SIS3153

I was faced with a similar decision as you: do I use the asyn "addr" field to specify the VME address, or do I encode it in the drvParams part of the link INP or OUT field?

https://epics-modules.github.io/asyn/asynDriver.html#generic-device-support-for-epics-records

I decided to use the drvParams field for 2 reasons:

• Each time an address is used in the addr field asynManager allocates resources for that specific address.  asynReport will report on each address.  This gets unwieldy if there are hundreds or thousand of addresses.
• asynManager treats the addr field as a signed integer, and negative numbers are not allowed except for -1.  For VME addresses this is a problem, because addresses from 0x80000000 and larger will be negative if interpreted as signed integers.

I adopted the following convention for the drvParams field:

field(INP,  "@asyn($(PORT), 0)ADDRESS_MODE VME_ADDRESS")
field(OUT,  "@asyn($(PORT), 0)ADDRESS_MODE VME_ADDRESS")

For example:

field(INP,  "@asyn($(PORT), 0)A32D16 0xB0000028")

My SIS3153 driver is based on asynPortDriver.  It implements the drvUserCreate method to parse the drvParams field.

https://github.com/epics-modules/SIS3153/blob/2adc1151b8204098ce6032707b7d19d73ba3f580/SIS3153App/src/drvSIS3153.cpp#L259

That routine converts the VME_ADDRESS field into an integer, and puts a pointer to that integer in the pasynUser->drvUser field.  It removes the VME_ADDRESS from the drvParams string, and passes the remainder of the string (ADDRESS_MODE) to the base class asynPortDriver::drvUserCreate.  That base class parses the ADDRESS_MODE because those are defined as the port driver parameters.

The driver constructor does not set the ASYN_MULTIDEVICE flag in the constructor, so asynPortDriver and asynManager treat the driver as only supporting addr=0.

The writeInt32 and readInt32 methods get the VME address to use as follows:

    addr = *(int *)pasynUser->drvUser;

Let me know if you have any questions.

Mark

---

From: Abdalla Ahmad <Abdalla.Ahmad at sesame.org.jo>
Sent: Sunday, November 10, 2024 1:22 AM
To: Mark Rivers <rivers at cars.uchicago.edu>; tech-talk at aps.anl.gov <Tech-talk at aps.anl.gov>
Subject: RE: Multiple asyn ports inside asynPortDriver class

 

It seemed to be an issue with the device itself, the device’s own software produced the same timeout behavior on continuous TCP requests. I have another question, right now I am relying on asyn’s address as the register’s address on the device but I think the driver could be designed in a better way, can you give pointers on how to parse asyn parameter strings and create corresponding asyn parameters?

 

Thanks!

Abdalla.

 

From: Mark Rivers <rivers at cars.uchicago.edu>
Sent: Thursday, November 7, 2024 4:30 PM
To: tech-talk at aps.anl.gov; Abdalla Ahmad <Abdalla.Ahmad at sesame.org.jo>
Subject: Re: Multiple asyn ports inside asynPortDriver class

 

Hi Abdallah,

 

UDP and TCP use the same port number, not just the same IP address and physical port?  I don't think I have run into that before, so I can't see whether it has been tested with asyn.  You can turn on asynTrace on each port, and send them to different files to see if there are issues.

 

Mark

 

---

From: Tech-talk <tech-talk-bounces at aps.anl.gov> on behalf of Abdalla Ahmad via Tech-talk <tech-talk at aps.anl.gov>
Sent: Thursday, November 7, 2024 6:14 AM
To: tech-talk at aps.anl.gov <Tech-talk at aps.anl.gov>
Subject: Multiple asyn ports inside asynPortDriver class

 

Hi

 

I am writing an EPICS driver for an Ethernet-based device in which the control interface is split between TCP and UDP layers of the same port number (i.e., registers IO is through UDP while waveforms is through TCP). I have to create two asyn ports using drvAsynIPPortConfigure function, so my question is are there any consideration regarding this approach? Because I am noticing some timeout issues, most likely from the device’s side, so I want to narrow the issue as much as I can. Thanks in advance.

 

Best Regards,

Abdalla Al-Dalleh

Control Engineer

SESAME

P.O. Box 7, Allan 19252, Jordan
Tel: +96253511348 , ext. 265

Fax: +96253511423

Email : abdalla.ahmad at sesame.org.jo
Website: www.sesame.org.jo