Experimental Physics and
Industrial Control System

"Mark Rivers" <[email protected]> · Fri, 21 May 2010 08:28:32 -0500

> Hm, you could try to prove that it happens. This may be difficult, though.

One way to test that would be with the areaDetector software.  It can generate large arrays (images in waveform records) at a high rate.  The waveform records are I/O Intr scanned whenever a new image arrives.  If this problem occurs CA clients should get images are mixes of old and new data.  I have never actually seen this, but because there is generally a high degree of correlation from image N to N+1 it would not be obvious.  There is a simulation driver which could be modified so that even images were all 0 and odd images were all 255, for example. Then such corruption would be easy to spot in a CA client.

Mark

________________________________

From: [email protected] on behalf of Benjamin Franksen
Sent: Fri 5/21/2010 8:17 AM
To: [email protected]
Subject: Re: Race conditions in SNL programs

Hi Dirk

On Friday 21 May 2010, Dirk Zimoch wrote:
> Locking and/or re-reading might be possible for SNL code accessing the
> variables but what about escaped C-Code? You probably have to take the
> lock before escaping to C but then nobody is allowed to do blocking
> things in that code.
>
> A layer of "shadow variables" might be safe. CA updates the shadow copy
> at any time while the state code executes. Only during the state
> transition and inside pvGet the data is transfered to the "user copy".
> (any back inside pvPut). CA has to be locked out only during these copy
> actions. Of course that requires twice the amount of memory for
> variables plus some time for the additional copy.

Yes. I came to the same conclusion when discussing this with Götz. I think
the additional overhead is the price to pay for correctness.

> Concerning double: On the new PPC 500v2 core it takes indeed two
> assembler instructions to load or store a double. Thus doubles are
> non-atomic.

Ah, thanks for a definite statement.

> By the way, I think arrays are already unsafe on the server!

Ouch!

> If the
> array record processes and updates the array at the same time as it is
> sent on the network by CA, the user gets a half old/hald new array. This
> is because only the pointer to the array is put into the queue for the
> CA thread, not the array data itself. (I am happy if somebody could
> prove me wrong.)

Hm, you could try to prove that it happens. This may be difficult, though.

Anyway, thanks for your input!

Cheers
Ben

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System