EPICS Controls Argonne National Laboratory

Experimental Physics and
Industrial Control System

2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  <20132014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024  Index 2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  <20132014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
<== Date ==> <== Thread ==>

Subject: Re: double buffered waveformRecord
From: Andrew Johnson <[email protected]>
To: [email protected], [email protected]
Date: Fri, 4 Jan 2013 16:36:13 -0600
Hi Michael,

On 2013-01-03 Michael Davidsaver wrote:
> 
> Recently I was looking at how array fields are handled in database
> processing and RSRV.  In the process I noticed a way to allow double
> buffering (and possible more creative schemes) of BPTR in the
> waveformRecord.
> 
> This involves playing games with the array offset returned by the
> get_array_info() record support function, and taking advantage of how
> this offset is used in dbConvert.c.  By doing this I can convince
> dbGet() and dbPut() to read/write from almost arbitrary locations which
> can change with each call.

I wouldn't regard setting *no_elements and *offset as "playing games", I think 
it's perfectly reasonable for a record to implement multiple-buffering by 
dividing the data buffer into a series of smaller segments and using *offset 
to select the current segment within the buffer.

> > static long get_array_info(DBADDR *paddr, long *no_elements, long
> > *offset) {
> > ...
> >     if(prec->bact) {
> >         /* cause caller to access buf[1] */
> >         *offset = (prec->buf[1] - prec->buf[0]) /
> > dbValueSize(prec->ftvl); } else {
> >         *offset = 0;
> >     }
> 
> http://bazaar.launchpad.net/~epics-core/epics-base/array-opt/revision/12397

If I've understood your code correctly, it's relying on *undefined behavior* 
because you're subtracting pointers to two differently allocated buffers to 
calculate the *offset value.  If the C compiler detects this it is quite at 
liberty to do anything it likes at that point.

If you kept the original single buffer allocation however and just split it 
into two parts logically there would be no undefined behavior.  I would think 
this might be slightly simpler to implement as well, although there's probably 
not much difference.

> There is also an example device support.
> 
> http://bazaar.launchpad.net/~epics-core/epics-base/array-opt/revision/12398

You will need to rename the record type before it has any chance of getting 
into Base BTW, but you probably guessed that already...

> Now that I have this working I'd like to create a mechanism which
> doesn't involve abusing 'offset'.  So far I've been thinking of allowing
> get_array_info() to provide a replacement pointer (in addition to offset
> and length) either by adding another argument, or allowing modification
> of the DBADDR argument.

I think there may be problems trying to change the pointer, because that's the 
fundamental value that the server code uses to identify the field that was 
subscribed to — it's the only field-specific parameter the record passes to 
db_post_events() so it must match the pointer given out when the subscription 
was created.

> The entire branch
> 
> https://code.launchpad.net/~epics-core/epics-base/array-opt
> 
> There is also some in-progress changes related to optimizing dbConvert.c.

Interesting; have you done any bench-marking of that optimization?

- Andrew
-- 
There is no such thing as a free lunch.  When invited for lunch,
it is best to check if you are there to eat, or to be eaten.
-- Clive Robinson


Replies:
Re: double buffered waveformRecord Michael Davidsaver
References:
double buffered waveformRecord Michael Davidsaver

Navigate by Date:
Prev: double buffered waveformRecord Michael Davidsaver
Next: Re: double buffered waveformRecord Michael Davidsaver
Index: 2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  <20132014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
Navigate by Thread:
Prev: double buffered waveformRecord Michael Davidsaver
Next: Re: double buffered waveformRecord Michael Davidsaver
Index: 2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  <20132014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
ANJ, 18 Nov 2013 Valid HTML 4.01! · Home · News · About · Base · Modules · Extensions · Distributions · Download ·
· Search · EPICS V4 · IRMIS · Talk · Bugs · Documents · Links · Licensing ·