Paul Sichta wrote:
>
> A few months ago the ability to specify the SHFT parameter for the mbbi
> record (for soft/raw soft device type) was added. I now realize that the
> mbbi record's MASK feature is not applied for soft or raw soft device
> types. Inspecting the source mbbiRecord.c confirms that the mask is not
> used in record support. Without the mask, higher-order bits (those beyond
> NOBT) affect the value and cause invalid states.
>
> 1) I propose that the bit mask in the mbbi record be used for all device
> types, to strip out bits "above" the bit range of interest.
> a) The mask is (now) created based on NOBT (number of bits).
> b) The mask would be applied to rval after the value has been shifted
> by SHFT number of bits. For hardware devices, the shifting occurs in
> device support (prior to the mbbi record). Hardware device types should have
> their SHFT field set to zero.
>
> A code segment from mbbiRecord.c is shown below, along with the proposed
> new line.
>
> recGblGetTimeStamp(pmbbi);
> if(status==0) { /* convert the value */
> unsigned long *pstate_values;
> short i;
> unsigned long rval = pmbbi->rval;
>
> if(pmbbi->shft>0) rval >>= pmbbi->shft;
> rval = rval & pmbbi->mask;
I just looked at some device support. Here is an example (camac)
/* set shft and mask fields to the bit position
(where the least significatn bit is 0) */
pmbbi->shft = fsd;
pmbbi->mask <<= fsd;
Here is another
/* For mbbi records, the final parameter specifies the input bit
shift,
with offset specifying the message byte number. */
prec->shft = pcanMbbi->inp.parameter;
prec->mask <<= pcanMbbi->inp.parameter;
Thus it appears that the mask should be applied before the shift.
Can we gurantee that no device support will be affected? Not clear.
Maybe it would be safer to change the soft raw device support.
/*
> proposed new line 25SEP98 */
> if (pmbbi->sdef){
> pstate_values = &(pmbbi->zrvl);
> pmbbi->val = 65535;
>
> 2) In an effort to figure out this problem, I recompiled a "test" version
> of mbbiRecord.c (ran gnumake in base/src/rec). The new mbbiRecord.o was
> installed in the target's bin directory. I ran gnumake in my application's
> src directory, and rebooted the IOC. The mbbiRecord operation was the
> same as before. Why didn't the new object file run on the IOC?
Maybe there is a dependence problem.
Try the following.
In your base/src/rec
touch mbbiRecord.c]
gnumake
Should see mbbiRecord compile and .o file installed.
In your application src directory
gnumake
Does the xxxLib get rebuilt? If not dependency problem. Do a
gnumake rebuild
Marty
- References:
- MBBI & mask Paul Sichta
- Navigate by Date:
- Prev:
Re: MBBI & mask Rozelle Wright
- Next:
beta12 & GPIB & Compiler Bug Marty Kraimer
- Index:
1994
1995
1996
1997
<1998>
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
- Navigate by Thread:
- Prev:
MBBI & mask Paul Sichta
- Next:
Re: MBBI & mask Rozelle Wright
- Index:
1994
1995
1996
1997
<1998>
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
|