Sorry, I still think I am right !
I run my tests on a soft Linux IOC. The asDumpRules
cmd seems to be asprules in my case. Here are my results with A=1 and A=2
following Andrew’s suggestion (A& … instead of A >> …).
With A=1 the 1rst RULE for WRITE is TRUE which is OK
With A=2 I A&2 = 2 which is NOT the TRUE value (TRUE
corresponds to 1 and not to !=0):
Simulateur >
Simulateur >dbpf
ASGVAL,1
DBR_LONG: 1 0x1
Simulateur >asprules
ASG(DEFAULT) {
INPA(ASGVAL) value=1.000000
RULE(1,READ,NOTRAPWRITE)
RULE(1,WRITE,NOTRAPWRITE)
{
UAG(tous)
HAG(hag)
CALC("A&1") result=TRUE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(develop)
HAG(hag)
CALC("A&2")
result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(physique)
HAG(hag)
CALC("A&4")
result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(gournay)
HAG(hag)
CALC("A&8")
result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(epics)
HAG(hag)
CALC("A&16")
result=FALSE
}
}
Simulateur >dbpf ASGVAL,2
DBR_LONG: 2 0x2
Simulateur >asprules
ASG(DEFAULT) {
INPA(ASGVAL) value=2.000000
RULE(1,READ,NOTRAPWRITE)
RULE(1,WRITE,NOTRAPWRITE) {
UAG(tous)
HAG(hag)
CALC("A&1")
result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE)
{
UAG(develop)
HAG(hag)
CALC("A&2") result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(physique)
HAG(hag)
CALC("A&4")
result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(gournay)
HAG(hag)
CALC("A&8")
result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(epics)
HAG(hag)
CALC("A&16")
result=FALSE
}
}
Simulateur >
If now I change the 2nd rule from A&2 to
A>>1, I have now TRUE:
Simulateur >dbpf
ASGVAL,2
DBR_LONG: 2 0x2
Simulateur >asprules
ASG(DEFAULT) {
INPA(ASGVAL) value=2.000000
RULE(1,READ,NOTRAPWRITE)
RULE(1,WRITE,NOTRAPWRITE) {
UAG(tous)
HAG(hag)
CALC("A&1") result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE)
{
UAG(develop)
HAG(hag)
CALC("A>>1")
result=TRUE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(physique)
HAG(hag)
CALC("A&4") result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(gournay)
HAG(hag)
CALC("A&8") result=FALSE
}
RULE(1,WRITE,NOTRAPWRITE) {
UAG(epics)
HAG(hag)
CALC("A&16") result=FALSE
}
}
J.F. Gournay
CEA Saclay
IRFU/SIS
-----Message d'origine-----
De : Martin L. Smith [mailto:[email protected]]
Envoyé : jeudi 10 septembre 2009 12:50
À : GOURNAY Jean-Francois
Cc : [email protected]
Objet : Re: Access security problem
Actually Andrew is correct. Take the
"A&2" as an example, when bit 1 is TRUE
then the result of the CALC is TRUE, when bit 1 is
FALSE then the result of
the CALC will be FALSE. TRUE being = 1 and FALSE being
= 0.
This may be why you are seeing this problem. If you
log into the IOC console
and issue this command you will see the value that CA
access security thinks
your CALC _expression_ evaluates to as shown below.
asDumpRules
ASG(ID01ds_ACTION) {
INPA(ID01ds:AccessSecurity) value=3.000000
RULE(1,READ,NOTRAPWRITE)
RULE(0,WRITE,TRAPWRITE) {
UAG(operator)
HAG(local)
CALC("A<3") result=FALSE
}
RULE(0,WRITE,TRAPWRITE) {
UAG(flcord)
HAG(flctrl,local)
CALC("A<2") result=FALSE
}
RULE(1,WRITE,TRAPWRITE) {
UAG(sec01)
HAG(sec01)
CALC("A<1") result=FALSE
}
RULE(1,WRITE,TRAPWRITE) {
UAG(appdev)
}
}
Marty
GOURNAY Jean-Francois wrote:
> Dear Andrew,
>
> I think you are wrong. Remember that the CALC
_expression_ must return TRUE or FALSE which will not be the case with A&2,
A&4 ...
> I guess my scheme is correct. Everything works
except this problem with this spiral2mgr user that I don't understand (see the
table below from my previous mail). We will try to overcome it in a way or
another ... unless somebody has another idea about it !!!
>
> J.F. Gournay
> CEA Saclay
> IRFU/SIS
>
> -----Message d'origine-----
> De : Andrew Johnson [mailto:[email protected]]
> Envoyé : mercredi 9 septembre 2009 16:56
> À : [email protected]
> Cc : GOURNAY Jean-Francois
> Objet : Re: Access security problem
>
> On Wednesday 09 September 2009 07:34:52 GOURNAY
Jean-Francois wrote:
>> I did more systematic tests and I have this
strange situation:
>>
>> ASGVAL UAG spiral2 spiral2mgr
gournay
>>
>> 1 tous W W
W OK
>> 2 develop - - (should be W)
- WRONG
>> 4 physique W -
- OK
>> 8 gournay - W (should be -)
W WRONG
>>
>> W : database WRITE access
>> - : no write access
>>
>> I modified the CALC expressions accordingly
to Martin suggestion :
>> CALC("A&1") instead of
CALC("A&1=1"). But I don't think I have to write
>> CALC (A>0&&A<2) as my A value
comes from a longout record.
>
> You're right there, but for historical reasons
the relative precedence of
> the ">>" and "="
operators are different in C and CALC, so you do need to
> change your expressions:
>
>> CALC("A&1=1")
> As Marty suggested "A&1" is
sufficient here
>
>> CALC("A>>1=1")
> CALC actually evaluates this as
"A>>(1=1)" which is not what you mean. You
> could use "(A>>1)=1" but I would
suggest "A&2" here instead.
>
>> CALC("A>>2=1")
> "A&4"
>
>> CALC("A>>3=1")
> "A&8"
>
> Using fewer operators will also parse and execute
faster.
>
>> This message sent yesterday didn't show up.
>
> That's because you actually sent it to
tech-talk-bounces@aps., not to
> tech-talk@aps., and I didn't get a chance to
forward it manually before you
> re-sent it.
>
> HTH,
>
> - Andrew
J.F. Gournay
CEA Saclay
IRFU/SIS