dtrace discuss - May 2010 - curthread->t_procp->p_model in a zone

I''d like for my provider to work properly with both 64 and 32 bit user
processes and for that I need to be able to distinguish them at
runtime.
>From what I''ve seen the way people are doing that is by looking at
the
value of

curthread->t_procp->p_model

This works fine in the global zone:
# dtrace -q -n ''BEGIN {printf("%x\n",
curthread->t_procp->p_model);}''
200000

but not very well in a non-global zone:

# dtrace -q -n ''BEGIN {printf("%x\n",
curthread->t_procp->p_model);}''
dtrace: error on enabled probe ID 1 (ID 1: dtrace:::BEGIN): invalid
kernel access in action #1 at DIF offset 0

I presume this is failing due to the lack of the dtrace_kernel privilege
(which cannot be granted to a zone).

Is there a better way to determine whether or not a user process is
64bit which works in a zone?

Thanks,

-M

Hey Mark,

Unfortunately we don''t have a good answer today for non-global zones or
DTrace consumers without the dtrace_kernel privilege. A good RFE would be to
add, say, pstatus_t or something else from proc(4) for less privileged DTrace
uses.

Adam

On May 11, 2010, at 7:36 AM, Mark Phalan wrote:
> 
> I''d like for my provider to work properly with both 64 and 32 bit
user
> processes and for that I need to be able to distinguish them at runtime.
>> From what I''ve seen the way people are doing that is by
looking at the
> value of
> 
> curthread->t_procp->p_model
> 
> This works fine in the global zone:
> # dtrace -q -n ''BEGIN {printf("%x\n",
curthread->t_procp->p_model);}''
> 200000
> 
> but not very well in a non-global zone:
> 
> # dtrace -q -n ''BEGIN {printf("%x\n",
curthread->t_procp->p_model);}''
> dtrace: error on enabled probe ID 1 (ID 1: dtrace:::BEGIN): invalid
> kernel access in action #1 at DIF offset 0
> 
> I presume this is failing due to the lack of the dtrace_kernel privilege
> (which cannot be granted to a zone).
> 
> Is there a better way to determine whether or not a user process is
> 64bit which works in a zone?
> 
> Thanks,
> 
> -M
> 
> _______________________________________________
> dtrace-discuss mailing list
> dtrace-discuss at opensolaris.org

--
Adam Leventhal, Fishworks                        http://blogs.sun.com/ahl