dtrace discuss - Jul 2008 - printa stddev error

Hi,
   Searched for similar errors and nothing came up. Anybody know what this is?
When using aggregate stddev, and then trying to print it at END, using printa, I
get this error:
dtrace: processing aborted: Invalid return value from callback
avg works fine. Tried on two recent builds of solaris (build 89 and build 94),
on two difference sparc systems.
Here''s a sample script:

#!/usr/sbin/dtrace -Fs

profile-1000
{
        @a["dev" ]= stddev(timestamp);
}

END
{
        printa("%3s\t%@d\n", @a);
}


--
This message posted from opensolaris.org

On Thu, Jul 24, 2008 at 2:12 PM, Chris Kiick <chris.kiick at sun.com>
wrote:
> Hi,
>   Searched for similar errors and nothing came up. Anybody know what this
is?
> When using aggregate stddev, and then trying to print it at END, using
printa, I get this error:
> dtrace: processing aborted: Invalid return value from callback
I''ll look at this when I get a chance (unless someone else gets there
first.)

Chad

On Thu, Jul 24, 2008 at 12:12:01PM -0700, Chris Kiick
wrote:
> Hi,
>    Searched for similar errors and nothing came up. Anybody know what this
is?
> When using aggregate stddev, and then trying to print it at END, using
printa, I get this error:
> dtrace: processing aborted: Invalid return value from callback
> avg works fine. Tried on two recent builds of solaris (build 89 and build
94), on two difference sparc systems.
> Here''s a sample script:
> 
Easier test case:

--- cut here ---
#!/usr/sbin/dtrace -Fs

BEGIN
{
        @ = stddev(100);
        @ = stddev(200);
        @ = stddev(300);

        printa("%@d", @);
}
--- cut here ---

Looks like we''re missing a printf handler for stddev:

usr/src/lib/libdtrace/common/dt_printf.c:
...
1213 static int
1214 pfprint_average(dtrace_hdl_t *dtp, FILE *fp, const char *format,
1215     const dt_pfargd_t *pfd, const void *addr, size_t size, uint64_t normal)
1216 {
1217         const uint64_t *data = addr;
1218 
1219         if (size != sizeof (uint64_t) * 2)
1220                 return (dt_set_errno(dtp, EDT_DMISMATCH));
1221 
1222         return (dt_printf(dtp, fp, format,
1223             data[0] ? data[1] / normal / data[0] : 0));
1224 }
...
1242 static int
1243 dt_printf_format(dtrace_hdl_t *dtp, FILE *fp, const dt_pfargv_t *pfv,
1244     const dtrace_recdesc_t *recs, uint_t nrecs, const void *buf,
1245     size_t len, const dtrace_aggdata_t **aggsdata, int naggvars)
1246 {
...
1414                 switch (rec->dtrd_action) {
1415                 case DTRACEAGG_AVG:
1416                         func = pfprint_average;
1417                         break;
...
1430                 default:
1431                         func = pfc->pfc_print;
1432                         break;
1433                 }

We need a matching entry for stddev.

Cheers,
- jonathan

On Fri, Jul 25, 2008 at 2:31 PM, Jonathan Adams <jonathan.adams at
sun.com> wrote:
> On Thu, Jul 24, 2008 at 12:12:01PM -0700, Chris Kiick wrote:
>> Hi,
>>    Searched for similar errors and nothing came up. Anybody know what
this is?
>> When using aggregate stddev, and then trying to print it at END, using
printa, I get this error:
>> dtrace: processing aborted: Invalid return value from callback
>> avg works fine. Tried on two recent builds of solaris (build 89 and
build 94), on two difference sparc systems.
>> Here''s a sample script:
>>
>
> Easier test case:
>
> --- cut here ---
> #!/usr/sbin/dtrace -Fs
>
> BEGIN
> {
>        @ = stddev(100);
>        @ = stddev(200);
>        @ = stddev(300);
>
>        printa("%@d", @);
> }
> --- cut here ---
>
> Looks like we''re missing a printf handler for stddev:
>
[ ... ]
>
> We need a matching entry for stddev.
Yep, I''m already there.

Chris, have you filed a bug for this?  If not, I can do so.  If so,
could you give me the number?

Thanks,
Chad

On Fri, Jul 25, 2008 at 3:08 PM, chris kiick <chris.kiick at sun.com>
wrote:
> On Fri, Jul 25, 2008 at 03:05:06PM -0500, Chad Mynhier wrote:
>> Chris, have you filed a bug for this?  If not, I can do so.  If so,
>> could you give me the number?
>>
>> Thanks,
>> Chad
>
> Hot off the press: CR6730130: dtrace missing printf handler for stdev
If anyone wants to sponsor this, I can have a webrev up later this
weekend (after a clean nightly build, bfu, etc.)

Chad

Chad Mynhier wrote:
> On Fri, Jul 25, 2008 at 3:08 PM, chris kiick <chris.kiick at sun.com>
wrote:
>> On Fri, Jul 25, 2008 at 03:05:06PM -0500, Chad Mynhier wrote:
>>> Chris, have you filed a bug for this?  If not, I can do so.  If so,
>>> could you give me the number?
>>>
>>> Thanks,
>>> Chad
>> Hot off the press: CR6730130: dtrace missing printf handler for stdev
> 
> If anyone wants to sponsor this, I can have a webrev up later this
> weekend (after a clean nightly build, bfu, etc.)
> 
> Chad
If no one opposes, I''d like to sponsor Chad on this one.

Rafael