Hi,

I just noticed the following:

Name  Mtu  Net/Dest      Address        Ipkts  Ierrs Opkts  Oerrs Collis Queue
lo0   8232 127.0.0.0     127.0.0.1      885409 0     885409 0     0      0
bge0  1500 172.23.7.0    172.23.7.1     0      0     0      0     0      0

All the counters of the bge interface are still zero - although there
is some activity on the network. It seems the kstat counters aren't
updated:

% kstat -n bge0
module: bge                             instance: 0
name:   bge0                            class:    net
        brdcstrcv                       0
        brdcstxmt                       0
        collisions                      0
        crtime                          66.537377628
        ierrors                         0
        ifspeed                         100000000
        ipackets                        0
        ipackets64                      0
        multircv                        0
        multixmt                        0
        norcvbuf                        0
        noxmtbuf                        0
        obytes                          0
        obytes64                        0
        oerrors                         0
        opackets                        0
        opackets64                      0
        rbytes                          0
        rbytes64                        0
        snaptime                        534806.471186194
        unknowns                        35

Is this in any way related, that the bge driver has been converted to
GLDv3:

New features in Solaris 10 1/06:
.. GLDv3: High performance, feature rich and easy-to-use framework for
  developing network device drivers. Solaris OS currently includes three
  GLD v3-based drivers: bge, e1000g and the 10 Gigabit xge driver.

I found the following bug description:

http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6270191

But why has this bug been closed? "Not a bug" - seems clearly a bug to
me.

-- 
Daniel

"Daniel Rock" <abuse@deadcafe.de> writes:

>I found the following bug description:

>http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6270191

>But why has this bug been closed? "Not a bug" - seems clearly a bug to
>me.


Check bug 6342824

(works for some hardware not for others; it is fixed in snv_30)

Casper
-- 
Expressed in this posting are my opinions.  They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

Casper H.S. Dik <Casper.Dik@sun.com> wrote:
> "Daniel Rock" <abuse@deadcafe.de> writes:
> 
>>I found the following bug description:
> 
>>http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6270191
> 
>>But why has this bug been closed? "Not a bug" - seems clearly a bug to
>>me.
> 
> 
> Check bug 6342824
> 
> (works for some hardware not for others; it is fixed in snv_30)

Followup this old thread again.

After applying patch 122028-01 kstat counters look ok at first sight, but
the 64 bit counters are not really 64 bits wide. Instead they wrap around
just like the 32 bit counters. Example:

% kstat -n bge0
[...]
        obytes                          4138495820
        obytes64                        4138497338
[...]
        snaptime                        11519.089478927


and a few seconds later:

% kstat -n bge0
[...]
        obytes                          30145664
        obytes64                        30147182
[...]
        snaptime                        11534.562332634


This drives my performance statistics program crazy, because it assumes that
the *64 kstat names really have 64 bit precision and do not wrap on 32 bit
boundaries.

It would have been better if the kstat names "*64" had been omitted instead
of returning bogus values.

-- 
Daniel

"Daniel Rock" <v200612@deadcafe.de> writes:

>Followup this old thread again.

>After applying patch 122028-01 kstat counters look ok at first sight, but
>the 64 bit counters are not really 64 bits wide. Instead they wrap around
>just like the 32 bit counters. Example:

>% kstat -n bge0
>[...]
>        obytes                          4138495820
>        obytes64                        4138497338
>[...]
>        snaptime                        11519.089478927


>and a few seconds later:

>% kstat -n bge0
>[...]
>        obytes                          30145664
>        obytes64                        30147182
>[...]
>        snaptime                        11534.562332634


>This drives my performance statistics program crazy, because it assumes that
>the *64 kstat names really have 64 bit precision and do not wrap on 32 bit
>boundaries.

>It would have been better if the kstat names "*64" had been omitted instead
>of returning bogus values.

Looks like another bug to me.

Casper