The Brussels dladm(1M) I was working from assumed that Brussels would
putback before UV, and there are obviously still some tweaks to be made
from what I brought in from that manpage. Thanks for looking through it.
Peter Memishian wrote:
> > Some devices do not support configurable datalinks or
> > aggregations. The fixed datalinks provided by such devices
> > can be viewed using dladm, but can not be configured.
>
> After UV, which devices would these be?
None; I've removed that paragraph.
>
> > For example, suppose that the bge0 device has a link name of
> > mgmt0 (because the administrator initially named it such by
> > typing "dladm rename-link bge0 mgmt0".) The bge0 device is
> > physically removed from the system due to failure, and replaced
> > with a ce0 device. To associate the newly added ce0 device
> > with the mgmt0 configuration previously associated with bge0,
> > the administrator types "dladm rename-link ce0 mgmt0"
>
> The above paragraph belongs in EXAMPLES.
Yes, I've done that and re-worded it slightly. See below in the context
of your other comment regarding examples.
>
> > dladm show-dev [-p] [-s [-i interval]] [-o field[,...]] [dev]
>
> Should we state that this is deprecated?
I've added "This subcommand is deprecated in favor of the show-phys
subcommand." at the end of the 1st paragraph.
> > Show device configuration information (the default) or
> > statistics, either for all network devices or for the
> > specified device dev.
>
> An aside: most of the stuff shown by show-dev seems redundant given
> show-link and show-ether. We should probably file an RFE to have it
> gutted, and have show-link display the speed.
I suppose we could have done that as part of UV, but it's a bit late to
do that now. I've filed:
6651255 dladm show-dev needs to be removed
>
> > -p, --parseable
> >
> > Display using a stable machine-parseable format.
>
> Doesn't seem like we document what that format is anywhere in the manpage.
I've added the following subsection after the subcommand descriptions and
before the "General Link Property" subsection:
Parseable Output Format
Many dladm subcommands have an option which displays output in a
machine-parseable format. The format is 0 or more lines with the
following syntax:
<key>="<value>" [<key>="<value>" ...]
Note that <value> is always surrounded explicitly by
double-quotes to make parsing of values with embedded space
characters easier. Also note that <value> may be the empty
string if the key is not associated with a value.
> > dladm show-phys [-pP] [-o field[,...]] [link]
> >
> > Show the physical device and attributes of all physical
> > links, or of the named physical link. Without -P, only
> > physical links which are available on the running system are
> > displayed.
>
> Another aside: IIRC, we had this display the STATE, SPEED, and DUPLEX so
> that we could depecate show-dev. But STATE is already covered by
> show-link, DUPLEX will be covered by show-ether, and MEDIA and SPEED
> probably belong in show-link. So in other words, I think we could
> simplify things a bit post-UV.
Would you like to throw this cleanup in with 6651255? There are no
manpage AIs from this aside, correct?
> >
> > dladm create-aggr [-t] [-R root-dir] [-P policy] [-L mode]
> > [-T time] [-u address] -l linkn ... link
>
> Need a paragraph break here.
This was not my doing. The man page has a paragraph break there, but the
tools to convert nroff to text make all kinds of stupid choices during
conversion. I was focusing more on the text to provide to Terry, and not
so much on the formatting, which I'm guessing Terry can correct using the
man page authoring tools.
Originally I was trying not to touch _any_ formatting, regardless of how
broken it looked since I wanted to provide useful low-noise diffs for
people to review (content-wise).
> > Combine a set of links into a single IEEE 802.3ad link
> > aggregation named "link". The use of an integer "key" to
> > generate a link name for the aggregation is also supported for
> > backward compatibility. Many of the aggr subcommands below
> > also support the use of a "key" to refer to a given
> > aggregation, but use of the aggregation link name is preferred.
> > See the NOTES section for more information on keys.
>
> > dladm modify-aggr [-t] [-R root-dir] [-P policy] [-L mode]
> > [-T time] [-u address] link
>
> Need a paragraph break here.
Same here.
>
> > Modify the parameters of the specified aggregation.
>
> > dladm delete-aggr [-t] [-R root-dir] link
>
> Need a paragraph break here.
Same here.
>
> > Deletes the specified aggregation.
>
> > dladm add-aggr [-t] [-R root-dir] -l linkn ... link
>
> The above should probably be "link ... linkn"
No, the last argument is the name of the aggregation. The arguments to
-l are the underlying links being added to the aggregation. This used to be:
-d dev ... key
It should probably be something like:
-l link1 [-l link2 ...] link
Is that better?
>
> > -l, --link
> >
> > Specifies a link to add to the aggregation. Multiple links
> > can be added by supplying multiple -l options.
>
> This doesn't seem to match the notation above.
It never has matched, this was a previously existing bug with -d. I've
fixed it as:
-l link ..., --link link ...
>
> > dladm remove-aggr [-t] [-R root-dir] -l linkn ... link
>
> Likewise.
Yep.
>
> > Removes links from the specified aggregation.
> >
> > -l, --link
> >
> > Specifies a link to remove from the aggregation. Multiple
> > links can be removed by supplying multiple -l options.
>
> Likewise.
Yep.
>
> > The MAC address of the link or port.
> >
> > PORTSTATE
> >
> > XXX
>
> Will be filled in soon?
What is it PORTSTATE?
>
>
> > VID
> >
> > The VLAN tag or ID associated with the VLAN.
>
> Might be clearer as "The VLAN tag (ID) associated with the VLAN."
Yes, fixed.
>
> > FLAGS
> >
> > A set of flags associated with the VLAN link. The
> > only flag possible is 'f', which denotes that the
> > VLAN was created using the -f option to create-vlan.
> > More flags may be defined in the future.
>
> Need to add 'i' (for implicitly created VLANs).
Ah yes. How about:
FLAGS
A set of flags associated with the VLAN link.
Possible flags are:
f The VLAN was created using the -f option to
create-vlan.
i The VLAN was implicitly created by plumbing an IP
interface on a PPA-hack DLPI device. These VLAN
links are automatically deleted on last close of
the DLPI device (i.e., when the IP interface
associated with the VLAN link is unplumbed.)
Additional flags may be defined in the future.
> > General Link Property
> > The following general link property is supported:
> >
>
> As Cathy mentioned, "autopush" needs to be covered.
I've added:
autopush
Specifies the set of STREAMS modules to push on the
stream associated with a link when its DLPI device is
opened. It is a space delimited list of modules.
This property is preferred over the more general
autopush(1M) command.
> > Ethernet Link Properties
> > The following MII Properties as documented in ieee802.3(5)
> > are supported in read-only mode:
> >
> > o link_duplex
> >
> > o link_up
> >
> > o adv_autoneg_cap
> >
> > o adv_1000fdx_cap
> >
> > o adv_1000hdx_cap
> >
> > o adv_100fdx_cap
> >
> > o adv_100hdx_cap
> >
> > o adv_10fdx_cap
> >
> > o adv_10hdx_cap
>
> Would be really nice to find a way to remove the blank lines in the above
> list. (This manpage is long enough already ;-)
Again, a lot of formatting is lost in translation.
>
> > EXAMPLES
>
> Would be nice to see some UV examples here :-)
I've added:
Example 7 Renaming a Link
To rename the bge0 link to mgmt0, enter the following command:
# dladm rename-link bge0 mgmt0
Example 8 Replacing a Network Card
Suppose that the bge0 device, whose link was named mgmt0 as shown
is the previous example, needs to be replaced with a ce0 device
due to a hardware failure. The bge0 NIC is physically removed,
and replaced with a new ce0 NIC. To associate the newly added
ce0 device with the mgmt0 configuration previously associated
with bge0, enter the following command:
# dladm rename-link ce0 mgmt0
> > tab() box; cw(2.75i) |cw(2.75i) lw(2.75i) |lw(2.75i) ATTRI-
> > BUTE TYPEATTRIBUTE VALUE _ AvailabilitySUNWcsu _ Interface
> > StabilityEvolving
> [ ... ]
> > tab() box; cw(2.75i) |cw(2.75i) lw(2.75i) |lw(2.75i) ATTRI-
> > BUTE TYPEATTRIBUTE VALUE _ AvailabilitySUNWcsr _ Interface
> > StabilityEvolving
>
> ?
See my previous comments regarding formatting.
Thanks, I'll send out an updated manpage momentarily.
-Seb
