ss is a no-frills tool to show internal information about STREAMS on your 
system.  It shows all queues from the stream head and those queues which are
not reachable from the stream head, how many bytes are in the queue,
as well as any flow control and (if possible) the responsible process 
or TCP/IP connection.  

Aside from being an educational tool, ss can be used to determine if STREAMS 
blocks aren't being released to the system because they're stuck in a 
full queue somewhere, possibly leading to STREAMS allocation failures.  
This can happen when TCP can't send any more data without exceeding 
the window size.  netstat -a will show this in the "Send-Q" column too.  
NOTE: STREAMS blocks that are allocated with allocb(K) and not freed (freeb(K))
and not stuck on a queue will *not* show up with ss -- there's no way to 
know which driver is still using them (or lost track of them).
Note that the Open Server 5.0.x TCP/IP stack caches these blocks - see below.

You can get the same information from crash(ADM)'s "queue" and "stream"
commands.  The flags are the same as crash and represent the following
from /usr/include/sys/stream.h:

en = QENAB
wr = QWANTR
ww = QWANTW
fl = QFULL
rr = QREADR
us = QUSE
ne = QNOENB

ss doesn't have any options.  It assumes /unix and /dev/kmem.

An early ODT3.0 version is buried on SLS net382e (ftp from ftp.sco.com in 
/SLS), although you must tar it off manually.

A totally unknown fact is that Open Server 5 TCP/IP caches 256 byte, 2K, and 4K
blocks controlled by the kernel parameter NSTRPAGES / 8  ***in each class***.  
As you increase NSTRPAGES, TCP/IP increases its internal cache.  
In other words, from a crash/strstat perspective, these blocks 
have been allocated but never returned to the system, so another driver
(IPX, MDI NIC, etc.) may not be able to obtain a STREAM block for
its own use.   When you shut down the stack with slink -u the blocks 
are returned via freeb/freemsg.  You can monitor how many blocks TCP/IP
is currently caching with the new-to-OSR5 undocumented netstat -ma flag.  
look at the

inet mblk cache: 256 = 0, 2048 = 125, 4096 = 15

line.  These blocks would otherwise be in the FREE column in crash's 
strstat output and available for another driver to use.

The cache size is controlled by a variable in /etc/conf/pack.d/socket/space.c  
called in_proto_tot.  By setting it to zero you will eliminate the cache. 

The ss program will not see any of TCP/IP's cached blocks as they aren't
on a queue.