Copy head@r302406 to stable/11 as part of the 11.0-RELEASE cycle.
Prune svn:mergeinfo from the new branch, as nothing has been merged
here.

Additional commits post-branch will follow.

Approved by: re (implicit)
Sponsored by: The FreeBSD Foundation

Add my copyright to flowtable.

Bring copyright notice to standard style.

o Remove at compile time the HASH_ALL code, that was never
tested and is unfinished. However, I've tested my version,
it works okay. As before it is unfinished: timeout aren't
driven by TCP session state. To enable the HASH_ALL mode,
one needs in kernel config:

options FLOWTABLE_HASH_ALL

o Reduce the alignment on flentry to 64 bytes. Without
the FLOWTABLE_HASH_ALL option, twice less memory would
be consumed by flows.
o API to ip_output()/ip6_output() got even more thin: 1 liner.
o Remove unused unions. Simply use fle->f_key[].
o Merge all IPv4 code into flowtable_lookup_ipv4(), and do same
flowtable_lookup_ipv6(). Stop copying data to on stack
sockaddr structures, simply use key[] on stack.
o Move code from flowtable_lookup_common() that actually works
on insertion into flowtable_insert().

Sponsored by: Netflix
Sponsored by: Nginx, Inc.

Whenever flowtable lookup fails, we do route lookup and then try to
insert flow entry. During the route lookup the critical section is
exited. It may happen, that after route lookup we will be executed
on an other CPU that already has such flowentry. Before this change
we simply freed the flowentry and returned to ip_output() with
failure.

Actually there is nothing wrong with using previously allocated
flow entry, updating it properly. Thus, make flowentry_insert()
return the new either old fle, and make use of it.

Count reuses as "collisions" and real inserts as "inserts".

Reviewed by: adrian
Sponsored by: Netflix
Sponsored by: Nginx, Inc.

Don't insert a flowtable entry if the lle isn't yet valid.

Some of the collisions that are occuring are due to flowtable lookups
that succeed but have an invalid lle - typically because the L2 adjacency
lookup hasn't completed. This would lead to a follow-up insert which
would then fail (ie, collision) and the code would fall through to doing
a slow-path L2/L3 lookup in the netinet/netinet6 code.

This patch simply aborts storing a new flowtable entry if the lle isn't
yet valid.

Whilst I'm here, add a new pcpu counter for the item so the number of
failures can be tracked separately from generic "collisions."

Reviewed by: glebius
MFC after: 10 days
Sponsored by: Netflix, Inc.

Remove unused FL_NOAUTO.

o Axe non-pcpu flowtable implementation. It wasn't enabled or used,
and probably is a leftover from first prototyping by Kip. The
non-pcpu implementation used mutexes, so it doubtfully worked
better than simple routing lookup.
o Use UMA_ZONE_PCPU zone for pointers instead of [MAXCPU] arrays,
use zpcpu_get() to access data in there.
o Substitute own single list implementation with SLIST(). This
has two functional side effects:
- new flows go into head of a list, before they went to tail.
- a bug when incorrect flow was deleted in flow cleaner is
fixed.
o Due to cache line alignment, there is no reason to keep
different zones for IPv4 and IPv6 flows. Both consume one
cache line, real size of allocation is equal.
o Rely on that f_hash, f_rt, f_lle are stable during fle
lifetime, remove useless volatile quilifiers.
o More INET/INET6 splitting.

Reviewed by: adrian
Sponsored by: Netflix
Sponsored by: Nginx, Inc.

Fix comment.

o Revamp API between flowtable and netinet, netinet6.
- ip_output() and ip_output6() simply call flowtable_lookup(),
passing mbuf and address family. That's the only code under
#ifdef FLOWTABLE in the protocols code now.
o Revamp statistics gathering and export.
- Remove hand made pcpu stats, and utilize counter(9).
- Snapshot of statistics is available via 'netstat -rs'.
- All sysctls are moved into net.flowtable namespace, since
spreading them over net.inet isn't correct.
o Properly separate at compile time INET and INET6 parts.
o General cleanup.
- Remove chain of multiple flowtables. We simply have one for
IPv4 and one for IPv6.
- Flowtables are allocated in flowtable.c, symbols are static.
- With proper argument to SYSINIT() we no longer need flowtable_ready.
- Hash salt doesn't need to be per-VNET.
- Removed rudimentary debugging, which use quite useless in dtrace era.

The runtime behavior of flowtable shouldn't be changed by this commit.

Sponsored by: Netflix
Sponsored by: Nginx, Inc.

allocate ipv6 flows from the ipv6 flow zone

reported by: rrs@

MFC after: 3 days

- restructure flowtable to support ipv6
- add a name argument to flowtable_alloc for printing with ddb commands
- extend ddb commands to print destination address or 4-tuples
- don't parse ports in ulp header if FL_HASH_ALL is not passed
- add kern_flowtable_insert to enable more generic use of flowtable
(e.g. system calls for adding entries)
- don't hash loopback addresses
- cleanup whitespace
- keep statistics per-cpu for per-cpu flowtables to avoid cache line contention
- add sysctls to accumulate stats and report aggregate

MFC after: 7 days

The flow-table associates TCP/UDP flows and IP destinations with
specific routes. When the routing table changes, for example,
when a new route with a more specific prefix is inserted into the
routing table, the flow-table is not updated to reflect that change.
As such existing connections cannot take advantage of the new path.
In some cases the path is broken. This patch will update the affected
flow-table entries when a more specific route is added. The route
entry is properly marked when a route is deleted from the table.
In this case, when the flow-table performs a search, the stale
entry is updated automatically. Therefore this patch is not
necessary for route deletion.

Submitted by: simon, phk
Reviewed by: bz, kmacy
MFC after: 3 days

- change the interface to flowtable_lookup so that we don't rely on
the mbuf for obtaining the fib index
- check that a cached flow corresponds to the same fib index as the
packet for which we are doing the lookup
- at interface detach time flush any flows referencing stale rtentrys
associated with the interface that is going away (fixes reported
panics)
- reduce the time between cleans in case the cleaner is running at
the time the eventhandler is called and the wakeup is missed less
time will elapse before the eventhandler returns
- separate per-vnet initialization from global initialization
(pointed out by jeli@)

Reviewed by: sam@
Approved by: re@

Remove unused VNET_SET() and related macros; only VNET_GET() is
ever actually used. Rename VNET_GET() to VNET() to shorten
variable references.

Discussed with: bz, julian
Reviewed by: bz
Approved by: re (kensmith, kib)

Build on Jeff Roberson's linker-set based dynamic per-CPU allocator
(DPCPU), as suggested by Peter Wemm, and implement a new per-virtual
network stack memory allocator. Modify vnet to use the allocator
instead of monolithic global container structures (vinet, ...). This
change solves many binary compatibility problems associated with
VIMAGE, and restores ELF symbols for virtualized global variables.

Each virtualized global variable exists as a "reference copy", and also
once per virtual network stack. Virtualized global variables are
tagged at compile-time, placing the in a special linker set, which is
loaded into a contiguous region of kernel memory. Virtualized global
variables in the base kernel are linked as normal, but those in modules
are copied and relocated to a reserved portion of the kernel's vnet
region with the help of a the kernel linker.

Virtualized global variables exist in per-vnet memory set up when the
network stack instance is created, and are initialized statically from
the reference copy. Run-time access occurs via an accessor macro, which
converts from the current vnet and requested symbol to a per-vnet
address. When "options VIMAGE" is not compiled into the kernel, normal
global ELF symbols will be used instead and indirection is avoided.

This change restores static initialization for network stack global
variables, restores support for non-global symbols and types, eliminates
the need for many subsystem constructors, eliminates large per-subsystem
structures that caused many binary compatibility issues both for
monitoring applications (netstat) and kernel modules, removes the
per-function INIT_VNET_*() macros throughout the stack, eliminates the
need for vnet_symmap ksym(2) munging, and eliminates duplicate
definitions of virtualized globals under VIMAGE_GLOBALS.

Bump __FreeBSD_version and update UPDATING.

Portions submitted by: bz
Reviewed by: bz, zec
Discussed with: gnn, jamie, jeff, jhb, julian, sam
Suggested by: peter
Approved by: re (kensmith)

V_irtualize flowtable state.

This change should make options VIMAGE kernel builds usable again,
to some extent at least.

Note that the size of struct vnet_inet has changed, though in
accordance with one-bump-per-day policy we didn't update the
__FreeBSD_version number, given that it has already been touched
by r194640 a few hours ago.
Reviewed by: bz
Approved by: julian (mentor)

Garbage collect an extern for a non-existent variable.
While here let the comment end in a '.' and mark the #endif of _KERNEL.

Reviewed by: rwatson (as part of a larger patch)

Move the kernel option FLOWTABLE chacking from the header file to the
actual implementation.
Remove the accessor functions for the compiled out case, just returning
"unavail" values. Remove the kernel conditional from the header file as
it is no longer needed, only leaving the externs.
Hide the improperly virtualized SYSCTL/TUNABLE for the flowtable size
under the kernel option as well.

Reviewed by: rwatson

revert to opt-in flowtable

make flowtable opt-out

Remove two unneeded, hidden includes.

After r193232 rt_tables in vnet.h are no longer indirectly dependent on
the ROUTETABLES kernel option thus there is no need to include opt_route.h
anymore in all consumers of vnet.h and no longer depend on it for module
builds.

Remove the hidden include in flowtable.h as well and leave the two
explicit #includes in ip_input.c and ip_output.c.

- Import infrastructure for caching flows as a means of accelerating L3 and L2 lookups
as well as providing stateful load balancing when used with RADIX_MPATH.
- Currently compiled in to i386 and amd64 but disabled by default, it can be enabled at
runtime with 'sysctl net.inet.flowtable.enable=1'.

- Embedded users can remove it entirely from the kernel by adding 'nooption FLOWTABLE' to
their kernel config files.

- A minimal hookup will be added to ip_output in a subsequent commit. I would like to see
more review before bringing in changes that require more churn.

Supported by: Bitgravity Inc.