relationships:
Sockets: filedesc->accept->sellck
Routing: radix node head->rtentry->ifaddr
UDP: udp->udpinp
TCP: tcp->tcpinp
SLIP: slip_mtx->slip sc_mtx
Drop in a place holder section for UNIX domain sockets. Various
sections to be expanded over the next few days.
sched_clock() rather than using callouts. This means we no longer have to
take the load of the callout thread into consideration while balancing and
should make the balancing decisions simpler and more accurate.
Tested on: x86/UP, amd64/SMP
global mutex, accept_mtx, which serializes access to the following
fields across all sockets:
so_qlen so_incqlen so_qstate
so_comp so_incomp so_list
so_head
While providing only coarse granularity, this approach avoids lock
order issues between sockets by avoiding ownership of the fields
by a specific socket and its per-socket mutexes.
While here, rewrite soclose(), sofree(), soaccept(), and
sonewconn() to add assertions, close additional races and address
lock order concerns. In particular:
- Reorganize the optimistic concurrency behavior in accept1() to
always allocate a file descriptor with falloc() so that if we do
find a socket, we don't have to encounter the "Oh, there wasn't
a socket" race that can occur if falloc() sleeps in the current
code, which broke inbound accept() ordering, not to mention
requiring backing out socket state changes in a way that raced
with the protocol level. We may want to add a lockless read of
the queue state if polling of empty queues proves to be important
to optimize.
- In accept1(), soref() the socket while holding the accept lock
so that the socket cannot be free'd in a race with the protocol
layer. Likewise in netgraph equivilents of the accept1() code.
- In sonewconn(), loop waiting for the queue to be small enough to
insert our new socket once we've committed to inserting it, or
races can occur that cause the incomplete socket queue to
overfill. In the previously implementation, it was sufficient
to simply tested once since calling soabort() didn't release
synchronization permitting another thread to insert a socket as
we discard a previous one.
- In soclose()/sofree()/et al, it is the responsibility of the
caller to remove a socket from the incomplete connection queue
before calling soabort(), which prevents soabort() from having
to walk into the accept socket to release the socket from its
queue, and avoids races when releasing the accept mutex to enter
soabort(), permitting soabort() to avoid lock ordering issues
with the caller.
- Generally cluster accept queue related operations together
throughout these functions in order to facilitate locking.
Annotate new locking in socketvar.h.
descriptors out of fdrop_locked() and into vn_closefile(). This
removes all knowledge of vnodes from fdrop_locked(), since the lock
behavior was specific to vnodes. This also removes the specific
requirement for Giant in fdrop_locked(), it's now only required by
code that it calls into.
Add GIANT_REQUIRED to vn_closefile() since VFS requires Giant.
nextpkt within the m_hdr was not being initialized to NULL for
!M_PKTHDR cases. *Maybe* this will fix weird socket buffer
inconsistency panics, but we'll see.
the socket is on an accept queue of a listen socket. This change
renames the flags to SQ_COMP and SQ_INCOMP, and moves them to a new
state field on the socket, so_qstate, as the locking for these flags
is substantially different for the locking on the remainder of the
flags in so_state.
them to behave the same as if the SS_NBIO socket flag had been set
for this call. The SS_NBIO flag for ordinary sockets is set by
fcntl(fd, F_SETFL, O_NONBLOCK).
Pass the MSG_NBIO flag to the soreceive() and sosend() calls in
fifo_read() and fifo_write() instead of frobbing the SS_NBIO flag
on the underlying socket for each I/O operation. The O_NONBLOCK
flag is a property of the descriptor, and unlike ordinary sockets,
fifos may be referenced by multiple descriptors.
mbuma is an Mbuf & Cluster allocator built on top of a number of
extensions to the UMA framework, all included herein.
Extensions to UMA worth noting:
- Better layering between slab <-> zone caches; introduce
Keg structure which splits off slab cache away from the
zone structure and allows multiple zones to be stacked
on top of a single Keg (single type of slab cache);
perhaps we should look into defining a subset API on
top of the Keg for special use by malloc(9),
for example.
- UMA_ZONE_REFCNT zones can now be added, and reference
counters automagically allocated for them within the end
of the associated slab structures. uma_find_refcnt()
does a kextract to fetch the slab struct reference from
the underlying page, and lookup the corresponding refcnt.
mbuma things worth noting:
- integrates mbuf & cluster allocations with extended UMA
and provides caches for commonly-allocated items; defines
several zones (two primary, one secondary) and two kegs.
- change up certain code paths that always used to do:
m_get() + m_clget() to instead just use m_getcl() and
try to take advantage of the newly defined secondary
Packet zone.
- netstat(1) and systat(1) quickly hacked up to do basic
stat reporting but additional stats work needs to be
done once some other details within UMA have been taken
care of and it becomes clearer to how stats will work
within the modified framework.
From the user perspective, one implication is that the
NMBCLUSTERS compile-time option is no longer used. The
maximum number of clusters is still capped off according
to maxusers, but it can be made unlimited by setting
the kern.ipc.nmbclusters boot-time tunable to zero.
Work should be done to write an appropriate sysctl
handler allowing dynamic tuning of kern.ipc.nmbclusters
at runtime.
Additional things worth noting/known issues (READ):
- One report of 'ips' (ServeRAID) driver acting really
slow in conjunction with mbuma. Need more data.
Latest report is that ips is equally sucking with
and without mbuma.
- Giant leak in NFS code sometimes occurs, can't
reproduce but currently analyzing; brueffer is
able to reproduce but THIS IS NOT an mbuma-specific
problem and currently occurs even WITHOUT mbuma.
- Issues in network locking: there is at least one
code path in the rip code where one or more locks
are acquired and we end up in m_prepend() with
M_WAITOK, which causes WITNESS to whine from within
UMA. Current temporary solution: force all UMA
allocations to be M_NOWAIT from within UMA for now
to avoid deadlocks unless WITNESS is defined and we
can determine with certainty that we're not holding
any locks when we're M_WAITOK.
- I've seen at least one weird socketbuffer empty-but-
mbuf-still-attached panic. I don't believe this
to be related to mbuma but please keep your eyes
open, turn on debugging, and capture crash dumps.
This change removes more code than it adds.
A paper is available detailing the change and considering
various performance issues, it was presented at BSDCan2004:
http://www.unixdaemons.com/~bmilekic/netbuf_bmilekic.pdf
Please read the paper for Future Work and implementation
details, as well as credits.
Testing and Debugging:
rwatson,
brueffer,
Ketrien I. Saihr-Kesenchedra,
...
Reviewed by: Lots of people (for different parts)
KERN_PROC_SESSION option which had been previously defined but
never implemented.
PR: bin/65803 (a very tiny piece of the PR)`
Submitted by: Cyrille Lefevre
of not clearing the flags for execv() syscall will result that a new
program runs in KSE thread mode without enabling it.
Submitted by: tjr
Modified by: davidxu
threatened in rev.1.10 of usr.sbin/kgmon/kgmon.c more than 2 years ago.
kgmon has been recovering from the missing initialization for too
long, but the fixup there is ifdefed for i386's and shouldn't be
needed for other arches.
high resolution kernel profiling (options GUPROF. "U" in GUPROF stands
for microseconds resolution, but the resolution is now smaller than 1
nanosecond on multi-GHz machines and the accuracy is heading towards
1 nanosecond too). Arches that support GUPROF must now provide certain
macros for the calibration. GUPROF is now only supported for i386's,
so the absence of the new macros for other arches doesn't break anything
that wasn't already broken. amd64's have uncommitted support for
GUPROF, and sparc64's have support that seems to be complete except
here (there was an #error for non-i386 cases; now there are undefined
macros).
Changed the asms a little:
- declare them as __volatile. They must not be moved, and exporting a
label across asms is technically incorrect, so try harder to stop gcc
moving them.
- don't put the non-clobbered register "bx" in the clobber list. The
clobber lists are still more conservative than necessary.
- drop the non-support for gcc-1. It just gave a better error message,
and this is not useful since compiling with gcc-1 would cause thousands
of worse error messages.
- drop the support for aout.
of kmupetext(). The declaration is misplaced in <machine/profile.h>
since it is not MD and not related to the lowest level of profiling.
It will be moved, but getting it via <sys/gmon.h> already works.
and cannot handle it going away, add an explicit reference to the kobj
class inside each linker class. Without this, a class with no modules
loaded will sit with an idle refcount of 0. Loading and unloading
a module with it causes a 0->1->0 transition which frees the ops table
and causes subsequent loads using that class to explode. Normally, the
"kernel" module will remain forever loaded and prevent this happening, but
if you have more than one linker class active, only one owns the "kernel".
This finishes making modules work for kldload(8) on amd64.
(nobits) tables to simplify some code. Try and shorten some of the very
wide lines. Somewhere along the way, I think I fixed the memory
corruption that caused panics after going multiuser.
elf_reloc() backends for two reasons. First, to support the possibility
of there being two elf linkers in the kernel (eg: amd64), and second, to
pass the relocbase explicitly (for relocating .o format kld files).
is "void *" (it isn't) or that the default promotion of pid_t is int.
Instead, assume that casting "struct foo *" to "void *" and printing the
result with %p is useful, and that all pid_t's are representable as longs.
Fixed some minor style bugs (mainly spelling errors in comments).
removes a specific thread from a sleep queue. sleepq_resume_thread()
resumes scheduling of a thread that has been previously removed from a
sleep queue.
- sleepq_catch_signals() just removes a thread from the queue it was just
added to when a pending signal is found.
- sleepq_signal() and sleepq_broadcast() remove threads from a queue,
drop the queue lock, and then resume all the previously removed threads.
This doesn't completely fix the sched_lock <-> sleepq chain LOR, but it
makes it a little better as we no longer call setrunnble() with a sleep
queue lock held meaning if setrunnable() tries to wakeup the swapper we
don't try to lock two sleep queue chains at the same time.
It's not quite correct from a posix Point Of view, but it is a lot better
than what was there before. This will be revisited later
when we decide what form our priority extensions will take. Posix doesn't
specify how a system scope thread can change its priority so you need to
add non-standard extensions to be able to do it..
For now make this slightly non standard to allow it to be done.
Submitted by: Dan Eischen originally, changed by myself.
