d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
8cb3aed24c
freebsd-skq/sys/kern/uipc_socket.c

3346 lines
87 KiB
C
Raw Normal View History

Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*-
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* Copyright (c) 1982, 1986, 1988, 1990, 1993
Canonicalize copyrights in some files I hold copyrights on: - Sort by date in license blocks, oldest copyright first. - All rights reserved after all copyrights, not just the first. - Use (c) to be consistent with other entries. MFC after: 3 days
2007-01-08 17:49:59 +00:00
* The Regents of the University of California.
As with socket consumer references (so_count), make sofree() return without GC'ing the socket if a strong protocol reference to the socket is present (SS_PROTOREF).
2006-03-15 12:45:35 +00:00
* Copyright (c) 2004 The FreeBSD Foundation
Update copyright date in light of soreceive_dgram(9).
2008-07-03 06:47:45 +00:00
* Copyright (c) 2004-2008 Robert N. M. Watson
Canonicalize copyrights in some files I hold copyrights on: - Sort by date in license blocks, oldest copyright first. - All rights reserved after all copyrights, not just the first. - Use (c) to be consistent with other entries. MFC after: 3 days
2007-01-08 17:49:59 +00:00
* All rights reserved.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)uipc_socket.c 8.3 (Berkeley) 4/15/94
*/
Add a somewhat sizable comment documenting the semantics of various kernel socket calls relating to the creation and destruction of sockets. This will eventually form the foundation of socket(9), but is currently in too much flux to do so. MFC after: 3 months
2006-04-01 10:43:02 +00:00
/*
* Comments on the socket life cycle:
*
* soalloc() sets of socket layer state for a socket, called only by
* socreate() and sonewconn(). Socket layer private.
*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* sodealloc() tears down socket layer state for a socket, called only by
Add a somewhat sizable comment documenting the semantics of various kernel socket calls relating to the creation and destruction of sockets. This will eventually form the foundation of socket(9), but is currently in too much flux to do so. MFC after: 3 months
2006-04-01 10:43:02 +00:00
* sofree() and sonewconn(). Socket layer private.
*
* pru_attach() associates protocol layer state with an allocated socket;
* called only once, may fail, aborting socket allocation. This is called
* from socreate() and sonewconn(). Socket layer private.
*
* pru_detach() disassociates protocol layer state from an attached socket,
* and will be called exactly once for sockets in which pru_attach() has
* been successfully called. If pru_attach() returned an error,
* pru_detach() will not be called. Socket layer private.
*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* pru_abort() and pru_close() notify the protocol layer that the last
* consumer of a socket is starting to tear down the socket, and that the
* protocol should terminate the connection. Historically, pru_abort() also
* detached protocol state from the socket state, but this is no longer the
* case.
*
Add a somewhat sizable comment documenting the semantics of various kernel socket calls relating to the creation and destruction of sockets. This will eventually form the foundation of socket(9), but is currently in too much flux to do so. MFC after: 3 months
2006-04-01 10:43:02 +00:00
* socreate() creates a socket and attaches protocol state. This is a public
* interface that may be used by socket layer consumers to create new
* sockets.
*
* sonewconn() creates a socket and attaches protocol state. This is a
* public interface that may be used by protocols to create new sockets when
* a new connection is received and will be available for accept() on a
* listen socket.
*
* soclose() destroys a socket after possibly waiting for it to disconnect.
* This is a public interface that socket consumers should use to close and
* release a socket when done with it.
*
* soabort() destroys a socket without waiting for it to disconnect (used
* only for incoming connections that are already partially or fully
* connected). This is used internally by the socket layer when clearing
* listen socket queues (due to overflow or close on the listen socket), but
* is also a public interface protocols may use to abort connections in
* their incomplete listen queues should they no longer be required. Sockets
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* placed in completed connection listen queues should not be aborted for
* reasons described in the comment above the soclose() implementation. This
* is not a general purpose close routine, and except in the specific
* circumstances described here, should not be used.
Add a somewhat sizable comment documenting the semantics of various kernel socket calls relating to the creation and destruction of sockets. This will eventually form the foundation of socket(9), but is currently in too much flux to do so. MFC after: 3 months
2006-04-01 10:43:02 +00:00
*
* sofree() will free a socket and its protocol state if all references on
* the socket have been released, and is the public interface to attempt to
* free a socket when a reference is removed. This is a socket layer private
* interface.
*
* NOTE: In addition to socreate() and soclose(), which provide a single
* socket reference to the consumer to be managed as required, there are two
* calls to explicitly manage socket references, soref(), and sorele().
* Currently, these are generally required only when transitioning a socket
* from a listen queue to a file descriptor, in order to prevent garbage
* collection of the socket at an untimely moment. For a number of reasons,
* these interfaces are not preferred, and should be avoided.
*/
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Accept filters broke kernels compiled without options INET. Make accept filters conditional on INET support to fix. Pointed out by: bde Tested and assisted by: Stephen J. Kiernan <sab@vegamuse.org>
2000-11-20 01:35:25 +00:00
#include "opt_inet.h"
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#include "opt_mac.h"
#include "opt_zero.h"
Allow 32bit get/setsockopt with SO_SNDTIMEO or SO_RECVTIMEO to work.
2005-10-27 04:26:35 +00:00
#include "opt_compat.h"
Accept filters broke kernels compiled without options INET. Make accept filters conditional on INET support to fix. Pointed out by: bde Tested and assisted by: Stephen J. Kiernan <sab@vegamuse.org>
2000-11-20 01:35:25 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
Don't #include <sys/fcntl.h> in <sys/file.h> if KERNEL is defined. Fixed everything that depended on getting fcntl.h stuff from the wrong place. Most things don't depend on file.h stuff at all.
1997-03-23 03:37:54 +00:00
#include <sys/fcntl.h>
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#include <sys/limits.h>
#include <sys/lock.h>
#include <sys/mac.h>
Include file cleanup; mac.h and malloc.h at one point had ordering relationship requirements, and no longer do. Reminded by: bde
2002-08-01 17:47:56 +00:00
#include <sys/malloc.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/mbuf.h>
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#include <sys/mutex.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/domain.h>
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
#include <sys/file.h> /* for struct knote */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/kernel.h>
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
#include <sys/event.h>
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
#include <sys/eventhandler.h>
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
#include <sys/poll.h>
Reviewed by: the cast of thousands This is the change to struct sockets that gets rid of so_uid and replaces it with a much more useful struct pcred *so_cred. This is here to be able to do socket-level credential checks (i.e. IPFW uid/gid support, to be added to HEAD soon). Along with this comes an update to pidentd which greatly simplifies the code necessary to get a uid from a socket. Soon to come: a sysctl() interface to finding individual sockets' credentials.
1999-06-17 23:54:50 +00:00
#include <sys/proc.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/resourcevar.h>
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
#include <net/route.h>
All of this is cosmetic. prototypes, #includes, printfs and so on. Makes GCC a lot more silent.
1994-10-02 17:35:40 +00:00
#include <sys/signalvar.h>
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
#include <sys/stat.h>
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
#include <sys/sx.h>
Make somaxconn (maximum backlog in a listen(2) request) and sb_max (maximum size of a socket buffer) tunable. Permit callers of listen(2) to specify a negative backlog, which is translated into somaxconn. Previously, a negative backlog was silently translated into 0.
1995-11-03 18:33:46 +00:00
#include <sys/sysctl.h>
Moved some #includes from <sys/param.h> nearer to where they are actually used.
1998-03-28 10:33:27 +00:00
#include <sys/uio.h>
o Modify jail to limit creation of sockets to UNIX domain sockets, TCP/IP (v4) sockets, and routing sockets. Previously, interaction with IPv6 was not well-defined, and might be inappropriate for some environments. Similarly, sysctl MIB entries providing interface information also give out only addresses from those protocol domains. For the time being, this functionality is enabled by default, and toggleable using the sysctl variable jail.socket_unixiproute_only. In the future, protocol domains will be able to determine whether or not they are ``jail aware''. o Further limitations on process use of getpriority() and setpriority() by jailed processes. Addresses problem described in kern/17878. Reviewed by: phk, jmg
2000-06-04 04:28:31 +00:00
#include <sys/jail.h>
Undo part of the tangle of having sys/lock.h and sys/mutex.h included in other "system" header files. Also help the deprecation of lockmgr.h by making it a sub-include of sys/lock.h and removing sys/lockmgr.h form kernel .c files. Sort sys/*.h includes where possible in affected files. OK'ed by: bde (with reservations)
2001-05-01 08:13:21 +00:00
Complete break-out of sys/sys/mac.h into sys/security/mac/mac_framework.h begun with a repo-copy of mac.h to mac_framework.h. sys/mac.h now contains the userspace and user<->kernel API and definitions, with all in-kernel interfaces moved to mac_framework.h, which is now included across most of the kernel instead. This change is the first step in a larger cleanup and sweep of MAC Framework interfaces in the kernel, and will not be MFC'd. Obtained from: TrustedBSD Project Sponsored by: SPARTA
2006-10-22 11:52:19 +00:00
#include <security/mac/mac_framework.h>
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#include <vm/uma.h>
Make somaxconn (maximum backlog in a listen(2) request) and sb_max (maximum size of a socket buffer) tunable. Permit callers of listen(2) to specify a negative backlog, which is translated into somaxconn. Previously, a negative backlog was silently translated into 0.
1995-11-03 18:33:46 +00:00
Allow 32bit get/setsockopt with SO_SNDTIMEO or SO_RECVTIMEO to work.
2005-10-27 04:26:35 +00:00
#ifdef COMPAT_IA32
#include <sys/mount.h>
#include <compat/freebsd32/freebsd32.h>
extern struct sysentvec ia32_freebsd_sysvec;
#endif
#include <machine/limits.h> explicitly in the few places that it is required.
1997-08-21 20:33:42 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
static int soreceive_rcvoob(struct socket *so, struct uio *uio,
int flags);
static void filt_sordetach(struct knote *kn);
static int filt_soread(struct knote *kn, long hint);
static void filt_sowdetach(struct knote *kn);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
static int filt_sowrite(struct knote *kn, long hint);
static int filt_solisten(struct knote *kn, long hint);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
static struct filterops solisten_filtops =
Extend kqueue down to the device layer. Backwards compatible approach suggested by: peter
2001-02-15 16:34:11 +00:00
{ 1, NULL, filt_sordetach, filt_solisten };
static struct filterops soread_filtops =
{ 1, NULL, filt_sordetach, filt_soread };
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
static struct filterops sowrite_filtops =
Extend kqueue down to the device layer. Backwards compatible approach suggested by: peter
2001-02-15 16:34:11 +00:00
{ 1, NULL, filt_sowdetach, filt_sowrite };
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
uma_zone_t socket_zone;
Convert socket structures to be type-stable and add a version number. Define a parameter which indicates the maximum number of sockets in a system, and use this to size the zone allocators used for sockets and for certain PCBs. Convert PF_LOCAL PCB structures to be type-stable and add a version number. Define an external format for infomation about socket structures and use it in several places. Define a mechanism to get all PF_LOCAL and PF_INET PCB lists through sysctl(3) without blocking network interrupts for an unreasonable length of time. This probably still has some bugs and/or race conditions, but it seems to work well enough on my machines. It is now possible for `netstat' to get almost all of its information via the sysctl(3) interface rather than reading kmem (changes to follow).
1998-05-15 20:11:40 +00:00
so_gen_t so_gencnt; /* generation count for sockets */
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
int maxsockets;
Last major round (Unless Bruce thinks of somthing :-) of malloc changes. Distribute all but the most fundamental malloc types. This time I also remembered the trick to making things static: Put "static" in front of them. A couple of finer points by: bde
1997-10-12 20:26:33 +00:00
MALLOC_DEFINE(M_SONAME, "soname", "socket name");
MALLOC_DEFINE(M_PCB, "pcb", "protocol control block");
Make somaxconn (maximum backlog in a listen(2) request) and sb_max (maximum size of a socket buffer) tunable. Permit callers of listen(2) to specify a negative backlog, which is translated into somaxconn. Previously, a negative backlog was silently translated into 0.
1995-11-03 18:33:46 +00:00
static int somaxconn = SOMAXCONN;
Rename somaxconn_sysctl() to sysctl_somaxconn() so that I will be able to claim that sofoo() functions all accept a socket as their first argument.
2007-02-15 10:11:00 +00:00
static int sysctl_somaxconn(SYSCTL_HANDLER_ARGS);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/* XXX: we dont have SYSCTL_USHORT */
SYSCTL_PROC(_kern_ipc, KIPC_SOMAXCONN, somaxconn, CTLTYPE_UINT | CTLFLAG_RW,
Rename somaxconn_sysctl() to sysctl_somaxconn() so that I will be able to claim that sofoo() functions all accept a socket as their first argument.
2007-02-15 10:11:00 +00:00
0, sizeof(int), sysctl_somaxconn, "I", "Maximum pending socket connection "
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
"queue size");
static int numopensockets;
SYSCTL_INT(_kern_ipc, OID_AUTO, numopensockets, CTLFLAG_RD,
&numopensockets, 0, "Number of open sockets");
#ifdef ZERO_COPY_SOCKETS
/* These aren't static because they're used in other files. */
int so_zero_copy_send = 1;
int so_zero_copy_receive = 1;
SYSCTL_NODE(_kern_ipc, OID_AUTO, zero_copy, CTLFLAG_RD, 0,
"Zero copy controls");
SYSCTL_INT(_kern_ipc_zero_copy, OID_AUTO, receive, CTLFLAG_RW,
&so_zero_copy_receive, 0, "Enable zero copy receive");
SYSCTL_INT(_kern_ipc_zero_copy, OID_AUTO, send, CTLFLAG_RW,
&so_zero_copy_send, 0, "Enable zero copy send");
#endif /* ZERO_COPY_SOCKETS */
/*
* accept_mtx locks down per-socket fields relating to accept queues. See
* socketvar.h for an annotation of the protected fields of struct socket.
*/
struct mtx accept_mtx;
MTX_SYSINIT(accept_mtx, &accept_mtx, "accept", MTX_DEF);
/*
* so_global_mtx protects so_gencnt, numopensockets, and the per-socket
* so_gencnt field.
*/
static struct mtx so_global_mtx;
MTX_SYSINIT(so_global_mtx, &so_global_mtx, "so_glabel", MTX_DEF);
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
/*
* General IPC sysctl name space, used by sockets and a variety of other IPC
* types.
*/
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
SYSCTL_NODE(_kern, KERN_IPC, ipc, CTLFLAG_RW, 0, "IPC");
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
/*
* Sysctl to get and set the maximum global sockets limit. Notify protocols
* of the change so that they can update their dependent limits as required.
*/
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
static int
sysctl_maxsockets(SYSCTL_HANDLER_ARGS)
{
int error, newmaxsockets;
newmaxsockets = maxsockets;
Despite several examples in the kernel, the third argument of sysctl_handle_int is not sizeof the int type you want to export. The type must always be an int or an unsigned int. Remove the instances where a sizeof(variable) is passed to stop people accidently cut and pasting these examples. In a few places this was sysctl_handle_int was being used on 64 bit types, which would truncate the value to be exported. In these cases use sysctl_handle_quad to export them and change the format to Q so that sysctl(1) can still print them.
2007-06-04 18:25:08 +00:00
error = sysctl_handle_int(oidp, &newmaxsockets, 0, req);
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
if (error == 0 && req->newptr) {
if (newmaxsockets > maxsockets) {
maxsockets = newmaxsockets;
if (maxsockets > ((maxfiles / 4) * 3)) {
maxfiles = (maxsockets * 5) / 4;
maxfilesperproc = (maxfiles * 9) / 10;
}
EVENTHANDLER_INVOKE(maxsockets_change);
} else
error = EINVAL;
}
return (error);
}
SYSCTL_PROC(_kern_ipc, OID_AUTO, maxsockets, CTLTYPE_INT|CTLFLAG_RW,
&maxsockets, 0, sysctl_maxsockets, "IU",
"Maximum number of sockets avaliable");
/*
Change comment on soabort() to more accurately describe how/when soabort() is used. Remove trailing white space.
2006-07-16 23:09:39 +00:00
* Initialise maxsockets.
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
*/
static void init_maxsockets(void *ignored)
{
TUNABLE_INT_FETCH("kern.ipc.maxsockets", &maxsockets);
maxsockets = imax(maxsockets, imax(maxfiles, nmbclusters));
}
SYSINIT(param, SI_SUB_TUNABLES, SI_ORDER_ANY, init_maxsockets, NULL);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Socket operation routines. These routines are called by the routines in
* sys_socket.c or from a system process, and implement the semantics of
* socket operations by switching out to the protocol specific routines.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Convert socket structures to be type-stable and add a version number. Define a parameter which indicates the maximum number of sockets in a system, and use this to size the zone allocators used for sockets and for certain PCBs. Convert PF_LOCAL PCB structures to be type-stable and add a version number. Define an external format for infomation about socket structures and use it in several places. Define a mechanism to get all PF_LOCAL and PF_INET PCB lists through sysctl(3) without blocking network interrupts for an unreasonable length of time. This probably still has some bugs and/or race conditions, but it seems to work well enough on my machines. It is now possible for `netstat' to get almost all of its information via the sysctl(3) interface rather than reading kmem (changes to follow).
1998-05-15 20:11:40 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Get a socket structure from our zone, and initialize it. Note that it
* would probably be better to allocate socket and PCB at the same time, but
* I'm not convinced that all the protocols can be easily modified to do
* this.
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
*
* soalloc() returns a socket with a ref count of 0.
Convert socket structures to be type-stable and add a version number. Define a parameter which indicates the maximum number of sockets in a system, and use this to size the zone allocators used for sockets and for certain PCBs. Convert PF_LOCAL PCB structures to be type-stable and add a version number. Define an external format for infomation about socket structures and use it in several places. Define a mechanism to get all PF_LOCAL and PF_INET PCB lists through sysctl(3) without blocking network interrupts for an unreasonable length of time. This probably still has some bugs and/or race conditions, but it seems to work well enough on my machines. It is now possible for `netstat' to get almost all of its information via the sysctl(3) interface rather than reading kmem (changes to follow).
1998-05-15 20:11:40 +00:00
*/
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
static struct socket *
Don't block on the socket zone limit during the socket() call which can easily lock up a system otherwise; instead, return ENOBUFS as documented in a manpage, thus reverting us to the FreeBSD 4.x behavior. Reviewed by: rwatson MFC after: 2 weeks
2007-02-26 10:45:21 +00:00
soalloc(void)
Convert socket structures to be type-stable and add a version number. Define a parameter which indicates the maximum number of sockets in a system, and use this to size the zone allocators used for sockets and for certain PCBs. Convert PF_LOCAL PCB structures to be type-stable and add a version number. Define an external format for infomation about socket structures and use it in several places. Define a mechanism to get all PF_LOCAL and PF_INET PCB lists through sysctl(3) without blocking network interrupts for an unreasonable length of time. This probably still has some bugs and/or race conditions, but it seems to work well enough on my machines. It is now possible for `netstat' to get almost all of its information via the sysctl(3) interface rather than reading kmem (changes to follow).
1998-05-15 20:11:40 +00:00
{
struct socket *so;
Don't block on the socket zone limit during the socket() call which can easily lock up a system otherwise; instead, return ENOBUFS as documented in a manpage, thus reverting us to the FreeBSD 4.x behavior. Reviewed by: rwatson MFC after: 2 weeks
2007-02-26 10:45:21 +00:00
so = uma_zalloc(socket_zone, M_NOWAIT | M_ZERO);
Rearrange code in soalloc() so that it's less indented by returning early if uma_zalloc() from the socket zone fails. No functional change. MFC after: 1 week
2006-06-08 22:33:18 +00:00
if (so == NULL)
return (NULL);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#ifdef MAC
Merge first in a series of TrustedBSD MAC Framework KPI changes from Mac OS X Leopard--rationalize naming for entry points to the following general forms: mac_<object>_<method/action> mac_<object>_check_<method/action> The previous naming scheme was inconsistent and mostly reversed from the new scheme. Also, make object types more consistent and remove spaces from object types that contain multiple parts ("posix_sem" -> "posixsem") to make mechanical parsing easier. Introduce a new "netinet" object type for certain IPv4/IPv6-related methods. Also simplify, slightly, some entry point names. All MAC policy modules will need to be recompiled, and modules not updates as part of this commit will need to be modified to conform to the new KPI. Sponsored by: SPARTA (original patches against Mac OS X) Obtained from: TrustedBSD Project, Apple Computer
2007-10-24 19:04:04 +00:00
if (mac_socket_init(so, M_NOWAIT) != 0) {
Rearrange code in soalloc() so that it's less indented by returning early if uma_zalloc() from the socket zone fails. No functional change. MFC after: 1 week
2006-06-08 22:33:18 +00:00
uma_zfree(socket_zone, so);
return (NULL);
Convert socket structures to be type-stable and add a version number. Define a parameter which indicates the maximum number of sockets in a system, and use this to size the zone allocators used for sockets and for certain PCBs. Convert PF_LOCAL PCB structures to be type-stable and add a version number. Define an external format for infomation about socket structures and use it in several places. Define a mechanism to get all PF_LOCAL and PF_INET PCB lists through sysctl(3) without blocking network interrupts for an unreasonable length of time. This probably still has some bugs and/or race conditions, but it seems to work well enough on my machines. It is now possible for `netstat' to get almost all of its information via the sysctl(3) interface rather than reading kmem (changes to follow).
1998-05-15 20:11:40 +00:00
}
Rearrange code in soalloc() so that it's less indented by returning early if uma_zalloc() from the socket zone fails. No functional change. MFC after: 1 week
2006-06-08 22:33:18 +00:00
#endif
SOCKBUF_LOCK_INIT(&so->so_snd, "so_snd");
SOCKBUF_LOCK_INIT(&so->so_rcv, "so_rcv");
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
sx_init(&so->so_snd.sb_sx, "so_snd_sx");
sx_init(&so->so_rcv.sb_sx, "so_rcv_sx");
Rearrange code in soalloc() so that it's less indented by returning early if uma_zalloc() from the socket zone fails. No functional change. MFC after: 1 week
2006-06-08 22:33:18 +00:00
TAILQ_INIT(&so->so_aiojobq);
mtx_lock(&so_global_mtx);
so->so_gencnt = ++so_gencnt;
++numopensockets;
mtx_unlock(&so_global_mtx);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
return (so);
Convert socket structures to be type-stable and add a version number. Define a parameter which indicates the maximum number of sockets in a system, and use this to size the zone allocators used for sockets and for certain PCBs. Convert PF_LOCAL PCB structures to be type-stable and add a version number. Define an external format for infomation about socket structures and use it in several places. Define a mechanism to get all PF_LOCAL and PF_INET PCB lists through sysctl(3) without blocking network interrupts for an unreasonable length of time. This probably still has some bugs and/or race conditions, but it seems to work well enough on my machines. It is now possible for `netstat' to get almost all of its information via the sysctl(3) interface rather than reading kmem (changes to follow).
1998-05-15 20:11:40 +00:00
}
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
/*
* Free the storage associated with a socket at the socket layer, tear down
* locks, labels, etc. All protocol state is assumed already to have been
* torn down (and possibly never set up) by the caller.
*/
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
static void
sodealloc(struct socket *so)
{
KASSERT(so->so_count == 0, ("sodealloc(): so_count %d", so->so_count));
KASSERT(so->so_pcb == NULL, ("sodealloc(): so_pcb != NULL"));
mtx_lock(&so_global_mtx);
so->so_gencnt = ++so_gencnt;
Move updated of 'numopensockets' from bottom of sodealloc() to the top, eliminating a second set of identical mutex operations at the bottom. This allows brief exceeding of the max sockets limit, but only by sockets in the last stages of being torn down.
2006-08-02 00:45:27 +00:00
--numopensockets; /* Could be below, but faster here. */
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
mtx_unlock(&so_global_mtx);
if (so->so_rcv.sb_hiwat)
(void)chgsbsize(so->so_cred->cr_uidinfo,
&so->so_rcv.sb_hiwat, 0, RLIM_INFINITY);
if (so->so_snd.sb_hiwat)
(void)chgsbsize(so->so_cred->cr_uidinfo,
&so->so_snd.sb_hiwat, 0, RLIM_INFINITY);
#ifdef INET
/* remove acccept filter if one is present. */
if (so->so_accf != NULL)
do_setopt_accept_filter(so, NULL);
#endif
#ifdef MAC
Merge first in a series of TrustedBSD MAC Framework KPI changes from Mac OS X Leopard--rationalize naming for entry points to the following general forms: mac_<object>_<method/action> mac_<object>_check_<method/action> The previous naming scheme was inconsistent and mostly reversed from the new scheme. Also, make object types more consistent and remove spaces from object types that contain multiple parts ("posix_sem" -> "posixsem") to make mechanical parsing easier. Introduce a new "netinet" object type for certain IPv4/IPv6-related methods. Also simplify, slightly, some entry point names. All MAC policy modules will need to be recompiled, and modules not updates as part of this commit will need to be modified to conform to the new KPI. Sponsored by: SPARTA (original patches against Mac OS X) Obtained from: TrustedBSD Project, Apple Computer
2007-10-24 19:04:04 +00:00
mac_socket_destroy(so);
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
#endif
crfree(so->so_cred);
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
sx_destroy(&so->so_snd.sb_sx);
sx_destroy(&so->so_rcv.sb_sx);
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
SOCKBUF_LOCK_DESTROY(&so->so_snd);
SOCKBUF_LOCK_DESTROY(&so->so_rcv);
uma_zfree(socket_zone, so);
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
* socreate returns a socket with a ref count of 1. The socket should be
* closed with soclose().
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
socreate(int dom, struct socket **aso, int type, int proto,
struct ucred *cred, struct thread *td)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct protosw *prp;
struct socket *so;
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (proto)
prp = pffindproto(dom, proto, type);
else
prp = pffindtype(dom, type);
o Modify jail to limit creation of sockets to UNIX domain sockets, TCP/IP (v4) sockets, and routing sockets. Previously, interaction with IPv6 was not well-defined, and might be inappropriate for some environments. Similarly, sysctl MIB entries providing interface information also give out only addresses from those protocol domains. For the time being, this functionality is enabled by default, and toggleable using the sysctl variable jail.socket_unixiproute_only. In the future, protocol domains will be able to determine whether or not they are ``jail aware''. o Further limitations on process use of getpriority() and setpriority() by jailed processes. Addresses problem described in kern/17878. Reviewed by: phk, jmg
2000-06-04 04:28:31 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (prp == NULL || prp->pr_usrreqs->pru_attach == NULL ||
prp->pr_usrreqs->pru_attach == pru_attach_notsupp)
Fix panic by moving the prp == 0 check up the order of sanity checks. Submitted by: Bart Thate <freebsd@1st.dudi.org> on -current Approved by: rwatson
2000-06-13 15:44:04 +00:00
return (EPROTONOSUPPORT);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (jailed(cred) && jail_socket_unixiproute_only &&
o Modify jail to limit creation of sockets to UNIX domain sockets, TCP/IP (v4) sockets, and routing sockets. Previously, interaction with IPv6 was not well-defined, and might be inappropriate for some environments. Similarly, sysctl MIB entries providing interface information also give out only addresses from those protocol domains. For the time being, this functionality is enabled by default, and toggleable using the sysctl variable jail.socket_unixiproute_only. In the future, protocol domains will be able to determine whether or not they are ``jail aware''. o Further limitations on process use of getpriority() and setpriority() by jailed processes. Addresses problem described in kern/17878. Reviewed by: phk, jmg
2000-06-04 04:28:31 +00:00
prp->pr_domain->dom_family != PF_LOCAL &&
prp->pr_domain->dom_family != PF_INET &&
prp->pr_domain->dom_family != PF_ROUTE) {
return (EPROTONOSUPPORT);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (prp->pr_type != type)
return (EPROTOTYPE);
Don't block on the socket zone limit during the socket() call which can easily lock up a system otherwise; instead, return ENOBUFS as documented in a manpage, thus reverting us to the FreeBSD 4.x behavior. Reviewed by: rwatson MFC after: 2 weeks
2007-02-26 10:45:21 +00:00
so = soalloc();
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (so == NULL)
Convert socket structures to be type-stable and add a version number. Define a parameter which indicates the maximum number of sockets in a system, and use this to size the zone allocators used for sockets and for certain PCBs. Convert PF_LOCAL PCB structures to be type-stable and add a version number. Define an external format for infomation about socket structures and use it in several places. Define a mechanism to get all PF_LOCAL and PF_INET PCB lists through sysctl(3) without blocking network interrupts for an unreasonable length of time. This probably still has some bugs and/or race conditions, but it seems to work well enough on my machines. It is now possible for `netstat' to get almost all of its information via the sysctl(3) interface rather than reading kmem (changes to follow).
1998-05-15 20:11:40 +00:00
return (ENOBUFS);
Changed socket code to use 4.4BSD queue macros. This includes removing the obsolete soqinsque and soqremque functions as well as collapsing so_q0len and so_qlen into a single queue length of unaccepted connections. Now the queue of unaccepted & complete connections is checked directly for queued sockets. The new code should be functionally equivilent to the old while being substantially faster - especially in cases where large numbers of connections are often queued for accept (e.g. http).
1996-03-11 15:37:44 +00:00
TAILQ_INIT(&so->so_incomp);
TAILQ_INIT(&so->so_comp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
so->so_type = type;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
so->so_cred = crhold(cred);
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
if ((prp->pr_domain->dom_family == PF_INET) ||
(prp->pr_domain->dom_family == PF_ROUTE))
so->so_fibnum = td->td_proc->p_fibnum;
else
so->so_fibnum = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
so->so_proto = prp;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#ifdef MAC
Merge first in a series of TrustedBSD MAC Framework KPI changes from Mac OS X Leopard--rationalize naming for entry points to the following general forms: mac_<object>_<method/action> mac_<object>_check_<method/action> The previous naming scheme was inconsistent and mostly reversed from the new scheme. Also, make object types more consistent and remove spaces from object types that contain multiple parts ("posix_sem" -> "posixsem") to make mechanical parsing easier. Introduce a new "netinet" object type for certain IPv4/IPv6-related methods. Also simplify, slightly, some entry point names. All MAC policy modules will need to be recompiled, and modules not updates as part of this commit will need to be modified to conform to the new KPI. Sponsored by: SPARTA (original patches against Mac OS X) Obtained from: TrustedBSD Project, Apple Computer
2007-10-24 19:04:04 +00:00
mac_socket_create(cred, so);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#endif
Fix the recent panics/LORs/hangs created by my kqueue commit by: - Introducing the possibility of using locks different than mutexes for the knlist locking. In order to do this, we add three arguments to knlist_init() to specify the functions to use to lock, unlock and check if the lock is owned. If these arguments are NULL, we assume mtx_lock, mtx_unlock and mtx_owned, respectively. - Using the vnode lock for the knlist locking, when doing kqueue operations on a vnode. This way, we don't have to lock the vnode while holding a mutex, in filt_vfsread. Reviewed by: jmg Approved by: re (scottl), scottl (mentor override) Pointyhat to: ssouhlal Will be happy: everyone
2005-07-01 16:28:32 +00:00
knlist_init(&so->so_rcv.sb_sel.si_note, SOCKBUF_MTX(&so->so_rcv),
NULL, NULL, NULL);
knlist_init(&so->so_snd.sb_sel.si_note, SOCKBUF_MTX(&so->so_snd),
NULL, NULL, NULL);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
so->so_count = 1;
Generic socket buffer auto sizing support, header defines, flag inheritance. MFC after: 1 month
2007-02-01 17:53:41 +00:00
/*
* Auto-sizing of socket buffers is managed by the protocols and
* the appropriate flags must be set in the pru_attach function.
*/
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
error = (*prp->pr_usrreqs->pru_attach)(so, proto, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (error) {
Before performing a sodealloc() when pru_attach() fails, assert that the socket refcount remains 1, and then drop to 0 before freeing the socket. PR: 101763 Reported by: Gleb Kozyrev <gkozyrev at ukr dot net>
2006-08-11 23:03:10 +00:00
KASSERT(so->so_count == 1, ("socreate: so_count %d",
so->so_count));
so->so_count = 0;
When pru_attach() fails, call sodealloc() on the socket rather than using sorele() and the full tear-down path. Since protocol state allocation failed, this is not required (and is arguably undesirable). This matches the behavior of sonewconn() under the same circumstances.
2006-07-11 21:56:58 +00:00
sodealloc(so);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
*aso = so;
return (0);
}
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
#ifdef REGRESSION
static int regression_sonewconn_earlytest = 1;
SYSCTL_INT(_regression, OID_AUTO, sonewconn_earlytest, CTLFLAG_RW,
&regression_sonewconn_earlytest, 0, "Perform early sonewconn limit test");
#endif
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* When an attempt at a new connection is noted on a socket which accepts
* connections, sonewconn is called. If the connection is possible (subject
* to space constraints, etc.) then we allocate a new structure, propoerly
* linked into the data structure of the original socket, and return this.
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
* Connstatus may be 0, or SO_ISCONFIRMING, or SO_ISCONNECTED.
*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Note: the ref count on the socket is 0 on return.
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
*/
struct socket *
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sonewconn(struct socket *head, int connstatus)
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
struct socket *so;
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
int over;
ACCEPT_LOCK();
over = (head->so_qlen > 3 * head->so_qlimit / 2);
ACCEPT_UNLOCK();
#ifdef REGRESSION
if (regression_sonewconn_earlytest && over)
#else
if (over)
#endif
return (NULL);
Don't block on the socket zone limit during the socket() call which can easily lock up a system otherwise; instead, return ENOBUFS as documented in a manpage, thus reverting us to the FreeBSD 4.x behavior. Reviewed by: rwatson MFC after: 2 weeks
2007-02-26 10:45:21 +00:00
so = soalloc();
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
if (so == NULL)
return (NULL);
if ((head->so_options & SO_ACCEPTFILTER) != 0)
connstatus = 0;
so->so_head = head;
so->so_type = head->so_type;
so->so_options = head->so_options &~ SO_ACCEPTCONN;
so->so_linger = head->so_linger;
so->so_state = head->so_state | SS_NOFDREF;
so->so_proto = head->so_proto;
so->so_cred = crhold(head->so_cred);
#ifdef MAC
SOCK_LOCK(head);
Merge first in a series of TrustedBSD MAC Framework KPI changes from Mac OS X Leopard--rationalize naming for entry points to the following general forms: mac_<object>_<method/action> mac_<object>_check_<method/action> The previous naming scheme was inconsistent and mostly reversed from the new scheme. Also, make object types more consistent and remove spaces from object types that contain multiple parts ("posix_sem" -> "posixsem") to make mechanical parsing easier. Introduce a new "netinet" object type for certain IPv4/IPv6-related methods. Also simplify, slightly, some entry point names. All MAC policy modules will need to be recompiled, and modules not updates as part of this commit will need to be modified to conform to the new KPI. Sponsored by: SPARTA (original patches against Mac OS X) Obtained from: TrustedBSD Project, Apple Computer
2007-10-24 19:04:04 +00:00
mac_socket_newconn(head, so);
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
SOCK_UNLOCK(head);
#endif
knlist_init(&so->so_rcv.sb_sel.si_note, SOCKBUF_MTX(&so->so_rcv),
NULL, NULL, NULL);
knlist_init(&so->so_snd.sb_sel.si_note, SOCKBUF_MTX(&so->so_snd),
NULL, NULL, NULL);
if (soreserve(so, head->so_snd.sb_hiwat, head->so_rcv.sb_hiwat) ||
(*so->so_proto->pr_usrreqs->pru_attach)(so, 0, NULL)) {
sodealloc(so);
return (NULL);
}
New sockets created by incoming connections into listen sockets should inherit all settings and options except listen specific options. Add the missing send/receive timeouts and low watermarks. Remove inheritance of the field so_timeo which is unused. Noticed by: phk Reviewed by: rwatson Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 days
2006-09-10 17:08:06 +00:00
so->so_rcv.sb_lowat = head->so_rcv.sb_lowat;
so->so_snd.sb_lowat = head->so_snd.sb_lowat;
so->so_rcv.sb_timeo = head->so_rcv.sb_timeo;
so->so_snd.sb_timeo = head->so_snd.sb_timeo;
Generic socket buffer auto sizing support, header defines, flag inheritance. MFC after: 1 month
2007-02-01 17:53:41 +00:00
so->so_rcv.sb_flags |= head->so_rcv.sb_flags & SB_AUTOSIZE;
so->so_snd.sb_flags |= head->so_snd.sb_flags & SB_AUTOSIZE;
Move some functions and definitions from uipc_socket2.c to uipc_socket.c: - Move sonewconn(), which creates new sockets for incoming connections on listen sockets, so that all socket allocate code is together in uipc_socket.c. - Move 'maxsockets' and associated sysctls to uipc_socket.c with the socket allocation code. - Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it to sysctl.h and remove lots of scattered implementations in various IPC modules. - Sort sodealloc() after soalloc() in uipc_socket.c for dependency order reasons. Statisticize soalloc() and sodealloc() as they are now required only in uipc_socket.c, and are internal to the socket implementation. After this change, socket allocation and deallocation is entirely centralized in one file, and uipc_socket2.c consists entirely of socket buffer manipulation and default protocol switch functions. MFC after: 1 month
2006-06-10 14:34:07 +00:00
so->so_state |= connstatus;
ACCEPT_LOCK();
if (connstatus) {
TAILQ_INSERT_TAIL(&head->so_comp, so, so_list);
so->so_qstate |= SQ_COMP;
head->so_qlen++;
} else {
/*
* Keep removing sockets from the head until there's room for
* us to insert on the tail. In pre-locking revisions, this
* was a simple if(), but as we could be racing with other
* threads and soabort() requires dropping locks, we must
* loop waiting for the condition to be true.
*/
while (head->so_incqlen > head->so_qlimit) {
struct socket *sp;
sp = TAILQ_FIRST(&head->so_incomp);
TAILQ_REMOVE(&head->so_incomp, sp, so_list);
head->so_incqlen--;
sp->so_qstate &= ~SQ_INCOMP;
sp->so_head = NULL;
ACCEPT_UNLOCK();
soabort(sp);
ACCEPT_LOCK();
}
TAILQ_INSERT_TAIL(&head->so_incomp, so, so_list);
so->so_qstate |= SQ_INCOMP;
head->so_incqlen++;
}
ACCEPT_UNLOCK();
if (connstatus) {
sorwakeup(head);
wakeup_one(&head->so_timeo);
}
return (so);
}
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sobind(struct socket *so, struct sockaddr *nam, struct thread *td)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
return ((*so->so_proto->pr_usrreqs->pru_bind)(so, nam, td));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
* solisten() transitions a socket from a non-listening state to a listening
* state, but can also be used to update the listen queue depth on an
* existing listen socket. The protocol will call back into the sockets
* layer using solisten_proto_check() and solisten_proto() to check and set
* socket-layer listen state. Call backs are used so that the protocol can
Fix spelling in a comment. MFC after: 3 days
2005-09-18 10:46:34 +00:00
* acquire both protocol and socket layer locks in whatever order is required
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
* by the protocol.
*
* Protocol implementors are advised to hold the socket lock across the
* socket-layer test and set to avoid races at the socket layer.
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
solisten(struct socket *so, int backlog, struct thread *td)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
Push the assignment of a new or updated so_qlimit from solisten() following the protocol pru_listen() call to solisten_proto(), so that it occurs under the socket lock acquisition that also sets SO_ACCEPTCONN. This requires passing the new backlog parameter to the protocol, which also allows the protocol to be aware of changes in queue limit should it wish to do something about the new queue limit. This continues a move towards the socket layer acting as a library for the protocol. Bump __FreeBSD_version due to a change in the in-kernel protocol interface. This change has been tested with IPv4 and UNIX domain sockets, but not other protocols.
2005-10-30 19:44:40 +00:00
return ((*so->so_proto->pr_usrreqs->pru_listen)(so, backlog, td));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
solisten_proto_check(struct socket *so)
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
{
SOCK_LOCK_ASSERT(so);
if (so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING |
SS_ISDISCONNECTING))
return (EINVAL);
return (0);
}
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
solisten_proto(struct socket *so, int backlog)
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
{
SOCK_LOCK_ASSERT(so);
Push the assignment of a new or updated so_qlimit from solisten() following the protocol pru_listen() call to solisten_proto(), so that it occurs under the socket lock acquisition that also sets SO_ACCEPTCONN. This requires passing the new backlog parameter to the protocol, which also allows the protocol to be aware of changes in queue limit should it wish to do something about the new queue limit. This continues a move towards the socket layer acting as a library for the protocol. Bump __FreeBSD_version due to a change in the in-kernel protocol interface. This change has been tested with IPv4 and UNIX domain sockets, but not other protocols.
2005-10-30 19:44:40 +00:00
if (backlog < 0 || backlog > somaxconn)
backlog = somaxconn;
so->so_qlimit = backlog;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
so->so_options |= SO_ACCEPTCONN;
}
/*
* Attempt to free a socket. This should really be sotryfree().
*
Chance protocol switch method pru_detach() so that it returns void rather than an error. Detaches do not "fail", they other occur or the protocol flags SS_PROTOREF to take ownership of the socket. soclose() no longer looks at so_pcb to see if it's NULL, relying entirely on the protocol to decide whether it's time to free the socket or not using SS_PROTOREF. so_pcb is now entirely owned and managed by the protocol code. Likewise, no longer test so_pcb in other socket functions, such as soreceive(), which have no business digging into protocol internals. Protocol detach routines no longer try to free the socket on detach, this is performed in the socket code if the protocol permits it. In rts_detach(), no longer test for rp != NULL in detach, and likewise in other protocols that don't permit a NULL so_pcb, reduce the incidence of testing for it during detach. netinet and netinet6 are not fully updated to this change, which will be in an upcoming commit. In their current state they may leak memory or panic. MFC after: 3 months
2006-04-01 15:42:02 +00:00
* sofree() will succeed if:
*
* - There are no outstanding file descriptor references or related consumers
* (so_count == 0).
*
* - The socket has been closed by user space, if ever open (SS_NOFDREF).
*
* - The protocol does not have an outstanding strong reference on the socket
* (SS_PROTOREF).
*
Add missing 'not' to SQ_COMP comment. MFC after: 3 months
2006-04-23 15:37:23 +00:00
* - The socket is not in a completed connection queue, so a process has been
Move handling of SQ_COMP exception case in sofree() to the top of the function along with the remainder of the reference checking code. Move comment from body to header with remainder of comments. Inclusion of a socket in a completed connection queue counts as a true reference, and should not be handled as an under-documented edge case. MFC after: 3 months
2006-04-23 15:33:38 +00:00
* notified that it is present. If it is removed, the user process may
* block in accept() despite select() saying the socket was ready.
*
Chance protocol switch method pru_detach() so that it returns void rather than an error. Detaches do not "fail", they other occur or the protocol flags SS_PROTOREF to take ownership of the socket. soclose() no longer looks at so_pcb to see if it's NULL, relying entirely on the protocol to decide whether it's time to free the socket or not using SS_PROTOREF. so_pcb is now entirely owned and managed by the protocol code. Likewise, no longer test so_pcb in other socket functions, such as soreceive(), which have no business digging into protocol internals. Protocol detach routines no longer try to free the socket on detach, this is performed in the socket code if the protocol permits it. In rts_detach(), no longer test for rp != NULL in detach, and likewise in other protocols that don't permit a NULL so_pcb, reduce the incidence of testing for it during detach. netinet and netinet6 are not fully updated to this change, which will be in an upcoming commit. In their current state they may leak memory or panic. MFC after: 3 months
2006-04-01 15:42:02 +00:00
* Otherwise, it will quietly abort so that a future call to sofree(), when
* conditions are right, can succeed.
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sofree(struct socket *so)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Reimplement socket buffer tear-down in sofree(): as the socket is no longer referenced by other threads (hence our freeing it), we don't need to set the can't send and can't receive flags, wake up the consumers, perform two levels of locking, etc. Implement a fast-path teardown, sbdestroy(), which flushes and releases each socket buffer. A manual dom_dispose of the receive buffer is still required explicitly to GC any in-flight file descriptors, etc, before flushing the buffer. This results in a 9% UP performance improvement and 16% SMP performance improvement on a tight loop of socket();close(); in micro-benchmarking, but will likely also affect CPU-bound macro-benchmark performance.
2006-08-01 10:30:26 +00:00
struct protosw *pr = so->so_proto;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct socket *head;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
ACCEPT_LOCK_ASSERT();
SOCK_LOCK_ASSERT(so);
Chance protocol switch method pru_detach() so that it returns void rather than an error. Detaches do not "fail", they other occur or the protocol flags SS_PROTOREF to take ownership of the socket. soclose() no longer looks at so_pcb to see if it's NULL, relying entirely on the protocol to decide whether it's time to free the socket or not using SS_PROTOREF. so_pcb is now entirely owned and managed by the protocol code. Likewise, no longer test so_pcb in other socket functions, such as soreceive(), which have no business digging into protocol internals. Protocol detach routines no longer try to free the socket on detach, this is performed in the socket code if the protocol permits it. In rts_detach(), no longer test for rp != NULL in detach, and likewise in other protocols that don't permit a NULL so_pcb, reduce the incidence of testing for it during detach. netinet and netinet6 are not fully updated to this change, which will be in an upcoming commit. In their current state they may leak memory or panic. MFC after: 3 months
2006-04-01 15:42:02 +00:00
if ((so->so_state & SS_NOFDREF) == 0 || so->so_count != 0 ||
Move handling of SQ_COMP exception case in sofree() to the top of the function along with the remainder of the reference checking code. Move comment from body to header with remainder of comments. Inclusion of a socket in a completed connection queue counts as a true reference, and should not be handled as an under-documented edge case. MFC after: 3 months
2006-04-23 15:33:38 +00:00
(so->so_state & SS_PROTOREF) || (so->so_qstate & SQ_COMP)) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_UNLOCK(so);
ACCEPT_UNLOCK();
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
}
head = so->so_head;
Integrate accept locking from rwatson_netperf, introducing a new 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.
2004-06-02 04:15:39 +00:00
if (head != NULL) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
KASSERT((so->so_qstate & SQ_COMP) != 0 ||
(so->so_qstate & SQ_INCOMP) != 0,
("sofree: so_head != NULL, but neither SQ_COMP nor "
"SQ_INCOMP"));
KASSERT((so->so_qstate & SQ_COMP) == 0 ||
(so->so_qstate & SQ_INCOMP) == 0,
("sofree: so->so_qstate is SQ_COMP and also SQ_INCOMP"));
TAILQ_REMOVE(&head->so_incomp, so, so_list);
head->so_incqlen--;
so->so_qstate &= ~SQ_INCOMP;
Changed socket code to use 4.4BSD queue macros. This includes removing the obsolete soqinsque and soqremque functions as well as collapsing so_q0len and so_qlen into a single queue length of unaccepted connections. Now the queue of unaccepted & complete connections is checked directly for queued sockets. The new code should be functionally equivilent to the old while being substantially faster - especially in cases where large numbers of connections are often queued for accept (e.g. http).
1996-03-11 15:37:44 +00:00
so->so_head = NULL;
Back out my lats commit of locking down a socket, it conflicts with hsu's work. Requested by: hsu
2002-05-31 11:52:35 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
KASSERT((so->so_qstate & SQ_COMP) == 0 &&
(so->so_qstate & SQ_INCOMP) == 0,
("sofree: so_head == NULL, but still SQ_COMP(%d) or SQ_INCOMP(%d)",
so->so_qstate & SQ_COMP, so->so_qstate & SQ_INCOMP));
Fix a race in soclose() where connections could be queued to the listening socket after the pass that cleans those queues. This results in these connections being orphaned (and leaked). The fix is to clean up the so queues after detaching the socket from the protocol. Thanks to ups and jhb for discussions and a thorough code review.
2006-11-22 23:54:29 +00:00
if (so->so_options & SO_ACCEPTCONN) {
KASSERT((TAILQ_EMPTY(&so->so_comp)), ("sofree: so_comp populated"));
KASSERT((TAILQ_EMPTY(&so->so_incomp)), ("sofree: so_comp populated"));
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_UNLOCK(so);
ACCEPT_UNLOCK();
Chance protocol switch method pru_detach() so that it returns void rather than an error. Detaches do not "fail", they other occur or the protocol flags SS_PROTOREF to take ownership of the socket. soclose() no longer looks at so_pcb to see if it's NULL, relying entirely on the protocol to decide whether it's time to free the socket or not using SS_PROTOREF. so_pcb is now entirely owned and managed by the protocol code. Likewise, no longer test so_pcb in other socket functions, such as soreceive(), which have no business digging into protocol internals. Protocol detach routines no longer try to free the socket on detach, this is performed in the socket code if the protocol permits it. In rts_detach(), no longer test for rp != NULL in detach, and likewise in other protocols that don't permit a NULL so_pcb, reduce the incidence of testing for it during detach. netinet and netinet6 are not fully updated to this change, which will be in an upcoming commit. In their current state they may leak memory or panic. MFC after: 3 months
2006-04-01 15:42:02 +00:00
Move the dom_dispose and pru_detach calls in sofree() earlier. Only after calling pru_detach we can be absolutely sure, that we don't have any references to the socket in the stack. This closes race between lockless sbdestroy() and data arriving on socket. Reviewed by: rwatson
2007-03-22 13:21:24 +00:00
if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose != NULL)
(*pr->pr_domain->dom_dispose)(so->so_rcv.sb_mb);
if (pr->pr_usrreqs->pru_detach != NULL)
(*pr->pr_usrreqs->pru_detach)(so);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
Reimplement socket buffer tear-down in sofree(): as the socket is no longer referenced by other threads (hence our freeing it), we don't need to set the can't send and can't receive flags, wake up the consumers, perform two levels of locking, etc. Implement a fast-path teardown, sbdestroy(), which flushes and releases each socket buffer. A manual dom_dispose of the receive buffer is still required explicitly to GC any in-flight file descriptors, etc, before flushing the buffer. This results in a 9% UP performance improvement and 16% SMP performance improvement on a tight loop of socket();close(); in micro-benchmarking, but will likely also affect CPU-bound macro-benchmark performance.
2006-08-01 10:30:26 +00:00
* From this point on, we assume that no other references to this
* socket exist anywhere else in the stack. Therefore, no locks need
* to be acquired or held.
*
* We used to do a lot of socket buffer and socket locking here, as
* well as invoke sorflush() and perform wakeups. The direct call to
* dom_dispose() and sbrelease_internal() are an inlining of what was
* necessary from sorflush().
*
* Notice that the socket buffer and kqueue state are torn down
* before calling pru_detach. This means that protocols shold not
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
* assume they can perform socket wakeups, etc, in their detach code.
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
*/
Reimplement socket buffer tear-down in sofree(): as the socket is no longer referenced by other threads (hence our freeing it), we don't need to set the can't send and can't receive flags, wake up the consumers, perform two levels of locking, etc. Implement a fast-path teardown, sbdestroy(), which flushes and releases each socket buffer. A manual dom_dispose of the receive buffer is still required explicitly to GC any in-flight file descriptors, etc, before flushing the buffer. This results in a 9% UP performance improvement and 16% SMP performance improvement on a tight loop of socket();close(); in micro-benchmarking, but will likely also affect CPU-bound macro-benchmark performance.
2006-08-01 10:30:26 +00:00
sbdestroy(&so->so_snd, so);
sbdestroy(&so->so_rcv, so);
Move destroying kqueue state from above pru_detach to below it in sofree(), as a number of protocols expect to be able to call soisdisconnected() during detach. That may not be a good assumption, but until I'm sure if it's a good assumption or not, allow it.
2006-08-02 18:37:44 +00:00
knlist_destroy(&so->so_rcv.sb_sel.si_note);
knlist_destroy(&so->so_snd.sb_sel.si_note);
Convert socket structures to be type-stable and add a version number. Define a parameter which indicates the maximum number of sockets in a system, and use this to size the zone allocators used for sockets and for certain PCBs. Convert PF_LOCAL PCB structures to be type-stable and add a version number. Define an external format for infomation about socket structures and use it in several places. Define a mechanism to get all PF_LOCAL and PF_INET PCB lists through sysctl(3) without blocking network interrupts for an unreasonable length of time. This probably still has some bugs and/or race conditions, but it seems to work well enough on my machines. It is now possible for `netstat' to get almost all of its information via the sysctl(3) interface rather than reading kmem (changes to follow).
1998-05-15 20:11:40 +00:00
sodealloc(so);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Close a socket on last file table reference removal. Initiate disconnect
* if connected. Free socket when disconnect complete.
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* This function will sorele() the socket. Note that soclose() may be called
* prior to the ref count reaching zero. The actual socket structure will
* not be freed until the ref count reaches zero.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soclose(struct socket *so)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
int error = 0;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
KASSERT(!(so->so_state & SS_NOFDREF), ("soclose: SS_NOFDREF on enter"));
Back out 1.68.2.26. It was a mis-guided change that was already backed out of HEAD and should not have been MFC'd. This will restore UDP socket functionality, which will correct the recent NFS problems. Submitted by: rwatson
2005-06-09 19:56:38 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
funsetown(&so->so_sigio);
Fix a race in soclose() where connections could be queued to the listening socket after the pass that cleans those queues. This results in these connections being orphaned (and leaked). The fix is to clean up the so queues after detaching the socket from the protocol. Thanks to ups and jhb for discussions and a thorough code review.
2006-11-22 23:54:29 +00:00
if (so->so_state & SS_ISCONNECTED) {
if ((so->so_state & SS_ISDISCONNECTING) == 0) {
error = sodisconnect(so);
if (error)
goto drop;
}
if (so->so_options & SO_LINGER) {
if ((so->so_state & SS_ISDISCONNECTING) &&
(so->so_state & SS_NBIO))
goto drop;
while (so->so_state & SS_ISCONNECTED) {
error = tsleep(&so->so_timeo,
PSOCK | PCATCH, "soclos", so->so_linger * hz);
if (error)
break;
}
}
}
drop:
if (so->so_proto->pr_usrreqs->pru_close != NULL)
(*so->so_proto->pr_usrreqs->pru_close)(so);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (so->so_options & SO_ACCEPTCONN) {
struct socket *sp;
ACCEPT_LOCK();
while ((sp = TAILQ_FIRST(&so->so_incomp)) != NULL) {
TAILQ_REMOVE(&so->so_incomp, sp, so_list);
so->so_incqlen--;
sp->so_qstate &= ~SQ_INCOMP;
sp->so_head = NULL;
ACCEPT_UNLOCK();
Change soabort() from returning int to returning void, since all consumers ignore the return value, soabort() is required to succeed, and protocols produce errors here to report multiple freeing of the pcb, which we hope to eliminate.
2006-03-16 07:03:14 +00:00
soabort(sp);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
ACCEPT_LOCK();
Fix for PR #1146: the "next" pointer must be cached before calling soabort since the struct containing it may be freed.
1996-04-16 03:50:08 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
while ((sp = TAILQ_FIRST(&so->so_comp)) != NULL) {
Port NetBSD's 19990120-accept bug fix. This works around the race condition where select(2) can return that a listening socket has a connected socket queued, the connection is broken, and the user calls accept(2), which then blocks because there are no connections queued. Reviewed by: wollman Obtained from: NetBSD (ftp://ftp.NetBSD.ORG/pub/NetBSD/misc/security/patches/19990120-accept)
1999-01-25 16:58:56 +00:00
TAILQ_REMOVE(&so->so_comp, sp, so_list);
Fix the port of the NetBSD 19990120-accept fix. I misread a piece of code when examining their fix, which caused my code (in rev 1.52) to: - panic("soaccept: !NOFDREF") - fatal trap 12, with tracebacks going thru soclose and soaccept
1999-02-02 07:23:28 +00:00
so->so_qlen--;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
sp->so_qstate &= ~SQ_COMP;
Fix the port of the NetBSD 19990120-accept fix. I misread a piece of code when examining their fix, which caused my code (in rev 1.52) to: - panic("soaccept: !NOFDREF") - fatal trap 12, with tracebacks going thru soclose and soaccept
1999-02-02 07:23:28 +00:00
sp->so_head = NULL;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
ACCEPT_UNLOCK();
Change soabort() from returning int to returning void, since all consumers ignore the return value, soabort() is required to succeed, and protocols produce errors here to report multiple freeing of the pcb, which we hope to eliminate.
2006-03-16 07:03:14 +00:00
soabort(sp);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
ACCEPT_LOCK();
Fix for PR #1146: the "next" pointer must be cached before calling soabort since the struct containing it may be freed.
1996-04-16 03:50:08 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
ACCEPT_UNLOCK();
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
ACCEPT_LOCK();
SOCK_LOCK(so);
KASSERT((so->so_state & SS_NOFDREF) == 0, ("soclose: NOFDREF"));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
so->so_state |= SS_NOFDREF;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
sorele(so);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
/*
Change comment on soabort() to more accurately describe how/when soabort() is used. Remove trailing white space.
2006-07-16 23:09:39 +00:00
* soabort() is used to abruptly tear down a connection, such as when a
* resource limit is reached (listen queue depth exceeded), or if a listen
* socket is closed while there are sockets waiting to be accepted.
Change protocol switch pru_abort() API so that it returns void rather than an int, as an error here is not meaningful. Modify soabort() to unconditionally free the socket on the return of pru_abort(), and modify most protocols to no longer conditionally free the socket, since the caller will do this. This commit likely leaves parts of netinet and netinet6 in a situation where they may panic or leak memory, as they have not are not fully updated by this commit. This will be corrected shortly in followup commits to these components. MFC after: 3 months
2006-04-01 15:15:05 +00:00
*
* This interface is tricky, because it is called on an unreferenced socket,
* and must be called only by a thread that has actually removed the socket
* from the listen queue it was on, or races with other threads are risked.
*
* This interface will call into the protocol code, so must not be called
* with any socket locks held. Protocols do call it while holding their own
* recursible protocol mutexes, but this is something that should be subject
* to review in the future.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Change soabort() from returning int to returning void, since all consumers ignore the return value, soabort() is required to succeed, and protocols produce errors here to report multiple freeing of the pcb, which we hope to eliminate.
2006-03-16 07:03:14 +00:00
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soabort(struct socket *so)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Change protocol switch pru_abort() API so that it returns void rather than an int, as an error here is not meaningful. Modify soabort() to unconditionally free the socket on the return of pru_abort(), and modify most protocols to no longer conditionally free the socket, since the caller will do this. This commit likely leaves parts of netinet and netinet6 in a situation where they may panic or leak memory, as they have not are not fully updated by this commit. This will be corrected shortly in followup commits to these components. MFC after: 3 months
2006-04-01 15:15:05 +00:00
/*
* In as much as is possible, assert that no references to this
* socket are held. This is not quite the same as asserting that the
* current thread is responsible for arranging for no references, but
* is as close as we can get for now.
*/
KASSERT(so->so_count == 0, ("soabort: so_count"));
Assert that sockets passed into soabort() not be SQ_COMP or SQ_INCOMP, since that removal should have been done a layer up. MFC after: 3 months
2006-04-23 18:15:54 +00:00
KASSERT((so->so_state & SS_PROTOREF) == 0, ("soabort: SS_PROTOREF"));
Change protocol switch pru_abort() API so that it returns void rather than an int, as an error here is not meaningful. Modify soabort() to unconditionally free the socket on the return of pru_abort(), and modify most protocols to no longer conditionally free the socket, since the caller will do this. This commit likely leaves parts of netinet and netinet6 in a situation where they may panic or leak memory, as they have not are not fully updated by this commit. This will be corrected shortly in followup commits to these components. MFC after: 3 months
2006-04-01 15:15:05 +00:00
KASSERT(so->so_state & SS_NOFDREF, ("soabort: !SS_NOFDREF"));
Assert that sockets passed into soabort() not be SQ_COMP or SQ_INCOMP, since that removal should have been done a layer up. MFC after: 3 months
2006-04-23 18:15:54 +00:00
KASSERT((so->so_state & SQ_COMP) == 0, ("soabort: SQ_COMP"));
KASSERT((so->so_state & SQ_INCOMP) == 0, ("soabort: SQ_INCOMP"));
Change protocol switch pru_abort() API so that it returns void rather than an int, as an error here is not meaningful. Modify soabort() to unconditionally free the socket on the return of pru_abort(), and modify most protocols to no longer conditionally free the socket, since the caller will do this. This commit likely leaves parts of netinet and netinet6 in a situation where they may panic or leak memory, as they have not are not fully updated by this commit. This will be corrected shortly in followup commits to these components. MFC after: 3 months
2006-04-01 15:15:05 +00:00
Change semantics of socket close and detach. Add a new protocol switch function, pru_close, to notify protocols that the file descriptor or other consumer of a socket is closing the socket. pru_abort is now a notification of close also, and no longer detaches. pru_detach is no longer used to notify of close, and will be called during socket tear-down by sofree() when all references to a socket evaporate after an earlier call to abort or close the socket. This means detach is now an unconditional teardown of a socket, whereas previously sockets could persist after detach of the protocol retained a reference. This faciliates sharing mutexes between layers of the network stack as the mutex is required during the checking and removal of references at the head of sofree(). With this change, pru_detach can now assume that the mutex will no longer be required by the socket layer after completion, whereas before this was not necessarily true. Reviewed by: gnn
2006-07-21 17:11:15 +00:00
if (so->so_proto->pr_usrreqs->pru_abort != NULL)
Several protocol switch functions (pru_abort, pru_detach, pru_sosetlabel) return void, so don't implement no-op versions of these functions. Instead, consistently check if those switch pointers are NULL before invoking them.
2006-07-11 23:18:28 +00:00
(*so->so_proto->pr_usrreqs->pru_abort)(so);
Change protocol switch pru_abort() API so that it returns void rather than an int, as an error here is not meaningful. Modify soabort() to unconditionally free the socket on the return of pru_abort(), and modify most protocols to no longer conditionally free the socket, since the caller will do this. This commit likely leaves parts of netinet and netinet6 in a situation where they may panic or leak memory, as they have not are not fully updated by this commit. This will be corrected shortly in followup commits to these components. MFC after: 3 months
2006-04-01 15:15:05 +00:00
ACCEPT_LOCK();
SOCK_LOCK(so);
sofree(so);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soaccept(struct socket *so, struct sockaddr **nam)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
int error;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_LOCK(so);
KASSERT((so->so_state & SS_NOFDREF) != 0, ("soaccept: !NOFDREF"));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
so->so_state &= ~SS_NOFDREF;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_UNLOCK(so);
Push the test for a disconnected socket when accept()ing down to the protocol layer. Not all protocols behave identically. This fixes the brokenness observed with unix-domain sockets (and postfix)
2001-03-09 08:16:40 +00:00
error = (*so->so_proto->pr_usrreqs->pru_accept)(so, nam);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soconnect(struct socket *so, struct sockaddr *nam, struct thread *td)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
int error;
Back out my lats commit of locking down a socket, it conflicts with hsu's work. Requested by: hsu
2002-05-31 11:52:35 +00:00
if (so->so_options & SO_ACCEPTCONN)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (EOPNOTSUPP);
/*
* If protocol is connection-based, can only connect once.
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Otherwise, if connected, try to disconnect first. This allows
* user to disconnect by connecting to, e.g., a null address.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Back out my lats commit of locking down a socket, it conflicts with hsu's work. Requested by: hsu
2002-05-31 11:52:35 +00:00
if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
(error = sodisconnect(so)))) {
Back out my lats commit of locking down a socket, it conflicts with hsu's work. Requested by: hsu
2002-05-31 11:52:35 +00:00
error = EISCONN;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
} else {
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Prevent accumulated error from previous connection from
* biting us.
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
*/
so->so_error = 0;
error = (*so->so_proto->pr_usrreqs->pru_connect)(so, nam, td);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soconnect2(struct socket *so1, struct socket *so2)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
return ((*so1->so_proto->pr_usrreqs->pru_connect2)(so1, so2));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sodisconnect(struct socket *so)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
int error;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if ((so->so_state & SS_ISCONNECTED) == 0)
return (ENOTCONN);
if (so->so_state & SS_ISDISCONNECTING)
return (EALREADY);
Modify the kernel to use the new pr_usrreqs interface rather than the old pr_usrreq mechanism which was poorly designed and error-prone. This commit renames pr_usrreq to pr_ousrreq so that old code which depended on it would break in an obvious manner. This commit also implements the new interface for TCP, although the old function is left as an example (#ifdef'ed out). This commit ALSO fixes a longstanding bug in the TCP timer processing (introduced by davidg on 1995/04/12) which caused timer processing on a TCB to always stop after a single timer had expired (because it misinterpreted the return value from tcp_usrreq() to indicate that the TCB had been deleted). Finally, some code related to polling has been deleted from if.c because it is not relevant t -current and doesn't look at all like my current code.
1996-07-11 16:32:50 +00:00
error = (*so->so_proto->pr_usrreqs->pru_disconnect)(so);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
Move zero copy statistics structure before sosend_copyin(). MFC after: 1 month Reported by: tinderbox, sam
2005-11-28 21:45:36 +00:00
#ifdef ZERO_COPY_SOCKETS
struct so_zerocopy_stats{
int size_ok;
int align_ok;
int found_ifp;
};
struct so_zerocopy_stats so_zerocp_stats = {0,0,0};
#include <netinet/in.h>
#include <net/route.h>
#include <netinet/in_pcb.h>
#include <vm/vm.h>
#include <vm/vm_page.h>
#include <vm/vm_object.h>
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
/*
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
* sosend_copyin() is only used if zero copy sockets are enabled. Otherwise
* sosend_dgram() and sosend_generic() use m_uiotombuf().
*
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
* sosend_copyin() accepts a uio and prepares an mbuf chain holding part or
* all of the data referenced by the uio. If desired, it uses zero-copy.
* *space will be updated to reflect data copied in.
*
* NB: If atomic I/O is requested, the caller must already have checked that
* space can hold resid bytes.
*
* NB: In the event of an error, the caller may need to free the partial
* chain pointed to by *mpp. The contents of both *uio and *space may be
* modified even in the case of an error.
*/
static int
sosend_copyin(struct uio *uio, struct mbuf **retmp, int atomic, long *space,
int flags)
{
struct mbuf *m, **mp, *top;
long len, resid;
int error;
#ifdef ZERO_COPY_SOCKETS
int cow_send;
#endif
*retmp = top = NULL;
mp = &top;
len = 0;
resid = uio->uio_resid;
error = 0;
do {
#ifdef ZERO_COPY_SOCKETS
cow_send = 0;
#endif /* ZERO_COPY_SOCKETS */
if (resid >= MINCLSIZE) {
#ifdef ZERO_COPY_SOCKETS
if (top == NULL) {
- Use m_gethdr(), m_get(), and m_clget() instead of the macros in sosend_copyin(). - Use M_WAITOK instead of M_TRYWAIT in sosend_copyin(). - Don't check for NULL from M_WAITOK and return ENOBUFS. M_WAITOK/M_TRYWAIT allocations don't fail with NULL. Reviewed by: andre Requested by: andre (2)
2007-03-12 19:27:36 +00:00
m = m_gethdr(M_WAITOK, MT_DATA);
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
m->m_pkthdr.len = 0;
Change comment on soabort() to more accurately describe how/when soabort() is used. Remove trailing white space.
2006-07-16 23:09:39 +00:00
m->m_pkthdr.rcvif = NULL;
- Use m_gethdr(), m_get(), and m_clget() instead of the macros in sosend_copyin(). - Use M_WAITOK instead of M_TRYWAIT in sosend_copyin(). - Don't check for NULL from M_WAITOK and return ENOBUFS. M_WAITOK/M_TRYWAIT allocations don't fail with NULL. Reviewed by: andre Requested by: andre (2)
2007-03-12 19:27:36 +00:00
} else
m = m_get(M_WAITOK, MT_DATA);
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
if (so_zero_copy_send &&
resid>=PAGE_SIZE &&
*space>=PAGE_SIZE &&
uio->uio_iov->iov_len>=PAGE_SIZE) {
so_zerocp_stats.size_ok++;
so_zerocp_stats.align_ok++;
cow_send = socow_setup(m, uio);
len = cow_send;
}
if (!cow_send) {
- Use m_gethdr(), m_get(), and m_clget() instead of the macros in sosend_copyin(). - Use M_WAITOK instead of M_TRYWAIT in sosend_copyin(). - Don't check for NULL from M_WAITOK and return ENOBUFS. M_WAITOK/M_TRYWAIT allocations don't fail with NULL. Reviewed by: andre Requested by: andre (2)
2007-03-12 19:27:36 +00:00
m_clget(m, M_WAITOK);
len = min(min(MCLBYTES, resid), *space);
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
}
#else /* ZERO_COPY_SOCKETS */
if (top == NULL) {
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
m = m_getcl(M_WAIT, MT_DATA, M_PKTHDR);
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
m->m_pkthdr.len = 0;
m->m_pkthdr.rcvif = NULL;
} else
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
m = m_getcl(M_WAIT, MT_DATA, 0);
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
len = min(min(MCLBYTES, resid), *space);
#endif /* ZERO_COPY_SOCKETS */
} else {
if (top == NULL) {
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
m = m_gethdr(M_WAIT, MT_DATA);
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
m->m_pkthdr.len = 0;
m->m_pkthdr.rcvif = NULL;
len = min(min(MHLEN, resid), *space);
/*
* For datagram protocols, leave room
* for protocol headers in first mbuf.
*/
if (atomic && m && len < MHLEN)
MH_ALIGN(m, len);
} else {
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
m = m_get(M_WAIT, MT_DATA);
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
len = min(min(MLEN, resid), *space);
}
}
if (m == NULL) {
error = ENOBUFS;
goto out;
}
*space -= len;
#ifdef ZERO_COPY_SOCKETS
if (cow_send)
error = 0;
else
#endif /* ZERO_COPY_SOCKETS */
error = uiomove(mtod(m, void *), (int)len, uio);
resid = uio->uio_resid;
m->m_len = len;
*mp = m;
top->m_pkthdr.len += len;
if (error)
goto out;
mp = &m->m_next;
if (resid <= 0) {
if (flags & MSG_EOR)
top->m_flags |= M_EOR;
break;
}
} while (*space > 0 && atomic);
out:
*retmp = top;
return (error);
}
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
#endif /*ZERO_COPY_SOCKETS*/
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
Correct two problems relating to sorflush(), which is called to flush read socket buffers in shutdown() and close(): - Call socantrcvmore() before sblock() to dislodge any threads that might be sleeping (potentially indefinitely) while holding sblock(), such as a thread blocked in recv(). - Flag the sblock() call as non-interruptible so that a signal delivered to the thread calling sorflush() doesn't cause sblock() to fail. The sblock() is required to ensure that all other socket consumer threads have, in fact, left, and do not enter, the socket buffer until we're done flushin it. To implement the latter, change the 'flags' argument to sblock() to accept two flags, SBL_WAIT and SBL_NOINTR, rather than one M_WAITOK flag. When SBL_NOINTR is set, it forces a non-interruptible sx acquisition, regardless of the setting of the disposition of SB_NOINTR on the socket buffer; without this change it would be possible for another thread to clear SB_NOINTR between when the socket buffer mutex is released and sblock() is invoked. Reviewed by: bz, kmacy Reported by: Jos Backus <jos at catnook dot com>
2008-01-31 08:22:24 +00:00
#define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? 0 : SBL_WAIT)
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sosend_dgram(struct socket *so, struct sockaddr *addr, struct uio *uio,
struct mbuf *top, struct mbuf *control, int flags, struct thread *td)
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
{
long space, resid;
int clen = 0, error, dontroute;
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
#ifdef ZERO_COPY_SOCKETS
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
int atomic = sosendallatonce(so) || top;
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
#endif
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
KASSERT(so->so_type == SOCK_DGRAM, ("sodgram_send: !SOCK_DGRAM"));
KASSERT(so->so_proto->pr_flags & PR_ATOMIC,
("sodgram_send: !PR_ATOMIC"));
if (uio != NULL)
resid = uio->uio_resid;
else
resid = top->m_pkthdr.len;
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* In theory resid should be unsigned. However, space must be
* signed, as it might be less than 0 if we over-committed, and we
* must use a signed comparison of space and resid. On the other
* hand, a negative resid causes us to loop sending 0-length
* segments to the protocol.
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
*
* Also check to make sure that MSG_EOR isn't used on SOCK_STREAM
* type sockets since that's an error.
*/
if (resid < 0) {
error = EINVAL;
goto out;
}
dontroute =
(flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0;
if (td != NULL)
- Move rusage from being per-process in struct pstats to per-thread in td_ru. This removes the requirement for per-process synchronization in statclock() and mi_switch(). This was previously supported by sched_lock which is going away. All modifications to rusage are now done in the context of the owning thread. reads proceed without locks. - Aggregate exiting threads rusage in thread_exit() such that the exiting thread's rusage is not lost. - Provide a new routine, rufetch() to fetch an aggregate of all rusage structures from all threads in a process. This routine must be used in any place requiring a rusage from a process prior to it's exit. The exited process's rusage is still available via p_ru. - Aggregate tick statistics only on demand via rufetch() or when a thread exits. Tick statistics are kept in the thread and protected by sched_lock until it exits. Initial patch by: attilio Reviewed by: attilio, bde (some objections), arch (mostly silent)
2007-06-01 01:12:45 +00:00
td->td_ru.ru_msgsnd++;
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
if (control != NULL)
clen = control->m_len;
SOCKBUF_LOCK(&so->so_snd);
if (so->so_snd.sb_state & SBS_CANTSENDMORE) {
SOCKBUF_UNLOCK(&so->so_snd);
error = EPIPE;
goto out;
}
if (so->so_error) {
error = so->so_error;
so->so_error = 0;
SOCKBUF_UNLOCK(&so->so_snd);
goto out;
}
if ((so->so_state & SS_ISCONNECTED) == 0) {
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* `sendto' and `sendmsg' is allowed on a connection-based
* socket if it supports implied connect. Return ENOTCONN if
* not connected and no address is supplied.
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
*/
if ((so->so_proto->pr_flags & PR_CONNREQUIRED) &&
(so->so_proto->pr_flags & PR_IMPLOPCL) == 0) {
if ((so->so_state & SS_ISCONFIRMING) == 0 &&
!(resid == 0 && clen != 0)) {
SOCKBUF_UNLOCK(&so->so_snd);
error = ENOTCONN;
goto out;
}
} else if (addr == NULL) {
if (so->so_proto->pr_flags & PR_CONNREQUIRED)
error = ENOTCONN;
else
error = EDESTADDRREQ;
SOCKBUF_UNLOCK(&so->so_snd);
goto out;
}
}
/*
* Do we need MSG_OOB support in SOCK_DGRAM? Signs here may be a
* problem and need fixing.
*/
space = sbspace(&so->so_snd);
if (flags & MSG_OOB)
space += 1024;
space -= clen;
Fix a lock leak in an error case. Reported by: netchild Reviewed by: rwatson
2006-09-13 06:58:40 +00:00
SOCKBUF_UNLOCK(&so->so_snd);
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
if (resid > space) {
error = EMSGSIZE;
goto out;
}
if (uio == NULL) {
resid = 0;
if (flags & MSG_EOR)
top->m_flags |= M_EOR;
} else {
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
#ifdef ZERO_COPY_SOCKETS
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
error = sosend_copyin(uio, &top, atomic, &space, flags);
if (error)
goto out;
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
#else
Unbreak writes of 0 bytes. Zero byte writes happen when only ancillary control data but no payload data is passed. Change m_uiotombuf() to return at least one empty mbuf if the requested length was zero. Add comment to sosend_dgram and sosend_generic(). Diagnoses by: jhb Regression test by: rwatson Pointy hat to. andre
2007-01-22 14:50:28 +00:00
/*
* Copy the data from userland into a mbuf chain.
* If no data is to be copied in, a single empty mbuf
* is returned.
*/
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
top = m_uiotombuf(uio, M_WAITOK, space, max_hdr,
(M_PKTHDR | ((flags & MSG_EOR) ? M_EOR : 0)));
if (top == NULL) {
error = EFAULT; /* only possible error */
goto out;
}
space -= resid - uio->uio_resid;
#endif
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
resid = uio->uio_resid;
}
KASSERT(resid == 0, ("sosend_dgram: resid != 0"));
/*
* XXXRW: Frobbing SO_DONTROUTE here is even worse without sblock
* than with.
*/
if (dontroute) {
SOCK_LOCK(so);
so->so_options |= SO_DONTROUTE;
SOCK_UNLOCK(so);
}
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* XXX all the SBS_CANTSENDMORE checks previously done could be out
* of date. We could have recieved a reset packet in an interrupt or
* maybe we slept while doing page faults in uiomove() etc. We could
* probably recheck again inside the locking protection here, but
* there are probably other places that this also happens. We must
* rethink this.
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
*/
error = (*so->so_proto->pr_usrreqs->pru_send)(so,
(flags & MSG_OOB) ? PRUS_OOB :
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* If the user set MSG_EOF, the protocol understands this flag and
* nothing left to send then use PRU_SEND_EOF instead of PRU_SEND.
Add sosend_dgram(), a greatly reduced and simplified version of sosend() intended for use solely with atomic datagram socket types, and relies on the previous break-out of sosend_copyin(). Changes to allow UDP to optionally use this instead of sosend() will be committed as a follow-up.
2006-01-13 10:22:01 +00:00
*/
((flags & MSG_EOF) &&
(so->so_proto->pr_flags & PR_IMPLOPCL) &&
(resid <= 0)) ?
PRUS_EOF :
/* If there is more to send set PRUS_MORETOCOME */
(resid > 0 && space > 0) ? PRUS_MORETOCOME : 0,
top, addr, control, td);
if (dontroute) {
SOCK_LOCK(so);
so->so_options &= ~SO_DONTROUTE;
SOCK_UNLOCK(so);
}
clen = 0;
control = NULL;
top = NULL;
out:
if (top != NULL)
m_freem(top);
if (control != NULL)
m_freem(control);
return (error);
}
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Send on a socket. If send must go all at once and message is larger than
* send buffering, then hard error. Lock against other senders. If must go
* all at once and not enough room now, then inform user that this would
* block and do nothing. Otherwise, if nonblocking, send as much as
* possible. The data to be sent is described by "uio" if nonzero, otherwise
* by the mbuf chain "top" (which must be null if uio is not). Data provided
* in mbuf chain must be small enough to send all at once.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Returns nonzero on error, timeout or signal; callers must check for short
* counts if EINTR/ERESTART are returned. Data and control buffers are freed
* on return.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sosend_generic(struct socket *so, struct sockaddr *addr, struct uio *uio,
struct mbuf *top, struct mbuf *control, int flags, struct thread *td)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
long space, resid;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
int clen = 0, error, dontroute;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
int atomic = sosendallatonce(so) || top;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (uio != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
resid = uio->uio_resid;
else
resid = top->m_pkthdr.len;
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* In theory resid should be unsigned. However, space must be
* signed, as it might be less than 0 if we over-committed, and we
* must use a signed comparison of space and resid. On the other
* hand, a negative resid causes us to loop sending 0-length
* segments to the protocol.
MF22: MSG_EOR bug fix. Submitted by: wollman
1997-11-09 05:07:40 +00:00
*
* Also check to make sure that MSG_EOR isn't used on SOCK_STREAM
* type sockets since that's an error.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Fix warnings in preparation for adding -Wall -Wcast-qual to the kernel compile
1999-01-27 21:50:00 +00:00
if (resid < 0 || (so->so_type == SOCK_STREAM && (flags & MSG_EOR))) {
MF22: MSG_EOR bug fix. Submitted by: wollman
1997-11-09 05:07:40 +00:00
error = EINVAL;
goto out;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
dontroute =
(flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
(so->so_proto->pr_flags & PR_ATOMIC);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (td != NULL)
- Move rusage from being per-process in struct pstats to per-thread in td_ru. This removes the requirement for per-process synchronization in statclock() and mi_switch(). This was previously supported by sched_lock which is going away. All modifications to rusage are now done in the context of the owning thread. reads proceed without locks. - Aggregate exiting threads rusage in thread_exit() such that the exiting thread's rusage is not lost. - Provide a new routine, rufetch() to fetch an aggregate of all rusage structures from all threads in a process. This routine must be used in any place requiring a rusage from a process prior to it's exit. The exited process's rusage is still available via p_ru. - Aggregate tick statistics only on demand via rufetch() or when a thread exits. Tick statistics are kept in the thread and protected by sched_lock until it exits. Initial patch by: attilio Reviewed by: attilio, bde (some objections), arch (mostly silent)
2007-06-01 01:12:45 +00:00
td->td_ru.ru_msgsnd++;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (control != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
clen = control->m_len;
All of this is cosmetic. prototypes, #includes, printfs and so on. Makes GCC a lot more silent.
1994-10-02 17:35:40 +00:00
error = sblock(&so->so_snd, SBLOCKWAIT(flags));
if (error)
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
goto out;
restart:
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
do {
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_LOCK(&so->so_snd);
if (so->so_snd.sb_state & SBS_CANTSENDMORE) {
SOCKBUF_UNLOCK(&so->so_snd);
error = EPIPE;
goto release;
}
Revert sosend() to its behavior from 4.3-Tahoe and before: if so_error is set, clear it before returning it. The behavior introduced in 4.3-Reno (to not clear so_error) causes potentially transient errors (e.g. ECONNREFUSED if the other end hasn't opened its socket yet) to be permanent on connected datagram sockets that are only used for writing. (soreceive() clears so_error before returning it, as does getsockopt(...,SO_ERROR,...).) Submitted by: Van Jacobson <van@ee.lbl.gov>, via a comment in the vat sources.
1998-02-19 19:38:20 +00:00
if (so->so_error) {
error = so->so_error;
so->so_error = 0;
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_UNLOCK(&so->so_snd);
Revert sosend() to its behavior from 4.3-Tahoe and before: if so_error is set, clear it before returning it. The behavior introduced in 4.3-Reno (to not clear so_error) causes potentially transient errors (e.g. ECONNREFUSED if the other end hasn't opened its socket yet) to be permanent on connected datagram sockets that are only used for writing. (soreceive() clears so_error before returning it, as does getsockopt(...,SO_ERROR,...).) Submitted by: Van Jacobson <van@ee.lbl.gov>, via a comment in the vat sources.
1998-02-19 19:38:20 +00:00
goto release;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if ((so->so_state & SS_ISCONNECTED) == 0) {
Merge in the socket-level support for Transaction TCP.
1995-02-07 02:01:16 +00:00
/*
* `sendto' and `sendmsg' is allowed on a connection-
* based socket if it supports implied connect.
* Return ENOTCONN if not connected and no address is
* supplied.
*/
if ((so->so_proto->pr_flags & PR_CONNREQUIRED) &&
(so->so_proto->pr_flags & PR_IMPLOPCL) == 0) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if ((so->so_state & SS_ISCONFIRMING) == 0 &&
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
!(resid == 0 && clen != 0)) {
SOCKBUF_UNLOCK(&so->so_snd);
error = ENOTCONN;
goto release;
}
} else if (addr == NULL) {
SOCKBUF_UNLOCK(&so->so_snd);
if (so->so_proto->pr_flags & PR_CONNREQUIRED)
error = ENOTCONN;
else
error = EDESTADDRREQ;
goto release;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
space = sbspace(&so->so_snd);
if (flags & MSG_OOB)
space += 1024;
All of this is cosmetic. prototypes, #includes, printfs and so on. Makes GCC a lot more silent.
1994-10-02 17:35:40 +00:00
if ((atomic && resid > so->so_snd.sb_hiwat) ||
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
clen > so->so_snd.sb_hiwat) {
SOCKBUF_UNLOCK(&so->so_snd);
error = EMSGSIZE;
goto release;
}
In sosend(), enforce the socket buffer limits regardless of whether the data was supplied as a uio or an mbuf. Previously the limit was ignored for mbuf data, and NFS could run the kernel out of mbufs when an ipfw rule blocked retransmissions.
2002-02-28 11:22:40 +00:00
if (space < resid + clen &&
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
(atomic || space < so->so_snd.sb_lowat || space < clen)) {
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
if ((so->so_state & SS_NBIO) || (flags & MSG_NBIO)) {
SOCKBUF_UNLOCK(&so->so_snd);
error = EWOULDBLOCK;
goto release;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = sbwait(&so->so_snd);
Add missing socket buffer unlock before returning to userland. Reviewed by: rwatson
2007-05-08 12:34:14 +00:00
SOCKBUF_UNLOCK(&so->so_snd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (error)
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
goto release;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto restart;
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_snd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
space -= clen;
do {
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
if (uio == NULL) {
resid = 0;
if (flags & MSG_EOR)
top->m_flags |= M_EOR;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
} else {
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
#ifdef ZERO_COPY_SOCKETS
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
error = sosend_copyin(uio, &top, atomic,
&space, flags);
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
if (error != 0)
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
goto release;
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
#else
Unbreak writes of 0 bytes. Zero byte writes happen when only ancillary control data but no payload data is passed. Change m_uiotombuf() to return at least one empty mbuf if the requested length was zero. Add comment to sosend_dgram and sosend_generic(). Diagnoses by: jhb Regression test by: rwatson Pointy hat to. andre
2007-01-22 14:50:28 +00:00
/*
* Copy the data from userland into a mbuf
* chain. If no data is to be copied in,
* a single empty mbuf is returned.
*/
Use the improved m_uiotombuf() function instead of home grown sosend_copyin() to do the userland to kernel copying in sosend_generic() and sosend_dgram(). sosend_copyin() is retained for ZERO_COPY_SOCKETS which are not yet supported by m_uiotombuf(). Benchmaring shows significant improvements (95% confidence): 66% less cpu (or 2.9 times better) with new sosend vs. old sosend (non-TSO) 65% less cpu (or 2.8 times better) with new sosend vs. old sosend (TSO) (Sender AMD Opteron 852 (2.6GHz) with em(4) PCI-X-133 interface and receiver DELL Poweredge SC1425 P-IV Xeon 3.2GHz with em(4) LOM connected back to back at 1000Base-TX full duplex.) Sponsored by: TCP/IP Optimization Fundraise 2005 MFC after: 3 month
2006-11-02 17:45:28 +00:00
top = m_uiotombuf(uio, M_WAITOK, space,
(atomic ? max_hdr : 0),
(atomic ? M_PKTHDR : 0) |
((flags & MSG_EOR) ? M_EOR : 0));
if (top == NULL) {
error = EFAULT; /* only possible error */
goto release;
}
space -= resid - uio->uio_resid;
#endif
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
resid = uio->uio_resid;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
if (dontroute) {
SOCK_LOCK(so);
so->so_options |= SO_DONTROUTE;
SOCK_UNLOCK(so);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
/*
* XXX all the SBS_CANTSENDMORE checks previously
* done could be out of date. We could have recieved
* a reset packet in an interrupt or maybe we slept
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* while doing page faults in uiomove() etc. We
* could probably recheck again inside the locking
* protection here, but there are probably other
* places that this also happens. We must rethink
* this.
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
*/
error = (*so->so_proto->pr_usrreqs->pru_send)(so,
(flags & MSG_OOB) ? PRUS_OOB :
Merge in the socket-level support for Transaction TCP.
1995-02-07 02:01:16 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* If the user set MSG_EOF, the protocol understands
* this flag and nothing left to send then use
* PRU_SEND_EOF instead of PRU_SEND.
Merge in the socket-level support for Transaction TCP.
1995-02-07 02:01:16 +00:00
*/
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
((flags & MSG_EOF) &&
(so->so_proto->pr_flags & PR_IMPLOPCL) &&
(resid <= 0)) ?
Add a flag, passed to pru_send routines, PRUS_MORETOCOME. This flag means that there is more data to be put into the socket buffer. Use it in TCP to reduce the interaction between mbuf sizes and the Nagle algorithm. Based on: "Justin C. Walker" <justin@apple.com>'s description of Apple's fix for this problem.
1999-01-20 17:32:01 +00:00
PRUS_EOF :
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
/* If there is more to send set PRUS_MORETOCOME. */
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
(resid > 0 && space > 0) ? PRUS_MORETOCOME : 0,
top, addr, control, td);
if (dontroute) {
SOCK_LOCK(so);
so->so_options &= ~SO_DONTROUTE;
SOCK_UNLOCK(so);
}
clen = 0;
control = NULL;
top = NULL;
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
if (error)
Break out functionality in sosend() responsible for building mbuf chains and copying in mbufs from the body of the send logic, creating a new function sosend_copyin(). This changes makes sosend() almost readable, and will allow the same logic to be used by tailored socket send routines. MFC after: 1 month Reviewed by: andre, glebius
2005-11-28 18:09:03 +00:00
goto release;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
} while (resid && space > 0);
} while (resid);
release:
sbunlock(&so->so_snd);
out:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (top != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
m_freem(top);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (control != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
m_freem(control);
return (error);
}
soreceive_generic(), and sopoll_generic(). Add new functions sosend(), soreceive(), and sopoll(), which are wrappers for pru_sosend, pru_soreceive, and pru_sopoll, and are now used univerally by socket consumers rather than either directly invoking the old so*() functions or directly invoking the protocol switch method (about an even split prior to this commit). This completes an architectural change that was begun in 1996 to permit protocols to provide substitute implementations, as now used by UDP. Consumers now uniformly invoke sosend(), soreceive(), and sopoll() to perform these operations on sockets -- in particular, distributed file systems and socket system calls. Architectural head nod: sam, gnn, wollman
2006-07-24 15:20:08 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
struct mbuf *top, struct mbuf *control, int flags, struct thread *td)
soreceive_generic(), and sopoll_generic(). Add new functions sosend(), soreceive(), and sopoll(), which are wrappers for pru_sosend, pru_soreceive, and pru_sopoll, and are now used univerally by socket consumers rather than either directly invoking the old so*() functions or directly invoking the protocol switch method (about an even split prior to this commit). This completes an architectural change that was begun in 1996 to permit protocols to provide substitute implementations, as now used by UDP. Consumers now uniformly invoke sosend(), soreceive(), and sopoll() to perform these operations on sockets -- in particular, distributed file systems and socket system calls. Architectural head nod: sam, gnn, wollman
2006-07-24 15:20:08 +00:00
{
/* XXXRW: Temporary debugging. */
KASSERT(so->so_proto->pr_usrreqs->pru_sosend != sosend,
("sosend: protocol calls sosend"));
return (so->so_proto->pr_usrreqs->pru_sosend(so, addr, uio, top,
control, flags, td));
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
* The part of soreceive() that implements reading non-inline out-of-band
* data from a socket. For more complete comments, see soreceive(), from
* which this code originated.
*
* Note that soreceive_rcvoob(), unlike the remainder of soreceive(), is
* unable to return an mbuf chain to the caller.
*/
static int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soreceive_rcvoob(struct socket *so, struct uio *uio, int flags)
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
{
struct protosw *pr = so->so_proto;
struct mbuf *m;
int error;
KASSERT(flags & MSG_OOB, ("soreceive_rcvoob: (flags & MSG_OOB) == 0"));
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
m = m_get(M_WAIT, MT_DATA);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
error = (*pr->pr_usrreqs->pru_rcvoob)(so, m, flags & MSG_PEEK);
if (error)
goto bad;
do {
#ifdef ZERO_COPY_SOCKETS
if (so_zero_copy_receive) {
int disposable;
if ((m->m_flags & M_EXT)
&& (m->m_ext.ext_type == EXT_DISPOSABLE))
disposable = 1;
else
disposable = 0;
error = uiomoveco(mtod(m, void *),
min(uio->uio_resid, m->m_len),
uio, disposable);
} else
#endif /* ZERO_COPY_SOCKETS */
error = uiomove(mtod(m, void *),
(int) min(uio->uio_resid, m->m_len), uio);
m = m_free(m);
} while (uio->uio_resid && error == 0 && m);
bad:
if (m != NULL)
m_freem(m);
return (error);
}
/*
* Following replacement or removal of the first mbuf on the first mbuf chain
* of a socket buffer, push necessary state changes back into the socket
* buffer so that other consumers see the values consistently. 'nextrecord'
* is the callers locally stored value of the original value of
* sb->sb_mb->m_nextpkt which must be restored when the lead mbuf changes.
* NOTE: 'nextrecord' may be NULL.
*/
static __inline void
sockbuf_pushsync(struct sockbuf *sb, struct mbuf *nextrecord)
{
SOCKBUF_LOCK_ASSERT(sb);
/*
* First, update for the new value of nextrecord. If necessary, make
* it the first record.
*/
if (sb->sb_mb != NULL)
sb->sb_mb->m_nextpkt = nextrecord;
else
sb->sb_mb = nextrecord;
/*
* Now update any dependent socket buffer fields to reflect the new
* state. This is an expanded inline of SB_EMPTY_FIXUP(), with the
* addition of a second clause that takes care of the case where
* sb_mb has been updated, but remains the last record.
*/
if (sb->sb_mb == NULL) {
sb->sb_mbtail = NULL;
sb->sb_lastrecord = NULL;
} else if (sb->sb_mb->m_nextpkt == NULL)
sb->sb_lastrecord = sb->sb_mb;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Implement receive operations on a socket. We depend on the way that
* records are added to the sockbuf by sbappend. In particular, each record
* (mbufs linked through m_next) must begin with an address if the protocol
* so specifies, followed by an optional mbuf or mbufs containing ancillary
* data, and then zero or more mbufs of data. In order to allow parallelism
* between network receive and copying to user space, as well as avoid
* sleeping with a mutex held, we release the socket buffer mutex during the
* user space copy. Although the sockbuf is locked, new data may still be
* appended, and thus we must maintain consistency of the sockbuf during that
* time.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* The caller may receive the data as a single mbuf chain by supplying an
* mbuf **mp0 for use in returning the chain. The uio is then used only for
* the count in uio_resid.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soreceive_generic(struct socket *so, struct sockaddr **psa, struct uio *uio,
struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct mbuf *m, **mp;
int flags, len, error, offset;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct protosw *pr = so->so_proto;
struct mbuf *nextrecord;
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int moff, type = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
int orig_resid = uio->uio_resid;
mp = mp0;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (psa != NULL)
*psa = NULL;
if (controlp != NULL)
*controlp = NULL;
if (flagsp != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
flags = *flagsp &~ MSG_EOR;
else
flags = 0;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (flags & MSG_OOB)
return (soreceive_rcvoob(so, uio, flags));
if (mp != NULL)
*mp = NULL;
if ((pr->pr_flags & PR_WANTRCVD) && (so->so_state & SS_ISCONFIRMING)
&& uio->uio_resid)
Modify the kernel to use the new pr_usrreqs interface rather than the old pr_usrreq mechanism which was poorly designed and error-prone. This commit renames pr_usrreq to pr_ousrreq so that old code which depended on it would break in an obvious manner. This commit also implements the new interface for TCP, although the old function is left as an example (#ifdef'ed out). This commit ALSO fixes a longstanding bug in the TCP timer processing (introduced by davidg on 1995/04/12) which caused timer processing on a TCB to always stop after a single timer had expired (because it misinterpreted the return value from tcp_usrreq() to indicate that the TCB had been deleted). Finally, some code related to polling has been deleted from if.c because it is not relevant t -current and doesn't look at all like my current code.
1996-07-11 16:32:50 +00:00
(*pr->pr_usrreqs->pru_rcvd)(so, 0);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
All of this is cosmetic. prototypes, #includes, printfs and so on. Makes GCC a lot more silent.
1994-10-02 17:35:40 +00:00
error = sblock(&so->so_rcv, SBLOCKWAIT(flags));
if (error)
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
restart:
SOCKBUF_LOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
m = so->so_rcv.sb_mb;
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* If we have less data than requested, block awaiting more (subject
* to any timeout) if:
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* 1. the current count is less than the low water mark, or
* 2. MSG_WAITALL is set, and it is possible to do the entire
* receive operation at once if we block (resid <= hiwat).
* 3. MSG_DONTWAIT is not set
* If MSG_WAITALL is set but resid is larger than the receive buffer,
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* we have to do the receive in sections, and thus risk returning a
* short count if a timeout or signal occurs after we start.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (m == NULL || (((flags & MSG_DONTWAIT) == 0 &&
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
so->so_rcv.sb_cc < uio->uio_resid) &&
(so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
m->m_nextpkt == NULL && (pr->pr_flags & PR_ATOMIC) == 0)) {
KASSERT(m != NULL || !so->so_rcv.sb_cc,
("receive: m == %p so->so_rcv.sb_cc == %u",
m, so->so_rcv.sb_cc));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (so->so_error) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (m != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto dontblock;
error = so->so_error;
if ((flags & MSG_PEEK) == 0)
so->so_error = 0;
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto release;
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
if (so->so_rcv.sb_state & SBS_CANTRCVMORE) {
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
if (m == NULL) {
SOCKBUF_UNLOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto release;
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
} else
goto dontblock;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
for (; m != NULL; m = m->m_next)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) {
m = so->so_rcv.sb_mb;
goto dontblock;
}
if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
(so->so_proto->pr_flags & PR_CONNREQUIRED)) {
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = ENOTCONN;
goto release;
}
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
if (uio->uio_resid == 0) {
SOCKBUF_UNLOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto release;
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if ((so->so_state & SS_NBIO) ||
(flags & (MSG_DONTWAIT|MSG_NBIO))) {
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = EWOULDBLOCK;
goto release;
}
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = sbwait(&so->so_rcv);
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (error)
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
goto release;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto restart;
}
dontblock:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
* From this point onward, we maintain 'nextrecord' as a cache of the
* pointer to the next record in the socket buffer. We must keep the
* various socket buffer pointers and local stack versions of the
* pointers in sync, pushing out modifications before dropping the
* socket buffer mutex, and re-reading them when picking it up.
*
* Otherwise, we will race with the network stack appending new data
* or records onto the socket buffer by using inconsistent/stale
* versions of the field, possibly resulting in socket buffer
* corruption.
*
* By holding the high-level sblock(), we prevent simultaneous
* readers from pulling off the front of the socket buffer.
*/
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
if (uio->uio_td)
- Move rusage from being per-process in struct pstats to per-thread in td_ru. This removes the requirement for per-process synchronization in statclock() and mi_switch(). This was previously supported by sched_lock which is going away. All modifications to rusage are now done in the context of the owning thread. reads proceed without locks. - Aggregate exiting threads rusage in thread_exit() such that the exiting thread's rusage is not lost. - Provide a new routine, rufetch() to fetch an aggregate of all rusage structures from all threads in a process. This routine must be used in any place requiring a rusage from a process prior to it's exit. The exited process's rusage is still available via p_ru. - Aggregate tick statistics only on demand via rufetch() or when a thread exits. Tick statistics are kept in the thread and protected by sched_lock until it exits. Initial patch by: attilio Reviewed by: attilio, bde (some objections), arch (mostly silent)
2007-06-01 01:12:45 +00:00
uio->uio_td->td_ru.ru_msgrcv++;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
KASSERT(m == so->so_rcv.sb_mb, ("soreceive: m != so->so_rcv.sb_mb"));
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nextrecord = m->m_nextpkt;
if (pr->pr_flags & PR_ADDR) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
KASSERT(m->m_type == MT_SONAME,
("m->m_type == %d", m->m_type));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
orig_resid = 0;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (psa != NULL)
*psa = sodupsockaddr(mtod(m, struct sockaddr *),
M_NOWAIT);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (flags & MSG_PEEK) {
m = m->m_next;
} else {
sbfree(&so->so_rcv, m);
Get rid of the twisted MFREE() macro entirely. Reviewed by: dg, bmilekic MFC after: 3 days
2002-02-05 02:00:56 +00:00
so->so_rcv.sb_mb = m_free(m);
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
m = so->so_rcv.sb_mb;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
sockbuf_pushsync(&so->so_rcv, nextrecord);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
* Process one or more MT_CONTROL mbufs present before any data mbufs
* in the first mbuf chain on the socket buffer. If MSG_PEEK, we
* just copy the data; if !MSG_PEEK, we call into the protocol to
* perform externalization (or freeing if controlp == NULL).
*/
if (m != NULL && m->m_type == MT_CONTROL) {
struct mbuf *cm = NULL, *cmn;
struct mbuf **cme = &cm;
do {
if (flags & MSG_PEEK) {
if (controlp != NULL) {
*controlp = m_copy(m, 0, m->m_len);
controlp = &(*controlp)->m_next;
}
m = m->m_next;
Back out 1.68.2.26. It was a mis-guided change that was already backed out of HEAD and should not have been MFC'd. This will restore UDP socket functionality, which will correct the recent NFS problems. Submitted by: rwatson
2005-06-09 19:56:38 +00:00
} else {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
sbfree(&so->so_rcv, m);
so->so_rcv.sb_mb = m->m_next;
m->m_next = NULL;
*cme = m;
cme = &(*cme)->m_next;
Use sockbuf_pushsync() to synchronize stack and socket buffer state in soreceive() after removing an MT_SONAME mbuf from the head of the socket buffer. When processing MT_CONTROL mbufs in soreceive(), first remove all of the MT_CONTROL mbufs from the head of the socket buffer to a local mbuf chain, then feed them into dom_externalize() as a set, which both avoids thrashing the socket buffer lock when handling multiple control mbufs, and also avoids races with other threads acting on the socket buffer when the socket buffer mutex is released to enter the externalize code. Existing races that might occur if the protocol externalize method blocked during processing have also been closed. Now that we synchronize socket buffer and stack state following modifications to the socket buffer, turn the manual synchronization that previously followed control mbuf processing with a set of assertions. This can eventually be removed. The soreceive() code is now substantially more MPSAFE.
2004-07-11 23:13:14 +00:00
m = so->so_rcv.sb_mb;
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
} while (m != NULL && m->m_type == MT_CONTROL);
if ((flags & MSG_PEEK) == 0)
sockbuf_pushsync(&so->so_rcv, nextrecord);
while (cm != NULL) {
cmn = cm->m_next;
cm->m_next = NULL;
if (pr->pr_domain->dom_externalize != NULL) {
SOCKBUF_UNLOCK(&so->so_rcv);
error = (*pr->pr_domain->dom_externalize)
(cm, controlp);
SOCKBUF_LOCK(&so->so_rcv);
} else if (controlp != NULL)
*controlp = cm;
else
m_freem(cm);
if (controlp != NULL) {
orig_resid = 0;
while (*controlp != NULL)
controlp = &(*controlp)->m_next;
}
cm = cmn;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fix a kernel panic based on receiving an ICMPv6 Packet too Big message. PR: 99779 Submitted by: Jinmei Tatuya Reviewed by: clement, rwatson MFC after: 1 week
2006-08-18 14:05:13 +00:00
if (m != NULL)
Fix for PR 83885. Make sure that there actually is a next packet before setting nextrecord to that field. PR: 83885 Submitted by: hirose@comm.yamaha.co.jp Obtained from: Patch suggested in the PR MFC after: 1 week
2005-07-28 10:10:01 +00:00
nextrecord = so->so_rcv.sb_mb->m_nextpkt;
else
Fix a kernel panic based on receiving an ICMPv6 Packet too Big message. PR: 99779 Submitted by: Jinmei Tatuya Reviewed by: clement, rwatson MFC after: 1 week
2006-08-18 14:05:13 +00:00
nextrecord = so->so_rcv.sb_mb;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
orig_resid = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (m != NULL) {
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
if ((flags & MSG_PEEK) == 0) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
KASSERT(m->m_nextpkt == nextrecord,
("soreceive: post-control, nextrecord !sync"));
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
if (nextrecord == NULL) {
KASSERT(so->so_rcv.sb_mb == m,
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
("soreceive: post-control, sb_mb!=m"));
KASSERT(so->so_rcv.sb_lastrecord == m,
("soreceive: post-control, lastrecord!=m"));
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
}
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
type = m->m_type;
if (type == MT_OOBDATA)
flags |= MSG_OOB;
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
} else {
if ((flags & MSG_PEEK) == 0) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
KASSERT(so->so_rcv.sb_mb == nextrecord,
("soreceive: sb_mb != nextrecord"));
if (so->so_rcv.sb_mb == NULL) {
KASSERT(so->so_rcv.sb_lastrecord == NULL,
("soreceive: sb_lastercord != NULL"));
}
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
* Now continue to read any data mbufs off of the head of the socket
* buffer until the read request is satisfied. Note that 'type' is
* used to store the type of any mbuf reads that have happened so far
* such that soreceive() can stop reading if the type changes, which
* causes soreceive() to return only one of regular data and inline
* out-of-band data in a single socket receive operation.
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
moff = 0;
offset = 0;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
while (m != NULL && uio->uio_resid > 0 && error == 0) {
/*
* If the type of mbuf has changed since the last mbuf
* examined ('type'), end the receive operation.
*/
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (m->m_type == MT_OOBDATA) {
if (type != MT_OOBDATA)
break;
} else if (type == MT_OOBDATA)
break;
Split DIAGNOSTIC -> DIAGNOSTIC, INVARIANTS, and INVARIANT_SUPPORT as discussed on -hackers. Introduce 'KASSERT(assertion, ("panic message", args))' for simple check + panic. Reviewed by: msmith
1999-01-08 17:31:30 +00:00
else
Retire MT_HEADER mbuf type and change its users to use MT_DATA. Having an additional MT_HEADER mbuf type is superfluous and redundant as nothing depends on it. It only adds a layer of confusion. The distinction between header mbuf's and data mbuf's is solely done through the m->m_flags M_PKTHDR flag. Non-native code is not changed in this commit. For compatibility MT_HEADER is mapped to MT_DATA. Sponsored by: TCP/IP Optimization Fundraise 2005
2005-11-02 13:46:32 +00:00
KASSERT(m->m_type == MT_DATA,
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
("m->m_type == %d", m->m_type));
so->so_rcv.sb_state &= ~SBS_RCVATMARK;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
len = uio->uio_resid;
if (so->so_oobmark && len > so->so_oobmark - offset)
len = so->so_oobmark - offset;
if (len > m->m_len - moff)
len = m->m_len - moff;
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* If mp is set, just pass back the mbufs. Otherwise copy
* them out via the uio, then free. Sockbuf must be
* consistent here (points to current mbuf, it points to next
* record) when we drop priority; we must note any additions
* to the sockbuf when we block interrupts again.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (mp == NULL) {
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
#ifdef ZERO_COPY_SOCKETS
if (so_zero_copy_receive) {
int disposable;
if ((m->m_flags & M_EXT)
&& (m->m_ext.ext_type == EXT_DISPOSABLE))
disposable = 1;
else
disposable = 0;
error = uiomoveco(mtod(m, char *) + moff,
(int)len, uio,
disposable);
} else
#endif /* ZERO_COPY_SOCKETS */
error = uiomove(mtod(m, char *) + moff, (int)len, uio);
SOCKBUF_LOCK(&so->so_rcv);
Fix a case where socket I/O atomicity is violated due to not dropping the entire record when a non-data mbuf is removed in the soreceive() path. This only triggers a panic directly when compiled with INVARIANTS. PR: 38495 Submitted by: James Juran MFC after: 1 week
2006-09-22 15:34:16 +00:00
if (error) {
/*
Drop all received data mbufs from a socket's queue if the MT_SONAME mbuf is dropped, to preserve the invariant in the PR_ADDR case. Add a regression test to detect this condition, but do not hook it up to the build for now. PR: kern/38495 Submitted by: James Juran Reviewed by: sam, rwatson Obtained from: NetBSD MFC after: 2 weeks
2006-12-23 21:07:07 +00:00
* The MT_SONAME mbuf has already been removed
* from the record, so it is necessary to
* remove the data mbufs, if any, to preserve
* the invariant in the case of PR_ADDR that
* requires MT_SONAME mbufs at the head of
* each record.
Fix a case where socket I/O atomicity is violated due to not dropping the entire record when a non-data mbuf is removed in the soreceive() path. This only triggers a panic directly when compiled with INVARIANTS. PR: 38495 Submitted by: James Juran MFC after: 1 week
2006-09-22 15:34:16 +00:00
*/
Diff reduction with RELENG_6, style(9): Remove unnecessary brace; && should be on end of line. No functional changes.
2007-02-03 03:57:45 +00:00
if (m && pr->pr_flags & PR_ATOMIC &&
((flags & MSG_PEEK) == 0))
Fix a case where socket I/O atomicity is violated due to not dropping the entire record when a non-data mbuf is removed in the soreceive() path. This only triggers a panic directly when compiled with INVARIANTS. PR: 38495 Submitted by: James Juran MFC after: 1 week
2006-09-22 15:34:16 +00:00
(void)sbdroprecord_locked(&so->so_rcv);
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
Check for error return from uiomove to prevent looping endlessly in soreceive(). Closes PR#2114. Submitted by: wpaul
1996-11-29 19:03:42 +00:00
goto release;
Fix a case where socket I/O atomicity is violated due to not dropping the entire record when a non-data mbuf is removed in the soreceive() path. This only triggers a panic directly when compiled with INVARIANTS. PR: 38495 Submitted by: James Juran MFC after: 1 week
2006-09-22 15:34:16 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
} else
uio->uio_resid -= len;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (len == m->m_len - moff) {
if (m->m_flags & M_EOR)
flags |= MSG_EOR;
if (flags & MSG_PEEK) {
m = m->m_next;
moff = 0;
} else {
nextrecord = m->m_nextpkt;
sbfree(&so->so_rcv, m);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (mp != NULL) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*mp = m;
mp = &m->m_next;
so->so_rcv.sb_mb = m = m->m_next;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
*mp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
} else {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
so->so_rcv.sb_mb = m_free(m);
m = so->so_rcv.sb_mb;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
In soreceive(), when a first mbuf is removed from socket buffer use sockbuf_pushsync(). Previous manipulation could lead to an inconsistent mbuf. Reviewed by: rwatson
2005-09-06 17:05:11 +00:00
sockbuf_pushsync(&so->so_rcv, nextrecord);
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
} else {
if (flags & MSG_PEEK)
moff += len;
else {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (mp != NULL) {
int copy_flag;
if (flags & MSG_DONTWAIT)
copy_flag = M_DONTWAIT;
else
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
copy_flag = M_WAIT;
if (copy_flag == M_WAIT)
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
*mp = m_copym(m, 0, len, copy_flag);
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
if (copy_flag == M_WAIT)
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(&so->so_rcv);
if (*mp == NULL) {
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* m_copym() couldn't
* allocate an mbuf. Adjust
* uio_resid back (it was
* adjusted down by len
* bytes, which we didn't end
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
* up "copying" over).
*/
uio->uio_resid += len;
break;
}
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
m->m_data += len;
m->m_len -= len;
so->so_rcv.sb_cc -= len;
}
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (so->so_oobmark) {
if ((flags & MSG_PEEK) == 0) {
so->so_oobmark -= len;
if (so->so_oobmark == 0) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
so->so_rcv.sb_state |= SBS_RCVATMARK;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
}
} else {
offset += len;
if (offset == so->so_oobmark)
break;
}
}
if (flags & MSG_EOR)
break;
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* If the MSG_WAITALL flag is set (for non-atomic socket), we
* must not quit until "uio->uio_resid == 0" or an error
* termination. If a signal/timeout occurs, return with a
* short count but without error. Keep sockbuf locked
* against other readers.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
!sosendallatonce(so) && nextrecord == NULL) {
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
if (so->so_error || so->so_rcv.sb_state & SBS_CANTRCVMORE)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
When doing a recv(.. MSG_WAITALL) for a message which is larger than the socket buffer size, the receive is done in sections. After completing a read, call pru_rcvd on the underlying protocol before blocking again. This allows the the protocol to take appropriate action, such as sending a TCP window update to the peer, if the window happened to close because the socket buffer was filled. If the protocol is not notified, a TCP transfer may stall until the remote end sends a window probe.
2001-03-16 22:37:06 +00:00
/*
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
* Notify the protocol that some data has been
* drained before blocking.
When doing a recv(.. MSG_WAITALL) for a message which is larger than the socket buffer size, the receive is done in sections. After completing a read, call pru_rcvd on the underlying protocol before blocking again. This allows the the protocol to take appropriate action, such as sending a TCP window update to the peer, if the window happened to close because the socket buffer was filled. If the protocol is not notified, a TCP transfer may stall until the remote end sends a window probe.
2001-03-16 22:37:06 +00:00
*/
Chance protocol switch method pru_detach() so that it returns void rather than an error. Detaches do not "fail", they other occur or the protocol flags SS_PROTOREF to take ownership of the socket. soclose() no longer looks at so_pcb to see if it's NULL, relying entirely on the protocol to decide whether it's time to free the socket or not using SS_PROTOREF. so_pcb is now entirely owned and managed by the protocol code. Likewise, no longer test so_pcb in other socket functions, such as soreceive(), which have no business digging into protocol internals. Protocol detach routines no longer try to free the socket on detach, this is performed in the socket code if the protocol permits it. In rts_detach(), no longer test for rp != NULL in detach, and likewise in other protocols that don't permit a NULL so_pcb, reduce the incidence of testing for it during detach. netinet and netinet6 are not fully updated to this change, which will be in an upcoming commit. In their current state they may leak memory or panic. MFC after: 3 months
2006-04-01 15:42:02 +00:00
if (pr->pr_flags & PR_WANTRCVD) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
When doing a recv(.. MSG_WAITALL) for a message which is larger than the socket buffer size, the receive is done in sections. After completing a read, call pru_rcvd on the underlying protocol before blocking again. This allows the the protocol to take appropriate action, such as sending a TCP window update to the peer, if the window happened to close because the socket buffer was filled. If the protocol is not notified, a TCP transfer may stall until the remote end sends a window probe.
2001-03-16 22:37:06 +00:00
(*pr->pr_usrreqs->pru_rcvd)(so, flags);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(&so->so_rcv);
}
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = sbwait(&so->so_rcv);
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
if (error) {
SOCKBUF_UNLOCK(&so->so_rcv);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
goto release;
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
}
All of this is cosmetic. prototypes, #includes, printfs and so on. Makes GCC a lot more silent.
1994-10-02 17:35:40 +00:00
m = so->so_rcv.sb_mb;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (m != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nextrecord = m->m_nextpkt;
}
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
if (m != NULL && pr->pr_flags & PR_ATOMIC) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
flags |= MSG_TRUNC;
Now socket buffer locks are being asserted at higher code blocks in soreceive(), remove some leaf assertions that are redundant.
2004-07-10 04:38:06 +00:00
if ((flags & MSG_PEEK) == 0)
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
(void) sbdroprecord_locked(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
if ((flags & MSG_PEEK) == 0) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (m == NULL) {
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
/*
* First part is an inline SB_EMPTY_FIXUP(). Second
* part makes sure sb_lastrecord is up-to-date if
* there is still data in the socket buffer.
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
so->so_rcv.sb_mb = nextrecord;
speedup stream socket recv handling by tracking the tail of the mbuf chain instead of walking the list for each append Submitted by: ps/jayanth Obtained from: netbsd (jason thorpe)
2003-10-28 05:47:40 +00:00
if (so->so_rcv.sb_mb == NULL) {
so->so_rcv.sb_mbtail = NULL;
so->so_rcv.sb_lastrecord = NULL;
} else if (nextrecord->m_nextpkt == NULL)
so->so_rcv.sb_lastrecord = nextrecord;
}
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* If soreceive() is being done from the socket callback,
* then don't need to generate ACK to peer to update window,
* since ACK will be generated on return to TCP.
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
*/
Change comment on soabort() to more accurately describe how/when soabort() is used. Remove trailing white space.
2006-07-16 23:09:39 +00:00
if (!(flags & MSG_SOCALLBCK) &&
Chance protocol switch method pru_detach() so that it returns void rather than an error. Detaches do not "fail", they other occur or the protocol flags SS_PROTOREF to take ownership of the socket. soclose() no longer looks at so_pcb to see if it's NULL, relying entirely on the protocol to decide whether it's time to free the socket or not using SS_PROTOREF. so_pcb is now entirely owned and managed by the protocol code. Likewise, no longer test so_pcb in other socket functions, such as soreceive(), which have no business digging into protocol internals. Protocol detach routines no longer try to free the socket on detach, this is performed in the socket code if the protocol permits it. In rts_detach(), no longer test for rp != NULL in detach, and likewise in other protocols that don't permit a NULL so_pcb, reduce the incidence of testing for it during detach. netinet and netinet6 are not fully updated to this change, which will be in an upcoming commit. In their current state they may leak memory or panic. MFC after: 3 months
2006-04-01 15:42:02 +00:00
(pr->pr_flags & PR_WANTRCVD)) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
Modify the kernel to use the new pr_usrreqs interface rather than the old pr_usrreq mechanism which was poorly designed and error-prone. This commit renames pr_usrreq to pr_ousrreq so that old code which depended on it would break in an obvious manner. This commit also implements the new interface for TCP, although the old function is left as an example (#ifdef'ed out). This commit ALSO fixes a longstanding bug in the TCP timer processing (introduced by davidg on 1995/04/12) which caused timer processing on a TCB to always stop after a single timer had expired (because it misinterpreted the return value from tcp_usrreq() to indicate that the TCB had been deleted). Finally, some code related to polling has been deleted from if.c because it is not relevant t -current and doesn't look at all like my current code.
1996-07-11 16:32:50 +00:00
(*pr->pr_usrreqs->pru_rcvd)(so, flags);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(&so->so_rcv);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (orig_resid == uio->uio_resid && orig_resid &&
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
(flags & MSG_EOR) == 0 && (so->so_rcv.sb_state & SBS_CANTRCVMORE) == 0) {
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto restart;
}
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (flagsp != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*flagsp |= flags;
release:
sbunlock(&so->so_rcv);
return (error);
}
Add soreceive_dgram(9), an optimized socket receive function for use by datagram-only protocols, such as UDP. This version removes use of sblock(), which is not required due to an inability to interlace data improperly with datagrams, as well as avoiding some of the larger loops and state management that don't apply on datagram sockets. This is experimental code, so hook it up only for UDPv4 for testing; if there are problems we may need to revise it or turn it off by default, but it offers *significant* performance improvements for threaded UDP applications such as BIND9, nsd, and memcached using UDP. Tested by: kris, ps
2008-07-02 23:23:27 +00:00
/*
* Optimized version of soreceive() for simple datagram cases from userspace;
* this is experimental, and while heavily tested, may contain errors.
*/
int
soreceive_dgram(struct socket *so, struct sockaddr **psa, struct uio *uio,
struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
{
struct mbuf *m, *m2;
int flags, len, error, offset;
struct protosw *pr = so->so_proto;
struct mbuf *nextrecord;
if (psa != NULL)
*psa = NULL;
if (controlp != NULL)
*controlp = NULL;
if (flagsp != NULL)
flags = *flagsp &~ MSG_EOR;
else
flags = 0;
/*
* For any complicated cases, fall back to the full
* soreceive_generic().
*/
if (mp0 != NULL || (flags & MSG_PEEK) || (flags & MSG_OOB))
return (soreceive_generic(so, psa, uio, mp0, controlp,
flagsp));
/*
* Enforce restrictions on use.
*/
KASSERT((pr->pr_flags & PR_WANTRCVD) == 0,
("soreceive_dgram: wantrcvd"));
KASSERT(pr->pr_flags & PR_ATOMIC, ("soreceive_dgram: !atomic"));
KASSERT((so->so_rcv.sb_state & SBS_RCVATMARK) == 0,
("soreceive_dgram: SBS_RCVATMARK"));
KASSERT((so->so_proto->pr_flags & PR_CONNREQUIRED) == 0,
("soreceive_dgram: P_CONNREQUIRED"));
restart:
SOCKBUF_LOCK(&so->so_rcv);
m = so->so_rcv.sb_mb;
/*
* If we have less data than requested, block awaiting more (subject
* to any timeout) if:
* 1. the current count is less than the low water mark, or
* 2. MSG_WAITALL is set, and it is possible to do the entire
* receive operation at once if we block (resid <= hiwat).
* 3. MSG_DONTWAIT is not set
* If MSG_WAITALL is set but resid is larger than the receive buffer,
* we have to do the receive in sections, and thus risk returning a
* short count if a timeout or signal occurs after we start.
*/
if (m == NULL) {
KASSERT(m != NULL || !so->so_rcv.sb_cc,
("receive: m == %p so->so_rcv.sb_cc == %u",
m, so->so_rcv.sb_cc));
if (so->so_error) {
if (m != NULL)
goto dontblock;
error = so->so_error;
so->so_error = 0;
SOCKBUF_UNLOCK(&so->so_rcv);
return (error);
}
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
if (so->so_rcv.sb_state & SBS_CANTRCVMORE) {
if (m == NULL) {
SOCKBUF_UNLOCK(&so->so_rcv);
return (0);
} else
goto dontblock;
}
if (uio->uio_resid == 0) {
SOCKBUF_UNLOCK(&so->so_rcv);
return (0);
}
if ((so->so_state & SS_NBIO) ||
(flags & (MSG_DONTWAIT|MSG_NBIO))) {
SOCKBUF_UNLOCK(&so->so_rcv);
error = EWOULDBLOCK;
return (error);
}
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
error = sbwait(&so->so_rcv);
SOCKBUF_UNLOCK(&so->so_rcv);
if (error)
return (error);
goto restart;
}
dontblock:
/*
* From this point onward, we maintain 'nextrecord' as a cache of the
* pointer to the next record in the socket buffer. We must keep the
* various socket buffer pointers and local stack versions of the
* pointers in sync, pushing out modifications before dropping the
* socket buffer mutex, and re-reading them when picking it up.
*
* Otherwise, we will race with the network stack appending new data
* or records onto the socket buffer by using inconsistent/stale
* versions of the field, possibly resulting in socket buffer
* corruption.
*
* By holding the high-level sblock(), we prevent simultaneous
* readers from pulling off the front of the socket buffer.
*/
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
if (uio->uio_td)
uio->uio_td->td_ru.ru_msgrcv++;
KASSERT(m == so->so_rcv.sb_mb, ("soreceive: m != so->so_rcv.sb_mb"));
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
nextrecord = m->m_nextpkt;
if (pr->pr_flags & PR_ADDR) {
KASSERT(m->m_type == MT_SONAME,
("m->m_type == %d", m->m_type));
if (psa != NULL)
*psa = sodupsockaddr(mtod(m, struct sockaddr *),
M_NOWAIT);
sbfree(&so->so_rcv, m);
so->so_rcv.sb_mb = m_free(m);
m = so->so_rcv.sb_mb;
sockbuf_pushsync(&so->so_rcv, nextrecord);
}
if (m == NULL) {
/* XXXRW: Can this happen? */
SOCKBUF_UNLOCK(&so->so_rcv);
return (0);
}
KASSERT(m->m_nextpkt == nextrecord,
("soreceive: post-control, nextrecord !sync"));
if (nextrecord == NULL) {
KASSERT(so->so_rcv.sb_mb == m,
("soreceive: post-control, sb_mb!=m"));
KASSERT(so->so_rcv.sb_lastrecord == m,
("soreceive: post-control, lastrecord!=m"));
}
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
KASSERT(m == so->so_rcv.sb_mb, ("soreceive_dgram: m not sb_mb"));
KASSERT(so->so_rcv.sb_mb->m_nextpkt == nextrecord,
("soreceive_dgram: m_nextpkt != nextrecord"));
/*
* Pull 'm' and its chain off the front of the packet queue.
*/
so->so_rcv.sb_mb = NULL;
sockbuf_pushsync(&so->so_rcv, nextrecord);
/*
* Walk 'm's chain and free that many bytes from the socket buffer.
*/
for (m2 = m; m2 != NULL; m2 = m2->m_next)
sbfree(&so->so_rcv, m2);
/*
* Do a few last checks before we let go of the lock.
*/
SBLASTRECORDCHK(&so->so_rcv);
SBLASTMBUFCHK(&so->so_rcv);
SOCKBUF_UNLOCK(&so->so_rcv);
/*
* Packet to copyout() is now in 'm' and it is disconnected from the
* queue.
*
* Process one or more MT_CONTROL mbufs present before any data mbufs
* in the first mbuf chain on the socket buffer. If MSG_PEEK, we
* just copy the data; if !MSG_PEEK, we call into the protocol to
* perform externalization (or freeing if controlp == NULL).
*/
if (m->m_type == MT_CONTROL) {
struct mbuf *cm = NULL, *cmn;
struct mbuf **cme = &cm;
do {
m2 = m->m_next;
m->m_next = NULL;
*cme = m;
cme = &(*cme)->m_next;
m = m2;
} while (m != NULL && m->m_type == MT_CONTROL);
while (cm != NULL) {
cmn = cm->m_next;
cm->m_next = NULL;
if (pr->pr_domain->dom_externalize != NULL) {
error = (*pr->pr_domain->dom_externalize)
(cm, controlp);
} else if (controlp != NULL)
*controlp = cm;
else
m_freem(cm);
if (controlp != NULL) {
while (*controlp != NULL)
controlp = &(*controlp)->m_next;
}
cm = cmn;
}
}
KASSERT(m->m_type == MT_DATA, ("soreceive_dgram: !data"));
offset = 0;
while (m != NULL && uio->uio_resid > 0) {
len = uio->uio_resid;
if (len > m->m_len)
len = m->m_len;
error = uiomove(mtod(m, char *), (int)len, uio);
if (error) {
m_freem(m);
return (error);
}
m = m_free(m);
}
if (m != NULL && pr->pr_flags & PR_ATOMIC)
flags |= MSG_TRUNC;
m_freem(m);
if (flagsp != NULL)
*flagsp |= flags;
return (0);
}
soreceive_generic(), and sopoll_generic(). Add new functions sosend(), soreceive(), and sopoll(), which are wrappers for pru_sosend, pru_soreceive, and pru_sopoll, and are now used univerally by socket consumers rather than either directly invoking the old so*() functions or directly invoking the protocol switch method (about an even split prior to this commit). This completes an architectural change that was begun in 1996 to permit protocols to provide substitute implementations, as now used by UDP. Consumers now uniformly invoke sosend(), soreceive(), and sopoll() to perform these operations on sockets -- in particular, distributed file systems and socket system calls. Architectural head nod: sam, gnn, wollman
2006-07-24 15:20:08 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soreceive(struct socket *so, struct sockaddr **psa, struct uio *uio,
struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
soreceive_generic(), and sopoll_generic(). Add new functions sosend(), soreceive(), and sopoll(), which are wrappers for pru_sosend, pru_soreceive, and pru_sopoll, and are now used univerally by socket consumers rather than either directly invoking the old so*() functions or directly invoking the protocol switch method (about an even split prior to this commit). This completes an architectural change that was begun in 1996 to permit protocols to provide substitute implementations, as now used by UDP. Consumers now uniformly invoke sosend(), soreceive(), and sopoll() to perform these operations on sockets -- in particular, distributed file systems and socket system calls. Architectural head nod: sam, gnn, wollman
2006-07-24 15:20:08 +00:00
{
/* XXXRW: Temporary debugging. */
KASSERT(so->so_proto->pr_usrreqs->pru_soreceive != soreceive,
("soreceive: protocol calls soreceive"));
return (so->so_proto->pr_usrreqs->pru_soreceive(so, psa, uio, mp0,
controlp, flagsp));
}
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soshutdown(struct socket *so, int how)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct protosw *pr = so->so_proto;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
In soshutdown(), use SHUT_{RD,WR,RDWR} instead of FREAD and FWRITE. Also, return EINVAL if `how' is invalid, as required by POSIX spec.
2001-02-27 13:48:07 +00:00
if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
return (EINVAL);
Add pru_flush routine so a transport can flush itself during Shutdown MFC after: 1 week
2008-04-14 18:06:04 +00:00
if (pr->pr_usrreqs->pru_flush != NULL) {
(*pr->pr_usrreqs->pru_flush)(so, how);
}
In soshutdown(), use SHUT_{RD,WR,RDWR} instead of FREAD and FWRITE. Also, return EINVAL if `how' is invalid, as required by POSIX spec.
2001-02-27 13:48:07 +00:00
if (how != SHUT_WR)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
sorflush(so);
In soshutdown(), use SHUT_{RD,WR,RDWR} instead of FREAD and FWRITE. Also, return EINVAL if `how' is invalid, as required by POSIX spec.
2001-02-27 13:48:07 +00:00
if (how != SHUT_RD)
Modify the kernel to use the new pr_usrreqs interface rather than the old pr_usrreq mechanism which was poorly designed and error-prone. This commit renames pr_usrreq to pr_ousrreq so that old code which depended on it would break in an obvious manner. This commit also implements the new interface for TCP, although the old function is left as an example (#ifdef'ed out). This commit ALSO fixes a longstanding bug in the TCP timer processing (introduced by davidg on 1995/04/12) which caused timer processing on a TCB to always stop after a single timer had expired (because it misinterpreted the return value from tcp_usrreq() to indicate that the TCB had been deleted). Finally, some code related to polling has been deleted from if.c because it is not relevant t -current and doesn't look at all like my current code.
1996-07-11 16:32:50 +00:00
return ((*pr->pr_usrreqs->pru_shutdown)(so));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (0);
}
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sorflush(struct socket *so)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct sockbuf *sb = &so->so_rcv;
struct protosw *pr = so->so_proto;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct sockbuf asb;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
Further clean up sorflush: - Expose sbrelease_internal(), a variant of sbrelease() with no expectations about the validity of locks in the socket buffer. - Use sbrelease_internel() in sorflush(), and as a result avoid intializing and destroying a socket buffer lock for the temporary stack copy of the actual buffer, asb. - Add a comment indicating why we do what we do, and remove an XXX since things have gotten less ugly in sorflush() lately. This makes socket close cleaner, and possibly also marginally faster. MFC after: 3 weeks
2008-02-04 12:25:13 +00:00
* In order to avoid calling dom_dispose with the socket buffer mutex
* held, and in order to generally avoid holding the lock for a long
* time, we make a copy of the socket buffer and clear the original
* (except locks, state). The new socket buffer copy won't have
* initialized locks so we can only call routines that won't use or
* assert those locks.
*
Correct two problems relating to sorflush(), which is called to flush read socket buffers in shutdown() and close(): - Call socantrcvmore() before sblock() to dislodge any threads that might be sleeping (potentially indefinitely) while holding sblock(), such as a thread blocked in recv(). - Flag the sblock() call as non-interruptible so that a signal delivered to the thread calling sorflush() doesn't cause sblock() to fail. The sblock() is required to ensure that all other socket consumer threads have, in fact, left, and do not enter, the socket buffer until we're done flushin it. To implement the latter, change the 'flags' argument to sblock() to accept two flags, SBL_WAIT and SBL_NOINTR, rather than one M_WAITOK flag. When SBL_NOINTR is set, it forces a non-interruptible sx acquisition, regardless of the setting of the disposition of SB_NOINTR on the socket buffer; without this change it would be possible for another thread to clear SB_NOINTR between when the socket buffer mutex is released and sblock() is invoked. Reviewed by: bz, kmacy Reported by: Jos Backus <jos at catnook dot com>
2008-01-31 08:22:24 +00:00
* Dislodge threads currently blocked in receive and wait to acquire
* a lock against other simultaneous readers before clearing the
* socket buffer. Don't let our acquire be interrupted by a signal
* despite any existing socket disposition on interruptable waiting.
*/
socantrcvmore(so);
(void) sblock(sb, SBL_WAIT | SBL_NOINTR);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Invalidate/clear most of the sockbuf structure, but leave selinfo
* and mutex data unchanged.
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
*/
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
SOCKBUF_LOCK(sb);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
bzero(&asb, offsetof(struct sockbuf, sb_startzero));
bcopy(&sb->sb_startzero, &asb.sb_startzero,
sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
bzero(&sb->sb_startzero,
sizeof(*sb) - offsetof(struct sockbuf, sb_startzero));
SOCKBUF_UNLOCK(sb);
sblock() implements a sleep lock by interlocking SB_WANT and SB_LOCK flags on each socket buffer with the socket buffer's mutex. This sleep lock is used to serialize I/O on sockets in order to prevent I/O interlacing. This change replaces the custom sleep lock with an sx(9) lock, which results in marginally better performance, better handling of contention during simultaneous socket I/O across multiple threads, and a cleaner separation between the different layers of locking in socket buffers. Specifically, the socket buffer mutex is now solely responsible for serializing simultaneous operation on the socket buffer data structure, and not for I/O serialization. While here, fix two historic bugs: (1) a bug allowing I/O to be occasionally interlaced during long I/O operations (discovere by Isilon). (2) a bug in which failed non-blocking acquisition of the socket buffer I/O serialization lock might be ignored (discovered by sam). SCTP portion of this patch submitted by rrs.
2007-05-03 14:42:42 +00:00
sbunlock(sb);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
Further clean up sorflush: - Expose sbrelease_internal(), a variant of sbrelease() with no expectations about the validity of locks in the socket buffer. - Use sbrelease_internel() in sorflush(), and as a result avoid intializing and destroying a socket buffer lock for the temporary stack copy of the actual buffer, asb. - Add a comment indicating why we do what we do, and remove an XXX since things have gotten less ugly in sorflush() lately. This makes socket close cleaner, and possibly also marginally faster. MFC after: 3 weeks
2008-02-04 12:25:13 +00:00
/*
* Dispose of special rights and flush the socket buffer. Don't call
* any unsafe routines (that rely on locks being initialized) on asb.
*/
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose != NULL)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
(*pr->pr_domain->dom_dispose)(asb.sb_mb);
Further clean up sorflush: - Expose sbrelease_internal(), a variant of sbrelease() with no expectations about the validity of locks in the socket buffer. - Use sbrelease_internel() in sorflush(), and as a result avoid intializing and destroying a socket buffer lock for the temporary stack copy of the actual buffer, asb. - Add a comment indicating why we do what we do, and remove an XXX since things have gotten less ugly in sorflush() lately. This makes socket close cleaner, and possibly also marginally faster. MFC after: 3 weeks
2008-02-04 12:25:13 +00:00
sbrelease_internal(&asb, so);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Perhaps this routine, and sooptcopyout(), below, ought to come in an
* additional variant to handle the case where the option value needs to be
* some kind of integer, but not a specific size. In addition to their use
* here, these functions are also called by the protocol-level pr_ctloutput()
* routines.
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sooptcopyin(struct sockopt *sopt, void *buf, size_t len, size_t minlen)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
size_t valsize;
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* If the user gives us more than we wanted, we ignore it, but if we
* don't get the minimum length the caller wants, we return EINVAL.
* On success, sopt->sopt_valsize is set to however much we actually
* retrieved.
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
*/
if ((valsize = sopt->sopt_valsize) < minlen)
return EINVAL;
if (valsize > len)
sopt->sopt_valsize = valsize = len;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (sopt->sopt_td != NULL)
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
return (copyin(sopt->sopt_val, buf, valsize));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
bcopy(sopt->sopt_val, buf, valsize);
Improve consistency of return() style. MFC after: 3 days
2006-02-12 15:00:27 +00:00
return (0);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Kernel version of setsockopt(2).
*
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
* XXX: optlen is size_t, not socklen_t
*/
int
so_setsockopt(struct socket *so, int level, int optname, void *optval,
size_t optlen)
{
struct sockopt sopt;
sopt.sopt_level = level;
sopt.sopt_name = optname;
sopt.sopt_dir = SOPT_SET;
sopt.sopt_val = optval;
sopt.sopt_valsize = optlen;
sopt.sopt_td = NULL;
return (sosetopt(so, &sopt));
}
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sosetopt(struct socket *so, struct sockopt *sopt)
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
{
int error, optval;
struct linger l;
struct timeval tv;
Realy fix overflow on SO_*TIMEO Submitted by: bde
1999-05-21 15:54:40 +00:00
u_long val;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#ifdef MAC
struct mac extmac;
#endif
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
error = 0;
if (sopt->sopt_level != SOL_SOCKET) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (so->so_proto && so->so_proto->pr_ctloutput)
return ((*so->so_proto->pr_ctloutput)
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
(so, sopt));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = ENOPROTOOPT;
} else {
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
switch (sopt->sopt_name) {
Accept filters broke kernels compiled without options INET. Make accept filters conditional on INET support to fix. Pointed out by: bde Tested and assisted by: Stephen J. Kiernan <sab@vegamuse.org>
2000-11-20 01:35:25 +00:00
#ifdef INET
case SO_ACCEPTFILTER:
error = do_setopt_accept_filter(so, sopt);
if (error)
goto bad;
break;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_LINGER:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
error = sooptcopyin(sopt, &l, sizeof l, sizeof l);
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto bad;
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_LOCK(so);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
so->so_linger = l.l_linger;
if (l.l_onoff)
so->so_options |= SO_LINGER;
else
so->so_options &= ~SO_LINGER;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_UNLOCK(so);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
break;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_DEBUG:
case SO_KEEPALIVE:
case SO_DONTROUTE:
case SO_USELOOPBACK:
case SO_BROADCAST:
case SO_REUSEADDR:
case SO_REUSEPORT:
case SO_OOBINLINE:
Make it possible to return more than one piece of control information (PR #1178). Define a new SO_TIMESTAMP socket option for datagram sockets to return packet-arrival timestamps as control information (PR #1179). Submitted by: Louis Mamakos <loiue@TransSys.com>
1996-05-09 20:15:26 +00:00
case SO_TIMESTAMP:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
case SO_BINTIME:
case SO_NOSIGPIPE:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
error = sooptcopyin(sopt, &optval, sizeof optval,
sizeof optval);
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto bad;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_LOCK(so);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
if (optval)
so->so_options |= sopt->sopt_name;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
else
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
so->so_options &= ~sopt->sopt_name;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_UNLOCK(so);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
case SO_SETFIB:
error = sooptcopyin(sopt, &optval, sizeof optval,
sizeof optval);
if (optval < 1 || optval > rt_numfibs) {
error = EINVAL;
goto bad;
}
if ((so->so_proto->pr_domain->dom_family == PF_INET) ||
(so->so_proto->pr_domain->dom_family == PF_ROUTE)) {
so->so_fibnum = optval;
} else {
so->so_fibnum = 0;
}
break;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_SNDBUF:
case SO_RCVBUF:
case SO_SNDLOWAT:
case SO_RCVLOWAT:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
error = sooptcopyin(sopt, &optval, sizeof optval,
sizeof optval);
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto bad;
Don't accept insane values for SO_(SND|RCV)BUF, and the low water marks. Specifically, don't allow a value < 1 for any of them (it doesn't make sense), and don't let the low water mark be greater than the corresponding high water mark. Pre-Approved by: wollman Obtained from: NetBSD
1997-06-27 15:28:54 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Values < 1 make no sense for any of these options,
* so disallow them.
Don't accept insane values for SO_(SND|RCV)BUF, and the low water marks. Specifically, don't allow a value < 1 for any of them (it doesn't make sense), and don't let the low water mark be greater than the corresponding high water mark. Pre-Approved by: wollman Obtained from: NetBSD
1997-06-27 15:28:54 +00:00
*/
if (optval < 1) {
error = EINVAL;
goto bad;
}
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
switch (sopt->sopt_name) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_SNDBUF:
case SO_RCVBUF:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
if (sbreserve(sopt->sopt_name == SO_SNDBUF ?
Implement RLIMIT_SBSIZE in the kernel. This is a per-uid sockbuf total usage limit.
1999-10-09 20:42:17 +00:00
&so->so_snd : &so->so_rcv, (u_long)optval,
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
so, curthread) == 0) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = ENOBUFS;
goto bad;
}
Generic socket buffer auto sizing support, header defines, flag inheritance. MFC after: 1 month
2007-02-01 17:53:41 +00:00
(sopt->sopt_name == SO_SNDBUF ? &so->so_snd :
&so->so_rcv)->sb_flags &= ~SB_AUTOSIZE;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
Don't accept insane values for SO_(SND|RCV)BUF, and the low water marks. Specifically, don't allow a value < 1 for any of them (it doesn't make sense), and don't let the low water mark be greater than the corresponding high water mark. Pre-Approved by: wollman Obtained from: NetBSD
1997-06-27 15:28:54 +00:00
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Make sure the low-water is never greater than the
* high-water.
Don't accept insane values for SO_(SND|RCV)BUF, and the low water marks. Specifically, don't allow a value < 1 for any of them (it doesn't make sense), and don't let the low water mark be greater than the corresponding high water mark. Pre-Approved by: wollman Obtained from: NetBSD
1997-06-27 15:28:54 +00:00
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_SNDLOWAT:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(&so->so_snd);
Don't accept insane values for SO_(SND|RCV)BUF, and the low water marks. Specifically, don't allow a value < 1 for any of them (it doesn't make sense), and don't let the low water mark be greater than the corresponding high water mark. Pre-Approved by: wollman Obtained from: NetBSD
1997-06-27 15:28:54 +00:00
so->so_snd.sb_lowat =
(optval > so->so_snd.sb_hiwat) ?
so->so_snd.sb_hiwat : optval;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_snd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
case SO_RCVLOWAT:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(&so->so_rcv);
Don't accept insane values for SO_(SND|RCV)BUF, and the low water marks. Specifically, don't allow a value < 1 for any of them (it doesn't make sense), and don't let the low water mark be greater than the corresponding high water mark. Pre-Approved by: wollman Obtained from: NetBSD
1997-06-27 15:28:54 +00:00
so->so_rcv.sb_lowat =
(optval > so->so_rcv.sb_hiwat) ?
so->so_rcv.sb_hiwat : optval;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
}
break;
case SO_SNDTIMEO:
case SO_RCVTIMEO:
Allow 32bit get/setsockopt with SO_SNDTIMEO or SO_RECVTIMEO to work.
2005-10-27 04:26:35 +00:00
#ifdef COMPAT_IA32
if (curthread->td_proc->p_sysent == &ia32_freebsd_sysvec) {
struct timeval32 tv32;
error = sooptcopyin(sopt, &tv32, sizeof tv32,
sizeof tv32);
CP(tv32, tv, tv_sec);
CP(tv32, tv, tv_usec);
} else
#endif
error = sooptcopyin(sopt, &tv, sizeof tv,
sizeof tv);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto bad;
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
Realy fix overflow on SO_*TIMEO Submitted by: bde
1999-05-21 15:54:40 +00:00
/* assert(hz > 0); */
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (tv.tv_sec < 0 || tv.tv_sec > INT_MAX / hz ||
Realy fix overflow on SO_*TIMEO Submitted by: bde
1999-05-21 15:54:40 +00:00
tv.tv_usec < 0 || tv.tv_usec >= 1000000) {
error = EDOM;
goto bad;
}
/* assert(tick > 0); */
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
/* assert(ULONG_MAX - INT_MAX >= 1000000); */
Realy fix overflow on SO_*TIMEO Submitted by: bde
1999-05-21 15:54:40 +00:00
val = (u_long)(tv.tv_sec * hz) + tv.tv_usec / tick;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (val > INT_MAX) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = EDOM;
goto bad;
}
Small SO_RCVTIMEO and SO_SNDTIMEO values are mistakenly taken to be zero. PR: kern/32827 Submitted by: Hartmut Brandt <brandt@fokus.gmd.de> Approved by: re (jhb) MFC after: 2 weeks
2002-11-27 13:34:04 +00:00
if (val == 0 && tv.tv_usec != 0)
val = 1;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
switch (sopt->sopt_name) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_SNDTIMEO:
so->so_snd.sb_timeo = val;
break;
case SO_RCVTIMEO:
so->so_rcv.sb_timeo = val;
break;
}
break;
Add three new read-only socket options, which allow regression tests and other applications to query the state of the stack regarding the accept queue on a listen socket: SO_LISTENQLIMIT Return the value of so_qlimit (socket backlog) SO_LISTENQLEN Return the value of so_qlen (complete sockets) SO_LISTENINCQLEN Return the value of so_incqlen (incomplete sockets) Minor white space tweaks to existing socket options to make them consistent. Discussed with: andre MFC after: 1 week
2005-09-18 21:08:03 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
case SO_LABEL:
#ifdef MAC
error = sooptcopyin(sopt, &extmac, sizeof extmac,
sizeof extmac);
if (error)
goto bad;
error = mac_setsockopt_label(sopt->sopt_td->td_ucred,
so, &extmac);
#else
error = EOPNOTSUPP;
#endif
break;
Add three new read-only socket options, which allow regression tests and other applications to query the state of the stack regarding the accept queue on a listen socket: SO_LISTENQLIMIT Return the value of so_qlimit (socket backlog) SO_LISTENQLEN Return the value of so_qlen (complete sockets) SO_LISTENINCQLEN Return the value of so_incqlen (incomplete sockets) Minor white space tweaks to existing socket options to make them consistent. Discussed with: andre MFC after: 1 week
2005-09-18 21:08:03 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
default:
error = ENOPROTOOPT;
break;
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (error == 0 && so->so_proto != NULL &&
so->so_proto->pr_ctloutput != NULL) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
(void) ((*so->so_proto->pr_ctloutput)
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
(so, sopt));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
}
bad:
return (error);
}
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
/*
* Helper routine for getsockopt.
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Make the second argument to sooptcopyout() constant in order to simplify the upcoming PIM patches. Submitted by: Pavlin Radoslavov <pavlin@icir.org>
2003-08-05 00:27:54 +00:00
sooptcopyout(struct sockopt *sopt, const void *buf, size_t len)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
int error;
size_t valsize;
error = 0;
/*
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
* Documented get behavior is that we always return a value, possibly
* truncated to fit in the user's buffer. Traditional behavior is
* that we always tell the user precisely how much we copied, rather
* than something useful like the total amount we had available for
* her. Note that this interface is not idempotent; the entire
* answer must generated ahead of time.
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
*/
Correctly set the return length regardless of the relative size of the user's buffer. Simplify the logic a bit. (Can we have a version of min() for size_t?)
1998-08-31 15:34:55 +00:00
valsize = min(len, sopt->sopt_valsize);
Bow to tradition and correctly implement the bogus-but-hallowed semantics of getsockopt never telling how much it might have copied if only the buffer were big enough.
1998-08-31 18:07:23 +00:00
sopt->sopt_valsize = valsize;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (sopt->sopt_val != NULL) {
if (sopt->sopt_td != NULL)
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
error = copyout(buf, sopt->sopt_val, valsize);
else
bcopy(buf, sopt->sopt_val, valsize);
}
Improve consistency of return() style. MFC after: 3 days
2006-02-12 15:00:27 +00:00
return (error);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
}
int
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sogetopt(struct socket *so, struct sockopt *sopt)
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
{
int error, optval;
struct linger l;
struct timeval tv;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#ifdef MAC
struct mac extmac;
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
error = 0;
if (sopt->sopt_level != SOL_SOCKET) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (so->so_proto && so->so_proto->pr_ctloutput) {
return ((*so->so_proto->pr_ctloutput)
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
(so, sopt));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
} else
return (ENOPROTOOPT);
} else {
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
switch (sopt->sopt_name) {
Accept filters broke kernels compiled without options INET. Make accept filters conditional on INET support to fix. Pointed out by: bde Tested and assisted by: Stephen J. Kiernan <sab@vegamuse.org>
2000-11-20 01:35:25 +00:00
#ifdef INET
return of the accept filter part II accept filters are now loadable as well as able to be compiled into the kernel. two accept filters are provided, one that returns sockets when data arrives the other when an http request is completed (doesn't work with 0.9 requests) Reviewed by: jmg
2000-06-20 01:09:23 +00:00
case SO_ACCEPTFILTER:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
error = do_getopt_accept_filter(so, sopt);
return of the accept filter part II accept filters are now loadable as well as able to be compiled into the kernel. two accept filters are provided, one that returns sockets when data arrives the other when an http request is completed (doesn't work with 0.9 requests) Reviewed by: jmg
2000-06-20 01:09:23 +00:00
break;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_LINGER:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_LOCK(so);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
l.l_onoff = so->so_options & SO_LINGER;
l.l_linger = so->so_linger;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCK_UNLOCK(so);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
error = sooptcopyout(sopt, &l, sizeof l);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
case SO_USELOOPBACK:
case SO_DONTROUTE:
case SO_DEBUG:
case SO_KEEPALIVE:
case SO_REUSEADDR:
case SO_REUSEPORT:
case SO_BROADCAST:
case SO_OOBINLINE:
Make getsockopt(..., SOL_SOCKET, SO_ACCEPTCONN, ...) work per IEEE Std 1003.1 (POSIX).
2005-08-01 21:15:09 +00:00
case SO_ACCEPTCONN:
Make it possible to return more than one piece of control information (PR #1178). Define a new SO_TIMESTAMP socket option for datagram sockets to return packet-arrival timestamps as control information (PR #1179). Submitted by: Louis Mamakos <loiue@TransSys.com>
1996-05-09 20:15:26 +00:00
case SO_TIMESTAMP:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
case SO_BINTIME:
case SO_NOSIGPIPE:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
optval = so->so_options & sopt->sopt_name;
integer:
error = sooptcopyout(sopt, &optval, sizeof optval);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
case SO_TYPE:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
optval = so->so_type;
goto integer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_ERROR:
When retrieving SO_ERROR via getsockopt(), hold the socket lock around the retrieval and replacement with 0. MFC after: 1 week
2006-06-18 19:02:49 +00:00
SOCK_LOCK(so);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
optval = so->so_error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
so->so_error = 0;
When retrieving SO_ERROR via getsockopt(), hold the socket lock around the retrieval and replacement with 0. MFC after: 1 week
2006-06-18 19:02:49 +00:00
SOCK_UNLOCK(so);
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
goto integer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_SNDBUF:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
optval = so->so_snd.sb_hiwat;
goto integer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_RCVBUF:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
optval = so->so_rcv.sb_hiwat;
goto integer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_SNDLOWAT:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
optval = so->so_snd.sb_lowat;
goto integer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_RCVLOWAT:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
optval = so->so_rcv.sb_lowat;
goto integer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case SO_SNDTIMEO:
case SO_RCVTIMEO:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
optval = (sopt->sopt_name == SO_SNDTIMEO ?
so->so_snd.sb_timeo : so->so_rcv.sb_timeo);
tv.tv_sec = optval / hz;
tv.tv_usec = (optval % hz) * tick;
Allow 32bit get/setsockopt with SO_SNDTIMEO or SO_RECVTIMEO to work.
2005-10-27 04:26:35 +00:00
#ifdef COMPAT_IA32
if (curthread->td_proc->p_sysent == &ia32_freebsd_sysvec) {
struct timeval32 tv32;
CP(tv, tv32, tv_sec);
CP(tv, tv32, tv_usec);
error = sooptcopyout(sopt, &tv32, sizeof tv32);
} else
#endif
error = sooptcopyout(sopt, &tv, sizeof tv);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
break;
Add three new read-only socket options, which allow regression tests and other applications to query the state of the stack regarding the accept queue on a listen socket: SO_LISTENQLIMIT Return the value of so_qlimit (socket backlog) SO_LISTENQLEN Return the value of so_qlen (complete sockets) SO_LISTENINCQLEN Return the value of so_incqlen (incomplete sockets) Minor white space tweaks to existing socket options to make them consistent. Discussed with: andre MFC after: 1 week
2005-09-18 21:08:03 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
case SO_LABEL:
#ifdef MAC
error = sooptcopyin(sopt, &extmac, sizeof(extmac),
sizeof(extmac));
if (error)
return (error);
error = mac_getsockopt_label(sopt->sopt_td->td_ucred,
so, &extmac);
if (error)
return (error);
error = sooptcopyout(sopt, &extmac, sizeof extmac);
#else
error = EOPNOTSUPP;
#endif
break;
Add three new read-only socket options, which allow regression tests and other applications to query the state of the stack regarding the accept queue on a listen socket: SO_LISTENQLIMIT Return the value of so_qlimit (socket backlog) SO_LISTENQLEN Return the value of so_qlen (complete sockets) SO_LISTENINCQLEN Return the value of so_incqlen (incomplete sockets) Minor white space tweaks to existing socket options to make them consistent. Discussed with: andre MFC after: 1 week
2005-09-18 21:08:03 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
case SO_PEERLABEL:
#ifdef MAC
error = sooptcopyin(sopt, &extmac, sizeof(extmac),
sizeof(extmac));
if (error)
return (error);
error = mac_getsockopt_peerlabel(
sopt->sopt_td->td_ucred, so, &extmac);
if (error)
return (error);
error = sooptcopyout(sopt, &extmac, sizeof extmac);
#else
error = EOPNOTSUPP;
#endif
break;
Add three new read-only socket options, which allow regression tests and other applications to query the state of the stack regarding the accept queue on a listen socket: SO_LISTENQLIMIT Return the value of so_qlimit (socket backlog) SO_LISTENQLEN Return the value of so_qlen (complete sockets) SO_LISTENINCQLEN Return the value of so_incqlen (incomplete sockets) Minor white space tweaks to existing socket options to make them consistent. Discussed with: andre MFC after: 1 week
2005-09-18 21:08:03 +00:00
case SO_LISTENQLIMIT:
optval = so->so_qlimit;
goto integer;
case SO_LISTENQLEN:
optval = so->so_qlen;
goto integer;
case SO_LISTENINCQLEN:
optval = so->so_incqlen;
goto integer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
default:
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
error = ENOPROTOOPT;
break;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Yow! Completely change the way socket options are handled, eliminating another specialized mbuf type in the process. Also clean up some of the cruft surrounding IPFW, multicast routing, RSVP, and other ill-explored corners.
1998-08-23 03:07:17 +00:00
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
}
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
/* XXX; prepare mbuf for (__FreeBSD__ < 3) routines. */
int
soopt_getm(struct sockopt *sopt, struct mbuf **mp)
{
struct mbuf *m, *m_prev;
int sopt_size = sopt->sopt_valsize;
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
MGET(m, sopt->sopt_td ? M_WAIT : M_DONTWAIT, MT_DATA);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (m == NULL)
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
return ENOBUFS;
if (sopt_size > MLEN) {
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
MCLGET(m, sopt->sopt_td ? M_WAIT : M_DONTWAIT);
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
if ((m->m_flags & M_EXT) == 0) {
m_free(m);
return ENOBUFS;
}
m->m_len = min(MCLBYTES, sopt_size);
} else {
m->m_len = min(MLEN, sopt_size);
}
sopt_size -= m->m_len;
*mp = m;
m_prev = m;
while (sopt_size) {
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
MGET(m, sopt->sopt_td ? M_WAIT : M_DONTWAIT, MT_DATA);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (m == NULL) {
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
m_freem(*mp);
return ENOBUFS;
}
if (sopt_size > MLEN) {
Replaced the misleading uses of a historical artefact M_TRYWAIT with M_WAIT. Removed dead code that assumed that M_TRYWAIT can return NULL; it's not true since the advent of MBUMA. Reviewed by: arch There are ongoing disputes as to whether we want to switch to directly using UMA flags M_WAITOK/M_NOWAIT for mbuf(9) allocation.
2008-03-25 09:39:02 +00:00
MCLGET(m, sopt->sopt_td != NULL ? M_WAIT :
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
M_DONTWAIT);
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
if ((m->m_flags & M_EXT) == 0) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
m_freem(m);
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
m_freem(*mp);
return ENOBUFS;
}
m->m_len = min(MCLBYTES, sopt_size);
} else {
m->m_len = min(MLEN, sopt_size);
}
sopt_size -= m->m_len;
m_prev->m_next = m;
m_prev = m;
}
Improve consistency of return() style. MFC after: 3 days
2006-02-12 15:00:27 +00:00
return (0);
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
}
/* XXX; copyin sopt data into mbuf chain for (__FreeBSD__ < 3) routines. */
int
soopt_mcopyin(struct sockopt *sopt, struct mbuf *m)
{
struct mbuf *m0 = m;
if (sopt->sopt_val == NULL)
Improve consistency of return() style. MFC after: 3 days
2006-02-12 15:00:27 +00:00
return (0);
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
while (m != NULL && sopt->sopt_valsize >= m->m_len) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (sopt->sopt_td != NULL) {
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
int error;
error = copyin(sopt->sopt_val, mtod(m, char *),
m->m_len);
if (error != 0) {
m_freem(m0);
return(error);
}
} else
bcopy(sopt->sopt_val, mtod(m, char *), m->m_len);
sopt->sopt_valsize -= m->m_len;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
sopt->sopt_val = (char *)sopt->sopt_val + m->m_len;
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
m = m->m_next;
}
if (m != NULL) /* should be allocated enoughly at ip6_sooptmcopyin() */
panic("ip6_sooptmcopyin");
Improve consistency of return() style. MFC after: 3 days
2006-02-12 15:00:27 +00:00
return (0);
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
}
/* XXX; copyout mbuf chain data into soopt for (__FreeBSD__ < 3) routines. */
int
soopt_mcopyout(struct sockopt *sopt, struct mbuf *m)
{
struct mbuf *m0 = m;
size_t valsize = 0;
if (sopt->sopt_val == NULL)
Improve consistency of return() style. MFC after: 3 days
2006-02-12 15:00:27 +00:00
return (0);
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
while (m != NULL && sopt->sopt_valsize >= m->m_len) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (sopt->sopt_td != NULL) {
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
int error;
error = copyout(mtod(m, char *), sopt->sopt_val,
m->m_len);
if (error != 0) {
m_freem(m0);
return(error);
}
} else
bcopy(mtod(m, char *), sopt->sopt_val, m->m_len);
sopt->sopt_valsize -= m->m_len;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
sopt->sopt_val = (char *)sopt->sopt_val + m->m_len;
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
valsize += m->m_len;
m = m->m_next;
}
if (m != NULL) {
/* enough soopt buffer should be given from user-land */
m_freem(m0);
return(EINVAL);
}
sopt->sopt_valsize = valsize;
Improve consistency of return() style. MFC after: 3 days
2006-02-12 15:00:27 +00:00
return (0);
KAME netinet6 basic part(no IPsec,no V6 Multicast Forwarding, no UDP/TCP for IPv6 yet) With this patch, you can assigne IPv6 addr automatically, and can reply to IPv6 ping. Reviewed by: freebsd-arch, cvs-committers Obtained from: KAME project
1999-11-22 02:45:11 +00:00
}
Update various uipc_socket.c comments, and reformat others.
2006-07-23 20:36:04 +00:00
/*
* sohasoutofband(): protocol notifies socket layer of the arrival of new
* out-of-band data, which will then notify socket consumers.
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
sohasoutofband(struct socket *so)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
if (so->so_sigio != NULL)
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
pgsigio(&so->so_sigio, SIGURG, 0);
selwakeuppri(&so->so_rcv.sb_sel, PSOCK);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
The long-awaited mega-massive-network-code- cleanup. Part I. This commit includes the following changes: 1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility glue for them is deleted, and the kernel will panic on boot if any are compiled in. 2) Certain protocol entry points are modified to take a process structure, so they they can easily tell whether or not it is possible to sleep, and also to access credentials. 3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt() call. Protocols should use the process pointer they are now passed. 4) The PF_LOCAL and PF_ROUTE families have been updated to use the new style, as has the `raw' skeleton family. 5) PF_LOCAL sockets now obey the process's umask when creating a socket in the filesystem. As a result, LINT is now broken. I'm hoping that some enterprising hacker with a bit more time will either make the broken bits work (should be easy for netipx) or dike them out.
1997-04-27 20:01:29 +00:00
int
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
sopoll(struct socket *so, int events, struct ucred *active_cred,
struct thread *td)
The long-awaited mega-massive-network-code- cleanup. Part I. This commit includes the following changes: 1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility glue for them is deleted, and the kernel will panic on boot if any are compiled in. 2) Certain protocol entry points are modified to take a process structure, so they they can easily tell whether or not it is possible to sleep, and also to access credentials. 3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt() call. Protocols should use the process pointer they are now passed. 4) The PF_LOCAL and PF_ROUTE families have been updated to use the new style, as has the `raw' skeleton family. 5) PF_LOCAL sockets now obey the process's umask when creating a socket in the filesystem. As a result, LINT is now broken. I'm hoping that some enterprising hacker with a bit more time will either make the broken bits work (should be easy for netipx) or dike them out.
1997-04-27 20:01:29 +00:00
{
soreceive_generic(), and sopoll_generic(). Add new functions sosend(), soreceive(), and sopoll(), which are wrappers for pru_sosend, pru_soreceive, and pru_sopoll, and are now used univerally by socket consumers rather than either directly invoking the old so*() functions or directly invoking the protocol switch method (about an even split prior to this commit). This completes an architectural change that was begun in 1996 to permit protocols to provide substitute implementations, as now used by UDP. Consumers now uniformly invoke sosend(), soreceive(), and sopoll() to perform these operations on sockets -- in particular, distributed file systems and socket system calls. Architectural head nod: sam, gnn, wollman
2006-07-24 15:20:08 +00:00
/* XXXRW: Temporary debugging. */
KASSERT(so->so_proto->pr_usrreqs->pru_sopoll != sopoll,
("sopoll: protocol calls sopoll"));
return (so->so_proto->pr_usrreqs->pru_sopoll(so, events, active_cred,
td));
}
int
sopoll_generic(struct socket *so, int events, struct ucred *active_cred,
struct thread *td)
{
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
int revents = 0;
The long-awaited mega-massive-network-code- cleanup. Part I. This commit includes the following changes: 1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility glue for them is deleted, and the kernel will panic on boot if any are compiled in. 2) Certain protocol entry points are modified to take a process structure, so they they can easily tell whether or not it is possible to sleep, and also to access credentials. 3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt() call. Protocols should use the process pointer they are now passed. 4) The PF_LOCAL and PF_ROUTE families have been updated to use the new style, as has the `raw' skeleton family. 5) PF_LOCAL sockets now obey the process's umask when creating a socket in the filesystem. As a result, LINT is now broken. I'm hoping that some enterprising hacker with a bit more time will either make the broken bits work (should be easy for netipx) or dike them out.
1997-04-27 20:01:29 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(&so->so_snd);
SOCKBUF_LOCK(&so->so_rcv);
Back out my lats commit of locking down a socket, it conflicts with hsu's work. Requested by: hsu
2002-05-31 11:52:35 +00:00
if (events & (POLLIN | POLLRDNORM))
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
if (soreadable(so))
revents |= events & (POLLIN | POLLRDNORM);
The long-awaited mega-massive-network-code- cleanup. Part I. This commit includes the following changes: 1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility glue for them is deleted, and the kernel will panic on boot if any are compiled in. 2) Certain protocol entry points are modified to take a process structure, so they they can easily tell whether or not it is possible to sleep, and also to access credentials. 3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt() call. Protocols should use the process pointer they are now passed. 4) The PF_LOCAL and PF_ROUTE families have been updated to use the new style, as has the `raw' skeleton family. 5) PF_LOCAL sockets now obey the process's umask when creating a socket in the filesystem. As a result, LINT is now broken. I'm hoping that some enterprising hacker with a bit more time will either make the broken bits work (should be easy for netipx) or dike them out.
1997-04-27 20:01:29 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (events & POLLINIGNEOF)
if (so->so_rcv.sb_cc >= so->so_rcv.sb_lowat ||
!TAILQ_EMPTY(&so->so_comp) || so->so_error)
revents |= POLLINIGNEOF;
Back out my lats commit of locking down a socket, it conflicts with hsu's work. Requested by: hsu
2002-05-31 11:52:35 +00:00
if (events & (POLLOUT | POLLWRNORM))
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
if (sowriteable(so))
revents |= events & (POLLOUT | POLLWRNORM);
Back out my lats commit of locking down a socket, it conflicts with hsu's work. Requested by: hsu
2002-05-31 11:52:35 +00:00
if (events & (POLLPRI | POLLRDBAND))
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (so->so_oobmark || (so->so_rcv.sb_state & SBS_RCVATMARK))
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
revents |= events & (POLLPRI | POLLRDBAND);
if (revents == 0) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (events &
(POLLIN | POLLINIGNEOF | POLLPRI | POLLRDNORM |
POLLRDBAND)) {
selrecord(td, &so->so_rcv.sb_sel);
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
so->so_rcv.sb_flags |= SB_SEL;
The long-awaited mega-massive-network-code- cleanup. Part I. This commit includes the following changes: 1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility glue for them is deleted, and the kernel will panic on boot if any are compiled in. 2) Certain protocol entry points are modified to take a process structure, so they they can easily tell whether or not it is possible to sleep, and also to access credentials. 3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt() call. Protocols should use the process pointer they are now passed. 4) The PF_LOCAL and PF_ROUTE families have been updated to use the new style, as has the `raw' skeleton family. 5) PF_LOCAL sockets now obey the process's umask when creating a socket in the filesystem. As a result, LINT is now broken. I'm hoping that some enterprising hacker with a bit more time will either make the broken bits work (should be easy for netipx) or dike them out.
1997-04-27 20:01:29 +00:00
}
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
if (events & (POLLOUT | POLLWRNORM)) {
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
selrecord(td, &so->so_snd.sb_sel);
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
so->so_snd.sb_flags |= SB_SEL;
The long-awaited mega-massive-network-code- cleanup. Part I. This commit includes the following changes: 1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility glue for them is deleted, and the kernel will panic on boot if any are compiled in. 2) Certain protocol entry points are modified to take a process structure, so they they can easily tell whether or not it is possible to sleep, and also to access credentials. 3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt() call. Protocols should use the process pointer they are now passed. 4) The PF_LOCAL and PF_ROUTE families have been updated to use the new style, as has the `raw' skeleton family. 5) PF_LOCAL sockets now obey the process's umask when creating a socket in the filesystem. As a result, LINT is now broken. I'm hoping that some enterprising hacker with a bit more time will either make the broken bits work (should be easy for netipx) or dike them out.
1997-04-27 20:01:29 +00:00
}
}
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
SOCKBUF_UNLOCK(&so->so_snd);
Extend select backend for sockets to work with a poll interface (more detail is passed back and forwards). This mostly came from NetBSD, except that our interfaces have changed a lot and this funciton is in a different part of the kernel. Obtained from: NetBSD
1997-09-14 02:34:14 +00:00
return (revents);
The long-awaited mega-massive-network-code- cleanup. Part I. This commit includes the following changes: 1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility glue for them is deleted, and the kernel will panic on boot if any are compiled in. 2) Certain protocol entry points are modified to take a process structure, so they they can easily tell whether or not it is possible to sleep, and also to access credentials. 3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt() call. Protocols should use the process pointer they are now passed. 4) The PF_LOCAL and PF_ROUTE families have been updated to use the new style, as has the `raw' skeleton family. 5) PF_LOCAL sockets now obey the process's umask when creating a socket in the filesystem. As a result, LINT is now broken. I'm hoping that some enterprising hacker with a bit more time will either make the broken bits work (should be easy for netipx) or dike them out.
1997-04-27 20:01:29 +00:00
}
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
Extend kqueue down to the device layer. Backwards compatible approach suggested by: peter
2001-02-15 16:34:11 +00:00
int
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
soo_kqfilter(struct file *fp, struct knote *kn)
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct socket *so = kn->kn_fp->f_data;
Extend kqueue down to the device layer. Backwards compatible approach suggested by: peter
2001-02-15 16:34:11 +00:00
struct sockbuf *sb;
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
Extend kqueue down to the device layer. Backwards compatible approach suggested by: peter
2001-02-15 16:34:11 +00:00
switch (kn->kn_filter) {
case EVFILT_READ:
if (so->so_options & SO_ACCEPTCONN)
kn->kn_fop = &solisten_filtops;
Back out 1.68.2.26. It was a mis-guided change that was already backed out of HEAD and should not have been MFC'd. This will restore UDP socket functionality, which will correct the recent NFS problems. Submitted by: rwatson
2005-06-09 19:56:38 +00:00
else
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
kn->kn_fop = &soread_filtops;
Extend kqueue down to the device layer. Backwards compatible approach suggested by: peter
2001-02-15 16:34:11 +00:00
sb = &so->so_rcv;
break;
case EVFILT_WRITE:
kn->kn_fop = &sowrite_filtops;
sb = &so->so_snd;
break;
default:
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
return (EINVAL);
Extend kqueue down to the device layer. Backwards compatible approach suggested by: peter
2001-02-15 16:34:11 +00:00
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(sb);
knlist_add(&sb->sb_sel.si_note, kn, 1);
Extend kqueue down to the device layer. Backwards compatible approach suggested by: peter
2001-02-15 16:34:11 +00:00
sb->sb_flags |= SB_KNOTE;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(sb);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
return (0);
}
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
/*
* Some routines that return EOPNOTSUPP for entry points that are not
* supported by a protocol. Fill in as needed.
*/
int
pru_accept_notsupp(struct socket *so, struct sockaddr **nam)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_attach_notsupp(struct socket *so, int proto, struct thread *td)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_bind_notsupp(struct socket *so, struct sockaddr *nam, struct thread *td)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_connect_notsupp(struct socket *so, struct sockaddr *nam, struct thread *td)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_connect2_notsupp(struct socket *so1, struct socket *so2)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_control_notsupp(struct socket *so, u_long cmd, caddr_t data,
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
struct ifnet *ifp, struct thread *td)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_disconnect_notsupp(struct socket *so)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_listen_notsupp(struct socket *so, int backlog, struct thread *td)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_peeraddr_notsupp(struct socket *so, struct sockaddr **nam)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_rcvd_notsupp(struct socket *so, int flags)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_rcvoob_notsupp(struct socket *so, struct mbuf *m, int flags)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_send_notsupp(struct socket *so, int flags, struct mbuf *m,
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
struct sockaddr *addr, struct mbuf *control, struct thread *td)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
/*
Following movement of functions from uipc_socket2.c to uipc_socket.c and uipc_sockbuf.c, clean up and update comments.
2007-03-26 17:05:09 +00:00
* This isn't really a ``null'' operation, but it's the default one and
* doesn't do anything destructive.
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
*/
int
pru_sense_null(struct socket *so, struct stat *sb)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
sb->st_blksize = so->so_snd.sb_hiwat;
return 0;
}
int
pru_shutdown_notsupp(struct socket *so)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_sockaddr_notsupp(struct socket *so, struct sockaddr **nam)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_sosend_notsupp(struct socket *so, struct sockaddr *addr, struct uio *uio,
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
struct mbuf *top, struct mbuf *control, int flags, struct thread *td)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_soreceive_notsupp(struct socket *so, struct sockaddr **paddr,
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
int
pru_sopoll_notsupp(struct socket *so, int events, struct ucred *cred,
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
struct thread *td)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
return EOPNOTSUPP;
}
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
static void
filt_sordetach(struct knote *kn)
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct socket *so = kn->kn_fp->f_data;
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(&so->so_rcv);
knlist_remove(&so->so_rcv.sb_sel.si_note, kn, 1);
if (knlist_empty(&so->so_rcv.sb_sel.si_note))
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
so->so_rcv.sb_flags &= ~SB_KNOTE;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_rcv);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
}
/*ARGSUSED*/
static int
filt_soread(struct knote *kn, long hint)
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct socket *so;
Conditional acquisition of socket buffer mutexes when testing socket buffers with kqueue filters is no longer required: the kqueue framework will guarantee that the mutex is held on entering the filter, either due to a call from the socket code already holding the mutex, or by explicitly acquiring it. This removes the last of the conditional socket locking.
2004-08-24 05:28:18 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
so = kn->kn_fp->f_data;
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
kn->kn_data = so->so_rcv.sb_cc - so->so_rcv.sb_ctl;
if (so->so_rcv.sb_state & SBS_CANTRCVMORE) {
kn->kn_flags |= EV_EOF;
When returning EV_EOF for the socket read/write filters, also return the current socket error in fflags. This may be useful for determining why a connect() request fails. Inspired by: "Jonathan Graehl" <jonathan@graehl.org>
2001-02-24 01:33:12 +00:00
kn->kn_fflags = so->so_error;
Conditional acquisition of socket buffer mutexes when testing socket buffers with kqueue filters is no longer required: the kqueue framework will guarantee that the mutex is held on entering the filter, either due to a call from the socket code already holding the mutex, or by explicitly acquiring it. This removes the last of the conditional socket locking.
2004-08-24 05:28:18 +00:00
return (1);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
} else if (so->so_error) /* temporary udp error */
Conditional acquisition of socket buffer mutexes when testing socket buffers with kqueue filters is no longer required: the kqueue framework will guarantee that the mutex is held on entering the filter, either due to a call from the socket code already holding the mutex, or by explicitly acquiring it. This removes the last of the conditional socket locking.
2004-08-24 05:28:18 +00:00
return (1);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
else if (kn->kn_sfflags & NOTE_LOWAT)
Conditional acquisition of socket buffer mutexes when testing socket buffers with kqueue filters is no longer required: the kqueue framework will guarantee that the mutex is held on entering the filter, either due to a call from the socket code already holding the mutex, or by explicitly acquiring it. This removes the last of the conditional socket locking.
2004-08-24 05:28:18 +00:00
return (kn->kn_data >= kn->kn_sdata);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
else
return (so->so_rcv.sb_cc >= so->so_rcv.sb_lowat);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
}
static void
filt_sowdetach(struct knote *kn)
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct socket *so = kn->kn_fp->f_data;
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_LOCK(&so->so_snd);
knlist_remove(&so->so_snd.sb_sel.si_note, kn, 1);
if (knlist_empty(&so->so_snd.sb_sel.si_note))
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
so->so_snd.sb_flags &= ~SB_KNOTE;
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
SOCKBUF_UNLOCK(&so->so_snd);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
}
/*ARGSUSED*/
static int
filt_sowrite(struct knote *kn, long hint)
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct socket *so;
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
so = kn->kn_fp->f_data;
SOCKBUF_LOCK_ASSERT(&so->so_snd);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
kn->kn_data = sbspace(&so->so_snd);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (so->so_snd.sb_state & SBS_CANTSENDMORE) {
kn->kn_flags |= EV_EOF;
When returning EV_EOF for the socket read/write filters, also return the current socket error in fflags. This may be useful for determining why a connect() request fails. Inspired by: "Jonathan Graehl" <jonathan@graehl.org>
2001-02-24 01:33:12 +00:00
kn->kn_fflags = so->so_error;
Conditional acquisition of socket buffer mutexes when testing socket buffers with kqueue filters is no longer required: the kqueue framework will guarantee that the mutex is held on entering the filter, either due to a call from the socket code already holding the mutex, or by explicitly acquiring it. This removes the last of the conditional socket locking.
2004-08-24 05:28:18 +00:00
return (1);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
} else if (so->so_error) /* temporary udp error */
Conditional acquisition of socket buffer mutexes when testing socket buffers with kqueue filters is no longer required: the kqueue framework will guarantee that the mutex is held on entering the filter, either due to a call from the socket code already holding the mutex, or by explicitly acquiring it. This removes the last of the conditional socket locking.
2004-08-24 05:28:18 +00:00
return (1);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
else if (((so->so_state & SS_ISCONNECTED) == 0) &&
Back out my lats commit of locking down a socket, it conflicts with hsu's work. Requested by: hsu
2002-05-31 11:52:35 +00:00
(so->so_proto->pr_flags & PR_CONNREQUIRED))
Conditional acquisition of socket buffer mutexes when testing socket buffers with kqueue filters is no longer required: the kqueue framework will guarantee that the mutex is held on entering the filter, either due to a call from the socket code already holding the mutex, or by explicitly acquiring it. This removes the last of the conditional socket locking.
2004-08-24 05:28:18 +00:00
return (0);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
else if (kn->kn_sfflags & NOTE_LOWAT)
Conditional acquisition of socket buffer mutexes when testing socket buffers with kqueue filters is no longer required: the kqueue framework will guarantee that the mutex is held on entering the filter, either due to a call from the socket code already holding the mutex, or by explicitly acquiring it. This removes the last of the conditional socket locking.
2004-08-24 05:28:18 +00:00
return (kn->kn_data >= kn->kn_sdata);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
else
return (kn->kn_data >= so->so_snd.sb_lowat);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
}
/*ARGSUSED*/
static int
filt_solisten(struct knote *kn, long hint)
{
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
struct socket *so = kn->kn_fp->f_data;
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
Make sure that sockets undergoing accept filtering are aborted in a LRU fashion when the listen queue fills up. Previously, there was no mechanism to kick out old sockets, leading to an easy DoS of daemons using accept filtering. Reviewed by: alfred MFC after: 3 days
2002-04-26 02:07:46 +00:00
kn->kn_data = so->so_qlen;
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
return (! TAILQ_EMPTY(&so->so_comp));
}
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
int
socheckuid(struct socket *so, uid_t uid)
{
if (so == NULL)
return (EPERM);
if (so->so_cred->cr_uid != uid)
return (EPERM);
return (0);
}
static int
Rename somaxconn_sysctl() to sysctl_somaxconn() so that I will be able to claim that sofoo() functions all accept a socket as their first argument.
2007-02-15 10:11:00 +00:00
sysctl_somaxconn(SYSCTL_HANDLER_ARGS)
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
{
int error;
int val;
val = somaxconn;
Despite several examples in the kernel, the third argument of sysctl_handle_int is not sizeof the int type you want to export. The type must always be an int or an unsigned int. Remove the instances where a sizeof(variable) is passed to stop people accidently cut and pasting these examples. In a few places this was sysctl_handle_int was being used on 64 bit types, which would truncate the value to be exported. In these cases use sysctl_handle_quad to export them and change the format to Q so that sysctl(1) can still print them.
2007-06-04 18:25:08 +00:00
error = sysctl_handle_int(oidp, &val, 0, req);
Drat! Committed from the wrong branch. Restore HEAD to its previous goodness.
2005-06-09 19:59:09 +00:00
if (error || !req->newptr )
return (error);
if (val < 1 || val > USHRT_MAX)
return (EINVAL);
somaxconn = val;
return (0);
}
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
/*
Following movement of functions from uipc_socket2.c to uipc_socket.c and uipc_sockbuf.c, clean up and update comments.
2007-03-26 17:05:09 +00:00
* These functions are used by protocols to notify the socket layer (and its
* consumers) of state changes in the sockets driven by protocol-side events.
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
*/
/*
Following movement of functions from uipc_socket2.c to uipc_socket.c and uipc_sockbuf.c, clean up and update comments.
2007-03-26 17:05:09 +00:00
* Procedures to manipulate state flags of socket and do appropriate wakeups.
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
*
Following movement of functions from uipc_socket2.c to uipc_socket.c and uipc_sockbuf.c, clean up and update comments.
2007-03-26 17:05:09 +00:00
* Normal sequence from the active (originating) side is that
* soisconnecting() is called during processing of connect() call, resulting
* in an eventual call to soisconnected() if/when the connection is
* established. When the connection is torn down soisdisconnecting() is
* called during processing of disconnect() call, and soisdisconnected() is
* called when the connection to the peer is totally severed. The semantics
* of these routines are such that connectionless protocols can call
* soisconnected() and soisdisconnected() only, bypassing the in-progress
* calls when setting up a ``connection'' takes no time.
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
*
Following movement of functions from uipc_socket2.c to uipc_socket.c and uipc_sockbuf.c, clean up and update comments.
2007-03-26 17:05:09 +00:00
* From the passive side, a socket is created with two queues of sockets:
* so_incomp for connections in progress and so_comp for connections already
* made and awaiting user acceptance. As a protocol is preparing incoming
* connections, it creates a socket structure queued on so_incomp by calling
* sonewconn(). When the connection is established, soisconnected() is
* called, and transfers the socket structure to so_comp, making it available
* to accept().
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
*
Following movement of functions from uipc_socket2.c to uipc_socket.c and uipc_sockbuf.c, clean up and update comments.
2007-03-26 17:05:09 +00:00
* If a socket is closed with sockets on either so_incomp or so_comp, these
* sockets are dropped.
*
* If higher-level protocols are implemented in the kernel, the wakeups done
* here will sometimes cause software-interrupt process scheduling.
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
*/
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soisconnecting(struct socket *so)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
SOCK_LOCK(so);
so->so_state &= ~(SS_ISCONNECTED|SS_ISDISCONNECTING);
so->so_state |= SS_ISCONNECTING;
SOCK_UNLOCK(so);
}
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soisconnected(struct socket *so)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
struct socket *head;
ACCEPT_LOCK();
SOCK_LOCK(so);
so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING|SS_ISCONFIRMING);
so->so_state |= SS_ISCONNECTED;
head = so->so_head;
if (head != NULL && (so->so_qstate & SQ_INCOMP)) {
if ((so->so_options & SO_ACCEPTFILTER) == 0) {
SOCK_UNLOCK(so);
TAILQ_REMOVE(&head->so_incomp, so, so_list);
head->so_incqlen--;
so->so_qstate &= ~SQ_INCOMP;
TAILQ_INSERT_TAIL(&head->so_comp, so, so_list);
head->so_qlen++;
so->so_qstate |= SQ_COMP;
ACCEPT_UNLOCK();
sorwakeup(head);
wakeup_one(&head->so_timeo);
} else {
ACCEPT_UNLOCK();
so->so_upcall =
head->so_accf->so_accept_filter->accf_callback;
so->so_upcallarg = head->so_accf->so_accept_filter_arg;
so->so_rcv.sb_flags |= SB_UPCALL;
so->so_options &= ~SO_ACCEPTFILTER;
SOCK_UNLOCK(so);
so->so_upcall(so, so->so_upcallarg, M_DONTWAIT);
}
return;
}
SOCK_UNLOCK(so);
ACCEPT_UNLOCK();
wakeup(&so->so_timeo);
sorwakeup(so);
sowwakeup(so);
}
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soisdisconnecting(struct socket *so)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
/*
* Note: This code assumes that SOCK_LOCK(so) and
* SOCKBUF_LOCK(&so->so_rcv) are the same.
*/
SOCKBUF_LOCK(&so->so_rcv);
so->so_state &= ~SS_ISCONNECTING;
so->so_state |= SS_ISDISCONNECTING;
so->so_rcv.sb_state |= SBS_CANTRCVMORE;
sorwakeup_locked(so);
SOCKBUF_LOCK(&so->so_snd);
so->so_snd.sb_state |= SBS_CANTSENDMORE;
sowwakeup_locked(so);
wakeup(&so->so_timeo);
}
void
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
soisdisconnected(struct socket *so)
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
{
/*
* Note: This code assumes that SOCK_LOCK(so) and
* SOCKBUF_LOCK(&so->so_rcv) are the same.
*/
SOCKBUF_LOCK(&so->so_rcv);
so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);
so->so_state |= SS_ISDISCONNECTED;
so->so_rcv.sb_state |= SBS_CANTRCVMORE;
sorwakeup_locked(so);
SOCKBUF_LOCK(&so->so_snd);
so->so_snd.sb_state |= SBS_CANTSENDMORE;
sbdrop_locked(&so->so_snd, so->so_snd.sb_cc);
sowwakeup_locked(so);
wakeup(&so->so_timeo);
}
/*
* Make a copy of a sockaddr in a malloced buffer of type M_SONAME.
*/
struct sockaddr *
sodupsockaddr(const struct sockaddr *sa, int mflags)
{
struct sockaddr *sa2;
sa2 = malloc(sa->sa_len, M_SONAME, mflags);
if (sa2)
bcopy(sa, sa2, sa->sa_len);
return sa2;
}
/*
Following movement of functions from uipc_socket2.c to uipc_socket.c and uipc_sockbuf.c, clean up and update comments.
2007-03-26 17:05:09 +00:00
* Create an external-format (``xsocket'') structure using the information in
* the kernel-format socket structure pointed to by so. This is done to
* reduce the spew of irrelevant information over this interface, to isolate
* user code from changes in the kernel structure, and potentially to provide
* information-hiding if we decide that some of this information should be
* hidden from users.
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
*/
void
sotoxsocket(struct socket *so, struct xsocket *xso)
{
Generally migrate to ANSI function headers, and remove 'register' use.
2007-05-16 20:41:08 +00:00
Complete removal of uipc_socket2.c by moving the last few functions to other C files: - Move sbcreatecontrol() and sbtoxsockbuf() to uipc_sockbuf.c. While sbcreatecontrol() is really an mbuf allocation routine, it does its work with awareness of the layout of socket buffer memory. - Move pru_*() protocol switch stubs to uipc_socket.c where the non-stub versions of several of these functions live. Likewise, move socket state transition calls (soisconnecting(), etc) to uipc_socket.c. Moveo sodupsockaddr() and sotoxsocket().
2007-03-26 08:59:03 +00:00
xso->xso_len = sizeof *xso;
xso->xso_so = so;
xso->so_type = so->so_type;
xso->so_options = so->so_options;
xso->so_linger = so->so_linger;
xso->so_state = so->so_state;
xso->so_pcb = so->so_pcb;
xso->xso_protocol = so->so_proto->pr_protocol;
xso->xso_family = so->so_proto->pr_domain->dom_family;
xso->so_qlen = so->so_qlen;
xso->so_incqlen = so->so_incqlen;
xso->so_qlimit = so->so_qlimit;
xso->so_timeo = so->so_timeo;
xso->so_error = so->so_error;
xso->so_pgid = so->so_sigio ? so->so_sigio->sio_pgid : 0;
xso->so_oobmark = so->so_oobmark;
sbtoxsockbuf(&so->so_snd, &xso->so_snd);
sbtoxsockbuf(&so->so_rcv, &xso->so_rcv);
xso->so_uid = so->so_cred->cr_uid;
}
Add accessor functions for socket fields. MFC after: 1 week
2008-07-21 00:49:34 +00:00
/*
* Socket accessor functions to provide external consumers with
* a safe interface to socket state
*
*/
void
so_listeners_apply_all(struct socket *so, void (*func)(struct socket *, void *), void *arg)
{
TAILQ_FOREACH(so, &so->so_comp, so_list)
func(so, arg);
}
struct sockbuf *
so_sockbuf_rcv(struct socket *so)
{
return (&so->so_rcv);
}
struct sockbuf *
so_sockbuf_snd(struct socket *so)
{
return (&so->so_snd);
}
int
so_state_get(const struct socket *so)
{
return (so->so_state);
}
void
so_state_set(struct socket *so, int val)
{
so->so_state = val;
}
int
so_options_get(const struct socket *so)
{
return (so->so_options);
}
void
so_options_set(struct socket *so, int val)
{
so->so_options = val;
}
int
so_error_get(const struct socket *so)
{
return (so->so_error);
}
void
so_error_set(struct socket *so, int val)
{
so->so_error = val;
}
int
so_linger_get(const struct socket *so)
{
return (so->so_linger);
}
void
so_linger_set(struct socket *so, int val)
{
so->so_linger = val;
}
struct protosw *
so_protosw_get(const struct socket *so)
{
return (so->so_proto);
}
void
so_protosw_set(struct socket *so, struct protosw *val)
{
so->so_proto = val;
}
void
so_sorwakeup(struct socket *so)
{
sorwakeup(so);
}
void
so_sowwakeup(struct socket *so)
{
sowwakeup(so);
}
void
so_sorwakeup_locked(struct socket *so)
{
sorwakeup_locked(so);
}
void
so_sowwakeup_locked(struct socket *so)
{
sowwakeup_locked(so);
}
void
so_lock(struct socket *so)
{
SOCK_LOCK(so);
}
void
so_unlock(struct socket *so)
{
SOCK_UNLOCK(so);
}
Reference in New Issue Copy Permalink