after demand of two people and no objections:
- included all registered protocols, so it's now a useful reference as well
- renamed title to reflect new content
(deleted word "IP", since all registered internet protocols are included now)
- corrected URL (file isn't hosted by iana.org anymore)
This means that the driver will add/delete routes when it knows it is
up/down, rather than have the generic code belive it is up if configured.
This is probably most useful for serial lines, although many PHY chips
could probably tell us if we're connected to the cable/hub as well.
mode by padding out the ``struct device'' to the maximum
device size.
Bump the ppp version number to indicate the transfer format
change.
This should make MP over tty and udp devices functional again.
two characters of $EDITOR. This allows things like "vim" and "vi -G"
(although nvi would fail...oh well).
- Avoid certain cases where the editor is passed an invalid line number.
Insure that device mappings get MAP_PREFAULT(_PARTIAL) set,
so that 4M page mappings are used when possible.
Reviewed by: Luoqi Chen <luoqi@watermarkgroup.com>
Do discard standard output from the sysctl for approxy_all, and echo
what this sysctl is doing in the usual way. This fix is probably
backwards. We should probably just use the standard sysctl output
in all cases (it needs to have a newline filtered out).
Echo what the sysctls for nfs_reserved_port_only and nfs_access_cache
are doing.
sizeof(ifr->ifr_addr) for the variable length field ifr->ifr_addr.sa_len.
Otherwise the increment will be wrong in certain cases.
Obtained from: Whistle source tree
For the record: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> suggests
SIOCGIFCONF should be dropped in favor of a sysctl mechanism.
sizeof(ifr->ifr_addr) for the variable length field ifr->ifr_addr.sa_len.
Otherwise the increment will be wrong in certain cases.
Obtained from: Whistle source tree
For the record: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> suggests
SIOCGIFCONF should be dropped in favor of a sysctl mechanism.
default.
Despite their name it doesn't keep TCP sessions alive, it kills
them if the other end has gone AWOL. This happens a lot with
clients which use NAT, dynamic IP assignment or which has a 2^32
* 10^-3 seconds upper bound on their uptime.
There is no detectable increase in network trafic because of this:
two minimal TCP packets every two hours for a live TCP connection.
Many servers already enable keepalives themselves.
The host requirements RFC is 10 years old, and doesn't know about
the loosing clients of todays InterNet.
txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge.
nfs_serv.c: don't call nfsm_adj() with negative values, windows clients
could crash servers when doing a readdir of a large directory.
nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin
loop trying to bind(). Don't clobber a mbuf pointer or we get panics
on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf.
nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB.
Obtained from: OpenBSD
corresponding variable `rc_wakeup_started' in rev.1.36 but broken
again in rev.1.37. This bug only caused excessive polling (it gave
NRC activations for each of the SWI handler and the timeout handler
instead of 1 of each).
Moved cdevsw attachment from the driver probe routine to the driver
attach routine.
flag to the kernel to mount a CDROM as the root filesystem. Alternatively,
the boot_cdrom env var can be set.
As Mike Smith noted, "-C is the "wrong" way to do this", but this is
an acceptable stopgap in lieu of a better way.
PR: bin/11884
Reviewed by: msmith@freebsd.org
to either enqueue or free their mbuf chains, but tcp_usr_send() was
dropping them on the floor if the tcpcb/inpcb has been torn down in the
middle of a send/write attempt. This has been responsible for a wide
variety of mbuf leak patterns, ranging from slow gradual leakage to rather
rapid exhaustion. This has been a problem since before 2.2 was branched
and appears to have been fixed in rev 1.16 and lost in 1.23/1.28.
Thanks to Jayanth Vijayaraghavan <jayanth@yahoo-inc.com> for checking
(extensively) into this on a live production 2.2.x system and that it
was the actual cause of the leak and looks like it fixes it. The machine
in question was loosing (from memory) about 150 mbufs per hour under
load and a change similar to this stopped it. (Don't blame Jayanth
for this patch though)
An alternative approach to this would be to recheck SS_CANTSENDMORE etc
inside the splnet() right before calling pru_send() after all the potential
sleeps, interrupts and delays have happened. However, this would mean
exposing knowledge of the tcp stack's reset handling and removal of the
pcb to the generic code. There are other things that call pru_send()
directly though.
Problem originally noted by: John Plevyak <jplevyak@inktomi.com>
