in stable/4. An MFC of slightly later code than that of r144411 (scottl).
This is a direct commit to stable/4.
Obtained from: Erich Chen <erich at areca com tw>
This is a direct commit to stable/4 since lnc(4) was deprecated
and does not exist beyond this branch.
Submitted by: Paul Meng <[email protected]>
PR: kern/77982
commas due to our deconstruction of the line in cf/feature/dnsbl.m4.
PR: 120038
Submitted by: mattijs vreeling
Revision Changes Path
1.36 +1 -1 src/etc/sendmail/freebsd.mc
MFC: You can never have enough documentation.
Document the SENDMAIL_ALIASES, SENDMAIL_MAP_SRC, SENDMAIL_MAP_TYPE, and
SENDMAIL_START_SCRIPT
.Xr make.conf 5
variables.
These are used in
.Pa /etc/mail/Makefile .
PR: conf/40548
Revision Changes Path
1.140 +40 -0 src/share/man/man5/make.conf.5
update. Unfortunately, this problem (the need to recompile filters) was
found after the MFC. The release engineering team has asked for this
instead of doing a full backout of sendmail 8.14 in the RELENG_[456]
branches.
Requested by: re (bmah)
read per call to bge_start()).
In packet blasting tests using ttcp with tiny udp packets on an A64-3200
with a 64-bit 5701 on a 32-bit 33MHz PCI bus, this gives a speedup
from 347 kpps to 623 kpps. sendto() has a lot of software overheads,
but even with these the single PCI write per call to bge_start() almost
doubled the per-packet time. This is partly because the software
overheads are so large that the CPU can't keep up with a Gbps NIC that
can actually get anywhere near Gbps speed for tiny packets (347 kppps
for tiny packets is only about 21% of wire speed). When the CPU can't
keep up, it gets further behind because it ends up calling bge_start()
at least once for every packet, so any overheads in bge_start() are
not amortized across multiple packets. Thus the PCI read had an
especially high overhead.
For larger packets, the speedup is closer to the 1.8% claimed in rev.
1.108.
Rev.1.108 claims to eliminate a PCI write but actually eliminates a
PCI read. The write of the tx product index is not so costly as its
read, and cannot be eliminated completely. It could be coalesced in
some cases.
- Fix probe for 5708S adapters.
- Use LIST_FOREACH() in bce_ifmedia_upd().
- Compile fixes for BCE_DEBUG.
- Don't reset the chip in bce_watchdog() if we are in the routine because
of pause frames.
- Fix multicast hash.
feature (should safe a heart attack or two when seeing the security feature
enabled when it isn't / shouldn't be).
This bug is not present in any more recent RELENG_X (the code is
different by an order of magnitude), so no MFC.
- Add support for the 5752 and 5780 phys.
- Handle 10/100 only "bge" phys correctly.
- MFC some parts of 1.6 that among other things fix use of non-autoneg
media with bce(4).
- Use bus_dmamap_load_mbuf_sg() rather than a private simple callback in
conjuction with bus_dmamap_load_mbuf().
- Use bus_alloc_resource_any().
- Remove bogus BUS_DMA_ALLOCNOW flag.
group to read kernel memory. [1]
Disable handling of GNUTYPE_NAMES tar file entries by default,
since they can be used to extract files outside of the cwd. [2]
Security: FreeBSD-SA-06:25.kmem [1]
Security: FreeBSD-SA-06:26.gtar [2]
to talk to them) if they aren't listed as the destination of any interrupts
in the MP Table instead of panic'ing when we fail to set the APIC ID. The
new APIC code in 5.x+ has a similar check.
compiled by RELENG_6's compiler:
1.26: pass constraints as a parameter since chars and shorts need
different ones.
1.27: fix pedantic syntax errors.
1.40: don't use the '+' constraint so much, since gcc-3.3.mumble broke
it for non-register operands.
1.40 essentially backs outs 1.22, and also avoids using the "0"
constraint, and also remumbers asm parameters. 1.22 had already been
merged to RELENG_4 in 1.9.2.2 in order to fix compiling with -O0 with
RELENG_4's compiler. This turned out to be not future-proof. I think
the "0" constaint would work with gcc-3, but not using it is a better
way to fix compiling with -O0 in RELENG_4. Problems in this area go
back to gcc-2.early or gcc-1 when the "0" and "+" constraints didn't
exist and there were no correct alternatives for them. 1.9.2.1
unnecessarily used an incorrect alternative that happened to work
except for -O0.
This change is essentially null when compiled by RELENG_4's compiler.
For GENERIC, LINT and one other configuration, it changes no object
files except atomic.o compiled with -O0; for the latter it gives
different register allocation and some unused instructions.
readable. gcc-2.8 was last sighted in FreeBSD-3 and wasn't the system
compiler in any version of FreeBSD-4. The next change will improve
portability and might even work without ifdefs for gcc-2.8.
Add missing parentheses to fix a segmentation fault that is easily
reproducable with `jot -s " " 400 1 | column -t'. The bug was present
in the the original CSRG 'column -t' added in 1989.
memory address can be 0 only with certain misconfigurations. We don't
try to detect misconfigurations of the kernel config file, but IIRC the
panic was caused by a hardware misconfiguration like the following:
- a cy PCI card physically present but with a hardware conflict
- then the cy PCI probe is called first but fails due to the conflict
- no cy ISA card present
- the cy ISA probe is called. Normally this uses the nonzero memory
address in the config file and just fails when there is no cy ISA
card present, but when there is a cy PCI card present, the ISA memory
resource is apparently lost. The (non-)presence of the cy ISA card
probably makes no difference here.
Reads worked correctly using the IU format. Writes only supported
signed integers using the CTLTYPE_INT flag. sysctl(1) parses args for
CTLTYPE_INT using (int)strtol() and has no range checking of course,
so read-write of frequencies larger than INT_MAX always blindly corrupts
the frequency to INT_MAX (2G-1).
This keeps writing of TSC frequencies working up to 4GHz.
new PCI IDs, lots of style changes, suspend/resume methods. Things not
included include the INTR_FAST + taskqueue changes, link flapping
workarounds in em_ioctl(), and jumbo frame fixups for architectures that
require strict alignment.
created this file way back in April 2001 (on all other active branches
the section containing this tag has been obsoleted so it was only a
RELENG_4 problem).
I have no idea how this file ever compiled before, but my doc
toolchain suddenly started complaining about it.
CCB instead of using BUS_DMA_NOWAIT(). Since bus_dmamap_load() doesn't
support BUS_DMA_NOWAIT in 4.x, this could result in data corruption if
sufficient bounce pages weren't available.
Submitted by: scottl
- Removed updates to if_ibytes, if_obytes, if_imcasts, and if_omcasts.
- Match code in 6.x and HEAD for handling excessively fragmented mbufs for
TX.
- Handle ifconfig up when HW is down.
in the MP Table, use the slot (aka device) of the _child_ PCI device
on the child bus of the PCI-PCI bridge instead of the slot of the
PCI-PCI bridge on the parent bus of the bridge.
Correctly identify the user running opiepasswd(1) when the login name
differs from the account name. [2]
Modify timeout handling logic in sendmail(8) to correct a reported
signal handling race condition. [3]
Security: FreeBSD-SA-06:11.ipsec [1]
Security: FreeBSD-SA-06:12.opie [2]
Security: FreeBSD-SA-06:13.sendmail [3]
this change restores is wrong, it turns out the "proper" method is worse
in practice for existing machines. This is probably due to the fact that
using fixed priority with a physical broadcast destination effectively
broadcasts interrupts to all CPUs. Fixing that would require a lot more
work than just restoring the previous "incorrect" behavior and would
probably be riskier as well.
PR: kern/94160
Reported by: Richard Wiwatowski rjwiwat at internode dot on dot net
too late in the boot process - "cold" is guaranteed to be false at
SI_SUB_INT_CONFIG_HOOKS, but it was being KASSERTed as true. With
usb_cold_explore() now running at the correct time, it does need
to use polling mode and splusb(), so back out 1.26.2.14.
PR: usb/81524
Tested by: Eugene Grosbein
contains incorrect fractional second values (outside the range
0-999999).
Prior to this change users could create files with values outside
that range. Moreover, on 32-bit machines tv_usec offsets larger than
4.3s would result in an unnormalized AND wrong timestamp value,
due to overflow.
The addition of a locking assertion to vm_page_zero_invalid() has
revealed a long-time bug: vm_pager_get_pages() assumes that m[reqpage]
contains a valid page upon return from pgo_getpages(). In the case of
the device pager this page has been freed and replaced by a fake page.
The fake page is properly inserted into the vm object but m[reqpage] is
left pointing to a freed page. For now, update m[reqpage] to point to
the fake page.
Submitted by: tegge
Approved by: alc, rwatson (mentor)
Correct insecure temporary file usage in ee. [06:02]
Correct a race condition when setting file permissions, sanitize file
names by default, and fix a buffer overflow when handling files
larger than 4GB in cpio. [06:03]
Security: FreeBSD-SA-06:01.texindex
Security: FreeBSD-SA-06:02.ee
Security: FreeBSD-SA-06:03.cpio
revision 1.67
It seems ciss should ignore overrun and underrun on a SCSI INQUIRY
command. This fixes some weird booting issues on newer versions
of the firmware on the MSA20.
revision 1.66
Update PCI ids to add the E200, E200i, P400, and P400i storage
controllers. Remove the E400 since it is not a real product.
revision 1.65
There's no reason to check the valence. This allows ciss to work
on the P600.
Bug fix: a numeric flag specification in the substitute command would
cause the next substitute flag to be ignored.
While working at it, detect and report overflows.
policy routed by ipfw(4) to local IP address, and the previous destination
of the packet was reachable via an interface with checksum offloading
capabilities.
Not applicable to RELENG_5, RELENG_6 and HEAD.
Submitted by: Mihail Balikov <misho interbgc.com>
This includes the recent change to improve handling of the -P flag
on read-only files, addition of the -I option for better handling
of recursive deletes, a test for removal of / in the argument list,
removal of clause 3 from the BSD license, some small code cleanups,
and related mdoc and comment changes and improvements.
The Makefile change removes a spurious SRCS=, and WARNS?=
the RDONLY option, so subsequent call of UFS_TRUNCATE (ffs_truncate)
would not panic the system. This fixes a panic that can happen
when mounting a corrputed filesystem read-only, and reading data
from it.
Reviewed by: mckusick
date: 2005/09/24 21:41:47; author: krion; state: Exp; lines: +1 -1
Do not bitch about bsdpan modules not having origin.
Based on PR: bin/82269
Submitted by: Steven Hartland <killing at multiplay dot co dot uk>
MFC after: 3 days
date: 2005/09/24 19:41:56; author: krion; state: Exp; lines: +1 -1
Remove '\n' since it's redundant in case if file doesn't exist while
running 'pkg_info -g'
Based on PR: bin/42609
Submitted by: Jeff King <peff-freebsd at peff dot net>
MFC after: 3 days
HEAD to RELENG_4:
Add a new rc.conf entry, kerberos5_server_flags, which allows the
administrator to specify additional start-up flags to the Kerberos
5 Authentication Server.
Don't free a struct inodedep if another process is allocating saved
inode memory for the same struct inodedep in
initiate_write_inodeblock().
Handle disappearing dependencies in softdep_disk_io_initiation().
Delay freeing disk space for file system blocks until all
dirty buffers are safely released. This fixes softdep
problems on truncation (deletion) of files with dirty
buffers.
Fix bogus check. It was possible to panic the kernel by giving 0 length.
This is actually a local DoS, as every user can use /dev/crypto if there
is crypto hardware in the system and cryptodev.ko is loaded (or compiled
into the kernel).
Reported by: Mike Tancsa <[email protected]>
Tested by: Mike Tancsa <[email protected]>
Security: Local DoS on systems with crypto HW and with /dev/crypto available
Interpret correctly the glob(3) return value.
Previously, interactive commands specifying a non-existent file or
directory used to display an "out of memory error".
Allow PLAY_MSF, PLAY_TRACK, PLAY_TRACK_REL, PAUSE, PLAY_12 commands to pass
through umass(4), in order to make cdcontrol(1) to issue commands to a USB
CD driver.
The command IDs were obtained from the CAM subsystem. This was tested
on half dozen of USB CD drivers from different vendors.
Suggested by: "intron" <intron at intron dot ac>
PR: usb/83439
Reviewed by: sanpei
date: 2005/07/13 10:40:07; author: delphij; state: Exp; lines: +7 -4
Better memory handling:
- It is acceptable to call free(3) when the given pointer itself
is NULL, so we do not need to determine NULL before passing
a pointer to free(3)
- Handle failure of malloc(3)
Submitted by: Dan Lukes <dan at obluda cz>
PR: bin/83352
Approved by: re (scottl)
Fix a bug when shell expansion is done against dangling symlinks, by
converting the stat() call to a lstat() call, which will cover the
situation. One can exercise this bug by referring a dangling link with
something like */the-link.
Submitted by: Simon 'corecode' Schubert [corecode fs ei tum de]
Obtained from: NetBSD via DragonFlyBSD (NetBSD rev. 1.51 and DragonFly
rev. 1.6)
instead of attempting the reverse, because the ftp.1 target is executed
before the cdrom.1 target, which is optional. This has been broken
ever since the order of the targets was swapped and cdrom.1 was made
optional, which happened a very long time ago.
Obtained from: Redhat, Steve Grubb via RedHat
Security: CAN-2005-0953, CAN-2005-1260
Security: FreeBSD-SA-05:14.bzip2
Approved by: obrien
Correct TCP connection stall denial-of-service vulnerabilities.
MFC: rev 1.270 of tcp_input.c, rev 1.25 of tcp_seq.h by ps: When a TCP
packets containing a timestamp is received, inadequate checking of
sequence numbers is performed, allowing an attacker to artificially
increase the internal "recent" timestamp for a connection.
A TCP packets with the SYN flag set is accepted for established
connections, allowing an attacker to overwrite certain TCP options.
Security: CAN-2005-0356, CAN-2005-2068
Security: FreeBSD-SA-05:15.tcp
Approved by: cperciva
- amr should now work reliably on systems with more than 4GB of ram.
- Assign a static slot number to each command instead of doing a
linear search for free slots each time a command is needed.
- Fixes for the ioctl path.
A good portition of these changes were submitted by
David Sze <dsze at alumni dot uwaterloo dot ca>
MFC:
revision 1.6
date: 2004/12/28 00:07:56; author: rik; state: Exp; lines: +1 -1
Add FR support to sppp (MFCronyx).
Silence on: net@, current@, hackers@.
No objections: joerg
Requested by: by many (mostly Cronyx) users for a long long time.
MFC after: 10 days
PR: kern/21771, kern/66348, kern/14848
MFC:
revision 1.118
date: 2005/01/07 01:45:34; author: imp; state: Exp; lines: +2 -1
/* -> /*- for license, minor formatting changes
----------------------------
revision 1.117
date: 2004/12/28 00:07:56; author: rik; state: Exp; lines: +38 -12
Add FR support to sppp (MFCronyx).
Silence on: net@, current@, hackers@.
No objections: joerg
Requested by: by many (mostly Cronyx) users for a long long time.
MFC after: 10 days
----------------------------
revision 1.116
date: 2004/12/14 18:18:54; author: rik; state: Exp; lines: +0 -2
Kill double inclusion for <netinet/in.h> and <netinet/in_systm.h>.
----------------------------
revision 1.111
date: 2004/06/25 10:25:33; author: rik; state: Exp; lines: +2 -1
Do not count loobacks as other fuilures.
As a result magic will not be rejected any more in case of loopback.
Discussed with: joerg@
----------------------------
revision 1.108
date: 2004/05/25 21:54:07; author: rik; state: Exp; lines: +2 -2
Keepalive timer should be added if we does not have any sppp consumers
before and should be deleted if we do not have any anymore.
----------------------------
revision 1.107
date: 2004/03/14 01:32:44; author: rwatson; state: Exp; lines: +2 -2
Compare spppq to NULL instead of using spppq as a boolean.
PR: kern/21771, kern/66348, kern/14848
MFC:
revision 1.976
date: 2004/12/28 00:07:56; author: rik; state: Exp; lines: +2 -0
Add FR support to sppp (MFCronyx).
Silence on: net@, current@, hackers@.
No objections: joerg
Requested by: by many (mostly Cronyx) users for a long long time.
MFC after: 10 days
PR: kern/21771, kern/66348, kern/14848
passed to kernel_sysctl() to fix an unaligned exception on Alpha. This
is not an MFC as the code in 5.x is very different.
Reported by: Eric Millbrandt emillbrandt at coldhaus dot com
Move couple of directories out of mtree and into their respective
ports. This mtree now specifies basic structure of X11BASE, similarly
to BSD.local.dist.
- Add missing manual page links for sysctl_add_oid(9).
The rman(9) parts of 1.212 are not MFC'ed since 4.X does not have
rman(9).
PR: docs/60088
Submitted by: le
date: 2005/05/24 08:26:44; author: cperciva; state: Exp; lines: +1 -0
Add an -I option to pkg_version:
-I Use only the index file for determining if a package is out of
date. This is much faster than using the version number from a
port's Makefile, at the expense of potentially giving an incor-
rect result if the index file is out of date.
Without this option, the index file is only consulted if "make -V PKGNAME"
cannot be used.
Belatedly document the disabling of Hyper-Threading. Security issues
are not normally documented in UPDATING on the stable branches, but
since this is a user-visible change, it needs to be here.
This is not an MFC since Hyper-Threading is enabled by default in HEAD.
Pointed out by: Erik Trulsson, via freebsd-stable
Revision Changes Path
1.342.2.30 +5 -0 src/UPDATING
make sure that the socket is either accepting connections or is connected
when attaching a knote to it... otherwise return EINVAL...
Suggested by: dclark at applmath.scu.edu
This code is only used when rlogin is compiled with Kerberos 4
support, and since the Kerberos code in rlogin has been removed from
FreeBSD 5.0 and newer, this is not an MFC.
It is not believed that this bug could be exploited as a security
vulnerability.
PR: bin/80661
Submitted by: Przemyslaw Frasunek <[email protected]>
Approved by: cperciva
- Changed from using explicit devices id to using descriptive labels.
- Added support for 82573 and 82546 Quad adapters.
- Corrected support for 82547EI and 82541ER (mac_type was not assigned)
- Removed #ifdef DBG_STATS and extraneous code.
- Corrected incorrect Packet Buffer assignment for 82541.
- Cleaned up a few typos and style nits.
PR: None
- Changed from using explicit devices id to using descriptive labels.
- Added support for 82573 and 82546 Quad adapters.
- Corrected support for 82547EI and 82541ER (mac_type was not assigned)
- Removed #ifdef DBG_STATS and extraneous code.
- Corrected incorrect Packet Buffer assignment for 82541.
- Cleaned up a few typos and style nits.
PR: None
problems for others. Rest assured you haven't seen the end of these..
As soon as I get back to a test environment I can use for finding the
problem I'm try again.
from getnameinfo(3). This improves API compatibility with other POSIX
complaint operating systems.
PR: standards/80008
Requested by: Xin Liu <[email protected]>
Reviewed by: freebsd-standards (das)
1. Copy a NULL-terminated string into a fixed-length buffer, and
2. copyout that buffer to userland,
we really ought to
0. Zero the entire buffer
first.
Security: FreeBSD-SA-05:08.kmem
- replace use of a spinlock with a mutex.
- enclose getservby{name,port}(3) call with mutex lock.
- protect _hostconf_init_done and _hostconf_init() call with mutex lock.
lib/libc/net/getaddrinfo.c: 1.41
lib/libc/net/name6.c: 1.33
lib/libc/net/getnameinfo.c: 1.14.6.1, 1.14.6.2
usable. In particular:
* disable unix password authentication (ie: do kerberos only)
* install k5su setuid by default
* allow instances for things like ksu to the www account instead
of only doing this for root.
This stuff is only enabled by make.conf switches.
Also bring in the code that probves the non EHCI controllers
after the EHCI controllers, thus ensuring that high speed
devices are not pingponged all aver theplace during startup.
the ``ip'' man page.
o put the word ``raw'' somewhere in
the title of the ``ip'' man page.
PR: docs/30873
Submitted by: Douglas De Couto <[email protected]>
Approved by: trhodes (mentor)
Correct multiple security related errors: a buffer overflow, NULL
pointer dereferences, possible use of uninitialized variables, and
memory leaks.
Security: CAN-2005-0753
Security: FreeBSD-SA-05:05.cvs
Approved by: peter
This gets the HW to work but not to deal with SATA errors. So if
the drive has a read error it will probably lock up. This is
a small subset of my patches that do deal with these errors and
other issues in FreeBSD and ata-raid.
Sponsored by: IronPort Systems
- Return POLLHUP rather than POLLIN/POLLOUT on TS_ZOMBIE ttys
- Remove unneeded POLLWRNORM (old bug)
- TS_ZOMBIE ttys will set POLLIN and POLLRDNORM
- Do not call selrecord in TS_ZOMBIE ttys
PR: kern/73821
that we can tsleep on. This bug is not present in 5.x since we have a
real thread context there. Suggested by jhb.
Submitted by: Dinesh Nair (dinesh alphaque.com)
Change the PATH for pkg_add to be the same used for pointyhat package builds.
This will give our users and porters a more consistent experience between
installing ports and installing packages.
revision 1.224
date: 2005/02/27 15:12:50; author: mux; state: Exp; lines: +10 -10
Fix a stupid bogon from myself, sc->revision wasn't initialized when
testing it to know whether we should enable the 82503 serial mode...
Move code to the right location and disallow the use of the 82503
serial mode if the sc->revision field is 0 again. This makes fxp(4)
work correctly with ATMEL 350 93C46 cards (3 port 82559 based with a
82555 PHY), as well as with the older ATMEL 220 93C46 (same flavour)
and with the even older 10Mbps-only 82557 cards with the 82503 serial
interface.
Tested by: Andre Albsmeier <[email protected]>, krion
Fix a problem in the Skinny ALG where a specially crafted packet could cause
a libalias application (e.g. natd, ppp, etc.) to crash. Note: Skinny support
is not enabled in natd or ppp by default.
Approved by: nectar (secteam)
Security: This fixes a remote DoS exploit
- In constructor(), don't initialize if_addrhead; if_attach()
will do it for us.
- In rcvdata(), don't try to emulate m_adj() and just let
ether_input() do its job.
- Add support for node removal (a-la hardware detach).
o ng_eiface.c#1.3
Allow a changed MAC address to show up in ifconfig by changing
it in the ifaddr list as well.
o ng_eiface.c#1.10
Check the correct set of interface flags and fix a memory leak.
o ng_eiface.c#1.12, ng_eiface.h#1.6
Use the (now standard) Ethernet address parse type.
u_int to the IPX address structure, which introduced strong alignment
requirements for the structure, which in turn caused problems for MARS.
Reported by: "Igor B. Bykhalo" <goshik at binep dot ac dot ru>
Casting the result of the 64 bit division to 32 bits (thus discarding the
upper 32 bits) and then looking at the truncated result to try and figure
out if the untruncated result would fit in 32 bits was utterly useless.
I am still not sure that it is right, but it has a chance of working now.
I'm not at all sure about the sign handling. NFSv3 only reports positive
values here, but correctness of handling the 63/64 bit signs on nfs
volumes is not a problem we'll likely have to deal with for some time. I
think the "most correct" test is for an unsigned division testing for
exceeding LONG_MAX, since we should never end up with a negative number to
compare against LONG_MIN.
in the kernel NFS code, df itself was broken. It tries to re-factor the
statfs blocks / blocksize into whatever the user requested. Thus, it must
be able to report >2TB file systems correctly, which means it must deal
with 64 bit block counts.
date: 2002/09/07 04:59:49; author: peter; state: Exp; lines: +1 -0
Supposedly linux has added a 6th syscall arg register (%ebp). I am not
100% sure if this is enough, but it will not harm anything.
This appears to fix some breakage with the emulation of linux's mmap2().
When that function became available recently the linux executables started
using it which broke installing the emulators/linux_base-8 port.
Pointer from: akiyama
MFS after: 1 day
The tcp syncache code was leaving the IPv6 flowlabel uninitialised
for the SYN|ACK packet and then letting in6_pcbconnect set the
flowlabel later. Arange for the syncache/syncookie code to set and
recall the flow label so that the flowlabel used for the SYN|ACK
is consistent. This is done by using some of the cookie (when tcp
cookies are enabeled) and by stashing the flowlabel in syncache.
commits, but libdisk is just plain lame. Being required to teach it about
each disk device name is part of the magic folklore that is just plain
wrong. This is one of the many many many reasons why libdisk & sysinstall
need to be thrown away and rebuilt from scratch. No amount of tweaking
and fixing them will make them better.
sources from HEAD, as opposed to a forward port of the original IBM 4.x
sources. As such, it benefits from the many reliability improvements
that have happened to the driver in HEAD. This is a RELENG_4_11 candidate.
became necessary due to the plans for using linux_base-8 for the 4.11
release. Among other things running "/compat/linux/bin/ls -l" dumped
core because it tried to use completely non-existent system calls. We
don't actually implement the system calls it's trying to use but we at
least need to have something in place that returns an error code.
Things being added:
- wrapper for mmap2(2) from
sys/compat/linux/linux_machdep.c v1.31
sys/compat/linux/linux_dummy.c v1.35
- add dummy entries for fadvise64(2) and *xattr(2) family of syscalls from
sys/compat/linux/linux_dummy.c v1.37
- add ENOATTR error code because that is what the *xattr(2) family of
syscalls returns, which in turn required tweaking the error code
translation table in the NFS server code[1].
- add $FreeBSD$ token to lib/libc/gen/errlst.c so it passes pre-commit
check script...
[1] Noticed by rwatson.
Requested by: nectar
Reviewed by: nectar, sobomax, re@
Tested by: nectar, ceri
MFS after: 2 days
date: 2005/01/04 19:30:14; author: mux; state: Exp; lines: +1 -1
Allow fxp(4) cards with a revision id of 0 to use the 82503 serial
interface as well. This is not an expected revision id per the
datasheet, but unfortunately there are such cards out there with
a 82557 chipset, and they want to use the 82503.
PR: kern/75739
Reported by: Andre Albsmeier <[email protected]>
- some SATA controllers were erroneously marked as SCSI, correct that
- move those controllers into the right place
- mention that the driver supports SATA
- bump .Dd
v1.50 of ata-lowlevel.c. From his log message:
Log:
Compensate for off by one bugs in disk firmware for 48BIT addressing cutover.
Requested by: marck
Glanced at by: sos
MFC after: 2 days
got to the point there wasn't enough space to put much more than the
basic gnome2-lite/kde-lite packages on disc1 which was painful. So
for 4.11 we're going to make a disc1-gnome and disc1-kde which allows
us to put back a bunch of packages that have been slowly stripped out
of disc1 due to size issues. Set the environment variable PKG_WINDOW
to select which base windowing system to include, default is gnome.
MFC after: 2 days
list, skip blocks that are too big by a factor of two or greater.
This avoids some cases of extremely inefficient memory use that can
occur when large (e.g. 64k) blocks on the free list get used when
allocating a 4k chunk of 64-byte fragments. Because fragments have
their own free list, the 60k difference got lost forever every time.
Replace the static "qdat" structure with a per-instance softc field
in all USB ethernet drivers. The qdat structure contains a pointer
to the interface's struct ifnet and is used to process incoming
packets, so simultaneous use of two similar devices caused crashes
and confusion.
PR: kern/59290
change wasn't included:
- Set the CAM status to CAM_SCSI_STATUS_ERROR rather than CAM_REQ_CMP
in case of a CHECK CONDITION.
Noticed by: Uwe Doering <[email protected]>
sizing for 88E8001 parts. Fix off-by-one in VPD data read. Correct handling
of jumbo frames for Yukon based devices.
PR: kern/75460
Committed on behalf of: bz
- call the driver 'bktr' instead of 'brooktree' throughout the document
- add a HARDWARE section
- add a more descriptive NAME section
- update for recently imported Pinnacle PCTV support
- add Branko Lankester, the original author of the PCTV patches,
to the AUTHORS section
- bump .Dd
Fix a bug where an aborted command was removed from the
queue but not from the hardware's cache of it, leading to the
hardware DMAing into "freed" memory buffers. (leading to crashes when
a usb ransaction was timed, but but eventually happenned)
date: 2004/07/28 18:23:08; author: alc; state: Exp; lines: +2 -2
Correct a very old error in both vm_object_madvise() (originating in
vm/vm_object.c revision 1.88) and vm_object_sync() (originating in
vm/vm_map.c revision 1.36): When descending a chain of backing objects,
both use the wrong object's backing offset. Consequently, both may
operate on the wrong pages.
Quoting Matt, "This could be responsible for all of the sporatic madvise
oddness that has been reported over the years."
Reviewed by: Matt Dillon
Reviewed by: alc
Helped by: DFly vm_map.c rev 1.31, vm_object.c rev 1.17
Slightly reorganise part of the ohci_softintr() function
to better keep track of the total amoutn transferred during a
transfer. Seems similar to some code in the NetBSD version.
I notice they have incorporated matches from him so I don't know which
direction it went.
Submitted by: [email protected]
Obtained from: patches to make the ueagle driver work
The change in current had a real bugfix but when I merged it in to 4.x
it disappeared and only the formatting changes remained..
Oh well it's a diff reduction..
to quote Matt:
Fix a very serious bug in contigmalloc() which we inherited from FreeBSD-4.x.
The contigmalloc() code incorrectly assumes that a page in PQ_CACHE can
be reused without having to do any further checks and it unconditionally
busies and frees such pages, and assumes that the page becomes PQ_FREE even
though it might actually have gone to a PQ_HOLD state. Additionally the
contigmalloc() code unconditionally sets m->object to NULL, ignoring the
fact that the page will be in the VM page bucket hash table if object
happens to not be NULL, leading to page bucket hash table corruption.
The fix is two fold. First, we add checks for m->busy, (m->flags & PG_BUSY),
m->wire_count, and m->hold_count and do not reuse a page with any of those
set. We do this for all pages, not just PQ_CACHE pages, though it is
believed that it only needs to be done for PQ_CACHE pages. Second, we
replace the m->object = NULL assignment with an assertion that it is
already NULL, since it had better be NULL and we cannot just set it to NULL
unconditionally without blowing up the VM page hash table.
Symptoms of the bug include:
* Filesystem corruption, in particular with slower disk drivers (e.g.
like the 'twe' driver), or in systems with drivers which use
contigmalloc() a lot (e.g. require bounce buffers).
Mangled directory entries, bad indirect blocks (containing data instead
of indirect block pointers), and files containing other file's data.
* 'page not found in hash' panic.
This is the last major VM issue in DragonFly, one that has plagued in
particular David Rhodus (who is a heavy user of the 'twe' driver) for over
a year. I would never have found this bug if not for DR's persistence and
the dozens of kernel cores he was able to provide me over the last year. We
finally got a core with a 'smoking gun', after having written a program
(/usr/src/test/debug/vmpageinfo.c) to run through all the VM pages and check
their hash table association for correctness it became obvious that pages
were being reused without being removed from the hash table which finally
led to contigmalloc*().
Many thanks to: David Rhodus! Free gift enclosed!
date: 2004/12/20 10:18:21; author: mux; state: Exp; lines: +2 -2
Only try to use the 82503 serial interface for the 82557 chipsets. The
datasheet says it is only valid for such chipsets and shouldn't be used
with others. This fixes some 82559 based cards which otherwise only
work at 10Mbit.
MFC after: 5 days
Tested by: krion
Never emit a message to stderr: use syslog instead.
When in inetd mode, this prevents bogus messages from
appearing on the control channel. When running as a
daemon, we shouldn't write to the terminal we used to
have at all.
buffer doesn't work for ipv6 packets, so use m_defrag() here instead as an
easy drop-in replacement.
PR: kern/70399
Submitted by: [email protected]
Reviewed by: darrenr
Approved by: re(kensmith)
Obtained from: [email protected]
revision 1.17
date: 2004/12/14 23:13:08; author: rik; state: Exp; lines: +1 -1
Bring back ability to use FR support with sppp from Cronyx.
Approved by: re (kensmith)
current problem of mfsroot.flp overflowing (note to self: can't speed up
test release builds for boot floppy sizes by using NODOC=). Address
original problem by removing sa (SCSI tape) and pass (SCSI passthrough)
devices.
Approved by: re (scottl)
Pointy hat: kensmith
lib/Makefile 1.155
libexec/Makefile 1.57
usr.bin/Makefile 1.234 (partly), 1.235
Build non-crypto telnet(1) and telnetd(8) if NO_OPENSSL is defined.
- Replace the check for the non-existence of src/secure and src/kerberosIV
with a check for the non-existence of src/crypto only. This is to bring
these checks in line with the demise of the NOSECURE knob and because
if src/crypto doesn't exist we won't build any crypto stuff including
not Kerberos IV (and also not Kerberos 5 for that matter).
Note: On HEAD all these checks for the existence of crypto sources even
have been completely removed.
Discussed with: ru (some months ago)
Reviewed by: ru (some months ago)
Approved by: markm (telnet and secure bits, some months ago)
(and therefore must not be used together with NOCRYPT or NO_OPENSSL).
Corresponds to:
share/man/man5/make.conf.5 1.73
share/examples/etc/make.conf 1.216
Reviewed by: ru (some months ago)
Unbreak world build if NO_OPENSSL is defined but MAKE_KERBEROS4 or
MAKE_KERBEROS5 are.
Reviewed by: ru (some months ago)
Approved by: markm (telnet and secure bits, some months ago)
Makefile.inc1 1.361
bin/ed/Makefile 1.22
lib/Makefile 1.157
lib/libcrypt/Makefile 1.31
lib/libfetch/Makefile 1.41
lib/libpam/modules/modules.inc 1.17
libexec/Makefile 1.61
share/examples/etc/make.conf 1.215
share/man/man5/make.conf.5 1.73
usr.bin/Makefile 1.239
usr.bin/fetch/Makefile 1.14
usr.sbin/pkg_install/Makefile 1.14
usr.sbin/pkg_install/add/Makefile 1.17
usr.sbin/pkg_install/info/Makefile 1.17
usr.sbin/pkg_install/version/Makefile 1.12
usr.sbin/ppp/Makefile 1.95, 1.96
usr.sbin/pppd/Makefile 1.25
usr.sbin/sendmail/Makefile 1.40
usr.sbin/tcpdump/tcpdump/Makefile 1.36
Retire the useless NOSECURE knob.
The motivation for this MFC is that MAKE_KERBEROS4 and MAKE_KERBEROS5
are meant to be buildable by various makefiles when NOSECURE is set but
this was broken because Kerberos depends on e.g. des(3) from OpenSSL
which wasn't built when NOSECURE was defined. Fixing the NOSECURE knob
however would have meant to make it do the same as the NOCRYPT knob.
If you had been using the NOSECURE build knob please now use the NOCRYPT
knob instead.
Discussed with: ru (some months ago)
Reviewed by: ru (some months ago)
Approved by: markm (telnet and secure bits, some months ago), kensmith (re)
supplied by the compiler and LINT is compiled with -no-builtin. Fix this
by defining abs() as a macro.
Another pointy hat for my collection...
Pointed out by: tinderbox, glebius, and ken.
1.42: Add a -k flag for querying whois.krnic.net (the National Internet
Development Agency of Korea) which hold details of IP address
allocations within Korea.
MFC whois.1:
1.33: Add a -k flag for querying whois.krnic.net (the National Internet
Development Agency of Korea) which hold details of IP address
allocations within Korea.
1.34: Bump document date.
Short MFC interval to get this into RELENG_4 before the freeze for
FreeBSD 4.11.
MFC ok'ed by: ceri
1.29: Mark a function as __printflike().
1.30: Leave the `copyright' variable exposed, since the main purpose of
it is to put a copyright in the object file.
1.35: Add support for identifying NORID (norwegian top level registry)
handles.
1.38: Add whois.iana.org to the whois(1) utility, under the -I option.
1.39: Add whois.abuse.net to whois(1).
1.40: Add -b to usage.
1.41: Add special case for the german whois nameserver.
MFC whois.1:
1.29: mdoc(7) police: markup polishing.
1.30: Add whois.iana.org to the whois(1) utility, under the -I option.
1.31: Date bump.
1.32: Add whois.abuse.net to whois(1).
- Set the CAM status to CAM_SCSI_STATUS_ERROR rather than CAM_REQ_CMP
in case of a CHECK CONDITION.
- Make this driver return SCSI status information.
- While here, factor out the clearing of the CAM status from every
element of the switch statement to only once before the switch.
MFC ok'ed by: scottl (some months ago)
for it. The change to struct if_data differs slightly from the one in
5.3 in that a pad is added after ifi_xmitquota rather then adding
ifi_link_state before ifi_recvquota. This is necessicary to maintain
the 4.x ABI.
added axe(4) needs to be here as well.
<quote>
# USB support
# (if you add any USB devices to this list, they must be added to the Alpha
# section of src/release/scripts/dokern.sh)
# USB Ethernet
</quote>
Forgotten by: rsm :-)
----------------------------
revision 1.25
date: 2004/07/07 19:36:12; author: simon; state: Exp; lines: +20 -1
Add a HARDWARE section.
For cx(4) this was based on information from the Hardware Notes.
OK'ed by: rik
----------------------------
revision 1.5
date: 2004/09/05 22:09:53; author: simon; state: Exp; lines: +16 -1
Add a HARDWARE section which lists supported devices based on the
Hardware Notes.
MFC after: 3 days
proxying applications. Modified methods of randomization which should
solve this problem are being worked on, and will hopefully make it in
before 4.11-release.
This commit simply changes the default sysctl setting, it does not remove
or change the randomization feature in any way.
When doing round-robin reads from a multi-plex volume, only switch to the
next plex if the sector to be read isn't nearby the last read sector.
Submitted by: Vsevolod Lobko <[email protected]> via ru@
Not only does this boost multi-access performance, but surprisingly it helps
single process sequential reads on my IDE disk mirror. My guess is that the
read-a-block skip-a-block read-a-block effect of strict round-robin defeats
some sort of on-disk caching, leading to poor performance. With this patch,
longer contiguous reads from a single disk are more common.
/sbin/vinum needs to be rebuilt as this patch changes the size of struct plex.
OK'd by: grog
date: 2004/08/03 02:01:44; author: hsu; state: Exp; lines: +1 -1
Fix bug with tracking the previous element in a list.
Found by: [email protected]
Submitted by: [email protected]
Fix off-by-one errors and potential buffer overruns
in handling file and link names that reach the allowed
maximum for old tar and ustar archive formats.
1) Fix a typo in a comment.
2) Merge macros that abstract locking and memory allocation.
3) Use the right bit of constant data when finishing hashing the name.
----------------------------
revision 1.35
date: 2004/12/03 16:57:15; author: rik; state: Exp; lines: +4 -2
Don not call pp_down()/pp_up() form XX_tlf()/XX_tls() in non PPP mode
to privent running of PPP's state machine in non PPP mode.
MFC: after 3 days.
----------------------------
revision 1.30
date: 2004/08/13 12:35:52; author: rik; state: Exp; lines: +6 -6
branches: 1.30.2;
Fix resource check while autodetection.
----------------------------
revision 1.17
date: 2004/12/03 16:57:15; author: rik; state: Exp; lines: +4 -2
Don not call pp_down()/pp_up() form XX_tlf()/XX_tls() in non PPP mode
to privent running of PPP's state machine in non PPP mode.
MFC: after 3 days.
----------------------------
revision 1.14
date: 2004/08/13 12:35:52; author: rik; state: Exp; lines: +6 -6
branches: 1.14.2;
Fix resource check while autodetection.
----------------------------
revision 1.14
date: 2004/12/03 16:57:15; author: rik; state: Exp; lines: +4 -2
Don not call pp_down()/pp_up() form XX_tlf()/XX_tls() in non PPP mode
to privent running of PPP's state machine in non PPP mode.
MFC: after 3 days.
the version now as a heads-up to users because this week typically means
increased developer activity and the branch is a tiny bit more volatile
than "usual".
tzdata on the 4-STABLE branch. For ongoing maintenance, users should
be aware that compiled zoneinfo formats are compatible across all
platforms and all versions of FreeBSD.
- Since divert protocol is not connection oriented, remove SS_ISCONNECTED flag
from divert sockets.
- Remove div_disconnect() method, since it shouldn't be called now.
- Remove div_abort() method. It was never called directly, since protocol
doesn't have listen queue. It was called only from div_disconnect(),
which is removed now.
- Document the DISABLE_VULNERABILITIES knob.
- Don't say this manual page is too long, it's not that long.
- s/man page/manual page/g
- Bump document date.
properly again. This is not a perfect fix but since this is only for 4.x
the 'not-goodness' of the patch will remain within this branch and should
never propogate.. "good enough for government work" as we say
in Australia.
PR: 69319
Submitted by: scott+freebsd at fishballoon dot org
little easier. Teach gdb how to look for thread info in core files. This
allows for more detailed post-mortem analysis of crashed programs as users
should now be able to use standard gdb commands to list all threads running
in the program at the time of the crash and switch between them to look at
their backtraces and inspect their frames.
Allow gdb to cleanly detach multithreaded programs.
Tested by: jhb
Approved by: obrien
Workaround a problem with Postfix MTA selection code when installing via
FTP by hard-coding the Postfix major and minor revision (for the current
stable version).
Approved by: trhodes
>alc 2004/05/21 21:53:52 PDT
>
> FreeBSD src repository
>
> Modified files:
> sys/vm vm_extern.h vm_fault.c vm_map.c vm_page.c
> Log:
> To date, unwiring a fictitious page has produced a panic. The reason
> being that PHYS_TO_VM_PAGE() returns the wrong vm_page for fictitious
> pages but unwiring uses PHYS_TO_VM_PAGE(). The resulting panic
> reported an unexpected wired count. Rather than attempting to fix
> PHYS_TO_VM_PAGE(), this fix takes advantage of the properties of
> fictitious pages. Specifically, fictitious pages will never be
> completely unwired. Therefore, we can keep a fictitious page's wired
> count forever set to one and thereby avoid the use of
> PHYS_TO_VM_PAGE() when we know that we're working with a fictitious
> page, just not which one.
>
> In collaboration with: green@, tegge@
> PR: kern/29915
Reviewed by: alc
This makes teh two systems disable each other's interrupts
while they are operating together.
It's not perfect as it doesn't fully cover dynamic devices
and teh effect is not reversible, but it is similar to the hack we have
been doing for ywars with slip and ppp.
Submitted by: lost the email.
pass1.c:1.43, pass2.c:1.26, pass3.c:1.15, pass4.c:1.15, pass5.c:1.4
with minor changes to match RELENG_4 code.
> truckman 2004-10-08 20:44:48 UTC
>
> FreeBSD src repository
>
> Modified files:
> sbin/fsck_ffs dir.c fsck.h fsutil.c inode.c main.c
> pass1.c pass2.c pass3.c pass4.c pass5.c
> Log:
> Eliminate linked list used to track inodes with an initial link
> count of zero and instead encode this information in the inode state.
> Pass 4 performed a linear search of this list for each inode in
> the file system, which performs poorly if the list is long.
>
> Reviewed by: sam & keramida (an earlier version of the patch), mckusick
> MFC after: 1 month
>
> Revision Changes Path
> 1.30 +2 -3 src/sbin/fsck_ffs/dir.c
> 1.33 +15 -15 src/sbin/fsck_ffs/fsck.h
> 1.25 +1 -3 src/sbin/fsck_ffs/fsutil.c
> 1.37 +2 -0 src/sbin/fsck_ffs/inode.c
> 1.42 +0 -8 src/sbin/fsck_ffs/main.c
> 1.43 +5 -16 src/sbin/fsck_ffs/pass1.c
> 1.26 +5 -1 src/sbin/fsck_ffs/pass2.c
> 1.15 +3 -3 src/sbin/fsck_ffs/pass3.c
> 1.15 +8 -11 src/sbin/fsck_ffs/pass4.c
> 1.40 +2 -0 src/sbin/fsck_ffs/pass5.c
Tested by: scrappy
of 'configXEnvironment', and add 'configXDesktop'.
Not applicable to 5.x or 6.x, since the X configuration stuff has been
ripped out of sysinstall(8) there.
PR: 62665
Submitted by: makham <[email protected]>
- Don't crash if the CROM is all zeros.
- Use new eui64(3) functions to print EUI-64s and to allow access to nodes
by EUI-64 and name.
- Add -m option, set default fwmem target.
Added support for Intel PRO/1000 GT Desktop Adapter(Device ID 8086 107C)
Removed support for Intel 82541ER
Added fix for 82547 which corrects an issue with Jumbo frames larger than 10k.
Added fix for vlan tagged frames not being properly bridged.
Corrected TBI workaround.
Corrected incorrect LED operation issues
functional and presently they are not back-ported, disable the
build of GB18030 locale by now so we won't get the 13K+ warnings
when doing ``make buildworld''.
Noticed by: ru
Approved by: murray (mentor)
The driver is designed for performing I/O under program control using
peripherals such as the Advantech PCL-724 card, which emulates the Intel
8255A PPI in mode 0. Other 8255A-based peripherals such as the BMCM
PIO24II card have also been reported to work.
PR: i386/46238
Approved by: re (kensmith)
The driver is designed for performing I/O under program control using
peripherals such as the Advantech PCL-724 card, which emulates the Intel
8255A PPI in mode 0. Other 8255A-based peripherals such as the BMC
Messsysteme PIO24II card have also been reported to work.
PR: i386/46238
Approved by: re (kensmith)
(this bug is specific to FreeBSD-4.x, so it's not committed to -CURRENT nor RELENG-5)
Reported in: freebsd-net ("Weird memory exhaustion with FreeBSD 4.10-STABLE" starting from Sep 23, 2004)
Obtained from: KAME
Approved by: re (scottl)
revision 1.10
date: 2003/09/18 17:00:03; author: jhb; state: Exp; lines: +1 -1
Fix a busted constant related to PCI configuration acess method #1. The
reserved bits in the port that must be zero are 24:30, not 20:30. Bits
16:23 are used to set the bus number. This meant that when we tested for
config mechanism #1, if the previous PCI configuration transaction sent
used a bus number greater than 15, one of the bits in 20:23 would be
non-zero and we would fail to use config mechanism #1 and thus fail to see
that PCI existed on the machine at all.
Approved by: jhb
date: 2004/09/15 08:19:03; author: delphij; state: Exp; lines: +7 -1
Clear all "active partition" bits before we set a new one.
This adds a safebelt that prevents users to mark more than
one "active" partitions, which will lead to a unbootable
machine, especially in multi-boot configurations.
Approved by: re (scottl), mentor (murray)
This is a direct commit to RELENG_4, because in 5.x the utility
of this change has been subsumed by ATAng and ports/sysutils/ataidle.
Approved by: re (kensmith)
Original patch by: Robert Sexton
date: 2002/10/06 09:09:27; author: charnier; state: Exp; lines: +11 -16
While removing a memory leak, rev 1.32 introduced a
free-memory-and-reuse-it-after. Correct both problems and
make rcp -r work again under /etc/malloc.conf -> AJ.
PR: bin/54969
Approved by: murray (mentor)
Increase PPTP_MAX_TIMEOUT up to 3 seconds. 10 proved too much for high
packet
loss links, and 1 second appeared to be too small for high latency links.
Apply EVL_VLANOFTAG() to a tag passed to vlan_input_tag() by a
VLAN-supporting Ethernet interface so the irrelevant bits (priority
etc) are cleared. Otherwise frames with non-zero dot1q priority
would not be directed to the right vlan interface if the parent
Ethernet interface didn't mask off the irrelevant bits by itself.
Refuse to configure a bogus VLAN ID on a vlan(4) interface.
When detaching an interface, don't leave an obsolete pointer to the
soon to be deleted struct ifnet around.
PR: kern/52260
Submitted by: Jinmei Tatuya <jinmei at isl.rdc.toshiba.co.jp>
relevant to RELENG_4, including documenting the following configuration
variables: DOC_LANG, DOCRELEASETAG, KERNEL_FLAGS, KERNELS, NOCDROM,
PORTSRELEASETAG, and WORLD_FLAGS.
Open a socket for a data transfer in active mode using euid
of the current user, not root, to allow matching the user's
FTP data traffic with an ipfw(8) rule.
struct ndblock's defn member instead of a duplicate allocated on the heap;
the address defn points to is significant, and is checked against the
address of "null" in certain conditionals.
PR: 59883
1.13 Add Epson Perfection 1670
1.14 Add Canon CanoScan LIDE 20 and 30 (and fix name of other Canon
scanners)
1.15 Update Document Date
Approved by: re (fake line to allow commits on RELENG_4 branch)
Do not attempt to clean up data that has not been initialized yet.
This fixes two kernel panics on boot when the xl driver fails to
allocate bus/port/memory resources.
- MFC of the package tools revision 20040629 changes.
- replace the perl version of pkg_version(1) by its C equivalent,
which means that `pkg_version -c' does no longer work
- keep the perl version of pkg_update(1) until a replacement is available
- include some man pages fixes
modules, allowing in particular /usr to be mounted read-only. This
change won't be MFS'ed, because semantics of installing modules in 5.x
is slightly different from 4.x.
PR: 68150
information being sent to the RADIUS server.
Logging of RADIUS accounting information moves to a ``set log [+-]radius''
level, along with the RADIUS alive info, and the version number is bumped
to 3.2 to reflect this.
Mostly submitted by: [email protected] (back in January)
This commit imports following commit to RELENG_4:
src/usr.sbin/kbdcontrol/kbdcontrol.1 rev 1.27
src/usr.sbin/kbdcontrol/kbdcontrol.c rev 1.34, 1.35
If my build sandbox works as expected, Jul/17/2004 or later RELENG_4
release build is broken iff old build environment is used. This is because
that even if a new keymap hits the tree, release build doesn't see src/
directory but /usr/share/syscons/keymaps.
About three years before, 5-current had a fix for that by sobomax, but
it was not MFCed yet. Let me give a chance to fix for RELENG_4.
Tested on: tora.jp.FreeBSD.org
(aka snapshots.jp.FreeBSD.org RELENG_4 buildbox)
Restore part of the word that got (accidentally) removed.
Removed whitespace at EOL.
Document the behaviour of the "-o file" option when "file" is a directory
(downloaded file(s) are created inside the directory).
Prevent abnormal termination of a child daemon process when created
by a parent that is a session leader (e.g., login shell) by ignoring
SIGHUP before calling fork(2) and then restoring SIGHUP's action
after setsid(3). Based on the patch by Martin Kammerhofer
<[email protected]>.
This has been in -current for 8 months with no reports of adverse
affects.
PR: bin/25462
Prompted-by: [email protected]
---snip---
New defines that paves the way for new midi features and ports.
Changes consists of an IOCTL (SNDCTL_SEQ_GETTIME) and a constant.
---snip---
PR: 59233
Approved by: matk
Requested by: lofi
ignore two more bogus return values sent by some hardware.
PR: kern/61482, kern/54188
Submitted by: Ulf Lilleengen <[email protected]>,
Victor Balada Diaz <[email protected]>
variable, but conveniently, because ncpaddr.ncpaddr_family !=
AF_INET, the call to ncpaddr_getip4addr() became a no-op leaving
the local address as it was (defaulting to whatever my hostname
resolves to).
Update the usage string in the limits(1) manpage to include -b for
sbsize. Also, correct the format string in getopt(3) usage to reflect
that -b takes an argument, and correct another case of RLIMIT_SBSIZE
having been forgotten.
ready to handle all its aspects. Failure to deal with HTTP/1.1,
with chunked transfer encoding in particular, led to broken transfers
and data corruption. Fall back to using HTTP/1.0, which is safe.
A point to note: While the "Host:" header had not been in the
original HTTP/1.0 standard, it became a de-facto standard for
HTTP/1.0 as well. It is understood by servers in HTTP/1.0
sessions if supported at all.
This issue doesn't apply to CURRENT since lukemftp handles HTTP/1.1
correctly.
PR: bin/68547
Reviewed by: vs
in -STABLE's make. The MAKE_SHELL description seems to have crept into
the merge in rev. 1.12.2.24.
PR: 65615
Submitted by: Matthew Seaman <[email protected]>
Deal with the possibility of negative available space from the file server
to avoid Bad Things(TM) happening (eg: df crashing with a floating point
exception).
to make.conf(5) to have a `deptree' target in all makefiles:
.NOTMAIN: deptree
deptree:
@echo 'Dependency tree for ${MAKE} in ${.CURDIR}:'
@l=1; \
for d in ${.MAKEFILE_LIST}; do \
case "$$d" in \
..) l=$$(($$l-2));; \
/*) printf '%-*s%s\n' $$l '' "`realpath \"$$d\"`"; \
l=$$(($$l+2));; \
*) printf '%-*s%s\n' $$l '' \
"`realpath \"${.OBJDIR}/$$d\"`"; \
l=$$(($$l+2));; \
esac \
done
1.128 Terminate program names if we hit a '/'.
1.127 Log file name may be prefixed by a '-' if it shouldn't be synced.
1.126 Be more careful about using using the file descriptor f_file.
1.125 String fixes (inspired by OpenBSD).
1.124 Warn about paths that are too long (OpenBSD).
1.123 Fix some not-in-practice bugs to do with log levels.
1.122 Update some comments.
1.121 Warns fixes.
SMP cases, and print separate messages for non-monotonic and negative
times).
Turned off the warning for zombie cases too.
Fixed some style bugs in the messages.
Only the warning about non-monotonic times is turned off. Unlike in
-current, these are only known to be caused by calling calcru() from
ttyinfo() for ^T processing (always if it is called for a zombie) so
turning off the warning is believed to only lose a cryptic warning
that the times printed for ^T are slightly out of date.
PR: 52490
This fixes races with schedcpu() and schedclock(). It corresponds to
sched locking before the sched_nice() call in -current. Lock assertions
made the races very obvious in -current.
Adjust the process's priority after calling resetpriority() as in
schedclock() to maintain the invariant that the current priority is
always _the_ process's user priority if it is a user priority. This
is to support uncommitted assertions that the priority is correct on
return to user mode. The priority is correct here (modulo bugs
elsewhere) but it needs to be set explicitly to satisfy the assertions.
processing -- EXCEPT that the default behavior is still to use the
"oldorder" processing. This allows people on 4.x-stable to try
out the "neworder" processing (either by `newsyslog -D neworder',
or by adding '<debug> neworder' to newsyslog.conf).
I plan to change the default to neworder in a week or two, and then
will remove all the oldorder code after a few more weeks.
PR: parts of this came from bin/25070 by Helge Oldach
debugging option via the newsyslog.conf file. Add a routine to change
file-attributes (chown, chmod, chflags), instead of having several
places doing the same sequence of system-calls. A few cosmetic/style
changes. These should not effect any users.
If specified, the matching log files will have the NODUMP flag set on
them after they are created.
[* = 1.87 newsyslog.c, 1.2 newsyslog.conf.5]
Submitted by: Sean Eric Fagan
routine that had been written some months ago for other processing.
This should get rid of a few subtle situations where an existing log
file would not exist (for a short time) while it is being rotated.
First release of ixgb driver for the Intel(R) PRO/10GbE Family of Adapters. This driver has
been developed for use with FreeBSD, version 4.8 and later.
Submitted by: Hema Joyce
Reviewed by: Prafulla Deuskar
Approved by: Prafulla Deuskar
sendfile(2)-related statistics that can be viewed through netstat -m:
- The number of times sendfile() had to do disk I/O
- The number of times sfbuf allocation failed
- The number of times sfbuf allocation had to wait
and three sysctl's: kern.ipc.nsfbufs, kern.ipc.nsfbufspeak,
kern.ipc.nsfbufsused that also can be viewed through netstat -m.
Submitted by: Igor Sysoev <is (at) rambler-co.ru>
the res_target structure. It was not a problem for now. But,
It should be fixed for a future.
PR: bin/67994
Submitted by: JINMEI Tatuya <[email protected]>
Fix the ordering in the description of the dlsym() lookup procedure to
reflect src/libexec/rtld-elf/rtld.c rev. 1.68 - the globally-loaded
objects (RTLD_GLOBAL) are searched before the local object's DAG's.
Bring the description for login_getclassbyname in sync with the function's
arguments. The function has as a second argument a struct passwd * pointer,
not a directory name.
driver tries to submit the same request repeatedly, on finding the
controller cmd queue to be full. This is an MFC from -CURRENT, and
is being done a bit fast because it addresses a 4.10 errata item, and
has already been tested for a week on a heavily used -STABLE machine
that was previously showing symptoms of the problem.
Submitted by:ps, vkashyap
Reviewed by:re
Approved by:re
the same as `ps -tpt', instead of being changed into `ps -tpT'. Also adjust
it so `ps t p0' is treated like `ps -t p0', instead of making it `ps -T p0'.
This includes the rewrite of the kludge-option processing, but it does NOT
remove the long-undocumented BACKWARD_COMPATIBILITY compile-time option.
If you delete the '#define BACKWARD_COMPATIBILITY' line, you'll get the
newer handling for "extra arguments", which is to treat them as requests
for processes.
Also brings in a number of style(9) improvements.
[* = 1.90 & 1.92 -> 1.98 ps.c]
PR: bin/52489 (for `ps tpt' and `ps t p0')
Submitted by: Jilles Tjoelker ( " )
- Remove entry which tells about periodical INDEX updates in
the CVS repository
- Add description for fetchindex target
Requested by: ru [1]
Approved by: josef (mentor)
pmap_prefault_pageorder array. This has two effects:
1. The resulting bogus contents of the array thwarts part of the
optimization effect pmap_prefault() is supposed to have.
2. The resulting array is only 7 elements long (auto-sized), while
pmap_prefault() expects it to be the intended 8 elements. So this
function in fact accesses memory beyond the end of the array.
Fortunately though, if the data at this location is out of bounds it
will be ignored.
This bug dates back more than 6 years. It has been introduced in
revision 1.178.
Submitted by: Uwe Doering <[email protected]>
PR: 67460
to handle that case better by adding any CPUs that appear to be logical
CPUs to the logical_cpus_mask bitmask. -CURRENT uses different code for
enumerating CPUs, so this is not an MFC.
Submitted by: Bruce Walker bmw at mxtreme dot com
1.126 Remove the requirement that the "discard" service be defined.
1.125 request_init pairs terminated by a 0 key.
1.124 Warn if inetd's config file doesn't exist.
1.123 Rename 'int log' to avoid shadowing.
the point the release goes out so we're adding it now to RELENG_4
and then will MFS it. This will probably be backed out so the entries
are empty in RELENG_4 some time after the 4.10 release.
Reviewed by: bmah
of msync( ...,MS_INVALIDATE) on NFS-mounted partitions. Given this
is a stable branch and the issue being addressed to begin with may
not concern some users we are providing a way to get the old behavior
if desired. The default will be to have the new behavior which is to
only flush pages from memory if they are not dirty. By changing the
sysctl vm.old_msync users can get the old behavior back. Because of the
unified VM/buffer-cache system what to do as a proper fix to this
issue is proving to be quite difficult. It is still being worked on
but we need this stop-gap solution for now.
Use the Generic 9005 mask when looking for generic 7901B
parts. This allows the driver to attach to 7901B parts
on motherboards using a non-Adaptec subvendor ID.
This also supports a recent Dell OEM part.
Submitted by: gibbs
1) only have a long timeout for the wi_cmd_ini command. All others should
much faster at 10ms. Some cards seem to timeout when the signal quality
is low, and a faster timeout means that retry will happen in a fraction of
a second, rather than seconds.
2) ifconfig passes -1 in i_val for the ssid and ignores the rest of the
parameters if it returns an error. wi used to return an error in this
case, which means none of the 802.11 parameters were being reported.
Since an works, fix the presumed bug in wi.
avoid writing a null character 1 byte past the end of cmdline[] when
libedit is being used for input, and avoid writing a null pointer 1
element past the end of margv[].
date: 2004/03/23 08:37:34; author: tjr; state: Exp; lines: +2 -1
Do not copy vm_exitingcnt to the new vmspace in vmspace_exec().
Copying it led to impossibly high values in the new vmspace,
causing it to never drop to 0 and be freed.
Requested/tested by: ps
Reviewed by: alc
MFS: immediately
status block. This makes the bge devices work much better when
interrupts are shared.
This change entails bringing in the atomic_readandclear_* functions
from -current. RELENG_4 already had the equivalent loadandclear
function for the i386, but it did not exist for the alpha.
compat packages are being removed because of security issues, they
are no longer buildable. The other packages being removed are due
to space issues, they don't fit. Packages chosen for removal were
either thought to be less popular with the majority of people, or
were large *and* you need a network connection to be able to use them
so downloading them post-install should be possible.
This will be merged into RELENG_4_10 immediately but note that a commit
to HEAD has not and will not be done for this now. -current package
sets are smaller (.tbz versus .tgz) so the package sets are at least
somewhat disjoint. And it's pointless to update this in HEAD until
a 5.X release comes.
Makefile to build the ISO's. Previous 4.X releases had a non-generic name
for the labels but were put together 'manually' instead of using the Makefile.
This will be insta-MFC'ed...
Fix a couple of camcontrol issues that popped up on sparc64:
- bzero the CCB header in getdevtree() and set the path properly, to
avoid having random garbage in the CCB header.
- if the lun isn't specified in a device specifier, it should default to
0, not whatever random data happens to be in the lun variable.
- move the prototype for getdevtree() out from under #ifndef MINIMALISTIC,
since it is used in both cases.
Approved by: ken (maintainer), marcel (mentor)
with mmap(2) where pages marked as PROT_NONE may become readable under
certain circumstances.
The commits to -current being MFC-ed with this are:
Revision Changes Path
1.333 +5 -4 src/sys/vm/vm_map.c
1.109 +1 -1 src/sys/vm/vm_map.h
Which were a result of a PR:
Submitted by: "Mark W. Krentel" <[email protected]>
PR: kern/64573
Reviewed by: alc
socket address length check fails. This is similar to the fix in
revision 1.99, except this version of the fix does not move the sanity
checks out from under the spl.
Submitted by: pjd
Pointy hat to: truckman
Being so close to the code freeze I went for the easy fix -- just remove
the bug -- rather than playing with convoluted casts to produce a
stylistically correct piece of code. This and whitespace can be
fixed after 4.10 by those interested.
Approved by: re
revision 1.74
date: 2003/01/30 15:00:17; author: phk; state: Exp; lines: +4 -0
Catch some cases where asking for ridiculously large allocations could
result in a segfault. Instead just return NULL.
Requested by: JINMEI Tatuya <jinmei (at) isl.rdc.toshiba.co.jp> and so@
Reviewed by: phk (but he's not responsible if this makes 4.10 go boom :-)
Approved by: re (scottl)
o split a sentence to make it more understandable
o mention that the acctfile has to exist for accton to work
o add reference to acct.5
Approved by: re (kensmith)
GNOME 2.6 and KDE 3.2.1.
MFC:
umct(4) driver added, and SIZE attribute in distinfo.
Update release note:
Use &man.twa.4;, and
usb(4) has early support for some USB2 devices[*].
Suggested by: julian[*]
Approved by: re (implicitly)
Diff reduction on several ethernet adapter drivers (name chenges etc)
MFC a Bill Paul Copyright notice.
Add several device IDs.
Reset a uhci controller the same way we do in -current
remove an un-needed include file
add code to probe a Clie 3 series
Approved by: re (scottl@)
A malicious CVS server could cause your CVS client to overwrite
arbitrary files (CAN-2004-0180).
When a CVS client uses the `-p' checkout option, the server could be
fooled into checking out files from outside the given $CVSROOT.
Submitted by: Derek Robert Price <[email protected]>
Approved by: re
Add 'umct', a driver for MCT Corp. and Belkin F5U109 USB-Serial
adapters. This is based on the ubsa driver by Alexander Kabaev along
with documentation gleaned from the Linux mct_u232 driver. I've had
this driver sitting in my tree for almost 6 months, and several others
have found it useful.
I've been using this with success on my RELENG_4 USB macines. MFC of
umct.4 to follow.
Approved by: re (scottl)
nge(4) VLAN support bugfix,
rl(4) polling support bugfix,
ste(4) polling support,
twa driver,
ps(1) supports more POSIX/SUSv3 compatible options, and
OpenSSL 0.9.7c -> 0.9.7d.
Update release note:
usb(4) initial support of EHCI and umodem(4) now uses /dev/ucom*.
Approved by: re (implicitly)
we convert ip_len into a network byte order; in_delayed_cksum() still
expects it in host byte order.
The symtom was the ``in_cksum_skip: out of data by %d'' complaints
from the kernel.
To add to the previous commit log. These fixes make tcpdump(1) happy
by not complaining about UDP/TCP checksum being bad for looped back
IP multicast when multicast router is deactivated.
Reported by: Vsevolod Lobko
1. Remove hyphen between prodoct and company name for Cronyx Sigma (cx (4)).
2. Add list of models for Cronyx Sigma.
3. Add Cronyx Tau (ctau(4)) entry.
when `ps' is compiled with LAZY_PS, so that there is only one PS_ARGS
string to modify when changing the option-list. Also get `-f' to show
up in the usage() statement when compiled with LAZY_PS. Also move the
'f' case in the select statement, so it in alphabetical order.
[* = 1.68 1.69 & 1.86 ps.c]
Approved by: re(scottl)
- Change `-p' to allow a list of process IDs, and `-t' to allow a list
of terminal names, instead of only a single value for each.
- Add the `-A' option of SUSv3, which is exactly the same as `-ax'.
- Add the `-G gidlist' (group id).
- Allow any of these "selector options" to be specified multiple times,
and have `ps' keep adding to a given list -- instead of replacing the
previously-specified values.
- Fix interactions between selector-options, so that: "If any are
specified, ... ps shall select the processes represented by the
inclusive OR of all the selection-criteria options." (from SUSv3)
- Add a `-X' option, which is the reverse of the `-x' option.
- various minor improvements in parsing and error handling.
This should result in no user-visible changes, except for adding the
new features. The man page is updated to describe the new features,
and to describe the sorting options in better and more complete detail.
[* = 1.67 1.72 1.76->1.79 1.84 & 1.85 ps.c, 1.71->1.74 ps.1]
Approved by: re(scottl)
resulted in panics if large numbers of events were created with short
timeouts.
This hasn't been fixed in -CURRENT yet; 5.x has a more serious race
condition which I need to fix first.
PR: kern/64121
Submitted by: Gerrit Nagelhout
Don't reject FAT file systems with a number of "Heads" greater than
255; USB keychains exist that use 256 as the number of heads. This
check has also been removed in Darwin (along with most of the other
head/sector sanity checks).
ttyz?)
2. Remove old control device /dev/cronyx in favour of /dev/cx? (it is safe to
do it befor rm cxconfig, cause it is not linked to system build process.)
Approved by: imp (mentor)
Problem:
When an NFS server is port-scanned nfsd sometimes exits. This has
happened 3 times the last few weeks.
Nfsd has been written to exit when accept(2) fails. Unfortunately
accept can sometimes make a "normal" return with errno ECONNABORTED
and in this case nfsd exits prematurely.
Solution:
Check for ECONNABORTED (and also EINTR, since nfsd uses signals)
and continue.
Submitted by: Bjoern Groenvall <[email protected]>
PR: 61084
Reminded by: Lewis Thompson <[email protected]>
After examining the remaining diffs after MFCing,
I've decided that -current has this right and 4.x has this wrong..
Set the buffer end address as startpage + offset + buflen
instead of startpage + buflen
After all, the buffer START is startpage + offset
* all
- s/__FUNCTION__/__func__/.
- Compatibility for RELENG_4 and DragonFly.
* firewire
- Fix kldload failure
- Add a PCI ID of NEC uPD72873.
- Fix a bug reported misc/64623.
- Timestamp just before queuing.
- Retry bus probe if it fails.
- Use device_printf() for debug message.
- Invalidiate CROM while update.
- Don't process minimum/invalid CROM.
* sbp
- Add ORB_SHORTAGE flag.
- Add sbp.tags tunable.
- Revive doorbell support. It's not enabled by default.
ut_time. Wtmp records with time==0 get "the time of the last valid
record", while records where time goes backwards (compared to the
previous record) are skipped. Also prints a message saying how many
records were changed or skipped due to these checks. This check was
inspired by a simpler check in OpenBSD's version.
debug-statements so the program compiles without errors or warnings
when DEBUG is defined on sparc64 with 64-bit time_t's (in -current).
Also have debug statements include the year when printing records
from a different year than 'now'. Also print out a special
timestamp in debug statements when ut_time==0.
for <x> being a comma-separated list of processes, and does a much
better checking for invalid-values of <x>, such as 'ps -p someword'.
[* = 1.65 ps.c, 1.67 ps.1]
(kern.bootp_cookie) in the list of paths used for diskless templates.
When foo.cpio.gz archives are present, do not copy the content from a
foo/ directory even if present. This restores the previous behaviour,
and is also important from the point of view of efficiency (saving
many RPCs to copy the individual files), which was the whole point for
introducing the cpio archive.
(these files are not anymore in -current but equivalent functionality
will be implemented there, and the maintainer has been already informed).
disk images). The change actually boils down to little more than
providing the equivalent of a 'virgin' label for the file, all
the remaining label manipulation being the same.
This saves the need for going through vn/mdconfig (and root privs)
just to label an image file.
RELENG_4 only commit because disklabel is not in RELENG_5/CURRENT anymore,
though I suspect a similar change could be applied to bsdlabel and friends.
mainly to greatly reduce the code-diffs in ps.c (but that dragged
in some similar changes to the other files). This should result
in no user-visible changes. This is done to make it easier to MFC
some user-visible changes that are coming up, but those MFC's will
be done at a later date.
speedup stream socket recv handling by tracking the tail of the
mbuf chain instead of walking the list for each append. This has
been pretty well tested at Yahoo!
Obtained from: netbsd (jason thorpe)
Reviewed by: silby
a prequisite of the vm_map.c rev 1.280 that dillon prematurely MFC'ed
on Jan 13th, 2003. Without this change, the kernel leaks vmspaces
to varying degrees depending on the circumstances. Specifically,
vm_exitingcnt was being copied from the parent to the child when it
might have been non-zero. This caused the child's vm_exitingcnt to
never hit zero. This was blowing up machines at work.
as this is more appropriate for modern disks.
Also, remove a duplicate variable assignment.
PR: kern/63839
Submitted by: Dmitry Morozovsky <[email protected]>
Approved by: grog (mentor)
Have edquota honour MAXLOGNAME rather than hardcoding the maximum
username length to 29 characters.
PR: misc/62944
Submitted by: David Hill <[email protected]>
Approved by: ru
0xFFFFFFFF, it means that the correct value is unknown. Since this value
is just a hint to improve performance, initially assume that the first
non-reserved cluster is free, then correct this assumption if necessary
before writing the FSInfo block back to disk.
PR: 62826
USB code. This will allow teh addition of USB-2 support to 4.10 and
simplify debugging as they had diverged too much.
Submitted by: [email protected]
Obtained from: NeBSD via -current
- New API functions for libnetgraph.
- Get rid of static buffers where possible.
- Fixed memory leak in NgSendAsciiMsg().
- WARNS fixes for ngctl(8).
- mdoc(7) cleanup for netgraph(3).
- Reduced gratuitous diffs to the trunk.
tcp6_usr_bind(), tcp_usr_connect(), and tcp6_usr_connect()
before checking whether the address is multicast so that the
proper errno value will be returned if sa_len is incorrect.
argument to avoid writing past the end of a dynamically
allocated buffer, even though the minimum allocation size
prevents an actual buffer overflow from occuring without
this check.
- deny attempts to delete or rename a file
if a guest user isn't allowed to modify files on the server;
- disallow downloading raw directory contents by real users
in order to work around buggy GUI clients which think that
trying to download a directory is a fair way to tell it
from a regular file;
- "(char *)0" --> "NULL" where possible.
getnetbyname fixes:
Do not choke on malformed network addresses.
Return n_name in static space, not on the function's stack.
Tinderbox fixes by: Ruslan Ermilov
* firewire
- Remove pending list.
- Ignore timeout for the FWXF_START state.
- Define M_FWMEM for debugging.
- Comment out DELAY() in fw_asybusy().
- Improve debugging messages.
- Fix register mis-alignment introduced in rev1.12.
- Add missing free() in exception handlers.
- Add tcode_str[] and improve debug message.
* sbp
- Freeze simq while bus reset.
- If max_speed is negative, use the maximum speed which the
ohci chip supports. The default max_speed is -1.
* sbp_targ.c
- Add speed in struct sbp_targ_login.
- Remove unnecessary htonl().
* if_fwe
- If tx_speed is negative, use the maximum speed which the
ohci chip supports. The default tx_speed is 2.
* all
- Remove __P().
When not creating a core dump due to resource limits specifying
a maximum dump size of 0, return a size-related error, rather
than returning success. Otherwise, waitpid() will incorrectly
return a status indicating that a core dump was created. Note
that the specific error doesn't actually matter, since it's lost.
MFC after: 2 weeks
PR: 60367
Submitted by: Valentin Nechayev <[email protected]>
This allows a 4.x cp binary to work correctly on 5.x kernels (for which
munmap() of a 0-length mapping returns an error).
Noticed by: bento's 4.x package build
ac97.c: 1.49: Add ID for ALC658 CODEC.
ac97.c: 1.48: Recognize the Avance Logic ALC655 codec found on some
ICH4/5-based motherboards, such as the Gigabyte I848P.
ac97.c: 1.46: Add Creative EV1938.
ac97.c: 1.45: When present use ogain instead of master for surround sound
channels.
ac97.c: 1.44: Additional VT1616 id.
ac97.c: 1.43: Apply Rudolf Cejka's patch for AD1980 hook & ac97_fix_auxout.
Associate AC97_MIX_AUXOUT with SOUND_MIXER_OGAIN rather than
SOUND_MIXER_MONITOR.
Add ac97_fix_tone to remove tone controls from mixer if invalid
ac97.h: 1.14: Apply Rudolf Cejka's patch for AD1980 hook & ac97_fix_auxout.
Associate AC97_MIX_AUXOUT with SOUND_MIXER_OGAIN rather than
SOUND_MIXER_MONITOR.
Add ac97_fix_tone to remove tone controls from mixer if invalid
ac97_patch.c: 1.2: Add patch for AD198x.
ac97_patch.h: 1.2: Add patch for AD198x.
Approved by: matk
gethostbyname2() is able to lookup AF_INET6.
PR: 23823
Noticed by: Andrew Arensburger <[email protected]>
This was mentioned again in:
PR: docs/59669
Submitted by: Volker Stolz <[email protected]>
(not so in HEAD). If you want to build/install RELENG_4 on HEAD, you
have to set MAKEDEVPATH to include /usr/bin (at least as long HEAD's
/bin/sh has no printf builtin).
Explained by: bde
date: 2003/05/16 01:13:16; author: rwatson; state: Exp; lines: +5 -1
Add a tunable/sysctl "hw.fxp_noflow" which disables flow control support
on if_fxp cards. When flow control is enabled, if the operating system
doesn't acknowledge the packet buffer filling, the card will begin to
generate ethernet quench packets, but appears to get into a feedback
loop of some sort, hosing local switches. This is a temporary workaround
for 5.1: the ability to configure flow control should probably be
exposed by some or another management interface on ethernet link layer
devices.
Requested by: Pawel Malachowski <[email protected]>
need to set np->n_size back to the desired size again after calling
nfs_meta_setsize(), since it could end up in nfs_loadattrcache() getting
called, which would change n_size back to the value it had before the
truncate request was issued. The result of this bug is that the size info
cached in the nfsnode becomes incorrect, lseek(fd, ofs, SEEK_END) seeks
past the end of the file, stat() returns the wrong size, etc.
PR: 41792
- Add support for SMB signing, which is required to connect to
Windows 2003 servers.
- Add support for mixed-case passwords.
- Add support for large (up to 64K) READ_ANDX/WRITE_ANDX requests.
- Increase TCP send/receive queue size to support these large requests.
- Receive packets in small pieces so TCP slowstart will get ACKs faster.
- Fix off-by-one bug preventing access to servers with NetBIOS names
exactly 15 characters long.
- Allow the [, ], and = characters in non-8.3 filenames.
- Only close files on the server when there are no more references to
the vnode; makes mmap()'d regions work properly after the local
file descriptor has been closed.
- Fix mishandling of files >4GB.
- Diff reduction to make future merges easier.
PR: 36038 46902 57123
it is marked as RTF_UP. This fixes a panic when an interface
is detached (ie. suspend/resume).
Not objected by: sam, iedowse
Fix in -current by: iedowse
- Titan VScom PCI-800L
- Oxford Semiconductor OX12PCI840
- Moxa Technologies, Smartio C104H/PCI
- Moxa Technologies, Smartio CP-104UL/PCI
- IC Book Labs Ironclad x8 Lite
- IC Book Labs Ironclad x8 Pro
This was prompted by the PR which included a patch for 4.x for the Titan
800L.
PR: kern/41555
Reported by: François Laissus <[email protected]>
1) Don't recurse on errors.
2) Remove unused printf argument.
3) Use skip_message for program and hostname matching (which allows multiple
comma-seperated names to be given).
(aka prescription flu medicine containing lots of hydrocodone) might not have
been one of my better ideas. Fix typo noticed by Nate Lawson. The other
files look OK.
I missed these from the MFC and caused the dysfunction reported on -stable,
I just didn't catch it due to lack of personal locale-use. Yes, when you
change an internal function it is also necessary to change the functions
which *call* it or bad things happen. Yes, I should have caught this myself.
I'm a little rusty in -stable, sorry. Won't happen again. :(
</big sigh>
broken and list the type of the pin as INT rather than ExtINT in the MP
Table. To workaround, force pin 0 of ioapic 0 to be ExtINT if it is of
type INT. This is not a MFC as -current has very different apic code now.
Current's mptable parser might need a similar workaround.
PR: i386/39234
Reported by: Henri Hennebert <[email protected]>
pannels, or 32 ports for the ech boards. The only possible way to
make an error is to have 8 + 8 + 8 + 16 or 8 + 16 + 16, so we detect
that more directly and cleanup properly (eg, ignore the last panel).
This was the intent of the too many ports attached to board message
before, and should be more robust.
# now, too port this to current.
than at the end where it suffered from a fencepost error.
However, I'm not entirely sure that it is still correct because it
looks like we access something in the ioaddr2 range. I need more
stallion pannels to know for sure. The code is less incorrect than it
was before at least when you have two pannels connected to the
EC8/32-AT. Two pannels are allowed and appear to work. Stallion says
that 3 cannot work, or when the agregate number of ports is > 32. I
think that these tests should be more explicit, but until I have more
hardware, this is less wrong.
Fir another mis-merge in ciss_cam_action_io when getting an error back
from ciss_start, only REQUEUE the request if the error is not EINPROGRESS
Pointy hat to: me
Remove extraneous semicolons. They are already provided by
the macro definition, and cause the generation of syntactically
incorrect code that gcc happens to accept.
Fix problem where initgroups would silently truncate groups
with more than NGROUP elements without providing the opportunity to
setgroups to fail and correctly retorn an error and set errno.
site). This only updates the stallion (stl) driver. The stli driver
isn't updated because the driver is corrupt on the stallion web site
:-(.
This fixes my EC8/32-AT card, which worked with the 1.0.0 driver, but
not the 2.0.0 driver. It also fails to break my EasyIO-8 PCI card
(well, I can't completely test it just yet).
This is being committed directly to stable because this driver doesn't
compile in current. I'm doing a current port, but it will commit that
later. Also, 2.0.0 hasn't made it into current either.
Add support for the AMD 8111
Fix support for the nForce (so it actually does something).
Add support for the nForc2 and nForce3.
Ideas from -current but different impl.
Sponsored by: Yahoo!
very well across buckets, especially in the case of cylinder group blocks
which are located at a sequence of locations that are a multiple of a large
power of two apart. In the case of large file systems, one or possibly
a few of the hash chains can get excessively long. Replace the existing
hash algorithm with a variation on the Fibonacci hash.
Tested by: Ken Marx <[email protected]>
Attempt to save the last dregs of emacs users' sanity by saving the
contents of the PR when an interrupt is received during the editor
session. This stops the use of ^G from deleting a filled PR from
underneath the user.
PR: bin/59201
Submitted by: Heikki Suonsivu <[email protected]>
to sendfile(2) being erroneously automatically restarted after a signal
is delivered. Fixed by converting ERESTART to EINTR prior to exiting.
Updated manual page to indicate the potential EINTR error, its cause
and consequences.
Add the following devices to the list of supported devices, to sync
manual page with the source code:
- HAL Corporation Crossam2+USB IR commander
- RATOC REX-USB60
- SOURCENEXT KeikaiDenwa 8 (with and without charger)
truncated. In environments where many tunnel or vlan interfaces are created,
interface names have high numbers which overflow the field width.
PRs: bin/52349, bin/35838
Submitted by: Mike Tancsa, Scot W. Hetzel
time during the boot process. This is needed in the case where NFS mounts
from servers reachable only via IPSEC are in /etc/fstab.
PR: conf/42497
Submitted by: Volker Stolz
compatibility, and marked the old value and RB_PROBEKBD as reserved
for internal use in boot blocks. Also unintentionally clobber a style
fix in 1.22).
The corresponding change to boot/boot2.c has not been not merged, since
RELENG_4's boot2 doesn't actually support RB_PAUSE (or even the 7 year
old flag RB_MUTE), so any plain merge would just add an incomplete set
of comments to code that should be rewritten so that it doesn't need
the comments. RELENG_4's boot/loader does support these flags, and
it doesn't need changing since it doesn't hard-code copies of them.
Add vendor ID to make Marvell chipset work. E.g. to be found
on SMC9452TX it seems
Submitted by: Jung-uk Kim <[email protected]>
Tested by: <Radu Bogdan 'veedee' Rusu> [email protected]
- Code cleanup
- In the receive routine handle the case where last descriptor could have
less than 4 bytes of data.
- Handle race between detach/ioctl routine.
- Add facilities for tuning "em" drivers interrupt delays without recompiling
the driver. These changes were done by John Polstra.
Submitted by: jdp (sysctl's for interrupt delay tuning)
- Detach black hole device on AC_PATH_DEREGISTERED.
- Pick up a correct path_id for the bus on AC_PATH_REGISTERED.
- Call cam_periph_invalidate() only if the periph is found.
This closes a local DOS whereby a user can put a process in an
uninterruptible sleep by calling rfork(RFNOWAIT|RFPPWAIT|RFPROC).
Reported by: Igor Serikov <[email protected]>
buildworld as src/etc/sendmail/freebsd.mc is missing. That example
was added 3 years ago, before the /etc/mail/ infrastructure was in
place for customized configurations. It is time to remove this example.
Revision Changes Path
1.11 +0 -1 src/share/examples/cvsup/refuse
1.10 +4 -6 src/share/examples/cvsup/refuse.README
directory. Otherwise a 2nd, 3rd etc. thread might end up using garbage
for address translation.
MFC (partial) rev 1.43[67]: Don't zero fill existing page table.
mention problems with machines that don't support emulation-booting
CDROMs.
While here, add date-stamps for errata items. A couple of people
asked for this and it seemed useful and easy to do.
and restructure them. This is mainly to fix up the "will trim at"
messages printed when running 'newsyslog -v'.
[* = 1.1->1.2 extern.h, 1.81->1.82 newsyslog.c, 1.1->1.4 ptimes.c,
1.15 Makefile]
the number of broken DNS servers out there in the world. Since we are
diverging from the sendmail.org submit.mc, it's time to make our own
freebsd.submit.mc.
- Update the page title to include information about which cards
are supported by the driver.
- Use a list to specify which cards are supported.
- Add the 3c592/3c597 EISA cards to the list of supported cards.
- Add the Dell PERC 4/Di to the list of supported devices; it
has been supported for the last 10 months.
- Make the device list compact, since it is getting rather large.
Don't read the MAC address from a copy of the EEPROM in the softc
that has been recorded earlier and overwrite it again later by
reading it directly from the EEPROM again.
Read the MAC address from the PAR0/PAR1 registers instead, which
are autoloaded on reboot.
routing problems.
- Enable cpu_idle_hlt for the SMP case by default since that significantly
helps HTT performance.
- Provide a facility for disabling individual CPUs by refusing to schedule
processes on them.
- Use the above to provide a sysctl for disabling or enabling all logical
CPUs. The default is to disable all logical CPUs.
- Remove the 'HTT' kernel option now that it is obsolete.
Submitted by: ps
Approved by: re (murray)
routing problems.
- Enable cpu_idle_hlt for the SMP case by default since that significantly
helps HTT performance.
- Provide a facility for disabling individual CPUs by refusing to schedule
processes on them.
- Use the above to provide a sysctl for disabling or enabling all logical
CPUs. The default is to disable all logical CPUs.
- Remove the 'HTT' kernel option now that it is obsolete.
Submitted by: ps
Approved by: re (murray)
checks for sleeping waiting for IRQs were not wrapped in spldrm().
PR: kern/57334
Submitted by: Shobaki sam. <[email protected]>
Approved by: re (murray)
Obtained from: DRI CVS (Keith Whitwell)
installation phase sometimes causes an error" bug. It also includes
"there's an error while loading firewire modules" bug.
Since the results of readdir(3)'s whims, either sbp.ko or if_fwe.ko is
picked up for load BEFORE firewire.ko. Unfortunately module_path is
not properly set, kldload(2) cannot find kernel modules, firewire.ko,
which is required to load sbp.ko/if_fwe.ko. This is why you'll see
"loading failed" message.
Actually these are mostly comes from following commits on the HEAD:
src/release/Makefile rev 1.721
src/usr.sbin/sysinstall/modules.c rev 1.5
Approved by: murray (as of re@, pre-approved)
disabling PSE on such processors was unnecessarily made twice.
But much worse, the second test was incorrect.
Symptoms reported by: Mike Tancsa <[email protected]> and others
``This commit looks wrong.'': tegge
Pointy hat to: nectar
data corruption will result otherwise. It seems that the bug is
aggravated by recent work related to PAE.
This change was originally based on rev 1.386, but now the test for
CPU type is correct and the display of the warning is postponed until
we have initialized the kernel message buffer (e.g. for dmesg).
peter@ believes that only certain revisions of the Pentium 4 processor
are affected. However, we are unsure which so we choose to be safe by
disabling PSE on all Pentium 4s.
The test can be disabled entirely with NO_PSE_HACK.
Approved by: re
o add missing {}'s that to safe_dmamap_uniform that caused extraneous copies
for partly-aligned operations through /dev/crypto (unlikely)
o add missing case in iov code that never showed up because of the above bug
Approved by: re (murray)
-stable they need to be 'if_ed' and 'if_ep'. ifconfig should be
modified, but this is a lower risk change this late in the release
cycle.
Approved by: re@
This could cause a reader of /proc/*/regs to spin, because read(2)
would never return end-of-file. In practice, this has much less
impact than I would expect. Tools such as truss(1) make only a
single, fixed-size read (into a `struct reg'). (Similarly for
write(2).)
The cause was a sort of typo: `uiomove_frombuf' was spelled
`uiomove'.
Noticed by: David Rhodus <[email protected]>
Approved by: re (implicitly)
Export PARALLEL_PACAKGE_BUILD to the environment as well to catch more
packages that should/should not be included on the first CD. This is needed
for the recent KDE space-saving change.
Approved by: re (bmah)
sys/vm/vm_map.c and affects the unwiring memory:
The map entries are processed in a loop, checking to make sure the
entry is wired and asserting it has a wired count. However, another
loop was inserted more-or-less in the middle of the of the unwiring
path. This loop picks up the "entry" loop variable from the first loop
without first setting it to start_entry. Naturally, the second loop
is never entered and the pages backing the entries are never unwired.
This can lead to a leak of wired pages.
Reviewed by: alc
Approved by: re (murray)
spl0() and hence assuming that the old interrupt mask was zero.
Although fork1() is normally called with a zero mask, one exception
to this is when it is called by kthread_create() before configure()
has unmasked interrupts for the first time. In that case it can
cause interrupt storm hangs during device configuration, because
interrupts can be delivered before the corresponding device drivers
are ready to receive them.
This should fix the boot-time USB hangs that have been reported.
Approved by: re (murray)
aic79xx_pci.c:
aic7xxx_pci.c:
When performing our register test, be careful
to avoid resetting the chip when pausing the
controller. The test reads the HCNTRL register
and then writes it back with the PAUSE bit
explicitly set. If the last write to the controller
before our probe is to reset it, the CHIPRST
bit will still be set, so we must mask it off
before the PAUSE operation. On some chip versions,
we cannot access registers for a few 100us after
a reset, so this inadvertant reset was causing PCI
errors to occur on the read to check for paused
status.
Approved by: re (murray)
attached network could exhaust kernel memory, and cause a system
panic, by sending a flood of spoofed ARP requests.
Approved by: security-officer, jake (mentor)
Reported by: Apple Product Security <[email protected]>
drivers (for one) seem to need this. This has existed in -current
for some time since both libthr and libkse also require this
behavior.
Requested by: kde@
Approved by: re (jhb, murray)
an initial FreeBSD install.
This MFCs the following revisions:
config.c revisions 1.211, 1.213, and 1.214
menus.c revision 1.374
sysinstall.h revision 1.247 and
variable.c revision 1.36
Approved by: re (murray)
calculation would only match a leap year every 400 years. The first
time this bug would cause any noticeable problems is in Feb 2004.
This is similar to the change made in revision 1.2 of ptimes.c.
Approved by: re (murray)
Document the fact that send(2) can return EPIPE (like when a socket is not
connected).
PR: docs/56683
Submitted by: Chris S.J. Peron <[email protected]>
Approved by: re (bmah)
+ fix aligmnent issues on 64-bit architectures (1.40);
+ better argument checking when enabling/disabling ipfw-related
sysctl variables (1.39)
+ fix handling of foo/0 as an alias for "any";
Approved by: re@
drivers out of cdregister() and daregister(), which are run from interrupt
context.
Files/revisions merged:
scsi_da.c: 1.157, 1.158
scsi_cd.c: 1.81
subr_taskqueue.c: 1.17
taskqueue.h: 1.8
cd.4: 1.33
da.4: 1.42
cd.9: 1.18
not merged: taskqueue.9 rev 1.7, 1.8 (not in RELENG_4)
This code is necessarily a little bit different than the version committed
to -current, mainly in the area of mutexes. (Which become splfoo() in
-stable.)
The sysctl code does blocking mallocs (M_WAITOK), which causes problems
if malloc(9) actually needs to sleep. (And causes a panic for people
running with INVARIANTS turned on.)
The eventual fix for this issue will involve moving the CAM probe process
inside a kernel thread. For now, though, I have fixed the issue by moving
dynamic sysctl variable creation for these two drivers to a task queue
running in a kernel thread.
The existing task queue in -stable (taskqueue_swi) runs in software interrupt
handlers, which wouldn't fix the problem at hand. So I have created a new
task queue, taskqueue_thread, that runs inside a kernel thread.
scsi_cd.c: Remove sysctl variable creation code from cdregister(), and
move it to a new function, cdsysctlinit(). Queue
cdsysctlinit() to the taskqueue_thread taskqueue once we
have fully registered the cd(4) driver instance.
scsi_da.c: Remove sysctl variable creation code from daregister(), and
move it to move it to a new function, dasysctlinit().
Queue dasysctlinit() to the taskqueue_thread taskqueue once
we have fully registered the da(4) instance.
taskqueue.h: Declare the new taskqueue_thread taskqueue, update some
comments.
subr_taskqueue.c:
Create the new kernel thread taskqueue.
cd.4: Update the cd(4) man page to talk about the minimum command
size sysctl/loader tunable. Also note that the changer
variables are available as loader tunables as well.
da.4: Update the da(4) man page to cover the retry_count,
default_timeout and minimum_cmd_size sysctl variables/loader
tunables. Remove references to /dev/r???, they aren't used
any longer.
cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY
quirk.
Approved by: re (scottl, murray)
sent on specific interfaces. This is required by aodvd, and may in future
help us in getting rid of the requirement for BPF from our import of isc-dhcp.
PR: kern/37486
Obtained from: BSD/OS
Approved by: re, jake (mentor)
(which occur after disk errors). The panic is due to attempts to acquire
the ATA channel lock multiple times (at different levels), with the second
attempt trying to sleep in an interrupt context. The fix is to indicate
to the other levels that we already have the lock.
The problem has been fixed in -current a different way and isn't portable
to -stable.
Reviewed by: [email protected]
Approved by: [email protected]
with the vendor sendmail distribution. By doing so, we avoid mergemaster
warnings after every 'make' in /etc/mail/.
Revision Changes Path
1.28 +11 -4 src/etc/sendmail/Makefile
Approved by: re (murray)
Add support for new devices.
Bug Fixes:
- Allow users to use LAA
- Remember promiscuous mode settings while bridging
- Allow gratuitous arp's to be sent
PR: 52966/54488
Approved by: re (murray)
Clarify the order of arguments passed to the pre-/post-install script
in the description of the pkg_create -i command-line option.
Approved by: re (bmah)
the existance of the EISA attachment. Since the EISA bus methods are
not DEVMETHODs, and thus can cause link failures if the eisa bus is not
compiled into the system, make the compiling of the EISA attachment be
optional. This is going into RELENG_4 first since it is needed to make
the install floppies fit, and does not represent the real solution that
should be attempted for HEAD.
This is functionally a no-op. If you have 4G or more memory, most likely
you will still have problem with dumping since PCI device windows are
not currently excluded.
Approved by: re
problem trying to use a regular 8139 NIC (D-Link DFE-538+/8139D chip)
on a machine with 1.2GB of RAM. A NIC with exactly the same chip rev
works in my test box without any problems. I suspect a strange busdma
interaction, but there's not enough time for me to properly debug/fix
it before the release is cut, and breaking 8139D support is a regression.
Eventually, I plan to put the 8139C+/8169 support in a separate driver
anyway.
Approved by: [email protected]
interrupts and temporary page table mappings performed outside
interrupt context without splvm() protection. Interrupt time async
completion callbacks for pageout operations triggered this conflict.
Approved by: re (murray)
based CP-168U board. It initializes and attaches in the same way
as the older (but higher performance) C168H. The only difference
is the board ID, which is 0x1681.
PR: kern/53548
Submitted by: [email protected]
Approved by: re
This fix has the side effect of making vsnprintf() properly return the
length of the string that would have been created if the buffer were
infinite length when the passed size is 1. Previously, it would
erroniously return 0 which causes auto-extending sbufs to misbehave in
userland apps linked against libsbuf.
Approved by: re (jhb)
Do it right this time. All davicom cards need DC_TX_ALIGN.
Avoid an unnecessary copy of a packet if it is already in a single mbuf.
Introduce an additional device flag for those NICs which require the
transmit buffers to be aligned to 32-bit boundaries.
Approved by: murray (re)
current (ata-chipset.c:1.32) that properly support the SiI 0680 chip.
Prior to this, heavy load on both channels at once would cause command
timeouts and sometimes data corruption.
Submitted by: sos
Approved by: re (murray) and sos
are the same for -current and -stable.
Also, temporarily re-enable all USB floppy quirks for 4.9R. Although I
received no input whether or not these devices work without the quirks,
installs often depend on USB floppies and it is unacceptable for them to
be broken for a release. They will be disabled again after the release.
- added the case of 802.11 to check link status.
- use strncpy just in case.
- __P() cleanup.
- improved the -a option. it can probe a interface automatically when
the interface wake up. it can be started anytime even when there is
no network interface on the list of intarfaces in the kernel.
- get a correct link ID for each interface at initialization
(using scope libraries if HAVE_SCOPELIB is defined).
- fill in sin6_scope_id correctly before sendmsg().
usr.sbin/rtsold/if.c: 1.10
usr.sbin/rtsold/probe.c: 1.10
usr.sbin/rtsold/rtsol.c: 1.10
usr.sbin/rtsold/rtsold.8: 1.12
usr.sbin/rtsold/rtsold.c: 1.15
usr.sbin/rtsold/rtsold.h: 1.7
Ispell sweep of share/man/man4.
Add VT6105/VT6105M Rhine III to the list of supported chips.
Change the manual page title to use the device family name (Rhine),
since the list of supported device id's won't fit on one line anymore.
will use a larger buffer-size for reading/writing spool files. This
update doesn't change the size which is used, it only makes the size
easier to change for someone willing to recompile lpd.
[* = 1.61 printjob.c, 1.31 recvjob.c]
Submitted by: John-Mark Gurney
were checked. FreeBSD's newsyslog was treating a username with leading
digits, such as '3com', as the uid matching those digits.
Obtained from: OpenBSD
fix a number of related problems along the way.
Files/revisions merged:
scsi_cd.c: 1.72, 1.77, 1.80
scsi_all.c: 1.39 (partial)
scsi_all.h: 1.22
scsi_cd.h: 1.7
scsi_da.c: 1.139
atapi-cam.c: 1.13
not merged: umass.c rev 1.76 (doesn't apply to -stable)
Because of the differences between -current and -stable, some of these
changes are not identical to the ones in -current.
Most notably, cd(4) in -stable uses the slice code, whereas cd(4) in
-current has no disklabel support at all. So there is extra code in the
-stable version of these patches to handle that.
- Automatically detect CDROM drives that can't handle 6 byte mode
sense and mode select, and adjust our command size accordingly.
We have to handle this in the cd(4) driver (where the buffers are
allocated), since the parameter list length is different for the
6 and 10 byte mode sense commands.
- Remove MODE_SENSE and MODE_SELECT translation in ATAPICAM, since
there's no way for that to work properly.
- Merge support for sending a path inquiry CCB to see if the SIM has set
the PIM_NO_6_BYTE flag. (scsi_cd.c rev 1.80)
- Add a quirk entry for CDROM drives that just hang when they get a 6
byte mode sense or mode select. The reason for the quirk must be
documented in a PR, and all quirks must be approved by
[email protected]. This is to make sure that we fully understand why
each quirk is needed. In most cases, the PIM_NO_6_BYTE flag set by the
SIM driver should eliminate the need for a quirk.
- Change the way the da(4) handles the no_6_byte sysctl. There is
now a per-drive sysctl to set the minimum command size for that
particular disk. (Since you could have multiple disks with
multiple requirements in one system.)
- Loader tunable support for all the sysctls in the da(4) and cd(4)
drivers.
- Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long
time ago).
- Add a media validation routine (cdcheckmedia()) to the cd(4)
driver, to fix some problems bde pointed out a long time ago. For
-stable, open() will still fail if there is no media in the drive. This
is because of issues with the slice code handling media changing
underneath it. If bde and I can get those issues worked out, there is
basic code in this set of patches to handle allowing the CDIOCCLOSE and
CDIOCEJECT ioctls without media in the drive.
- The media validation routine also reads the table of contents off
the drive. We use the table of contents to implement the
CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The
PLAY AUDIO TRACK INDEX command that we previously used was
deprecated after SCSI-2. It works in every SCSI CDROM I've tried,
but doesn't seem to work on ATAPI CDROM drives. We still use the
play audio track index command if we don't have a valid TOC, but
I suppose it'll fail anyway in that case.
- Add _len() versions of scsi_mode_sense() and scsi_mode_select() so
that we can specify the minimum command length.
Approved by: re
happen at the end of the day. This is much better than we had before.
However, in reviewing the current code, it appears that some bugs in
extreme cases may exist in this version. Unless you setting the time
of your system by a small amount in the actual leapsecond, or have an
extremely loaded machine, you are unlikely to encounter problems. A
more complete fix may follow after the 4.9 release.
kernel, add 'device acpica' to your kernel config. Due to the way
config(8) works in -stable, I was unable to use 'device acpi' to enable
compilation of this driver. Note that there is no kernel module, you
must compile ACPI into your kernel as a static driver.
Sponsored by: The Weather Channel
support for PCI interrupt routing or enumeration of ISA bridges or
Host to PCI bridges. While functional on some machines, this driver
should be considered experimental and should be tested prior to being
deployed in a production environment.
Sponsored by: The Weather Channel
- logged the fact to dump the status.
- dump the content of route information options
- set LOG_PERROR for openlog() when running foreground.
- variable shared with signal handler needs to be "volatile sig_atomic_t".
- use arc4random.
- supported a string notation for xxflags.
- deprecate routes#N, as it is hard to keep consistency with
rtprefixN. accept any number of "rtprefix", "rtrefix0",
..., "rtprefix99".
- deprecate "addrs#N", as it is difficult for users to keep
consistency with "addrN".
accept 100 prefix info in maximum - like "addr", "addr0"
... "addr99". WARNS=2 clean on netbsd.
old configuration file should work just fine.
behavior change:
previously, we rejected "addrN" if there's "addr", and we rejected
"addr" if there is "addrN". now we accept both without problem.
- when an advertised prefix configured from the kernel has been added
or invalidated, notice the change in a short delay.
- when invalidating a prefix, do not bark even if there is
inconsistency about prefix lifetimes.
- wrap more specific route info code into ROUTEINFO.
usr.sbin/rtadvd/Makefile: 1.7, 1.8
usr.sbin/rtadvd/config.c: 1.23
usr.sbin/rtadvd/config.h: 1.6
usr.sbin/rtadvd/dump.c: 1.10 - 1.12
usr.sbin/rtadvd/rtadvd.8: 1.14
usr.sbin/rtadvd/rtadvd.c: 1.20 - 1.23
usr.sbin/rtadvd/rtadvd.conf.5: 1.12
usr.sbin/rtadvd/rtadvd.conf: 1.4
usr.sbin/rtadvd/rtadvd.h: 1.9
by the sysctl hw.fore.hfaN.shape, which can be set to 0 (don't shape),
1 (shape at most one channel) and 2 (try to shape all channels). This
works only with firmware version > 4, which is the default now.
Submitted by: Vincent Jardin <[email protected]>
stable when loaded with 4.X.Y firmware. This difference manifests itself
in the card returning receive buffer chains that start with a large buffer
under certain circumstances. This uncovered an inherent bug in the
mbuf handling of the atm stack: the driver prefixes 8 bytes to the mbuf
that contain a function address to dispatch this buffer to and a cookie
for this function. When the driver finds, that it has no space at the
head of the mbuf chain, it allocates a new packet header mbuf and
prepends it to the chain and moves the header to the new mbuf.
atm_intr() strips the additional 8 bytes from the mbuf chain and, if it
finds the first mbuf to become empty, removes that mbuf from the chain, but
does not care to move the header to the next mbuf. Here the packet
header is lost.
Work around:
1. Allocate all large buffers for the Fore card with a packet header.
2. When prefixing the new mbuf to the chain, don't move the packet header,
but initialize it with the correct values. This results in an mbuf
chain with two packet headers. When atm_intr strips the first mbuf,
everything is ok.
It would probably be better to fix the problem instead of working around it,
but this could break other drivers (hea and proatm).
- Allow access to phy registers via ioctl(2)(fwdev).
- Drop too short packets(if_fwe).
- Fill reserved fields of transmitting packet hear with zero(fwohci).
- Include order bit in sdev->type(sbp)
adapters. The old firmware (3.0.1) can still be used by specifying the
'-3' option to fore_dnld.
Document the -r option that resets the adapter before the download.
The newer firmware version allows traffic shaping.
fore_intr.c:1.10, fore_var.h:1.8: Make the driver to work with
firmware version 4.1.12 (other 4.X.Y should also do it).
Three problems have been encountered:
1. The initialisation command does not work in interrupt mode. Whether this
is a firmware bug or a feature is not clear. The original Fore drivers
execute the initialize command always in polling mode, so it appears
that this behaviour is expected. When we detect a 4.X.Y firmware do
busy wait on the command status.
2. The command code of the GET_PROM has changed. This is an unofficial
command anyway. What was GET_PROM in 3.X.Y is CLEAR_STATS in 4.X.Y
(although unimplemented in the firmware). We need to use the correct
code depending on the firmware.
3. The 4.X.Y can set the error flag in command status without also setting
the completion flag (as the documentation says). Check both variants.
An additional field in the per-card structure fu_ft4 is TRUE when we have
detected a 4.X.Y firmware. Otherwise it is false. The behaviour of the
driver when using 3.X.Y should be identical to the previous behaviour .
This change will enable traffic shaping of CBR channels.
- Merge the power management subsystem for handling multiple power
management drivers as well as power management events.
- Merge the pmtimer(4) driver to preserve the i386 clock chips across
suspend and resume.
- Enhance the suspend and resume support for the AT PIC interrupt
controller to match that of -current.
- Update the apm(4) driver to work with the above changes.
Discussed with: re
and bus_dmamap_load_uio into -stable. Also MFC 1.20 while
I'm here (teach bus_dmamem_free() about contigfree()).
The resulting kernel builds and boots, but I don't have any
if_xl or if_rl devices in my -stable alpha to test with.
The resulting code looks just like -current (which is well tested),
so I don't anticipate problems.
Relocate a call to enable inetd so that it is set regardless of
whether the user chooses to edit inetd.conf.
PR: 39311
Reported by: Martin Faxer <[email protected]>
Fix warning when calling w(1) when logged in via xdm/kdm. This is
really a problem with utmp/wtmp, but takes the same approach as who(1).
Reviewed by: knu (mentor), mini, silence on -audit
Approved by: knu (mentor), mini
Spotted in RELENG_4 by: Eric S <[email protected]>
If stat() on the terminal specified in utmp fails due to ENOENT, don't
print a warning, and set the idletime variable for the entry to -1;
then pick up the -1 later in sprint() and lprint() and ignore those
idle times by printing just whitespace. When third party applications,
such as kdm, insert utmp entries, they sometimes use strings like ":0",
which can't be stat()'d and currently result in warnings that are
not helpful to the user.
Requested by: Jonathan Chen <[email protected]>
- __FUNCTION__ --> __func__
- Process the "O bit."
If the value of OtherConfigFlag changes from FALSE to TRUE, the
host should invoke the stateful autoconfiguration protocol,
requesting information.
[RFC 2462 Section 5.5.3]
- stop use of register.
- KNF, correct typos and cleanup spaces.
- drop the code for the environment where getifaddrs(3) is not
supported.
- Drop MIP6 code. We don't support MIP6 yet.
usr.sbin/rtadvd/advcap.c: 1.6 - 1.8
usr.sbin/rtadvd/config.c: 1.16 - 1.18
usr.sbin/rtadvd/dump.c: 1.5 - 1.7
usr.sbin/rtadvd/if.c: 1.7, 1.8
usr.sbin/rtadvd/rrenum.c: 1.6, 1.7
usr.sbin/rtadvd/rtadvd.c: 1.12 - 1.14
usr.sbin/rtadvd/rtadvd.h: 1.7, 1.8
usr.sbin/rtadvd/timer.c: 1.6, 1.7
usr.sbin/rtadvd/timer.h: 1.6
usr.sbin/rtsold/Makefile: 1.10
usr.sbin/rtsold/dump.c: 1.5 - 1.7
usr.sbin/rtsold/if.c: 1.6 - 1.8
usr.sbin/rtsold/probe.c: 1.6, 1.7
usr.sbin/rtsold/rtsock.c: 1.2
usr.sbin/rtsold/rtsol.c: 1.6 - 1.8
usr.sbin/rtsold/rtsold.8: 1.11
usr.sbin/rtsold/rtsold.c: 1.6 - 1.9
usr.sbin/rtsold/rtsold.h: 1.5, 1.6
strchr(3) 1.9 + 1.10, strrchr(3) 1.10
* Merge index(3) and rindex(3) to index(3) since the two functions are
almost identical.
* Merge strchr(3) and strrchr(3) to strchr(3) since the two functions
are almost identical.
* Make the wording of index(3) and strchr(3) more similar.
* mdoc(7) cleanup.
Use kmem_alloc_nofault() rather than kmem_alloc_pageable() in procfs_rwmem().
Use vm_page_hold() in place of vm_page_wire() since the page can be freed.
Don't hold extra reference to the containing vm object while page is mapped.
structure to control whether to accept RAs per-interface basis.
the new stuff ensures the backward compatibility;
- the kernel does not accept RAs on any interfaces by default.
- since the default value of the flag bit is on, the kernel accepts
RAs on all interfaces when net.inet6.ip6.accept_rtadv is 1.
sys/netinet6/nd6.c: 1.25
sys/netinet6/nd6.h: 1.11
sys/netinet6/nd6_rtr.c: 1.16
usr.sbin/ndp/ndp.8: 1.10
usr.sbin/ndp/ndp.c: 1.9
if_xl.c 1.149
if_xlreg.h 1.46
As few modifications as possible were made in this MFC, which includes
busdma support for PAE users, Vlan support for older pre-905B cards,
and other various cleanups.
timeouts on Dell 1750 servers, and fix spurious 'gigabit link up'
messages due to not ignoring the 'MI_COMPLETE' bit in the MAC
status register. (I'm MFCing these fairly quickly since they're
fixes for pretty serious/annoying bugs.)
work was sponsored by DARPA, Network Associates Laboratories.
Major changes include:
- introduction of vm_paddr_t to represent physical address
- support for page directory pointer table
- removal of (recursive) alternative address space
Reviewed by: jake
OID as the other protocol family sub-trees do, that is equal to the
protocol-family identifier. Make the ATM layer debugging flags
available under this tree.
Submitted by: Vincent Jardin <[email protected]>
to such devices. If a device fails due to this commit, add:
options DA_OLD_QUIRKS
to the kernel config and recompile. Then send the output of "camcontrol
inquiry da0" to [email protected] so the quirk can be re-enabled.
Fix warnings: a variable that was unused, one that was unused unless
sun is defined, an unused function parameter and u_longs printed with
the wrong format.
PR: bin/39818
Submitted by: [email protected]
some style bugs on the same lines).
This completes making some RELENG_4 kernels work when compiled by gcc-3.3.
I intended to make more than some work, but gave up on LINT. GENERIC
compliles but has not been tested at runtime.
to return EAGAIN on a non-blocking pipe in cases where select(2)
returns the file descriptor as ready for write. This in turns causes
libc_r, for one, to busy wait in such cases.
before using it, and use up to date comments about this).
Adjustments for RELENG_4 (n/a in -current).
locore.s:
Don't staticize tmpstk since it is needed in swtch.s. Adjust comments
some more to cover the BDE_DEBUGEGR case.
swtch.s:
Don't hide the name of tmpstk here either.
This is part of making RELENG_4 kernels work when compiled by gcc-3.3.
Fixing bss clearing is the only nontrivial part.
altering the remote directory contents.
Not applicable directly to CURRENT since the latter uses
the contributed lukemftp as the default ftp client.
Luke Mewburn has applied this fix to his ftp client, too,
so CURRENT's will get fixed upon the next import of lukemftp.
Submitted by: demon
Reviewed by: lukem (at) netbsd.org
consistently match style(9) guidelines in lpr/filters* and lpr/lptest.
This include Makefile changes so these programs can find lp.cdefs.h in
lpr/common_source.
set this flag to indicate they do not wish to receive 6 byte commands.
The da(4) driver now honors this flag. Note that the cd(4) portion of the
commit is not MFCd now because it depends on other code which needs to be
MFCd first.
fix 'sccsid' lines to consistently match style(9) guidelines. Note
that this means you will have to add '-a' to the 'strings' command
when searching for rcs ids, eg:
strings -a /usr/sbin/lpd | grep '$FreeBSD'
[* = 1.32->1.33 common.c, 1.9 ctlinfo.c, 1.3 ctlinfo.h,
1.24->1.25 display.c, 1.24 lp.h, 1.4 matchjobs.c, 1.2 matchjobs.h,
1.8 net.c, 1.10->1.11 printcap.c, 1.4->1.5 request.c,
1.18->1.19 rmjob.c, 1.11->1.12 startdaemon.c]
- Removed dead declarations
- Made objects that should have been declared as static, static.
The changes use STATIC instead of static, following the existing
convention in the rest of the code.
alias.c: 1.17
cd.c: 1.33
exec.c: 1.23
exec.h: 1.11
expand.c: 1.44
input.c: 1.20
jobs.c: 1.64
memalloc.c: 1.25
nodes.c.pat: 1.14
output.c: 1.18
parser.c: 1.49
redir.c: 1.24
var.c: 1.25
and the driver supports UBR, CBR and VBR. The driver will not be brought
over the current, because current has the patm driver (which uses new
ATM infrastructure and hence is hard to MFC). This driver is reported to be
very stable.
PR: kern/46790
-current, which has int size refcounts.)
4.x uses char refcounts, which can be overflowed, causing neat panics.
This patch works around the problem by hooking into every m_* function
that increases the refcount and doing two things:
1. Panicing if the refcount went negative (perhaps due to some external
function messing with the refcount.)
2. Making a full copy of the resulting chain in order to keep the
refcount under a reasonable threshold if necessary.
The kern.ipc.m_clreflimithits sysctl has been made available so that
the number of times a copy had to be performed can be monitored. Luckily,
this should not be a common occurance.
/etc/defaults/rc.conf and boot-time support to /etc/rc.network
in RELENG_4. This permits the Heimdal password change daemon
to be automatically started when the system is booted; this
is present in -CURRENT but appears not to have been merged with
Heimdal support.
Merge all of OLDCARD stuff from current to stable. This includes:
o Improved CIS parsing
o read from cis file
o probe only and exit
o doc fixes
o Migrate manucactuer Id stuff into the kernel.
o warnings improvements.
* implement a '-n' option to do a syntax-check only of ipfw2 rules;
* allow spaces after commas in ipfw rules;
* support for comma-separated address lists e.g.
ipfw add allow ip from not 10.0.0.0/8, 192.168.0.0/16, 1.2.3.4 to me
(note the possibility to put a 'not' in front of the entire list,
which was not possible with "or blocks");
* allow comments in ipfw rules which are stored together with rules and
appear upon an 'ipfw show':
ipfw add allow udp from any to any 53 // nameserver
* allow set 31 to be used for ordinary (non-default) rules, but with
the special feature that rules in set 31 cannot be disabled and
are not affected by a 'flush' command (so they must be deleted
explicitly). This permits a flexible form of "persistent" rules
which should survive across firewall reloads.
* allow ranges to be specified in the "ipfw show" and "ipfw list"
commands (the same ought to be done for ""ipfw delete"):
ipfw show 100-1000 2000 3000-5500
I believe the kernel side of these changes is entirely backward
compatible with the old /sbin/ipfw[2], though of course you need
to update the userland command to use the new features.
Update if_dc to use m_defrag, removing the semi-duplicate dc_coal
function.
Also, use m_defrag where appropriate to defrag long mbuf chains
in the same fashion as was done in if_sis.c. Before this change,
if_dc would blow up and take down the interface if fed a really long
mbuf chain.
instead of SENDMAIL_MC but don't remove on it 'make clean' as the
user may not have the original .mc file and removing it could be
dangerous (e.g., make SENDMAIL_CF=/etc/mail/sendmail.cf clean).
Revision Changes Path
1.26 +5 -2 src/etc/sendmail/Makefile
defined. The only two files installed in this case are aliases (which
I believe other MTAs may use) and mailer.conf (which isn't sendmail,
it belongs to mailwrapper).
Revision Changes Path
1.315 +4 -0 src/etc/Makefile
`hostname`.submit.mc which is templated from freebsd.submit.mc if the
default file does not exist. This makes the building of the submit.cf
behavior identical to that of the the sendmail.cf.
Revision Changes Path
1.35 +13 -4 src/etc/mail/Makefile
(though probably not a good idea in general) to set the various
SENDMAIL_*_MC variables to /etc/mail/sendmail.mc or /etc/mail/submit.mc.
Revision Changes Path
1.217 +0 -9 src/share/examples/etc/make.conf
1.76 +0 -21 src/share/man/man5/make.conf.5
- Stop 'make clean' from removing SENDMAIL_CF
- install and distribute targets should not attempt to build anything
- SENDMAIL_ADDITIONAL_CF were not installed in the distribution case
- If SENDMAIL_SET_USER_ID was defined, submit.cf was needlessly installed
in the distribution case
- Collapse install and distribution target into one to remove code
duplication
Revision Changes Path
1.23 +47 -86 src/etc/sendmail/Makefile
- Fix a bug where a truncate operation involving truncate() or ftruncate() on
- an MSDOSFS file system either failed, silently corrupted the file, or
- sometimes corrupted the neighboring file.
MFC msdosfs_vnops.c revision 1.139:
- If bread() returns a zero-length buffer, as can happen after a
- failed write, return an error instead of looping forever.
Do not do manually what sendmail(8) can do better automatically.
Tell sendmail to clean up its own host status cache.
The error condition handling could probably be done better.
Requested by: gshapiro
MFC: rev. 1.5 Fix the type of the NULL arg to execl()
MFC: rev. 1.7 de-__P()
MFC: rev. 1.8 Use __FBSDID over rcsid[]. Protect copyright[] where needed.
MFC: rev. 1.9 Explicitly compare the values returned by chdir(2) and
chroot(2) to -1.
Among other things, this includes the following:
+ pass to the preprocessor all command-line options after -p
(except the last one, the ruleset file)
+ add the "verrevpath" option
+ support strong alignment architectures such as alpha and sparc64;
+ support multiple values and ranges for "iplen", "ipttl", "ipid" options.
+ support range notations such as 1.2.3.4/24{5,6,7,10-20,60-90}
for sets of IP addresses
The changes (also those in sys/netinet/ip_dummynet.c) are all
IPFW2-specific, which is entirely optional in RELENG_4 so there
are no ABI issues for those using the standard ipfw[1].
Note, however, that ipfw2 users MUST REBUILD /sbin/ipfw
together with the new kernel.
may want to pass the value to upper layer protocol such as DHCPv6
for prefix delegation.
usr.sbin/ppp/command.c: 1.296
usr.sbin/ppp/radius.c: 1.42
usr.sbin/ppp/radius.h: 1.18
of Framed-IPv6-Route is user defined, it follows Framed-IP-route.
usr.sbin/ppp/ipv6cp.c: 1.14
usr.sbin/ppp/radius.c: 1.40
usr.sbin/ppp/radius.h: 1.15, 1.16
1.20: correct a typo s/ a / an / in a comment.
1.24: pass correct arguments to printf()
and move a variable declaration and a couple of lines to match
more closely the version in -current.
ipfw[2] and dummynet.
There ought to be a better way to do this, something similar to
device_printf() which automatically generates the subsystem names!
The previous logmessage was referring to another file :( It should have been:
comment out "grep" and "date", the image didn't fit into a 1.44 floppy anymore.
Remove unwanted information about future changes to the packet
queueing discipline and remove the IF_ENQ_DROP macro from the
SYNOPSIS because it was nuked when jlemon did the locking work
on the interface queues.
The original commit to HEAD was ...
Approved by: des (mentor)
Bring in the pci(9) manual page, which documents some of the
vital functions of the PCI bus interface.
Written by: Bruce M. Simpson <[email protected]>
Approved by: des (mentor)
"pkg_create incorrectly does not add trailing '\n' when it receives
either COMMENT or DESCR from the command line. When a port is
installed, one gets both +COMMENT and +DESCR files with a trailing
'\n' character. However, +COMMENT does not contain a trailing '\n'
when it is installed from a package due to this behavior of pkg_create.
Therefore, make sure it behaves exactly the same regardless of
where it got its information; either command line or files. The
modified functions are used by pkg_create."
PR: 52097
"pkg_create incorrectly does not add trailing '\n' when it receives
either COMMENT or DESCR from the command line. When a port is
installed, one gets both +COMMENT and +DESCR files with a trailing
'\n' character. However, +COMMENT does not contain a trailing '\n'
when it is installed from a package due to this behavior of pkg_create.
Therefore, make sure it behaves exactly the same regardless of
where it got its information; either command line or files. The
modified functions are used by pkg_create."
PR: 52097
bzero(ptr, sizeof(DC_RXLEN * 5));
which should obviously be:
bzero(ptr, DC_RXLEN * 5);
Looks like this bug may have reduced the effectiveness of the
workaround for the hardware bug in the PNIC chips.
Relevant HEAD revision:
if_dc.c: 1.110
- Document ``target'' and ``unit'' of cd(4)
for config.
PR: docs/35941
Submitted by: Gary W. Swearingen <[email protected]>
- Mdoc dept., nuke hard sentence breaks.
Approved by: des (mentor)
o add hash operations
o fixup AES operations
o add -c option to compare/verify operation result
o make multi-threaded reporting to be more meaningful
o fix default block size selection
completes and is intended for callers to check operations that may complete
``prematurely'' because of direct callbacks
o correct crypto_dispatch's handling of operations where the driver returns
ERESTART: the return value must be zero and not ERESTART, otherwise the
caller may free the crypto request prematurely (fixes panic for ops submitted
through /dev/crypto)
o Don't reinitialize fields that are already initialized by getpbuf().
o Sufficient access checks are performed by vmapbuf() that calling
useracc() is pointless (and costly). Remove the call to useracc().
libc_r. This merges the following revisions from the trunk:
1.28 src/include/pthread.h
1.20 src/lib/libc_r/man/Makefile.inc
1.1, 1.2 src/lib/libc_r/man/pthread_getconcurrency.3
1.36 src/lib/libc_r/uthread/Makefile.inc
1.1 src/lib/libc_r/uthread/uthread_concurrency.c
PR: standards/50848
Make the reservation of KVA space for kernel map entries a function
of the KVA space's size in addition to the amount of physical memory
and reduce it by a factor of two.
Update amr(4) to note that LSILogic MegaRAID 320-[1/2] cards work
with this driver.
PR: docs/50139
Submitted by: Jeroen Ruigrok van der Werven (asmodai)
Approved by: des (mentor)
rev. 1.62 ip_dummynet.c
o Protect set_fs_param() by splimp(9).
o Flush accumulated credits for idle pipes.
o Flush accumulated credits when change pipe characteristics.
rev. 1.26 ip_dummynet.h
o Change dn_flow_queue.numbytes type to unsigned long.
rev. 1.63 ip_dummynet.c
o Fix indentation.
which were rotated and then compressed with bzip2 instead of gzip.
Before this fix, any file which had a time-interval specified for
'when' and also specified the 'J' flag would be rotated every time
newsyslog was run.
PR: bin/51519
if_bge.c:
- Support for the 5704
- Write twice to send producer index workaround
- Leave LED set to hardware default
- ONEDMA_AT_ONCE workaround
- Disable AUTOPOLL when accessing the PHY
- Check the UPDATED bit in the status block
brgphy.c:
- Support for the 5704
- Loop the phy when changing media types
- For 1000BaseT operation, autonegotiation must be enabled
- DSP patches
1. Hostnames were not treated case insensitively in all cases.
2. The method for stripping hostnames when reading the syslog.conf
differed from that when finding the hostname of an incoming
request. This lead to a broken match check. In my case, it
meant I had to have '@scooter.smi.example.com.example.com'
to have 'logger.example.com' properly save messages from
'scooter.smi.sendmail.com'.
3. Add paranoia to cfline() such that it doesn't try to access
memory outside of the bounds of the f_host string.
4. While I am here, get rid of an outdated comment, argv[{0,1,2}]
are now checked for NULL after the strdup() calls.
Revision Changes Path
1.114 +24 -17 src/usr.sbin/syslogd/syslogd.c
- Add a description of b0 / b1 fields.
- Do not use 'entry' to refer to both 'entry' and 'field'.
- Do not confuse people with heading 'Name' and entry 'Name'.
PR: 48104
Submitted by: Gary W. Swearingen <[email protected]> (original version)
rev. 1.7 jail.c
o Add missing includes and sort includes.
rev. 1.9 Makefile
rev. 1.42 jail.8
rev. 1.9 - 1.11 jail.c
o Add -u <username> flag to jail(8): set user context before exec.
Fix race between a process registering a NOTE_EXIT EVFILT_PROC event
and the target process exiting which causes attempts to register the
kevent to randomly fail depending on whether the target runs to
completion before the parent can call kevent(2).
Sponsored by: NTT Multimedia Communications Labs
entries in the config file. If the -C option is specified once,
then newsyslog will create any entries which specify the 'C' flag.
If -C is given twice, then newsyslog will create all missing log
files.
[* = 1.67 newsyslog.c, 1.42 newsyslog.8]
The -t flag in the retired apmconf.8 is the same as "-h false" in the
apm.8. Since the -t flag in the apm.8 is something different entirely,
the reference to -t was quite confusing. Fix that.
we could control the use of dbxout_really_begin_function() based on if the
target format is a.out or ELF. The goof resulted in our target logic never
being used and thus we always called dbxout_really_begin_function() twice
in the ELF case and never in the a.out case.
PR: 41671
Submitted by: Mark Kettenis <[email protected]>
entries for a filename pattern, and also wrt filenames given on
the command line. Now if a file is listed as a specific entry, it
will not *also* be processed by an entry that specifies a pattern.
Also, filename-pattern entries will now only match existing files
(ignoring directories, etc).
pid-file will contain the pid for a process group. This means the
file must contain a negative value (as would be needed in the 'kill'
commmand). Parts of this change are from OpenBSD, even though
OpenBSD does not actually have a 'U' flag.
[* = 1.64 newsyslog.c, 1.41 newsyslog.8]
PR: bin/28435
should not send a signal to any processes. Also add a config-file flag
of 'N' or 'n', which indicates that the given logfile has no process which
needs a signal when it is rotated. Both of these are based on changes
NetBSD has made, although the implementation is somewhat different.
Also add a command-line option of '-R somename', which indicates that
newsyslog should rotate all files given on the command, even if they
don't seem to need to be rotated.
This also brings in some other minor improvements, such as improving the
interactions between specific log files given on the command line, and
config-file entries which specify a filename-pattern (flag 'G').
[* = 1.40 newsyslog.8, 1.55->1.63 & 1.65 newsyslog.c]
PR: bin/36553 (2nd half), 29363 (a small part of)
Obtained from: NetBSD (for option -s and flag 'N')
- Busdma conversion.
- Use natural byte order for IEEE1394 packet headers.
SBP related:
- Fix for multiple unit directories in a configuration ROM.
- Increase size of tagged queue.
and many changes and cleanups.
file is specified on the command line, but that filename is not listed in
the newsyslog.conf file. This action is set by having a rule in the
config file with a filename of "<default>".
[* = 1.50 + 1.54 newsyslog.c, 1.39 newsyslog.8]
of the month, give an example of how to rotate logs at the beginning of
the month. Also, Note the equivalence of certain time formats.
(mfc-ing two updates done July 2001)
tested on -stable with a Rage 128 so far. Hooking it up for building into the
kernel will follow, but modules won't be built by default until it is confirmed
to be stable by the people who had reported problems so far.
Add TAG_VENDOR_INDENTIFIER (option 60) to our DHCP request done by the
kernel BOOTP option. The format will be:
FreeBSD:<MACHINE>:<osrelease>
this way people can tune their DHCP server to server up root file systems
via the OS, machine type and version.
PR: 50249 (since MFC reminder seemed to be broken)
for another reader or writer if one arrived and departed while we were
waiting (or a little earlier). This should fix some races, but it is
reported to not actually fix the original problem with sendmail.
The -C option's behavior still has an issue of not affecting
the value of .CURDIR and .OBJDIR variables, as it should be.
This will be addressed later.
PR: bin/47149, bin/50981
In src/lib/libz/gzio.c the function gzprintf does not check if the
amount of bytes (supposed to be) written by vsnprintf exceeds the
size of the buffer.
again. Apparently some code in pmap reads the value of this variable
before it is properly initialized. That is a bug (!) but not a critical
one and it is easier to make this change than hunt down that bug. This
fixes booting on 80386 machines.
Submitted by: Erik Trulsson <[email protected]>
Pointy hat to: jhb
1) "Friday" option to tell calendar which is the day before the weekend, so
it knows when to show you the events for the start of next week.
2) Make sure directories exist before installing the calendar files.
3) Misc datafile updates, additions and corrections.
1.108 Clear up a few warnings.
1.113 When printing the wait status, break it into a signal and a exit status.
1.118 Call login_close so that we don't leak memory.
1.119 Use close-on-exec rather than closing the discriptors manually.
negotiation, if we need ppp.link{up,down}, we couldn't disable
IPCP. Now, if IPCP is disabled, ppp.link{up,down} is invoked
at the end of IPV6CP negotiation.
that allows the destination string to be NULL, and returns the number of
bytes (wcstombs) or wide characters (mbstowcs) that would be required to
store the result of the conversion without storing anything.
PR: 17694
- Mark the process leader as having an advisory lock
- Check if process leader is marked as having advisory lock when
closing file
- Check that file is still open after lock has been obtained
- Don't allow file descriptor table sharing between processes
with different leaders
PR: 10265
without the awi(4), and 860 bytes with awi(4) back. That would be enough
to put awi(4) back, but would leave only 6K of free space on kern.flp to
be useful.
smbfs_rename(). One of the vnodes is on different mount and is possibly on
a different kind of filesystem; treating it as an smbfs vnode then writing
to it will probably corrupt it.
PR: 48381
Move security advisories section to the top, rather than having it
sandwiched between kernel and userland.
New release notes: SA-03:07, sendmail-8.12.9.
Approved by: re (implicitly)
- Clear PCIM_CMD_MWRICEN:
some chips seem to have problem with write invalidate.
clearing this bit fixes SBP timeout problem.
Tested by: Michael Reifenberger <[email protected]>
- Moderate value for latency timer.
Approved by: re (jhb)
the system scheduler was committed, access to these facilities was
limited to the root user (for unclear reasons, perhaps lack of
understanding about the potential impact of the interfacs on
system operation, or due to bugs). However, the JDK requires the
ability to monitor scheduler parameters and selection for
linux-threaded processes; a return of EPERM causes some applications
to fail as a result (in particular, the JDK). In 5.x-CURRENT,
we've replaced the access control with centralized access control
primitives, giving these interfaces controls similar to those
applied for rtprio(), setpriority(), etc, resulting in uniform
enforcement.
In order to enable proper JDK operation for 4.8-RELEASE, work around
the lack of proper access control by permitting the use of two
system calls: sched_getparam() and sched_getscheduler(), for non-root
processes when the call is made on the current process (either using
a 0 pid argument, or curproc->p_pid). While we're here, fix a
bug that caused the result of the call to be returned in the
target process, not in the subject process (ouch!), but that
previously only affected root-owned processes. These fixes are
deemed to be the lowest impact approachin the release; a backport
of the 5.x-CURRENT access control primitives might also be appropriate
in a non-release scenario. This fix doesn't permit the calls to
succeed on other linuxthreads in the same linuxthread process, but
despite those failures the JDK appears to operate properly, so
we've opted not to broaden the scope to permit the p->p_leader ==
targetp->p_leader case at this point.
The "wrong process" return value may apply to other system calls
due to overloading of the subject process pointer with the target
process, but doesn't currently affect non-root processes (and is
fairly uncommon as usually processes are interested in frobbing their
own scheduler details, not other processes, hence it not really
showing up before). This should be fixed in the post 4.8-RELEASE
time frame. These bugs should not be present in 5.x due to
process-locking and security-related changes made in that branch
well prior to 5.0-RELEASE.
Submitted by: mbr (collaborative)
Approved by: re (murray)
This corrects problems encountered when FAST_IPSEC is used together
/dev/crypto.
Reviewed by: Jonathan Stone <[email protected]>
Approved by: re (murray, jb)
We don't have cat(1) and kill(1) on the miniroot disks.
This fix will be committed to the ISC repo later, but for now
take this file out of the vendor tree.
Reviewed by: phk
Approved by: re (murray)
2003-03-23 23:33:59 +00:00
4201 changed files with 551401 additions and 144816 deletions
Loaded 100 of 4201 files, more files were not shown because too many files have changed in this diff.
Show more
Reference in new issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Devin Teske