dragonfly.git
19 years agoI guess I should bumb document date too.
David Rhodus [Tue, 14 Sep 2004 00:22:15 +0000 (00:22 +0000)]
I guess I should bumb document date too.

19 years agoFix printf example.
David Rhodus [Tue, 14 Sep 2004 00:20:26 +0000 (00:20 +0000)]
Fix printf example.

Merged from fbsd.

19 years agoFix typo.
David Rhodus [Tue, 14 Sep 2004 00:13:36 +0000 (00:13 +0000)]
Fix typo.

Noted by: Steve Mynott <steve.mynott@gmail.com>

19 years agodo not send icmp response if the original packet is encrypted.
David Rhodus [Tue, 14 Sep 2004 00:11:07 +0000 (00:11 +0000)]
do not send icmp response if the original packet is encrypted.

Obtained from: KAME
Merged from fbsd.

19 years agoBring in fix from gzip 1.3.3 to avoid crashes when processing certain
David Rhodus [Tue, 14 Sep 2004 00:03:32 +0000 (00:03 +0000)]
Bring in fix from gzip 1.3.3 to avoid crashes when processing certain
corrupt input files:

1999-06-25  Paul Eggert  <eggert@twinsun.com>

        * inflate.c (huft_build):
        Set n to length of v, to detect improper tables.
        Don't accidentally grow j past z.

Merged from fbsd.

19 years agoadd in support Frontier Labs NEX IA+ Digital Audio Player with USB CF
David Rhodus [Mon, 13 Sep 2004 23:59:16 +0000 (23:59 +0000)]
add in support Frontier Labs NEX IA+ Digital Audio Player with USB CF
card reader/writer

Merged from fbsd.

19 years agoAdd in support for the iRiver iFP MP3 player.
David Rhodus [Mon, 13 Sep 2004 23:57:41 +0000 (23:57 +0000)]
Add in support for the iRiver iFP MP3 player.

Merged from fbsd.

19 years agoyar 2004-08-28 12:49:58 UTC
David Rhodus [Mon, 13 Sep 2004 23:52:57 +0000 (23:52 +0000)]
yar         2004-08-28 12:49:58 UTC

 FreeBSD src repository

 Modified files:        (Branch: RELENG_4)
   libexec/ftpd         ftpd.c
 Log:
 MFC 1.156:
 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.

19 years agoAvoid leaving an obsolete pointer while the interface is detaching
David Rhodus [Mon, 13 Sep 2004 23:45:57 +0000 (23:45 +0000)]
Avoid leaving an obsolete pointer while the interface is detaching
and the struct ifnet is being deleted.

Merged from fbsd pr: kern/52260

19 years agoChange the array of char to an array of struct pollfd to avoid an
David Rhodus [Mon, 13 Sep 2004 23:41:18 +0000 (23:41 +0000)]
Change the array of char to an array of struct pollfd to avoid an
address being migaligned.

Merged from fbsd pr: kern/85214

19 years agoANSIfication/cleanup, no functional changes.
Matthew Dillon [Mon, 13 Sep 2004 23:18:20 +0000 (23:18 +0000)]
ANSIfication/cleanup, no functional changes.

19 years agoAllow the syncache to run lock-free in parallel on multiple processors
Jeffrey Hsu [Mon, 13 Sep 2004 19:01:50 +0000 (19:01 +0000)]
Allow the syncache to run lock-free in parallel on multiple processors
by giving each cpu its own syncache hashtable and timer queues.

Joint work with: dillon
Testing provided by: drhodus (earlier version)

19 years agoClean up struct session hold/rele management. The tty half-closed support
Matthew Dillon [Mon, 13 Sep 2004 16:22:41 +0000 (16:22 +0000)]
Clean up struct session hold/rele management.  The tty half-closed support
(i.e. showing 'p0-' in the ps output for the tty instead of '??' after a
process has detached) had an issue where the tty would be left with a
reference to the freed session structure in certain situations because the
session structure's ref-counting code was not properly implementing the
release case.  Consolidate the disparate session ref-counting code into
real sess_hold() and sess_rele() functions and ensure that any tty reference
to the session is cleared before the session structure is free()'d.

NOTE: Joerg noticed a 0xdeadc1de (deadcode) panic related to this issue
which means that prior to this fix it was possible for the bug to cause
memory corruption in certain situations.

NOTE: Linux does not implement half-closed tty sessions like BSD.  Add code
to implement fully-closed tty sessions, and document the whole mess, but
leave it conditionalized-out for now.

Reported-by: Joerg Sonnenberger <joerg@britannica.bec.de>
19 years agoscalb() takes two doubles, not (double, int) (though there appear to be
Matthew Dillon [Mon, 13 Sep 2004 15:53:29 +0000 (15:53 +0000)]
scalb() takes two doubles, not (double, int) (though there appear to be
several versions of scalb() in the tree, it's a real mess).

19 years agoChange hackers@freebsd.org to kernel@crater.dragonflybsd.org.
Jeroen Ruigrok/asmodai [Mon, 13 Sep 2004 13:30:11 +0000 (13:30 +0000)]
Change hackers@freebsd.org to kernel@crater.dragonflybsd.org.

19 years agoMake sio.S aware of when -mrtd is or is not being used, so it can implement
Matthew Dillon [Sun, 12 Sep 2004 18:37:15 +0000 (18:37 +0000)]
Make sio.S aware of when -mrtd is or is not being used, so it can implement
the correct return semantics.

Sio-bug-found-after-agonizing-research-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>
19 years agoRemove a duplicated PCIR_BAR definition.
Joerg Sonnenberger [Sun, 12 Sep 2004 17:40:30 +0000 (17:40 +0000)]
Remove a duplicated PCIR_BAR definition.

19 years agoFix a badly written preprocessor macro.
Matthew Dillon [Sun, 12 Sep 2004 17:19:58 +0000 (17:19 +0000)]
Fix a badly written preprocessor macro.

Adventure-doesn't-work-reported-by: Steve Mynott <steve.mynott@gmail.com>

19 years agoMinor style cleanup. Use err(3) since we are interested in the error message.
Joerg Sonnenberger [Sun, 12 Sep 2004 11:54:00 +0000 (11:54 +0000)]
Minor style cleanup. Use err(3) since we are interested in the error message.

Submitted-by: Liam J. Foy
Make it WARNS=6 clean.

19 years agoImplement a convenient lwkt_initport_null_rport() call which initializes
Matthew Dillon [Fri, 10 Sep 2004 18:23:56 +0000 (18:23 +0000)]
Implement a convenient lwkt_initport_null_rport() call which initializes
a message port with a non-queueing sink-null replyport rather then a default
replyport, rather then rolling our own in the net code and other places where
we may need it.

19 years agoFrom FreeBSD 5:
Joerg Sonnenberger [Fri, 10 Sep 2004 16:40:31 +0000 (16:40 +0000)]
From FreeBSD 5:
  Define some generic eventhandler priorities for use with eventhandlers and
  redefine the existing priority macros for specific eventhandlers in terms
  of the generic priorities.

19 years agoFrom FreeBSD 5:
Joerg Sonnenberger [Fri, 10 Sep 2004 14:02:01 +0000 (14:02 +0000)]
From FreeBSD 5:
  Bring eventhandler callbacks for pf.
  This enables pf to track dynamic address changes on interfaces (dailup) with
  the "on (<ifname>)"-syntax. This also brings hooks in anticipation of
  tracking cloned interfaces, which will be in future versions of pf.

The events are called attach and detach instead of arrival / departue.

19 years agoAdd GID_MAX and UID_MAX.
Joerg Sonnenberger [Fri, 10 Sep 2004 01:45:37 +0000 (01:45 +0000)]
Add GID_MAX and UID_MAX.

19 years agoWith vnode locking now mandatory a number of bugs have cropped up in the
Matthew Dillon [Thu, 9 Sep 2004 20:52:21 +0000 (20:52 +0000)]
With vnode locking now mandatory a number of bugs have cropped up in the
linux emulation code.  Neither the linux path translation code nor the
linprocfs vfs code was following the locking API properly, resulting in
panics.

Reported-by: Andrew Atrens <atrens@nortelnetworks.com> and others
19 years agoTemporary hack to remove historical schg protection on 'cu' to avoid
Matthew Dillon [Thu, 9 Sep 2004 07:24:41 +0000 (07:24 +0000)]
Temporary hack to remove historical schg protection on 'cu' to avoid
an install failure.  Our LINKS rules do not properly deal with the schg bit.

Submitted-by: YONETANI Tomokazu <qhwt+dragonfly-bugs@les.ath.cx>
19 years agoFix some warnings in the code about non-ISO C prototypes and void *
Joerg Sonnenberger [Thu, 9 Sep 2004 03:47:08 +0000 (03:47 +0000)]
Fix some warnings in the code about non-ISO C prototypes and void *
arithmetic.

19 years agoDon't sleep and commit. Fix the last commit.
Joerg Sonnenberger [Wed, 8 Sep 2004 03:21:14 +0000 (03:21 +0000)]
Don't sleep and commit. Fix the last commit.

19 years agoTransform the PCI probe ifs into the normal loop. Merge the controller
Joerg Sonnenberger [Wed, 8 Sep 2004 03:19:19 +0000 (03:19 +0000)]
Transform the PCI probe ifs into the normal loop. Merge the controller
specific function selection into the probe code as well.

19 years agoRemove void casts of function return values. Remove empty lines for missing
Joerg Sonnenberger [Wed, 8 Sep 2004 03:01:11 +0000 (03:01 +0000)]
Remove void casts of function return values. Remove empty lines for missing
local variables.

Side note: Find the guy who introduced (void)free(nbuf) and shoot him.

19 years agoRemove !TIOCSTI code and TIOCEXT conditionals. This have been in the
Joerg Sonnenberger [Tue, 7 Sep 2004 22:59:07 +0000 (22:59 +0000)]
Remove !TIOCSTI code and TIOCEXT conditionals. This have been in the
(FreeBSD) tree for ever (literally!).

19 years agoMake edstop and writeback static.
Joerg Sonnenberger [Tue, 7 Sep 2004 22:42:26 +0000 (22:42 +0000)]
Make edstop and writeback static.

19 years agoMake mark/unmark static. Use getprogname() instead of __progname.
Joerg Sonnenberger [Tue, 7 Sep 2004 22:33:35 +0000 (22:33 +0000)]
Make mark/unmark static. Use getprogname() instead of __progname.

19 years agoSync with FreeBSD. Most importantly, add handling for -e, -H and -F
Joerg Sonnenberger [Tue, 7 Sep 2004 21:31:45 +0000 (21:31 +0000)]
Sync with FreeBSD. Most importantly, add handling for -e, -H and -F
for SUS compliance.

Obtained-from: FreeBSD
Suggested-by: Douwe Kiela <virtus@wanadoo.nl>
19 years agoAdd wlan and crypto for the new wi driver.
Joerg Sonnenberger [Mon, 6 Sep 2004 17:23:22 +0000 (17:23 +0000)]
Add wlan and crypto for the new wi driver.

19 years agoSync driver:
Joerg Sonnenberger [Mon, 6 Sep 2004 16:39:47 +0000 (16:39 +0000)]
Sync driver:
- use a specific malloc type instead of M_DEVBUF
- use local variables to hold longer subexpressions
- merge two consecutive loops in the init code
- add support for newer Adaptec controllers
- use bus_alloc_resource_any

Submitted-by: YONETANI Tomokazu <qhwt+dragonfly-submit@les.ath.cx>
Obtained-from: FreeBSD

19 years agoMinor style cleanups.
Joerg Sonnenberger [Mon, 6 Sep 2004 16:30:10 +0000 (16:30 +0000)]
Minor style cleanups.

Submitted-by: Yonetani Tomokazu <qhwt+dragonfly-submit@les.ath.cx>
Additionally cleanup Makefile as well.

19 years agoImport the new wi(4) driver based on the generic 802.11 layer.
Joerg Sonnenberger [Mon, 6 Sep 2004 13:52:24 +0000 (13:52 +0000)]
Import the new wi(4) driver based on the generic 802.11 layer.

Obtained-from: FreeBSD

19 years agoAdd const for argument of argmatch to fix warnings.
Joerg Sonnenberger [Mon, 6 Sep 2004 13:24:05 +0000 (13:24 +0000)]
Add const for argument of argmatch to fix warnings.

19 years agoThe mount options matching code was incorrectly testing for string
Matthew Dillon [Mon, 6 Sep 2004 01:19:07 +0000 (01:19 +0000)]
The mount options matching code was incorrectly testing for string
termination or '=' of m->option[len], where len = strlen(m->option).  Also,
at that particular point in the code the opt[] string has already been
stripped of any '='.  The original code must have been intended to check
an unstripped opt[] string but with opt[] stripped it really just needs to
do a case case-insensitive match.

Noticed-by: Johannes Hofmann <Johannes.Hofmann@gmx.de>
19 years agoNULL out p_stats when detaching the underlying thread stack, since the
Matthew Dillon [Sun, 5 Sep 2004 21:25:53 +0000 (21:25 +0000)]
NULL out p_stats when detaching the underlying thread stack, since the
pointer will no longer be valid.  This should solve occassional panics
that occur when procfs tries to read an invalid p_stats.

Crash-Reported-by: Rum <rumko@precious.networklinux.net>
19 years agoLeave the USB keyboard registered with the keyboard subsystem even if the
Matthew Dillon [Sun, 5 Sep 2004 21:23:34 +0000 (21:23 +0000)]
Leave the USB keyboard registered with the keyboard subsystem even if the
USB keyboard is unplugged or reset.  USB keyboard resets can occur under a
number of conditions, including during system boot and when other devices
on the same controller are pulled or create problems for the controller.  If
we were to deregister the keyboard then the console would disconnect from it
and the keyboard would become unusable.

Use the new keyboard preference feature and give a USB keyboard a higher
priority then a standard atkbd keyboard.  ATKBD keyboards often probe even
when no keyboard exists, especially if the BIOS was set to USB keyboard
compatibility mode.  This way if there is a USB keyboard connected up, it
will get priority over the AT keyboard even if the system thinks an AT
keyboard exists.

19 years agoUse the new keyboard preference feature.
Matthew Dillon [Sun, 5 Sep 2004 21:19:41 +0000 (21:19 +0000)]
Use the new keyboard preference feature.

19 years agoAdd a keyboard preference to kbd_init_struct(). When a wildcard keyboard is
Matthew Dillon [Sun, 5 Sep 2004 21:19:19 +0000 (21:19 +0000)]
Add a keyboard preference to kbd_init_struct().  When a wildcard keyboard is
allocated the highest priority unallocated keyboard will be returned.

Add kbd_reinit_struct() so USB can leave its keyboard registered.

19 years agoSave current version of wi(4) as owi before switching to generic 802.11
Joerg Sonnenberger [Sun, 5 Sep 2004 13:34:56 +0000 (13:34 +0000)]
Save current version of wi(4) as owi before switching to generic 802.11
version.

Small modifications to allow kernels with both devices included.

if_owi.c: copy of if_wi.c, 1.15
if_owi_pccard.c: copy of if_wi_pccard.c, 1.5
if_owi_pci.c: copy of if_wi_pci.c, 1.4
if_wireg.h: copy of if_wireg.h, 1.3
if_wivar.h: copy of if_wivar.h, 1.4
owi_hostap.c: copy of wi_hostap.c, 1.9
wi_hostap.h: copy of wi_hostap.h, 1.2

19 years agoCreate the dma maps before allocating the memory. Use WAITOK for the
Joerg Sonnenberger [Sun, 5 Sep 2004 13:06:18 +0000 (13:06 +0000)]
Create the dma maps before allocating the memory. Use WAITOK for the
RX and TX buffers.

19 years agoRemove the unit included in the softc, it is not used beside the call to
Joerg Sonnenberger [Sun, 5 Sep 2004 12:36:16 +0000 (12:36 +0000)]
Remove the unit included in the softc, it is not used beside the call to
if_initname and can be fetched via device_get_unit anyway. Don't wire the
interface name to "nv", use the driver name instead.

19 years agoSet OPENSSH_USE_POSIX_THREADS to 1 to work around a privilege separation
Matthew Dillon [Sun, 5 Sep 2004 11:53:58 +0000 (11:53 +0000)]
Set OPENSSH_USE_POSIX_THREADS to 1 to work around a privilege separation
bug with PAM which causes dead sshd processes in [pam] mode to hang around
forever.

19 years agoRemove function casts to (void).
Matthew Dillon [Sun, 5 Sep 2004 02:20:15 +0000 (02:20 +0000)]
Remove function casts to (void).

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
19 years agoRemove #ifdef sgi and related conditionals. Remove improper libc prototypes.
Matthew Dillon [Sun, 5 Sep 2004 02:16:48 +0000 (02:16 +0000)]
Remove #ifdef sgi and related conditionals.  Remove improper libc prototypes.

Inspired-by: "Douwe Kiela" <virtus@wanadoo.nl>
19 years agoMake timed WARNS 2 compatible.
Matthew Dillon [Sun, 5 Sep 2004 02:09:24 +0000 (02:09 +0000)]
Make timed WARNS 2 compatible.

19 years agoAvoid WARNS=2 error by renaming the 'print' variable to 'printerr' to avoid
Matthew Dillon [Sun, 5 Sep 2004 02:02:25 +0000 (02:02 +0000)]
Avoid WARNS=2 error by renaming the 'print' variable to 'printerr' to avoid
collision with the print() function in the same program, and renaming
the 'adjtime' variable to 'adjusttime' to avoid a conflict with adjtime().

Obtained-from: NetBSD
Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
19 years agoReplace all occurences of strcpy by the safe strlcpy where needed.
Matthew Dillon [Sun, 5 Sep 2004 01:59:44 +0000 (01:59 +0000)]
Replace all occurences of strcpy by the safe strlcpy where needed.
Replace all occurences of strncpy by the safer strlcpy where needed.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
19 years agoMinor non-functional syntax changes. Remove (void) casts, and remove
Matthew Dillon [Sun, 5 Sep 2004 01:18:27 +0000 (01:18 +0000)]
Minor non-functional syntax changes.  Remove (void) casts, and remove
the register keyword.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
19 years agoMisc syntax cleanup.
Matthew Dillon [Sun, 5 Sep 2004 00:06:43 +0000 (00:06 +0000)]
Misc syntax cleanup.

19 years agoCorrect a bug introduced in a recent commit. This fixes touch -t.
Matthew Dillon [Sun, 5 Sep 2004 00:03:30 +0000 (00:03 +0000)]
Correct a bug introduced in a recent commit.  This fixes touch -t.

Submitted-by: YONETANI Tomokazu <qhwt+dragonfly-bugs@les.ath.cx>
19 years agoFix a bug in sillyrename handling in nfs_inactive(). The code was improperly
Matthew Dillon [Sat, 4 Sep 2004 23:12:55 +0000 (23:12 +0000)]
Fix a bug in sillyrename handling in nfs_inactive().  The code was improperly
ignoring the lock state of the passed vp and recursing nfs_inactive() by
calling vrele() from within nfs_inactive().  Since NFS uses real vnode
locking now, this resulted in a panic.

KDE startup problems reported by:  Emiel Kollof <coolvibe@hackerheaven.org>

19 years agoAdd noops for vga_pxlmouse_direct and vga_pxlmouse_planar in the case where
Eirik Nygaard [Sat, 4 Sep 2004 12:03:14 +0000 (12:03 +0000)]
Add noops for vga_pxlmouse_direct and vga_pxlmouse_planar in the case where
SC_NO_CUTPASTE is defined. Also remove the old vga_pxlmouse noop.

19 years agoAdd VESA mode support for syscons. The vesa.ko module must be loaded and
Matthew Dillon [Sat, 4 Sep 2004 06:15:08 +0000 (06:15 +0000)]
Add VESA mode support for syscons.  The vesa.ko module must be loaded and
syscons (sc0) must be compiled with 'options SC_PIXEL_MODE'.
'vidcontrol -i mode' may then be used to list available modes and
'vidcontrol MODE_n' may be used to set the mode.  This patch adds support
for 15, 16, 24, and 32 bit modes and also does a bit of related cleanup and
restructuring.

Submitted-by: Sascha Wildner <saw@online.de>
Testing-and-Discussion: Sarunas Vancevicius, YONETANI Tomokazu and Joerg Anslik

19 years agoReenable pkill. There wasn't actually a problem with it, it turned out to
Matthew Dillon [Sat, 4 Sep 2004 04:00:39 +0000 (04:00 +0000)]
Reenable pkill.  There wasn't actually a problem with it, it turned out to
be stale source on my part.

19 years agoWith the 'cu' program removed from the system (long ago), enable 'cu'
Matthew Dillon [Fri, 3 Sep 2004 20:48:16 +0000 (20:48 +0000)]
With the 'cu' program removed from the system (long ago), enable 'cu'
emulation in 'tip' to replace the functionality.

From discussions with: Jeroen Ruigrok/asmodai <asmodai@wxs.nl>, others

19 years agoRemove some unnecessary extern's of libc functions.
Matthew Dillon [Fri, 3 Sep 2004 20:38:01 +0000 (20:38 +0000)]
Remove some unnecessary extern's of libc functions.

19 years agopkill does not compile cleanly, remove it from the build.
Matthew Dillon [Fri, 3 Sep 2004 19:47:20 +0000 (19:47 +0000)]
pkill does not compile cleanly, remove it from the build.

19 years agoAdd the -8 option to finger for 8-bit output (from NetBSD).
Matthew Dillon [Fri, 3 Sep 2004 19:13:23 +0000 (19:13 +0000)]
Add the -8 option to finger for 8-bit output (from NetBSD).

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
19 years agoMinor style cleanups. Start a list of obsolete functions that should no
Matthew Dillon [Fri, 3 Sep 2004 18:40:13 +0000 (18:40 +0000)]
Minor style cleanups.  Start a list of obsolete functions that should no
longer be used, and add a warning about the use of unbounded string functions.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
Additional-work-by: Matt Dillon
19 years agoFix the way warnings are printed. Use warnx(3) if the message is provided,
Joerg Sonnenberger [Fri, 3 Sep 2004 14:30:41 +0000 (14:30 +0000)]
Fix the way warnings are printed. Use warnx(3) if the message is provided,
default to warn(3) otherwise.

Noticed-by: corecode
19 years agoMove the simple pager from db_ps.c into the more obvious file db_output.c and
Eirik Nygaard [Fri, 3 Sep 2004 08:50:47 +0000 (08:50 +0000)]
Move the simple pager from db_ps.c into the more obvious file db_output.c and
remove the static declaration of it so other files also can take use of it.

19 years agosendsys.h is no longer needed.
Eirik Nygaard [Thu, 2 Sep 2004 11:21:40 +0000 (11:21 +0000)]
sendsys.h is no longer needed.

19 years agoRedo the sysmsg test programs to match the changes done in the kernel to
Eirik Nygaard [Thu, 2 Sep 2004 11:21:12 +0000 (11:21 +0000)]
Redo the sysmsg test programs to match the changes done in the kernel to
properly support synchronous and asynchronous syscall messages.

19 years agoUpdate per latest manual pages after 'man-update'.
Jeroen Ruigrok/asmodai [Thu, 2 Sep 2004 09:35:38 +0000 (09:35 +0000)]
Update per latest manual pages after 'man-update'.

19 years agoUpdate per latest manual pages after running 'man-update'.
Jeroen Ruigrok/asmodai [Thu, 2 Sep 2004 09:31:49 +0000 (09:31 +0000)]
Update per latest manual pages after running 'man-update'.

19 years agoUpdate per latest manual pages.
Jeroen Ruigrok/asmodai [Thu, 2 Sep 2004 09:27:41 +0000 (09:27 +0000)]
Update per latest manual pages.

19 years agoCommit manual pages after running 'man-update' and add new manual pages.
Jeroen Ruigrok/asmodai [Thu, 2 Sep 2004 09:26:58 +0000 (09:26 +0000)]
Commit manual pages after running 'man-update' and add new manual pages.

19 years agoSwitch from OpenSSL 0.9.7a to OpenSSL 0.9.7d.
Jeroen Ruigrok/asmodai [Thu, 2 Sep 2004 08:57:05 +0000 (08:57 +0000)]
Switch from OpenSSL 0.9.7a to OpenSSL 0.9.7d.

19 years agoFix it properly.
Jeroen Ruigrok/asmodai [Thu, 2 Sep 2004 05:45:20 +0000 (05:45 +0000)]
Fix it properly.

Submitted by: Simon 'corecode' Schubert

19 years agoFix a problem introduced by the last commit.
David Rhodus [Wed, 1 Sep 2004 18:49:25 +0000 (18:49 +0000)]
Fix a problem introduced by the last commit.

19 years agoSynchronize with our current code.
Jeroen Ruigrok/asmodai [Wed, 1 Sep 2004 14:30:05 +0000 (14:30 +0000)]
Synchronize with our current code.
Overhaul some other parts to keep it concise.

19 years agoFirst stab at getting the Silicon Image (SiI) SATA controllers 3112 and
Jeroen Ruigrok/asmodai [Wed, 1 Sep 2004 14:13:55 +0000 (14:13 +0000)]
First stab at getting the Silicon Image (SiI) SATA controllers 3112 and
3124 working.

19 years agoMLINK sata(4) to ata(4).
Jeroen Ruigrok/asmodai [Wed, 1 Sep 2004 12:11:16 +0000 (12:11 +0000)]
MLINK sata(4) to ata(4).

19 years agoMerge from vendor branch OPENSSL:
Jeroen Ruigrok/asmodai [Tue, 31 Aug 2004 20:02:04 +0000 (20:02 +0000)]
Merge from vendor branch OPENSSL:
Add OpenSSL 0.9.7d.

19 years agoAdd OpenSSL 0.9.7d.
Jeroen Ruigrok/asmodai [Tue, 31 Aug 2004 20:02:04 +0000 (20:02 +0000)]
Add OpenSSL 0.9.7d.

19 years agoRemove OpenSSH-3.8.1p1 from base.
Scott Ullrich [Tue, 31 Aug 2004 00:36:10 +0000 (00:36 +0000)]
Remove OpenSSH-3.8.1p1 from base.

19 years agoMerge from vendor branch OPENSSH:
Scott Ullrich [Mon, 30 Aug 2004 21:59:58 +0000 (21:59 +0000)]
Merge from vendor branch OPENSSH:
Import OpenSSH-3.9p1 from openbsd.org

Assistance-from: Chris Pressey, Jörg Sonnenberger
Testing-by: Guillermo García Rojas C, Chris Pressey & Jeroen Ruigrok van der Werven

19 years agoImport OpenSSH-3.9p1 from openbsd.org
Scott Ullrich [Mon, 30 Aug 2004 21:59:58 +0000 (21:59 +0000)]
Import OpenSSH-3.9p1 from openbsd.org

Assistance-from: Chris Pressey, Jörg Sonnenberger
Testing-by: Guillermo García Rojas C, Chris Pressey & Jeroen Ruigrok van der Werven

19 years agoImport OpenSSH-3.9p1 from openbsd.org
Scott Ullrich [Mon, 30 Aug 2004 21:59:58 +0000 (21:59 +0000)]
Import OpenSSH-3.9p1 from openbsd.org

Assistance-from: Chris Pressey, Jörg Sonnenberger
Testing-by: Guillermo García Rojas C, Chris Pressey & Jeroen Ruigrok van der Werven

19 years agoUse strchr instead of index, and strrchr instead of rindex because the str*chr
Eirik Nygaard [Mon, 30 Aug 2004 19:27:22 +0000 (19:27 +0000)]
Use strchr instead of index, and strrchr instead of rindex because the str*chr
functions are standard compliant.

index() and rindex() do the same thing as repectively strchr() and strrchr() do.

Submitted by: Douwe Kiela <virtus@wanadoo.nl>

19 years agoUse strchr instead of index, and strrchr instead of rindex because the str*chr
Eirik Nygaard [Mon, 30 Aug 2004 18:06:50 +0000 (18:06 +0000)]
Use strchr instead of index, and strrchr instead of rindex because the str*chr
functions are standard compliant.

index() and rindex() do the same thing as repectively strchr() and strrchr() do.

Submitted by: Douwe Kiela <virtus@wanadoo.nl>

19 years agoAdd 136: UDPLite [UDP for error prone networks]
Jeroen Ruigrok/asmodai [Mon, 30 Aug 2004 15:10:50 +0000 (15:10 +0000)]
Add 136: UDPLite [UDP for error prone networks]
Add 253, 254: Free for experimentation and testing.
In effect this means that divert (254) is on a slot that
more or less suits it.

19 years ago-1, not -15.
Jeroen Ruigrok/asmodai [Mon, 30 Aug 2004 06:17:16 +0000 (06:17 +0000)]
-1, not -15.

19 years agoAdd missing entries for fo_FO, lt_LT, Argentina.
Jeroen Ruigrok/asmodai [Mon, 30 Aug 2004 05:17:14 +0000 (05:17 +0000)]
Add missing entries for fo_FO, lt_LT, Argentina.

19 years agoUse proper filenames for the Faroese locales.
Jeroen Ruigrok/asmodai [Sun, 29 Aug 2004 16:14:25 +0000 (16:14 +0000)]
Use proper filenames for the Faroese locales.

19 years agoAdd OpenSSL 0.9.7d.
Jeroen Ruigrok/asmodai [Sun, 29 Aug 2004 12:45:27 +0000 (12:45 +0000)]
Add OpenSSL 0.9.7d.

19 years agoMerge from vendor branch OPENSSL:
Jeroen Ruigrok/asmodai [Sun, 29 Aug 2004 12:45:27 +0000 (12:45 +0000)]
Merge from vendor branch OPENSSL:
Add OpenSSL 0.9.7d.

19 years agoForced commit to note that site_perl/5.005/i386-freebsd is now:
Jeroen Ruigrok/asmodai [Sun, 29 Aug 2004 09:57:46 +0000 (09:57 +0000)]
Forced commit to note that site_perl/5.005/i386-freebsd is now:
site_perl/5.005/dragonfly.

19 years agoGet rid of the alpha entries.
Jeroen Ruigrok/asmodai [Sun, 29 Aug 2004 09:46:37 +0000 (09:46 +0000)]
Get rid of the alpha entries.

19 years agoFix a bug that was causing a 'lockmgr: draining against myself' panic when
Matthew Dillon [Sat, 28 Aug 2004 23:04:23 +0000 (23:04 +0000)]
Fix a bug that was causing a 'lockmgr: draining against myself' panic when
you operated on a NULL mount of a NFS partition.  NFS's lookup() code was not
properly setting the CNP_PDIRUNLOCK flag.  NULLFS's lookup code depends on
this flag to determine the proper lock state for the nullfs layer's vnode.

19 years agoAdd first stab at Faroese locale.
Jeroen Ruigrok/asmodai [Sat, 28 Aug 2004 22:39:04 +0000 (22:39 +0000)]
Add first stab at Faroese locale.

19 years agoFix handling of the recycling of vnodes from a failed hash collision.
Matthew Dillon [Sat, 28 Aug 2004 21:32:28 +0000 (21:32 +0000)]
Fix handling of the recycling of vnodes from a failed hash collision.

Fix a hash table scan that was causing the root mount point to collide
when the same source directory is null-mounted multiple times.

Reported-by: esmith <esmith@patmedia.net>
19 years agoVFS messaging/interfacing work stage 4/99. This stage goes a long ways
Matthew Dillon [Sat, 28 Aug 2004 19:02:35 +0000 (19:02 +0000)]
VFS messaging/interfacing work stage 4/99.  This stage goes a long ways
towards allowing us to move the vnode locking into a kernel layer.  It
gets rid of a lot of cruft from FreeBSD-4.  FreeBSD-5 has done some of this
stuff too (such as changing the default locking to stdlock from nolock),
but DragonFly is going further.

* Consolidate vnode locks into the vnode structure, add an embedded v_lock,
  and getting rid of both v_vnlock and v_data based head-of-structure locks.

* Change the default vops to use a standard vnode lock rather then a fake
  non-lock.

* Get rid of vop_nolock() and friends, we no longer support non-locking
  vnodes.

* Get rid of vop_sharedlock(), we no longer support non standard shared-only
  locks (only NFS was using it and the mount-crossing lookup code should
  now prevent races to root from dead NFS volumes).

* Integrate lock initialization into getnewvnode().  We do not yet
  incorporate automatically locking into getnewvnode().  getnewvnode()
  now has two additional arguments, lktimeout and lkflags, for lock
  structure initialization.

* Change the sync vnode lock from nolock to stdlock.  This may require more
  tuning down the line.  Fix various sync_inactive() to properly unlock
  the lock as per the VOP API.

* Properly flag the 'rename' vop operation regarding required tdvp and tvp
  unlocks (the flags are only used by nullfs).

* Get rid of all inode-embedded vnode locks

* Remove manual lockinit and use new getnewvnode() args instead.
  Lock the vnode prior to doing anything that might block in
  order to avoid synclist access before the vnode has been properly
  initialize.

* Generally change inode hash insertion to also check
  for a hash collision and return failure if it occurs,
  rather then doing (often non-atomic) relookups and
  other checks.  These sorts of collisions can occur
  if a vnode is being destroyed at the same time a new
  vnode is being created from an inode.  A new vnode is
  not generally accessible, except by the sync code (from
  the mountlist) until it's underlying inode has been hashed
  so dealing with a hash collision should be as simple as
  throwing away the vnode with a vput().

* Do not initialize a new vnode's v_data until after
  the associated inode has been successfully added to
  the hash, and make the xxx_inactive() and xxx_reclaim()
  code friendly towards vnodes with a NULL v_data.

* NFS now uses standard locks rather then shared-only locks.

* PROCFS now uses standard locks rather then non-locks, and PROCFS's
  lookup code now understands VOP lookup semantics.  PROCFS now uses
  a real hash table for its node search rather then a single singly-linked
  list (which should better scale to systems with thousands of processes).

* NULLFS should now properly handle lookup() and rename() locks.  NULLFS's
  node handling code has been rewritten.  NULLFS's bypass code now understands
  vnode unlocks (rename case).

* UFS no longer needs the ffs_inode_hash_lock hacks.  It now uses the new
  collision-on-hash-add methodology.   This will speed up UFS when operating
  on lots of small files (reported by David Rhodus).

19 years agoSince ip_input() truncates the packet to ip->ip_len prior to entering the
Matthew Dillon [Sat, 28 Aug 2004 18:33:03 +0000 (18:33 +0000)]
Since ip_input() truncates the packet to ip->ip_len prior to entering the
protocol stack, ip_demux must incorporate an ip_len check in its tcp/udp
prechecks to avoid an assertion in the tcp/udp stacks if the packet is
malformed.

This is a temporary hack until Jeff and I can come up with a better way
to do per-protocol mbuf checks.  Basically the issue involved is that we
want to pull up the entire tcp/udp/etc... header in order to be able to
trivially demux an IP packet (choose which protocol thread to route it to).
Since we have to m_pullup the packet in the demux we would rather do all
header length checks in the demux as well so as not have to repeat them in
the protocol code.

Reported-by: Sven Willenberger <sven@dmv.com>
In-discussion-with: Jeffrey Hsu

19 years agoGet rid of COPY, its functionality has been superseded by install -C.
Jeroen Ruigrok/asmodai [Sat, 28 Aug 2004 17:57:51 +0000 (17:57 +0000)]
Get rid of COPY, its functionality has been superseded by install -C.

Inspired by: a FreeBSD commit message