14 years agoDon't prematurely check for non-active descriptors by looking up the
Simon Schubert [Tue, 1 Aug 2006 08:25:58 +0000 (08:25 +0000)]
Don't prematurely check for non-active descriptors by looking up the
entry in _thread_fd_table, as it might have not been populated before.
_FD_LOCK already takes care of filtering out non-existing descriptors.

Reported-by: Thomas E. Spanjaard
14 years agoUpdate the documentation to reflect current state: d_reclen is dead.
Simon Schubert [Mon, 31 Jul 2006 19:29:41 +0000 (19:29 +0000)]
Update the documentation to reflect current state:  d_reclen is dead.

14 years agoMove include under _KERNEL protection.
Simon Schubert [Mon, 31 Jul 2006 12:17:39 +0000 (12:17 +0000)]
Move include under _KERNEL protection.

This reduces namespace pollution and allows C++ sources including
netinet/if_ether.h to be compiled again.

Reported-by: Joe Talbott <josepht@cstone.net>
14 years agoAdd amd64 support
Simon Schubert [Mon, 31 Jul 2006 12:12:08 +0000 (12:12 +0000)]
Add amd64 support

14 years agoFix an incorrect #ifndef label. Also remove a now unnecessary
Matthew Dillon [Sun, 30 Jul 2006 16:49:34 +0000 (16:49 +0000)]
Fix an incorrect #ifndef label.  Also remove a now unnecessary
#include <sys/msgport.h>.

Reported-by: Joe Talbott <josepht@ntelos.net>
14 years agoRather than calling mircotime() in catchpacket(), make catchpacket()
Sepherosa Ziehau [Sun, 30 Jul 2006 09:39:27 +0000 (09:39 +0000)]
Rather than calling mircotime() in catchpacket(), make catchpacket()
take a timeval indicating when the packet was captured. Move
microtime() to the calling functions and grab the timestamp as soon
as we know that we're going to call catchpacket at least once.

This means that we call microtime() once per matched packet, as
opposed to once per matched packet per bpf listener. It also means
that we return the same timestamp to all bpf listeners, rather than
slightly different ones.

It would be more accurate to call microtime() even earlier for all
packets, but microtime() can be costly, so this didn't seem like a good idea.

Obtained-from: FreeBSD (dwmalone@freebsd.org)
Remined-and-Reviewed-by: joerg
14 years agoruserpass() takes 4 arguments but is called with only 3. Since we
Sascha Wildner [Sun, 30 Jul 2006 07:50:28 +0000 (07:50 +0000)]
ruserpass() takes 4 arguments but is called with only 3. Since we
don't actually use acct, just take a dummy pointer.

14 years agoRemove duplicate code line.
Matthew Dillon [Sun, 30 Jul 2006 06:49:39 +0000 (06:49 +0000)]
Remove duplicate code line.

Reported-by: "Bill Marquette" <bill.marquette@gmail.com>
Taken-from: OpenBSD

14 years agoRemove brconfig(8) from the build.
Sascha Wildner [Sat, 29 Jul 2006 18:56:28 +0000 (18:56 +0000)]
Remove brconfig(8) from the build.

14 years agoRetire brconfig. Its functionality is covered in ifconfig.
Simon Schubert [Sat, 29 Jul 2006 12:11:11 +0000 (12:11 +0000)]
Retire brconfig.  Its functionality is covered in ifconfig.

14 years agoLast commit changed the function arguments of userland version of
YONETANI Tomokazu [Sat, 29 Jul 2006 03:49:01 +0000 (03:49 +0000)]
Last commit changed the function arguments of userland version of
iplioctl() to use (struct dev_ioctl_args *) only for ip_fil.c.
Adjust it to make it work again.

14 years agoWhy is ip_fil.h trying to declare kernel procedures for userland #include's?
Matthew Dillon [Fri, 28 Jul 2006 07:31:08 +0000 (07:31 +0000)]
Why is ip_fil.h trying to declare kernel procedures for userland #include's?
Remove them.

14 years agoFor the moment adjust dd to find the new location of the device type
Matthew Dillon [Fri, 28 Jul 2006 06:24:03 +0000 (06:24 +0000)]
For the moment adjust dd to find the new location of the device type
flags (D_*)

Reported-by: YONETANI Tomokazu <qhwt+dfly@les.ath.cx>
14 years agoMASSIVE reorganization of the device operations vector. Change cdevsw
Matthew Dillon [Fri, 28 Jul 2006 02:17:41 +0000 (02:17 +0000)]
MASSIVE reorganization of the device operations vector.  Change cdevsw
to dev_ops.  dev_ops is a syslink-compatible operations vector structure
similar to the vop_ops structure used by vnodes.

Remove a huge number of instances where a thread pointer is still being
passed as an argument to various device ops and other related routines.
The device OPEN and IOCTL calls now take a ucred instead of a thread pointer,
and the CLOSE call no longer takes a thread pointer.

14 years agoFix a minor bug that prevented compilation.
Matthew Dillon [Fri, 28 Jul 2006 02:11:53 +0000 (02:11 +0000)]
Fix a minor bug that prevented compilation.

14 years agoMake a few more #define's visible when compiling with _KERNEL_STRUCTURES
Matthew Dillon [Fri, 28 Jul 2006 02:11:31 +0000 (02:11 +0000)]
Make a few more #define's visible when compiling with _KERNEL_STRUCTURES

14 years agoUpdate amd64 bits of libc's math routines
Simon Schubert [Thu, 27 Jul 2006 00:46:57 +0000 (00:46 +0000)]
Update amd64 bits of libc's math routines

Obtained-from: NetBSD

14 years agoCorrect TLS and PIC addressing
Simon Schubert [Thu, 27 Jul 2006 00:44:27 +0000 (00:44 +0000)]
Correct TLS and PIC addressing

14 years agoRemove get/make/signalcontext.
Simon Schubert [Thu, 27 Jul 2006 00:44:20 +0000 (00:44 +0000)]
Remove get/make/signalcontext.
SUSv3 calls it "obsolescent" and i386 doesn't have it either.

14 years agoadd header glue
Simon Schubert [Thu, 27 Jul 2006 00:43:56 +0000 (00:43 +0000)]
add header glue

14 years agoAlways create __sys_name syscalls with weak symbols _name and name
Simon Schubert [Thu, 27 Jul 2006 00:43:42 +0000 (00:43 +0000)]
Always create __sys_name syscalls with weak symbols _name and name

Obtained-from: FreeBSD

14 years agoAdd amd64 atomic_lock implementation
Simon Schubert [Thu, 27 Jul 2006 00:43:35 +0000 (00:43 +0000)]
Add amd64 atomic_lock implementation

Obtained-from: FreeBSD

14 years agoThis is suitable for amd64 as well
Simon Schubert [Thu, 27 Jul 2006 00:43:17 +0000 (00:43 +0000)]
This is suitable for amd64 as well

14 years agoTell malloc.c about amd64
Simon Schubert [Thu, 27 Jul 2006 00:43:09 +0000 (00:43 +0000)]
Tell malloc.c about amd64

14 years agoAdd amd64 defines
Simon Schubert [Thu, 27 Jul 2006 00:43:02 +0000 (00:43 +0000)]
Add amd64 defines

Obtained-from: FreeBSD

14 years agoinclude the correct headers
Simon Schubert [Thu, 27 Jul 2006 00:42:54 +0000 (00:42 +0000)]
include the correct headers

14 years agoImport amd64 MD headers
Simon Schubert [Thu, 27 Jul 2006 00:42:46 +0000 (00:42 +0000)]
Import amd64 MD headers

Obtained-from: FreeBSD and our sys/i386

14 years agoCompile libc_r for all archs
Simon Schubert [Thu, 27 Jul 2006 00:41:38 +0000 (00:41 +0000)]
Compile libc_r for all archs

14 years agoBuild battd only for i386 (uses apm)
Simon Schubert [Thu, 27 Jul 2006 00:41:31 +0000 (00:41 +0000)]
Build battd only for i386 (uses apm)

14 years agoAdd defines for amd64
Simon Schubert [Thu, 27 Jul 2006 00:41:24 +0000 (00:41 +0000)]
Add defines for amd64

Obtained-from: FreeBSD

14 years agoPull up changes from i386
Simon Schubert [Thu, 27 Jul 2006 00:41:17 +0000 (00:41 +0000)]
Pull up changes from i386

14 years agoUse %zd for size_t printf()s
Simon Schubert [Thu, 27 Jul 2006 00:41:10 +0000 (00:41 +0000)]
Use %zd for size_t printf()s

14 years agoremove i386-only conditionalisation
Simon Schubert [Thu, 27 Jul 2006 00:41:03 +0000 (00:41 +0000)]
remove i386-only conditionalisation

14 years agoDo obscure va_list handling also on amd64
Simon Schubert [Thu, 27 Jul 2006 00:40:55 +0000 (00:40 +0000)]
Do obscure va_list handling also on amd64

Obtained-from: FreeBSD

14 years agoAdd amd64 support
Simon Schubert [Thu, 27 Jul 2006 00:40:35 +0000 (00:40 +0000)]
Add amd64 support

Obtained-from: FreeBSD

14 years agoAdd amd64 target sources
Simon Schubert [Thu, 27 Jul 2006 00:39:53 +0000 (00:39 +0000)]
Add amd64 target sources

Obtained-from: FreeBSD

14 years agoAdd correct register offsets
Simon Schubert [Thu, 27 Jul 2006 00:39:40 +0000 (00:39 +0000)]
Add correct register offsets

14 years agoRemove unneeded patches
Simon Schubert [Thu, 27 Jul 2006 00:39:33 +0000 (00:39 +0000)]
Remove unneeded patches

14 years agoRestructure Makefiles to accomodate multiple archs
Simon Schubert [Thu, 27 Jul 2006 00:35:02 +0000 (00:35 +0000)]
Restructure Makefiles to accomodate multiple archs

14 years agoRestructure Makefiles to accomodate multiple archs
Simon Schubert [Thu, 27 Jul 2006 00:30:10 +0000 (00:30 +0000)]
Restructure Makefiles to accomodate multiple archs

14 years agoLink libgcc_pic for shared libs
Simon Schubert [Thu, 27 Jul 2006 00:29:57 +0000 (00:29 +0000)]
Link libgcc_pic for shared libs

14 years agoBuild amd64 with the i486 files
Simon Schubert [Thu, 27 Jul 2006 00:29:49 +0000 (00:29 +0000)]
Build amd64 with the i486 files

14 years agoMerge from vendor branch GDB:
Simon Schubert [Thu, 27 Jul 2006 00:29:42 +0000 (00:29 +0000)]
Merge from vendor branch GDB:
restore erroneously removed header

14 years agorestore erroneously removed header
Simon Schubert [Thu, 27 Jul 2006 00:29:42 +0000 (00:29 +0000)]
restore erroneously removed header

14 years agoComment out empty BUGS section.
Sascha Wildner [Wed, 26 Jul 2006 20:21:07 +0000 (20:21 +0000)]
Comment out empty BUGS section.

14 years agoDocument the recently added WITHOUT_SRCS variable.
Sascha Wildner [Wed, 26 Jul 2006 20:18:54 +0000 (20:18 +0000)]
Document the recently added WITHOUT_SRCS variable.

14 years agoNOUUCP was removed a long time ago from the system.
Sascha Wildner [Wed, 26 Jul 2006 18:26:13 +0000 (18:26 +0000)]
NOUUCP was removed a long time ago from the system.

14 years agoUpdate upgrade procedure and adjust disk usage.
Sascha Wildner [Wed, 26 Jul 2006 18:23:16 +0000 (18:23 +0000)]
Update upgrade procedure and adjust disk usage.

14 years agoFix a bug where the VMAYHAVELOCKS flag on a vnode may get lost, resulting
Matthew Dillon [Tue, 25 Jul 2006 20:01:50 +0000 (20:01 +0000)]
Fix a bug where the VMAYHAVELOCKS flag on a vnode may get lost, resulting
in an assertion during exit.

Fix a bug where F_NOEND is not properly set in one of the merge cases.  This
only effects very rare F_GETLK cases.

Reported-by: Joerg Sonnenberger <joerg@britannica.bec.de>
14 years agoAdd code to dump lockf locks associated with a vnode.
Matthew Dillon [Tue, 25 Jul 2006 20:00:56 +0000 (20:00 +0000)]
Add code to dump lockf locks associated with a vnode.

14 years agoRemove devd specific documentation.
Sascha Wildner [Sun, 23 Jul 2006 07:47:12 +0000 (07:47 +0000)]
Remove devd specific documentation.

14 years agoFix date format.
Sascha Wildner [Sun, 23 Jul 2006 06:01:46 +0000 (06:01 +0000)]
Fix date format.

14 years agoGet rid of .Pp before .Sh
Sascha Wildner [Sat, 22 Jul 2006 23:55:30 +0000 (23:55 +0000)]
Get rid of .Pp before .Sh

14 years agoUse .Cd for kernel configuration options.
Sascha Wildner [Sat, 22 Jul 2006 23:52:23 +0000 (23:52 +0000)]
Use .Cd for kernel configuration options.

14 years agoUse full month name for .Dd
Sascha Wildner [Sat, 22 Jul 2006 18:49:25 +0000 (18:49 +0000)]
Use full month name for .Dd

14 years ago- Move error exit in front of tsleep_interlock()
Simon Schubert [Fri, 21 Jul 2006 00:51:13 +0000 (00:51 +0000)]
- Move error exit in front of tsleep_interlock()
- invoke tsleep_interlock and tsleep in a critical section
- release the serializer on error exit

14 years agoDefault kdump's data limit to 64 bytes and document how it can be disabled.
Simon Schubert [Thu, 20 Jul 2006 22:57:47 +0000 (22:57 +0000)]
Default kdump's data limit to 64 bytes and document how it can be disabled.

14 years agoCheck the the ops mount pointer is not NULL before indirecting through it.
Matthew Dillon [Thu, 20 Jul 2006 20:16:24 +0000 (20:16 +0000)]
Check the the ops mount pointer is not NULL before indirecting through it.
The mount pointer can be NULL if the vnode is dead.  This can occur if a
tty is revoke()'d.

Reported-by: walt <wa1ter@myrealbox.com>, Thomas Schlesinger <schlesinger@netcologne.de>
14 years agoPull in fixes from FreeBSD to make -e handling work better.
Simon Schubert [Thu, 20 Jul 2006 17:01:22 +0000 (17:01 +0000)]
Pull in fixes from FreeBSD to make -e handling work better.

Obtained-from: FreeBSD
Noticed-by: joerg, Jiawei Ye <leafy@moderntimes.com.cn> (indirect)
14 years agoFix bug introduced in rev 1.33 and make a clearer code flow.
Simon Schubert [Thu, 20 Jul 2006 16:22:48 +0000 (16:22 +0000)]
Fix bug introduced in rev 1.33 and make a clearer code flow.

Noticed-by: y0netani
14 years agoregen
Sepherosa Ziehau [Wed, 19 Jul 2006 13:38:56 +0000 (13:38 +0000)]

14 years agoAdd device ids for various NVIDIA's GigE
Sepherosa Ziehau [Wed, 19 Jul 2006 13:34:53 +0000 (13:34 +0000)]
Add device ids for various NVIDIA's GigE

14 years agoMention all arguments and fix wording.
Sascha Wildner [Wed, 19 Jul 2006 07:49:21 +0000 (07:49 +0000)]
Mention all arguments and fix wording.

14 years agoIntroduce sys/syslink.h, the beginnings of a VOP-compatible RPC-like
Matthew Dillon [Wed, 19 Jul 2006 06:08:14 +0000 (06:08 +0000)]
Introduce sys/syslink.h, the beginnings of a VOP-compatible RPC-like
communications infrastructure that will be used for userland VFS and
communications between hosts in a cluster.

Begin merging the vnode operations vector code with syslink by replacing
vnodeop_desc with syslink_desc.  Also get rid of a lot of junk related
to vnodeop_desc that is no longer used.

14 years agoGet rid of the weird coda VOP function arguments and void casts and
Matthew Dillon [Wed, 19 Jul 2006 05:59:54 +0000 (05:59 +0000)]
Get rid of the weird coda VOP function arguments and void casts and
replace with the proper arguments.

14 years agoRemove several layers in the vnode operations vector init code. Declare
Matthew Dillon [Tue, 18 Jul 2006 22:22:16 +0000 (22:22 +0000)]
Remove several layers in the vnode operations vector init code.  Declare
the operations vector directly instead of via a descriptor array.  Remove
most of the recalculation code, it stopped being needed over a year ago.

This work is similar to what FreeBSD now does, but was developed along a
different line.  Ultimately our vop_ops will become SYSLINK ops for userland
VFS and clustering support.

14 years agoAdd a fairly bad hack to detect ripouts that might occur during a list
Matthew Dillon [Tue, 18 Jul 2006 02:03:11 +0000 (02:03 +0000)]
Add a fairly bad hack to detect ripouts that might occur during a list
scan.  This does not fix any currently known bugs.

14 years agoAdjust HEAD version from 1.5 to 1.7.
Matthew Dillon [Mon, 17 Jul 2006 00:12:01 +0000 (00:12 +0000)]
Adjust HEAD version from 1.5 to 1.7.

14 years agoAdd a new target for cvs administration of the 1.6 slip tag. Add a cvsup
Matthew Dillon [Mon, 17 Jul 2006 00:04:15 +0000 (00:04 +0000)]
Add a new target for cvs administration of the 1.6 slip tag.  Add a cvsup
file example that will maintain a 1.6 release.

14 years agoBump sub-versions and DragonFly_version in preparation for branching.
Matthew Dillon [Mon, 17 Jul 2006 00:00:04 +0000 (00:00 +0000)]
Bump sub-versions and DragonFly_version in preparation for branching.

14 years agoAdd a note on where to find the release engineering document.
Matthew Dillon [Sun, 16 Jul 2006 23:06:56 +0000 (23:06 +0000)]
Add a note on where to find the release engineering document.

14 years agoAttempt to fix an occassional panic in pf_purge_expired_states() by
Matthew Dillon [Sun, 16 Jul 2006 22:42:23 +0000 (22:42 +0000)]
Attempt to fix an occassional panic in pf_purge_expired_states() by
replacing an unsafe loop iteration with a safe one.

Reported-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>
Note: Simon's crash only occurs once a month or so, so this patch is
not definitive.

14 years agoUse pbufs instead of ebufs.
Matthew Dillon [Sun, 16 Jul 2006 22:39:42 +0000 (22:39 +0000)]
Use pbufs instead of ebufs.

Submitted-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>
14 years agoMake dlsym() search all (recursively) loaded objects and not just
Simon Schubert [Sun, 16 Jul 2006 22:15:38 +0000 (22:15 +0000)]
Make dlsym() search all (recursively) loaded objects and not just
the specified one alone.

Submitted-by: Kostik Belousov <kostikbel@gmail.com> via freebsd-hackers
14 years agoAdd missing prototype.
Matthew Dillon [Sun, 16 Jul 2006 21:44:18 +0000 (21:44 +0000)]
Add missing prototype.

14 years agolibssl needs libcrypto, so link it
Simon Schubert [Sun, 16 Jul 2006 12:28:11 +0000 (12:28 +0000)]
libssl needs libcrypto, so link it

14 years agoA broken pipe error is sometimes reported by zcat if the user quits out
Matthew Dillon [Sat, 15 Jul 2006 07:33:05 +0000 (07:33 +0000)]
A broken pipe error is sometimes reported by zcat if the user quits out
of the pager without reading the entire manual page.  It is unclear why
this happens (EPIPE should kill zcat, not cause an error to be reported).
For now, just quiet error reporting when piping zcat to a pager.

Reported-by: Petr Janda <elekktretterr@exemail.com.au>
14 years agoAdd an ASCII copy of the handbook to the system. It can't hurt to have it
Sascha Wildner [Fri, 14 Jul 2006 21:42:07 +0000 (21:42 +0000)]
Add an ASCII copy of the handbook to the system. It can't hurt to have it
around on the LiveCD. It will be updated periodically.

Also, get rid of the /usr/share/doc/ncurses directory (which we don't use).

Handbook-provided-by: Justin C. Sherrill <justin@shiningsilence.com>
                      (with contributions by many others)

Inclusion-OK'd-by: Matt Dillon

14 years agoFix two bugs in cluster_read:
Simon Schubert [Fri, 14 Jul 2006 19:23:39 +0000 (19:23 +0000)]
Fix two bugs in cluster_read:
- brelse rbp with B_INVAL set in case of a VOP_BMAP() error

- move the read-ahead after starting the real read for bp, so that it
  can be skipped if an error occured in the reading of bp.

  Before this commit rbp might have been a CLUSTER and didn't get special
  treatment for freeing, which resulted in a panic.

Reported-by: swildner
14 years agoAssign the packet length to the pkthdr and mbuf.
Simon Schubert [Fri, 14 Jul 2006 18:57:34 +0000 (18:57 +0000)]
Assign the packet length to the pkthdr and mbuf.
This was erroneously removed in rev 1.12 and led to no packet reception.

Reported-by: Vincent Hourdin
14 years agoRemove trailing whitespace.
Sascha Wildner [Thu, 13 Jul 2006 13:34:12 +0000 (13:34 +0000)]
Remove trailing whitespace.

14 years agoFor crash dump debugging, we use kgdb.
Sascha Wildner [Thu, 13 Jul 2006 12:38:29 +0000 (12:38 +0000)]
For crash dump debugging, we use kgdb.

14 years agoCompiling the nv(4) driver into the kernel is not supported in DragonFly.
Sascha Wildner [Thu, 13 Jul 2006 12:37:06 +0000 (12:37 +0000)]
Compiling the nv(4) driver into the kernel is not supported in DragonFly.

Noticed-by: sephe
14 years ago- Port ath(4) driver from FreeBSD, which supports various Atheros chip based
Sepherosa Ziehau [Thu, 13 Jul 2006 09:15:22 +0000 (09:15 +0000)]
- Port ath(4) driver from FreeBSD, which supports various Atheros chip based
  WiFi NIC
- Hook ath(4) into module building
- Add ath(4) and ath_hal(4) manpage
  With-help-from: swildner

Thank Sam Leffler and many other people for their greate work, also thank
Sam Leffler for his kind explaining about some caveats concerning ath_hal(4).

An early version of this driver, which works under DragonFly, was submitted by:
Andrew Atrens <atrens@nortel.com>
and was later modified to work on DragonFly 1.4.x and resubmitted by:
Adrian Michael Nida <nida@musc.edu>

Thank them for their submission.

Obtained-from: FreeBSD (mainly sam@freebsd.org)
Tested-by: many
Approved-by: dillon
14 years agoMerge from vendor branch ATHEROS:
Sepherosa Ziehau [Thu, 13 Jul 2006 08:46:40 +0000 (08:46 +0000)]
Merge from vendor branch ATHEROS:
Initial import of Atheros Hardware Access Layer (HAL) version

14 years agoInitial import of Atheros Hardware Access Layer (HAL) version vendor/ATHEROS
Sepherosa Ziehau [Thu, 13 Jul 2006 08:46:40 +0000 (08:46 +0000)]
Initial import of Atheros Hardware Access Layer (HAL) version

14 years agoCopy /etc/localtime into the bind chroot so that log messages come from
Simon Schubert [Wed, 12 Jul 2006 23:53:36 +0000 (23:53 +0000)]
Copy /etc/localtime into the bind chroot so that log messages come from
the correct time zone.

14 years ago* To distinguish the separator from commit msg lines consisting of
Sascha Wildner [Wed, 12 Jul 2006 21:40:49 +0000 (21:40 +0000)]
* To distinguish the separator from commit msg lines consisting of
  28 -'s (as the case in FreeBSD's commit logs) look ahead to the
  next line and ensure that it begins with "revision".

* The revision line might contain info about CVS locks. Add this
  to the regexp.

14 years agoSort references by section.
Sascha Wildner [Tue, 11 Jul 2006 11:41:35 +0000 (11:41 +0000)]
Sort references by section.

14 years agoCorrect a problem with the user process scheduler's estimated cpu
Matthew Dillon [Tue, 11 Jul 2006 01:01:50 +0000 (01:01 +0000)]
Correct a problem with the user process scheduler's estimated cpu
calculation.  lwp_slptime would sometimes not get reset to 0 after a
wakeup and cause a cpu-bound process to get more cpu then it deserves.

Move the estcpu call for calculations based on slptime to the tsleep code
and then unconditionally set lwp_slptime back to 0 after the process
wakes up again.

14 years agoMatch revision separator only if preceded by a newline.
Sascha Wildner [Mon, 10 Jul 2006 23:07:45 +0000 (23:07 +0000)]
Match revision separator only if preceded by a newline.

This prevents the script from choking on

CVS ----------------------------------------------------------------------

lines accidentally left in commit messages.

perl-clue: corecode

14 years agoUpdate the manual pages for the kernel random number generator.
Matthew Dillon [Mon, 10 Jul 2006 22:52:31 +0000 (22:52 +0000)]
Update the manual pages for the kernel random number generator.

14 years agoTurn on the new kern.seedenable sysctl when seeding the PRNG.
Matthew Dillon [Mon, 10 Jul 2006 22:19:14 +0000 (22:19 +0000)]
Turn on the new kern.seedenable sysctl when seeding the PRNG.

14 years agoChange the seeder array from a modulo to a logical AND, improving performance
Matthew Dillon [Mon, 10 Jul 2006 21:06:09 +0000 (21:06 +0000)]
Change the seeder array from a modulo to a logical AND, improving performance
and allowing us to overflow the index variable without overflowing the

Implement seeding via writes to /dev/random.  Seeding is only allowed
if the kern.seedenable sysctl is set to 1 (defaults to 0), and the securelevel
is less then or equal to 0.  Note that the random number generator also
self-seeds to some degree.

Writing to /dev/random is no longer a sink null.  Writing to /dev/urandom
is no longer allowed at all.

Modify the seeding code to mix a few of the bits with PNG feedback to reduce
the possibility of seeding-based attacks.

Intentionally feed /dev/urandom back into /dev/random for a bit, then run
DIEHARD to test for any obvious screwups.  Passes DIEHARD.

14 years agoSync with FreeBSD (makes it actually work).
Sascha Wildner [Mon, 10 Jul 2006 20:34:47 +0000 (20:34 +0000)]
Sync with FreeBSD (makes it actually work).

14 years agoCleanup, no functional changes.
Matthew Dillon [Mon, 10 Jul 2006 20:33:36 +0000 (20:33 +0000)]
Cleanup, no functional changes.

14 years agoDisassociate the VM object after calling VOP_INACTIVE instead of before.
Matthew Dillon [Mon, 10 Jul 2006 04:42:56 +0000 (04:42 +0000)]
Disassociate the VM object after calling VOP_INACTIVE instead of before.
VOP_INACTIVE may have to do some work on the vnode that requires a
functional buffer cache.  For example, UFS may have to truncate a removed

14 years agoDon't use B_PAGING with geteblk() bufs
Simon Schubert [Sun, 9 Jul 2006 22:55:45 +0000 (22:55 +0000)]
Don't use B_PAGING with geteblk() bufs

14 years agoCorrectly handle dumping kernel threads
Simon Schubert [Sun, 9 Jul 2006 01:38:57 +0000 (01:38 +0000)]
Correctly handle dumping kernel threads

14 years agoUse bzip instead of gzip, rename the tar file to make it more obvious that
Matthew Dillon [Sat, 8 Jul 2006 18:12:05 +0000 (18:12 +0000)]
Use bzip instead of gzip, rename the tar file to make it more obvious that
only kernel sources are included, and allow WITHOUT_SRCS to be defined to
prevent inclusion of the kernel sources in the ISO.

Suggested-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>