]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoRemove rants/whines about the rune interface being superior to the
Tim J. Robbins [Sun, 6 Oct 2002 06:03:23 +0000 (06:03 +0000)]
Remove rants/whines about the rune interface being superior to the
ISO C interface.

21 years agoRemove a completely incorrect statement from the Return Values section.
Tim J. Robbins [Sun, 6 Oct 2002 05:58:24 +0000 (05:58 +0000)]
Remove a completely incorrect statement from the Return Values section.
Add cross-references to the restartable mulitybte functions (mbrlen(3) etc.)

21 years agoIt appears CPU_MAXID should be 1 more than the number of CPU_* defines.
David E. O'Brien [Sun, 6 Oct 2002 05:25:10 +0000 (05:25 +0000)]
It appears CPU_MAXID should be 1 more than the number of CPU_* defines.

21 years agoCPU_MAXID should
David E. O'Brien [Sun, 6 Oct 2002 05:24:13 +0000 (05:24 +0000)]
CPU_MAXID should

21 years agoUse %jd in format string and cast argument to intmax_t instead of using
Tim J. Robbins [Sun, 6 Oct 2002 03:50:06 +0000 (03:50 +0000)]
Use %jd in format string and cast argument to intmax_t instead of using
%qd to print struct stat.st_size. Remove WARNS=0 and WFORMAT=0 from
Makefile.

Tested on: alpha, sparc64, i386

21 years agoFix format string errors relating mainly to the use of %qu to print off_t's.
Tim J. Robbins [Sun, 6 Oct 2002 03:20:27 +0000 (03:20 +0000)]
Fix format string errors relating mainly to the use of %qu to print off_t's.
Instead use %ju and cast the argument.

WFORMAT=0 is still required in the Makefile because gcc warns about
some strftime() calls (I don't think this behaviour is useful.)

Tested on: sparc64, alpha, i386

21 years agoINclude <sys/mutex.h>'s prerequisite <sys/lock.h> instead of depending on
Bruce Evans [Sun, 6 Oct 2002 03:18:14 +0000 (03:18 +0000)]
INclude <sys/mutex.h>'s prerequisite <sys/lock.h> instead of depending on
namespace pollution 2 layers deep in <sys/eventhandler.h>.

21 years agoSync from MAC tree: break out the single mmap entry point into
Robert Watson [Sun, 6 Oct 2002 02:46:26 +0000 (02:46 +0000)]
Sync from MAC tree: break out the single mmap entry point into
seperate entry points for each occasion:

mac_check_vnode_mmap() Check at initial mapping
mac_check_vnode_mprotect() Check at mapping protection change
mac_check_vnode_mmap_downgrade() Determine if a mapping downgrade
should take place following
subject relabel.

Implement mmap() and mprotect() entry points for labeled vnode
policies.  These entry points are currently not hooked up to the
VM system in the base tree.  These changes improve the consistency
of the access control interface and offer more flexibility regarding
limiting access to vnode mmaping.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoMake this work under GEOM.
Matthew N. Dodd [Sun, 6 Oct 2002 00:19:38 +0000 (00:19 +0000)]
Make this work under GEOM.

21 years agoMake stpcpy() only visiable w/in __BSD_VISIBLE.
David E. O'Brien [Sat, 5 Oct 2002 22:07:28 +0000 (22:07 +0000)]
Make stpcpy() only visiable w/in __BSD_VISIBLE.

21 years agoDon't be noisy if DIOCWLABEL fails, it's probably because we're writing
Poul-Henning Kamp [Sat, 5 Oct 2002 21:57:05 +0000 (21:57 +0000)]
Don't be noisy if DIOCWLABEL fails, it's probably because we're writing
the first label to the raw disk.

Sponsored by: DARPA & NAI Labs.

21 years agoThis patch got lost in my trees: Pass setattr down to device drivers
Poul-Henning Kamp [Sat, 5 Oct 2002 21:55:31 +0000 (21:55 +0000)]
This patch got lost in my trees:  Pass setattr down to device drivers
as well.

Detected by: scottl
Sponsored by: DARPA & NAI Labs.

21 years agoModify label allocation semantics for sockets: pass in soalloc's malloc
Robert Watson [Sat, 5 Oct 2002 21:23:47 +0000 (21:23 +0000)]
Modify label allocation semantics for sockets: pass in soalloc's malloc
flags so that we can call malloc with M_NOWAIT if necessary, avoiding
potential sleeps while holding mutexes in the TCP syncache code.
Similar to the existing support for mbuf label allocation: if we can't
allocate all the necessary label store in each policy, we back out
the label allocation and fail the socket creation.  Sync from MAC tree.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoThe msdosfs module now builds on sparc64. No idea if it works
Maxime Henrion [Sat, 5 Oct 2002 20:18:34 +0000 (20:18 +0000)]
The msdosfs module now builds on sparc64.  No idea if it works
though.

21 years agoAdd two extern's for adjkerntz and wall_cmos_clock, all other
Maxime Henrion [Sat, 5 Oct 2002 20:14:49 +0000 (20:14 +0000)]
Add two extern's for adjkerntz and wall_cmos_clock, all other
archs have them there, alghough the variable are declared in
subr_clock.c.  These should probably be moved into some MI
place.

Approved by: jake

21 years agoMake sure that the accounting credential is saved along with the vp
Robert Watson [Sat, 5 Oct 2002 20:05:23 +0000 (20:05 +0000)]
Make sure that the accounting credential is saved along with the vp
when accounting is suspended--otherwise when accounting is restored,
we may incorrectly assume the credential is valid.

Panics experienced by: juli

21 years agoImplement mac_create_devfs_symlink() for policies that interact with
Robert Watson [Sat, 5 Oct 2002 18:56:25 +0000 (18:56 +0000)]
Implement mac_create_devfs_symlink() for policies that interact with
vnode labels.  Sync from MAC tree.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoFix argument order mistake when decoding disklabels from on-disk format.
Poul-Henning Kamp [Sat, 5 Oct 2002 18:52:06 +0000 (18:52 +0000)]
Fix argument order mistake when decoding disklabels from on-disk format.

Detected by: jhay
Sponsored by: DARPA & NAI Labs.

21 years agoIntegrate a devfs/MAC fix from the MAC tree: avoid a race condition during
Robert Watson [Sat, 5 Oct 2002 18:40:10 +0000 (18:40 +0000)]
Integrate a devfs/MAC fix from the MAC tree: avoid a race condition during
devfs VOP symlink creation by introducing a new entry point to determine
the label of the devfs_dirent prior to allocation of a vnode for the
symlink.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoRemove matcd from MAKEDEV.
Scott Long [Sat, 5 Oct 2002 18:28:48 +0000 (18:28 +0000)]
Remove matcd from MAKEDEV.

X-MFC after: ASAP

21 years agoMerge implementation of mpo_check_vnode_link() for various appropriate
Robert Watson [Sat, 5 Oct 2002 18:25:48 +0000 (18:25 +0000)]
Merge implementation of mpo_check_vnode_link() for various appropriate
file-system aware MAC policies.  Sync to MAC tree.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoMerge support for mac_check_vnode_link(), a MAC framework/policy entry
Robert Watson [Sat, 5 Oct 2002 18:11:36 +0000 (18:11 +0000)]
Merge support for mac_check_vnode_link(), a MAC framework/policy entry
point that instruments the creation of hard links.  Policy implementations
to follow.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoWhile the MAC API has supported the ability to handle M_NOWAIT passed
Robert Watson [Sat, 5 Oct 2002 17:44:49 +0000 (17:44 +0000)]
While the MAC API has supported the ability to handle M_NOWAIT passed
to mbuf label initialization, that functionality was never merged to
the main tree.  Go ahead and merge that functionality now.  Note that
this requires policy modules to accept the case where the label
element may be destroyed even if init has not succeeded on it (in
the event that policy failed the init).  This will shortly also
apply to sockets.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoRearrange object and label init/destroy functions to match the
Robert Watson [Sat, 5 Oct 2002 17:38:45 +0000 (17:38 +0000)]
Rearrange object and label init/destroy functions to match the
order used in mac_policy.h and elsewhere.  Sort order is basically
"by operation category", then "alphabetically by object". Sync to
MAC tree.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoAdd hifn(4) and ubsec(4) devices for i386/pc98.
Bruce A. Mah [Sat, 5 Oct 2002 17:23:18 +0000 (17:23 +0000)]
Add hifn(4) and ubsec(4) devices for i386/pc98.

21 years agoNew release notes: crypto(4) and related info
Bruce A. Mah [Sat, 5 Oct 2002 17:22:22 +0000 (17:22 +0000)]
New release notes:  crypto(4) and related info

Updated release notes:  Update KSE info and add cross-reference to
kse(2) manpage.

Fix a couple of minor typos while I'm here.

21 years agoSync to MAC tree: use 'flag' instead of 'how' for mac_init_mbuf();
Robert Watson [Sat, 5 Oct 2002 17:18:43 +0000 (17:18 +0000)]
Sync to MAC tree: use 'flag' instead of 'how' for mac_init_mbuf();
remove a slightly less than useful comment.

21 years agoDon't allow dev_stdclone(9) to accept minors larger than the system is
Brian Feldman [Sat, 5 Oct 2002 17:10:28 +0000 (17:10 +0000)]
Don't allow dev_stdclone(9) to accept minors larger than the system is
able to handle (0xffffff).

21 years agoAnother big diff, little functional change: move label internalization,
Robert Watson [Sat, 5 Oct 2002 16:57:16 +0000 (16:57 +0000)]
Another big diff, little functional change: move label internalization,
externalization, and cred label life cycle events to entirely above
devfs and vnode events.  Sync from MAC tree.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoMove all object label init/destroy routines to the head of the
Robert Watson [Sat, 5 Oct 2002 16:54:59 +0000 (16:54 +0000)]
Move all object label init/destroy routines to the head of the
entry points to better match the entry point ordering in mac_policy.h.
Big diff, no functional change; merge from the MAC tree.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoSynch from TrustedBSD MAC tree:
Robert Watson [Sat, 5 Oct 2002 16:46:03 +0000 (16:46 +0000)]
Synch from TrustedBSD MAC tree:

- If a policy isn't registered when a policy module unloads, silently
  succeed.

- Hold the policy list lock across more of the validity tests to avoid
  races.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoNB: This commit does *NOT* make GEOM the default in FreeBSD
Poul-Henning Kamp [Sat, 5 Oct 2002 16:35:33 +0000 (16:35 +0000)]
NB: This commit does *NOT* make GEOM the default in FreeBSD
NB: But it will enable it in all kernels not having options "NO_GEOM"

Put the GEOM related options into the intended order.

Add "options NO_GEOM" to all kernel configs apart from NOTES.

In some order of controlled fashion, the NO_GEOM options will be
removed, architecture by architecture in the coming days.

There are currently three known issues which may force people to
need the NO_GEOM option:

boot0cfg/fdisk:
        Tries to update the MBR while it is being used to control
        slices.  GEOM does not allow this as a direct operation.

SCSI floppy drives:
        Appearantly the scsi-da driver return "EBUSY" if no media
        is inserted.  This is wrong, it should return ENXIO.

PC98:
        It is unclear if GEOM correctly recognizes all variants of
        PC98 disklabels.  (Help Wanted!  I have neither docs nor HW)

These issues are all being worked.

Sponsored by: DARPA & NAI Labs.

21 years agoCosmetic line wrap synchronization.
Robert Watson [Sat, 5 Oct 2002 16:33:46 +0000 (16:33 +0000)]
Cosmetic line wrap synchronization.

21 years agoPush the debugging obect label counters into security.mac.debug.counters
Robert Watson [Sat, 5 Oct 2002 16:30:53 +0000 (16:30 +0000)]
Push the debugging obect label counters into security.mac.debug.counters
rather than directly under security.mac.debug.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoDon't use dkunit() to find our softc when we can hang it off the dev_t.
Poul-Henning Kamp [Sat, 5 Oct 2002 15:49:39 +0000 (15:49 +0000)]
Don't use dkunit() to find our softc when we can hang it off the dev_t.

This removes yet a dependency on the old disklabel stuff.

Sponsored by: DARPA & NAI Labs.

21 years agoMerge the last couple of my changes to fd.c into the pc98 version.
Poul-Henning Kamp [Sat, 5 Oct 2002 15:15:43 +0000 (15:15 +0000)]
Merge the last couple of my changes to fd.c into the pc98 version.

Sponsored by: DARPA & NAI Labs

21 years agoBegin another merge from the TrustedBSD MAC branch:
Robert Watson [Sat, 5 Oct 2002 15:10:00 +0000 (15:10 +0000)]
Begin another merge from the TrustedBSD MAC branch:

- Change mpo_init_foo(obj, label) and mpo_destroy_foo(obj, label) policy
  entry points to mpo_init_foo_label(label) and
  mpo_destroy_foo_label(label).  This will permit the use of the same
  entry points for holding temporary type-specific label during
  internalization and externalization, as well as for caching purposes.
- Because of this, break out mpo_{init,destroy}_socket() and
  mpo_{init,destroy}_mount() into seperate entry points for socket
  main/peer labels and mount main/fs labels.
- Since the prototype for label initialization is the same across almost
  all entry points, implement these entry points using common
  implementations for Biba, MLS, and Test, reducing the number of
  almost identical looking functions.

This simplifies policy implementation, as well as preparing us for the
merge of the new flexible userland API for managing labels on objects.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoFix building of minimal kernels without npx by rearranging ifdefs.
Daniel Eischen [Sat, 5 Oct 2002 14:36:14 +0000 (14:36 +0000)]
Fix building of minimal kernels without npx by rearranging ifdefs.
Also fix some style bugs in surrounding code, and add a comment
about FP state restoral that seems questionable.

Submitted by: bde

21 years agoUse in-tree libbsdxml instead of ports based libexpat.
Poul-Henning Kamp [Sat, 5 Oct 2002 13:16:07 +0000 (13:16 +0000)]
Use in-tree libbsdxml instead of ports based libexpat.

Sponsored by: DARPA & NAI Labs.

21 years agoMatch the new g_io_deliver() API in the kernel.
Poul-Henning Kamp [Sat, 5 Oct 2002 13:12:10 +0000 (13:12 +0000)]
Match the new g_io_deliver() API in the kernel.

Sponsored by: DARPA & NAI Labs.

21 years agoPass sbuf to the kernel routines to match new kernel code.
Poul-Henning Kamp [Sat, 5 Oct 2002 13:11:01 +0000 (13:11 +0000)]
Pass sbuf to the kernel routines to match new kernel code.

Sponsored by: DARPA & NAI Labs

21 years agoFix problem introduced in rev.1.406, which can cause already unlocked
Maxim Sobolev [Sat, 5 Oct 2002 12:56:10 +0000 (12:56 +0000)]
Fix problem introduced in rev.1.406, which can cause already unlocked
mutex being unlocked again causing system panic.

21 years agoPlug memoryleaks detected by FlexeLint.
Poul-Henning Kamp [Sat, 5 Oct 2002 12:00:11 +0000 (12:00 +0000)]
Plug memoryleaks detected by FlexeLint.

21 years agoIf dsgetlabel() returns a label with a size of zero in diskdumpconf(),
Brian Somers [Sat, 5 Oct 2002 11:24:21 +0000 (11:24 +0000)]
If dsgetlabel() returns a label with a size of zero in diskdumpconf(),
treat it as an invalid partition.

This fixes a bug where ``dumpon <device>'' will configure the dump
device at a random offset on the disk if <device> isn't a valid
partition.

Reviewed by: phk

21 years agoFix namespace issues by using visibility conditionals from
Mike Barcroft [Sat, 5 Oct 2002 05:48:50 +0000 (05:48 +0000)]
Fix namespace issues by using visibility conditionals from
<sys/cdefs.h>.  Sort function prototypes.

21 years agoFix namespace issues by using visibility conditionals from
Mike Barcroft [Sat, 5 Oct 2002 05:47:56 +0000 (05:47 +0000)]
Fix namespace issues by using visibility conditionals from
<sys/cdefs.h>.

21 years agoChange <sys/_sigset.h> to typedef __sigset_t instead of sigset_t, so
Mike Barcroft [Sat, 5 Oct 2002 05:40:48 +0000 (05:40 +0000)]
Change <sys/_sigset.h> to typedef __sigset_t instead of sigset_t, so
that headers that include it can conditionalize sigset_t's visibility.

21 years agoPut an easy-to-miss assignment into the proper place. It was stray in the
Juli Mallett [Sat, 5 Oct 2002 04:49:46 +0000 (04:49 +0000)]
Put an easy-to-miss assignment into the proper place.  It was stray in the
middle of a block of code, with no clear assignment.  While here, move one
nearby assignment out of declaration.

21 years agoRemove bogus duplicate assignment of local variables.
Juli Mallett [Sat, 5 Oct 2002 04:35:59 +0000 (04:35 +0000)]
Remove bogus duplicate assignment of local variables.

21 years agoGet Vinum up and running with GEOM:
Robert Watson [Sat, 5 Oct 2002 03:44:00 +0000 (03:44 +0000)]
Get Vinum up and running with GEOM:

(1) Use namei() and devfs to discover devices rather than a hard-coded
    MAKEDEV implementation.  Once rootfs is in place, this will allow
    Vinum to be used for the root file system partition.

(2) Pass FREAD to device opens so that GEOM will return sector size
    rather than an error on attempts to read label data.

(3) Avoid clobbering return values from close_drive() and masking this
    failure, resulting in a later divide by zero due to not having
    updated the Vinum-cached sector size.

(4) Ignore failures from DIOCWLABEL as that appears not to be required
    in the GEOM environment.

We've done testing in simple Vinum environments, but those with more
complex environments might want to give this a spin in DP2 and make
sure everything is up to speed.

Fixes in collaboration with: iedowse
Reviewed by: grog

21 years agoRefinement on previous fix for mutex destruction: make sure we don't
Robert Watson [Sat, 5 Oct 2002 03:33:38 +0000 (03:33 +0000)]
Refinement on previous fix for mutex destruction: make sure we don't
release the mutex multiple times for multi-plex volumes.

Following further consultation with: grog

21 years agoRemove a panic on vinum module unload: make sure to destroy all mutexes
Robert Watson [Sat, 5 Oct 2002 03:07:23 +0000 (03:07 +0000)]
Remove a panic on vinum module unload: make sure to destroy all mutexes
before freeing so that WITNESS doesn't dereference mutex data pointers
and page fault.  It's now possible to unload vinum.ko with a GENERIC
kernel on 5.0-CURRENT without panic.

Debugged/fixed with the aid of: jake, grog

21 years agoDefine _MACHINE.
Juli Mallett [Sat, 5 Oct 2002 02:56:49 +0000 (02:56 +0000)]
Define _MACHINE.

21 years agoOnce again, remove the i386-specific hacks to save and restore
Daniel Eischen [Sat, 5 Oct 2002 02:22:26 +0000 (02:22 +0000)]
Once again, remove the i386-specific hacks to save and restore
the FPU state on receiving and returning from a signal.
The FPU save and restore macros are no longer needed, but
remain defined in case we need to use them again (something
else breaks).  They'll be removed permanently once new
syscalls are added to handle the new i386 ucontext size.

21 years agoMake sure that ACPI PCI driver probe routine call pci_cfgregopen()
Mitsuru IWASAKI [Sat, 5 Oct 2002 02:16:49 +0000 (02:16 +0000)]
Make sure that ACPI PCI driver probe routine call pci_cfgregopen()
before start accessing PCI config space.

Reviewed by: jhb

21 years agoOops, forgot to add main file for my previous commit :-)
Mitsuru IWASAKI [Sat, 5 Oct 2002 02:06:00 +0000 (02:06 +0000)]
Oops, forgot to add main file for my previous commit :-)

21 years agoAdd code for ACPI PCI link object manipulation.
Mitsuru IWASAKI [Sat, 5 Oct 2002 02:01:05 +0000 (02:01 +0000)]
Add code for ACPI PCI link object manipulation.
This allocate the best IRQ to boot-disable devices (have IRQ 0).
Allocated IRQ will be used for PCI interrupt routing when ACPI is
enabled.

Note that verbose messaging enabled for the time being so that
people can easily notice the strange behavior if it happened.

21 years agostyle(9) <machine/setjmp.h> headers so they look mostly the same.
Mike Barcroft [Fri, 4 Oct 2002 22:10:06 +0000 (22:10 +0000)]
style(9) <machine/setjmp.h> headers so they look mostly the same.

21 years agoAdd restrict type-qualifier to sem_getvalue().
Mike Barcroft [Fri, 4 Oct 2002 21:32:00 +0000 (21:32 +0000)]
Add restrict type-qualifier to sem_getvalue().

21 years agoo Adjust the SEM_VALUE_MAX macro so that <machine/limits.h> isn't
Mike Barcroft [Fri, 4 Oct 2002 21:31:33 +0000 (21:31 +0000)]
o Adjust the SEM_VALUE_MAX macro so that <machine/limits.h> isn't
  needed.
o Remove unneeded includes which only add namespace pollution.
o Sort function prototypes.
o Add restrict type-qualifier to sem_getvalue().

21 years agoadd support for /dev/crypto
Sam Leffler [Fri, 4 Oct 2002 20:44:46 +0000 (20:44 +0000)]
add support for /dev/crypto

21 years agoinstall sys/opencrypto include files in /usr/include/crypto
Sam Leffler [Fri, 4 Oct 2002 20:44:28 +0000 (20:44 +0000)]
install sys/opencrypto include files in /usr/include/crypto

21 years agomanual pages for the new kernel crypto support (need work)
Sam Leffler [Fri, 4 Oct 2002 20:43:30 +0000 (20:43 +0000)]
manual pages for the new kernel crypto support (need work)

Obtained from: openbsd

21 years agohookup new crypto support to the config/build process
Sam Leffler [Fri, 4 Oct 2002 20:42:36 +0000 (20:42 +0000)]
hookup new crypto support to the config/build process

21 years agoNew bus_dma interfaces for use by crypto device drivers:
Sam Leffler [Fri, 4 Oct 2002 20:40:39 +0000 (20:40 +0000)]
New bus_dma interfaces for use by crypto device drivers:

o bus_dmamap_load_mbuf
o bus_dmamap_load_uio

Test on i386.  Known to compile on alpha and sparc64, but not tested.
Otherwise untried.

21 years agoadd crypto interface to the MFILES list
Sam Leffler [Fri, 4 Oct 2002 20:36:39 +0000 (20:36 +0000)]
add crypto interface to the MFILES list

21 years agomajor 70 is for /dev/crypto (to be consistent with openbsd)
Sam Leffler [Fri, 4 Oct 2002 20:36:13 +0000 (20:36 +0000)]
major 70 is for /dev/crypto (to be consistent with openbsd)

21 years agomodule for ubsec crypto driver
Sam Leffler [Fri, 4 Oct 2002 20:35:33 +0000 (20:35 +0000)]
module for ubsec crypto driver

21 years agomodule for hifn driver
Sam Leffler [Fri, 4 Oct 2002 20:35:23 +0000 (20:35 +0000)]
module for hifn driver

21 years agomodule for /dev/crypto support
Sam Leffler [Fri, 4 Oct 2002 20:35:02 +0000 (20:35 +0000)]
module for /dev/crypto support

21 years agoadd DTYPE_CRYPTO for use by /dev/crypto support
Sam Leffler [Fri, 4 Oct 2002 20:34:30 +0000 (20:34 +0000)]
add DTYPE_CRYPTO for use by /dev/crypto support

21 years agoadd RANDOM_PURE for use by crypto drivers that harvest data from h/w RNG's
Sam Leffler [Fri, 4 Oct 2002 20:34:08 +0000 (20:34 +0000)]
add RANDOM_PURE for use by crypto drivers that harvest data from h/w RNG's

21 years agoCrypto device driver for Broadcom-based cards. Known to work with 582x-based
Sam Leffler [Fri, 4 Oct 2002 20:33:20 +0000 (20:33 +0000)]
Crypto device driver for Broadcom-based cards.  Known to work with 582x-based
cards.  Supposed to work with several others.

Obtained from: openbsd

21 years agoCrypto device driver for hifn-based cards. Tested with 7951-, and 7811-based
Sam Leffler [Fri, 4 Oct 2002 20:32:37 +0000 (20:32 +0000)]
Crypto device driver for hifn-based cards.  Tested with 7951-, and 7811-based
cards.  Claimed to support many others.

Obtained from:  openbsd

21 years agoIn-kernel crypto framework derived from openbsd. This facility provides
Sam Leffler [Fri, 4 Oct 2002 20:31:23 +0000 (20:31 +0000)]
In-kernel crypto framework derived from openbsd.  This facility provides
a consistent interface to h/w and s/w crypto algorithms for use by the
kernel and (for h/w at least) by user-mode apps.  Access for user-level
code is through a /dev/crypto device that'll eventually be used by openssl
to (potentially) accelerate many applications.  Coming soon is an IPsec
that makes use of this service to accelerate ESP, AH, and IPCOMP protocols.

Included here is the "core" crypto support, /dev/crypto driver, various
crypto algorithms that are not already present in the KAME crypto area,
and support routines used by crypto device drivers.

Obtained from: openbsd

21 years agoGive make(1) the ability to use KQUEUE to wait for worker processes
Poul-Henning Kamp [Fri, 4 Oct 2002 20:30:03 +0000 (20:30 +0000)]
Give make(1) the ability to use KQUEUE to wait for worker processes
instead of polling for them.

Unfortunately we cannot enable it yet because it panics the kernel
somewhere in kqueue.

Submitted by: Stefan Farfeleder <e0026813@stud3.tuwien.ac.at>

21 years agoFix a bogon in previous commit. bcopy() from the malloc'd memory that we
John Baldwin [Fri, 4 Oct 2002 20:19:36 +0000 (20:19 +0000)]
Fix a bogon in previous commit.  bcopy() from the malloc'd memory that we
already copied into, rather than doing the bcopy() from the userland
pointer.  "Oops."

21 years agoResolve conflicts arising from the ACPI CA 20021002 import.
Mitsuru IWASAKI [Fri, 4 Oct 2002 20:09:19 +0000 (20:09 +0000)]
Resolve conflicts arising from the ACPI CA 20021002 import.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r104470,
Mitsuru IWASAKI [Fri, 4 Oct 2002 20:07:58 +0000 (20:07 +0000)]
This commit was generated by cvs2svn to compensate for changes in r104470,
which included commits to RCS files with non-trunk default branches.

21 years agoVendor import of the Intel ACPI CA 20021002 drop.
Mitsuru IWASAKI [Fri, 4 Oct 2002 20:07:58 +0000 (20:07 +0000)]
Vendor import of the Intel ACPI CA 20021002 drop.

21 years agoWhite-space change only. Move closer to style(9).
Poul-Henning Kamp [Fri, 4 Oct 2002 18:10:39 +0000 (18:10 +0000)]
White-space change only.  Move closer to style(9).

21 years agoRemove matcd(4) device.
Bruce A. Mah [Fri, 4 Oct 2002 16:53:39 +0000 (16:53 +0000)]
Remove matcd(4) device.

21 years agoNew release notes: matcd(4) removal, libbsdxml.
Bruce A. Mah [Fri, 4 Oct 2002 16:53:12 +0000 (16:53 +0000)]
New release notes:  matcd(4) removal, libbsdxml.

While here, fix a minor style nit.

21 years agoFix off-by-one error processing the line after an empty line when the
Tim J. Robbins [Fri, 4 Oct 2002 16:45:25 +0000 (16:45 +0000)]
Fix off-by-one error processing the line after an empty line when the
-a option is not specified.

Submitted by: schweikh
MFC after: 1 week

21 years agoSort in ``phone directory'' order (except for LIBC_*).
Ruslan Ermilov [Fri, 4 Oct 2002 16:25:08 +0000 (16:25 +0000)]
Sort in ``phone directory'' order (except for LIBC_*).

21 years agoAdd a couple of missing ':' in panic msgs.
Giorgos Keramidas [Fri, 4 Oct 2002 15:58:03 +0000 (15:58 +0000)]
Add a couple of missing ':' in panic msgs.

Suggested by: gallatin

21 years agoFix typo in panic msg.
Giorgos Keramidas [Fri, 4 Oct 2002 15:56:02 +0000 (15:56 +0000)]
Fix typo in panic msg.
Reviewed by: gallatin

21 years agoMinor tweaks to make this manpage readable.
Ruslan Ermilov [Fri, 4 Oct 2002 15:07:39 +0000 (15:07 +0000)]
Minor tweaks to make this manpage readable.

21 years agoStyle.
Ruslan Ermilov [Fri, 4 Oct 2002 15:06:44 +0000 (15:06 +0000)]
Style.

Approved by: phk

21 years agoAdd another temporary hack to allow running older i386 binaries.
Daniel Eischen [Fri, 4 Oct 2002 14:50:55 +0000 (14:50 +0000)]
Add another temporary hack to allow running older i386 binaries.
This will be removed when new versions of syscalls sigreturn()
and sigaction() are added (mini is working on this but is in
the middle of a move).

This should fix the problem of cvsupd dying.

21 years agoInclude <sys/diskpc98.h>
Yoshihiro Takahashi [Fri, 4 Oct 2002 14:05:00 +0000 (14:05 +0000)]
Include <sys/diskpc98.h>

21 years agoConnect libbsdxml (nee libexpat) to the build.
Poul-Henning Kamp [Fri, 4 Oct 2002 13:40:39 +0000 (13:40 +0000)]
Connect libbsdxml (nee libexpat) to the build.

21 years agoReflect the fact that we install our libexpat as libbsdxml.
Poul-Henning Kamp [Fri, 4 Oct 2002 13:36:12 +0000 (13:36 +0000)]
Reflect the fact that we install our libexpat as libbsdxml.

21 years agoUnhook the SCSI CD driver from the disk "mini-layer" and/or GEOM.
Poul-Henning Kamp [Fri, 4 Oct 2002 13:33:30 +0000 (13:33 +0000)]
Unhook the SCSI CD driver from the disk "mini-layer" and/or GEOM.

SCSI disks are too square pegs for the round holes in both of these.

And since atapi-cd has clearly shown that there are better acccess
models for CD media than trying to pretend to be a classical disk,
we stop the masquerade rather than patch up the costume.

But do implement the DIOCGMEDIASIZE and DIOCGSECTORSIZE so it will
be possible to manually attach to GEOM, should some the need arise.

Ideally, this driver should do media-detection and call make_dev()
when a CD is inserted and destroy_dev() when it is removed, this
would allow our future devd(8) to automount etc etc but coding that
takes SCSI-clue beyond anything I posses.

Tested on:      sparc64

21 years agoUse the -mno-align-long-strings on i386's to debloat the kernel a little.
Bruce Evans [Fri, 4 Oct 2002 12:54:36 +0000 (12:54 +0000)]
Use the -mno-align-long-strings on i386's to debloat the kernel a little.
This reduces the size of GENERIC's text space by 73999 bytes (about 2%).
The bloat is from approximately 3437 strings longer than 31 characters
being padded to a 32-byte boundary.

21 years agoCorrect the regressive part of my last commit to these files:
Robert Drehmel [Fri, 4 Oct 2002 11:31:00 +0000 (11:31 +0000)]
Correct the regressive part of my last commit to these files:
use the .Fn macro instead of the .Fo ... .Fc combination to
format function prototypes.

Reminded by: bde

21 years agoOk, concensus was to install libexpat under a "private brandname", and after
Poul-Henning Kamp [Fri, 4 Oct 2002 11:20:18 +0000 (11:20 +0000)]
Ok, concensus was to install libexpat under a "private brandname", and after
some deliberation the name "libbsdxml" was chosen since it conveys the two
most important attributes:  "Private to FreeBSD" and "XML".

Add a skeleton man-page to give the credit and point for further
documentation.  (If somebody wants to write a true mdoc manpage for
this I am sure both the eXpat people and I will be grateful).

(Still not connected to the build)

21 years agoProperly isolate the locking domains of sysctl from the topology lock
Poul-Henning Kamp [Fri, 4 Oct 2002 10:38:36 +0000 (10:38 +0000)]
Properly isolate the locking domains of sysctl from the topology lock
for the sysctls which report the configuration.

Sponsored by: DARPA & NAI Labs.

21 years agoImplement the "kern.disks" sysctl in GEOM.
Poul-Henning Kamp [Fri, 4 Oct 2002 10:15:26 +0000 (10:15 +0000)]
Implement the "kern.disks" sysctl in GEOM.

This makes "mdconfig -l" work again.

Sponsored by: DARPA & NAI Labs.

21 years agoProperly conditionalize a debugging printf.
Poul-Henning Kamp [Fri, 4 Oct 2002 10:06:19 +0000 (10:06 +0000)]
Properly conditionalize a debugging printf.

Sponsored by: DARPA & NAI Labs.