]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoAdd support for -R for file relabel operations.
Robert Watson [Mon, 2 Dec 2002 03:22:57 +0000 (03:22 +0000)]
Add support for -R for file relabel operations.

Add 'setfsmac' link, which permits labels to be provided in a label
specification file, making it easier to provide initial file system
labeling specifications.  This is used by the new mac_lomac to
provide initial system labeling and policy, and by sebsd, the port
of SELinux FLASK/TE to FreeBSD.

Approved by: re (jhb)
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoUpdate bmake glue after 1.11.2.1-20021201 import. options.h is no longer
Peter Wemm [Mon, 2 Dec 2002 03:20:25 +0000 (03:20 +0000)]
Update bmake glue after 1.11.2.1-20021201 import.  options.h is no longer
touched by configure in the autoconf system and a standard version is
now used.

Approved by:    re

21 years agoMerge 1.11.2 -> 1.11.2.1-20021201 changes onto mainline
Peter Wemm [Mon, 2 Dec 2002 03:17:49 +0000 (03:17 +0000)]
Merge 1.11.2 -> 1.11.2.1-20021201 changes onto mainline

Approved by: re

21 years agoThis commit was generated by cvs2svn to compensate for changes in r107484,
Peter Wemm [Mon, 2 Dec 2002 03:13:37 +0000 (03:13 +0000)]
This commit was generated by cvs2svn to compensate for changes in r107484,
which included commits to RCS files with non-trunk default branches.

21 years agoImport cvs-1.11.2.1 as of 2002/12/01 onto vendor branch. This fixes all
Peter Wemm [Mon, 2 Dec 2002 03:13:37 +0000 (03:13 +0000)]
Import cvs-1.11.2.1 as of 2002/12/01 onto vendor branch.  This fixes all
of the bugs that I know of.  We've been running a slightly older version
of this on freefall/repoman, where it was afflicted by a silly merge error
on my part (fixed).

Approved by:  re

21 years agoFix a dumb bug that broke net booting on sparc64. The wrong length was
Jake Burkholder [Mon, 2 Dec 2002 01:46:22 +0000 (01:46 +0000)]
Fix a dumb bug that broke net booting on sparc64.  The wrong length was
passed to strncmp.

Noticed by: tmm
Approved by: re
Pointy hat to: jake

21 years agoo Newer EFI implementations require that a GPT is preceeded by
Marcel Moolenaar [Mon, 2 Dec 2002 01:42:03 +0000 (01:42 +0000)]
o  Newer EFI implementations require that a GPT is preceeded by
   a PMBR. Make sure the create command creates a PMBR as well
   (if not already present).
o  When parsing the MBR, explicitly check for a PMBR and create
   a PMBR map node if one is found.
o  When parsing the MBR, recurse to handle extended partitions.
   This allows us to flatten nested MBRs when migrating to a
   GPT.
o  Have the migrate command bail out if it encounters a partition
   it doesn't know how to migrate. This avoids data loss.
o  Change the output of the show command so that the UUIDs of the
   GPT partitions fit on the same line.
o  Show when partitions are extended partitions and add the PMBR
   type.

Approved by: re (blanket)

21 years agoMFi386
Alan Cox [Mon, 2 Dec 2002 01:12:05 +0000 (01:12 +0000)]
MFi386
 Hold the page queues lock around vm_page_unhold() in vunmapbuf().

Approved by: re (blanket)

21 years agoReplace rev 1.33 with a real fix. The problem was integer overflows
Peter Wemm [Mon, 2 Dec 2002 01:05:08 +0000 (01:05 +0000)]
Replace rev 1.33 with a real fix.  The problem was integer overflows
when trying to store the year in a signed int.  The maximum time_t on ia64
is around 292 billion years in the future, but 'int' and struct tm.tm_year
can only represent then ext 2.1 billion years or so.

This solves the problem of mktime/localtime looping on ia64.  Unfortunately,
the standards say that tm_year is an 'int', so we are still stuck with a
y2147483647 bug.  bash2's configure script looks for bugs in mktime() and
fails on ia64 because of this.  However, mktime() on FreeBSD fails the test
normally anyway so this is no big loss.

This change does not affect any other platforms besides ia64.

Approved by: re

21 years agoDocument the MLS policy
Chris Costello [Mon, 2 Dec 2002 01:04:37 +0000 (01:04 +0000)]
Document the MLS policy

Approved by: re
Sponsored by: DARPA, Network Associates Labs
Obtained from: TrustedBSD Project

21 years agoSpelling: "current" -> "currently"
Chris Costello [Mon, 2 Dec 2002 00:00:43 +0000 (00:00 +0000)]
Spelling: "current" -> "currently"

Approved by: re
Sponsored by: DARPA, Network Associates Labs

21 years agoAlways initialize the UPA target module id in the interrupt mapping
Thomas Moestl [Sun, 1 Dec 2002 23:30:26 +0000 (23:30 +0000)]
Always initialize the UPA target module id in the interrupt mapping
register to the one of the processor doing the interrupt setup. This
is required since this field is preinitialized to 0, but there exist
machines which have no processor with a MID of 0 (e.g. e450s with 1 or 2
processors).

Add some more macros for handle the interrupt mapping registers, and
rename some existing ones for consistency.

Approved by: re

21 years ago1.) Do not look for PCI INOs in the FFB interrupt mapping registers; they
Thomas Moestl [Sun, 1 Dec 2002 23:21:15 +0000 (23:21 +0000)]
1.) Do not look for PCI INOs in the FFB interrupt mapping registers; they
    are nevers used for PCI interrupts, but can cause false matches since
    they are fully programmable.
2.) Skip the mapping registers for slot a2 and a3 on "psycho" bridges,
    since they are not present there. Again, this could cause false matches,
    which would result in the interrupt being delivered at most once.

Submitted by: jake (2)
Approved by: re

21 years agoTeach mdmfs how to pass UFS version numbers through to newfs. Because
Robert Watson [Sun, 1 Dec 2002 23:19:57 +0000 (23:19 +0000)]
Teach mdmfs how to pass UFS version numbers through to newfs.  Because
of an argument name collision with -O, use -v, and default to whatever
the newfs default is for the platform (generally, UFS1).  This is
required to support diskless workstations that use UFS2 for their
mdmfs file systems.

Reviewed by: dd, bmah
Approved by: re (bmah)
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoRemove some long-dead cruft in the interrupt handling code which was
Thomas Moestl [Sun, 1 Dec 2002 23:13:59 +0000 (23:13 +0000)]
Remove some long-dead cruft in the interrupt handling code which was
never used in FreeBSD.

Approved by: re

21 years agoIn general, prefer WARNS to CFLAGS+=-Wall. Tend towards a more
Robert Watson [Sun, 1 Dec 2002 23:10:28 +0000 (23:10 +0000)]
In general, prefer WARNS to CFLAGS+=-Wall.  Tend towards a more
BSD-esque Makefile style.

Submitted by: obrien
Approved by: re

21 years agoReverse the quirk table entry for swizzling on a missing interrupt map;
Thomas Moestl [Sun, 1 Dec 2002 23:06:14 +0000 (23:06 +0000)]
Reverse the quirk table entry for swizzling on a missing interrupt map;
this is now done on all machines except for some known problematic ones.

Add an additional guard to make sure that the interrupt numbers are
in the correct range before swizzling. This should catch any remaining
models for which the swizzle is inappropriate.

Correct the swizzle calculation to account for the fact that the parent
interrupt numbers to be swizzled are 1-based.

Approved by: re

21 years agoFix some comments describing psycho registers.
Thomas Moestl [Sun, 1 Dec 2002 23:00:41 +0000 (23:00 +0000)]
Fix some comments describing psycho registers.

Approved by: re

21 years agoDo not panic when a dmamap is unloaded more then once, but just silently
Thomas Moestl [Sun, 1 Dec 2002 22:59:29 +0000 (22:59 +0000)]
Do not panic when a dmamap is unloaded more then once, but just silently
ignore it. This is non-fatal on the other architectures, and some
drivers seem to do this.

Approved by: re

21 years agoThe Bluetooth stack apparently works on pc98, so enable items in
Bruce A. Mah [Sun, 1 Dec 2002 19:40:37 +0000 (19:40 +0000)]
The Bluetooth stack apparently works on pc98, so enable items in
the pc98 release documentation.

Pointed out by: nyan
Approved by: re (implicitly)

21 years agoFix minor SGML markup glitch.
Bruce A. Mah [Sun, 1 Dec 2002 19:35:30 +0000 (19:35 +0000)]
Fix minor SGML markup glitch.

Pointed out by: ue
Approved by: re (implicitly)

21 years agoHold the page queues lock when calling pmap_protect(); it updates fields
Alan Cox [Sun, 1 Dec 2002 18:57:56 +0000 (18:57 +0000)]
Hold the page queues lock when calling pmap_protect(); it updates fields
of the vm_page structure.  Make the style of the pmap_protect() calls
consistent.

Approved by: re (blanket)

21 years agoMake the following changes to the semi-automated package split:
Bruce A. Mah [Sun, 1 Dec 2002 18:03:42 +0000 (18:03 +0000)]
Make the following changes to the semi-automated package split:

1.  On disc1, print/acroread4 for alpha and print/acroread5 for i386
replace print/acroread, which was not compatible with the linux_base
that we ship by default.  Remove print/acroread4 from disc3 as
redundant.

2.  www/linux-netscape47-* were removed due to security holes;
substitute www/netscape48-* on disc1.

3.  Add www/opera to disc1.

Approved by: re (murray)

21 years agoBacked out previous commit (alignment suitable for RELENG_4) as planned
Bruce Evans [Sun, 1 Dec 2002 17:36:18 +0000 (17:36 +0000)]
Backed out previous commit (alignment suitable for RELENG_4) as planned
since it has been MFC'ed.  See the log message for the previous commit
for more details.  The alignment bug in gcc-3 has not been fixed, but
it is not very serious and the previous commit just moved it (as intended).

Approved by: re (murray)

21 years agoMFbed: translation updates that were not yet available on cvsup2.de
Udo Erdelhoff [Sun, 1 Dec 2002 16:20:37 +0000 (16:20 +0000)]
MFbed: translation updates that were not yet available on cvsup2.de
hardware/common/dev.sgml: 1.116 -> 1.117
relnotes/common/new.sgml: 1.460 -> 1.461

21 years agoAdd Makefile.inc to include ../Makefile.inc.
Yoshihiro Takahashi [Sun, 1 Dec 2002 16:07:53 +0000 (16:07 +0000)]
Add Makefile.inc to include ../Makefile.inc.

Approved by: re (rwatson)

21 years agoDe-danglify.
Poul-Henning Kamp [Sun, 1 Dec 2002 15:59:33 +0000 (15:59 +0000)]
De-danglify.

Submitted by: ceri
Approved by: re (blanket)

21 years agoSynchronize usage() with reality.
Poul-Henning Kamp [Sun, 1 Dec 2002 15:58:28 +0000 (15:58 +0000)]
Synchronize usage() with reality.
Semi-automatic handling of /dev prefix for device names.

Sponsored by: DARPA & NAI Labs.
Approved by: re (blanket)

21 years agoDe-danglify the manual page.
Poul-Henning Kamp [Sun, 1 Dec 2002 15:57:02 +0000 (15:57 +0000)]
De-danglify the manual page.

Submitted by: ceri
Approved by: re (blanket)

21 years agoUse more mnemonic argument names in the access functions.
Poul-Henning Kamp [Sun, 1 Dec 2002 15:54:39 +0000 (15:54 +0000)]
Use more mnemonic argument names in the access functions.

Sponsored by: DARPA & NAI Labs
Approved by: re (blanket)

21 years agoFix a cut&past-o.
Poul-Henning Kamp [Sun, 1 Dec 2002 15:53:31 +0000 (15:53 +0000)]
Fix a cut&past-o.

Spotted by: yar
Approved by: re (blanket)

21 years agoConceiveably, there may exist an algorithm which can tell if a sequence of bytes
Poul-Henning Kamp [Sun, 1 Dec 2002 15:50:48 +0000 (15:50 +0000)]
Conceiveably, there may exist an algorithm which can tell if a sequence of bytes
are the output of AES/128/CBC or ARC4RANDOM.  Encrypt the random data with which
we wipe when we get a BIO_DELETE to make such an algorithm useful.

Sponsored by: DARPA & NAI Labs
Approved by: re (blanket)

21 years agoUse unsigned for an index.
Poul-Henning Kamp [Sun, 1 Dec 2002 15:47:49 +0000 (15:47 +0000)]
Use unsigned for an index.

Sponsored by: DARPA & NAI Labs.
Approved by: re (blanket).

21 years agoIn ata_reinit, when ata_getparam fails for a new device (indicating that
Thomas Quinot [Sun, 1 Dec 2002 14:20:08 +0000 (14:20 +0000)]
In ata_reinit, when ata_getparam fails for a new device (indicating that
the device is not actually present), clear the corresponding bit in
the ch->devices bitmap. This resolves a panic that occurred with ATAPI/CAM
after an APM suspend/resume, when the ATA hardware would erroneously
report an extra ATAPI device.

Approved by: re, sos
Reviewed by: roberto
MFC after: 7 days

21 years agoBootstrapping aid from pre-kqueue(2) systems, e.g. 4.0-RELEASE.
Ruslan Ermilov [Sun, 1 Dec 2002 13:38:25 +0000 (13:38 +0000)]
Bootstrapping aid from pre-kqueue(2) systems, e.g. 4.0-RELEASE.

Submitted by: jmallett
Approved by: re (bmah)

21 years agoMFbed: track changes in the english version and general cleanup
Udo Erdelhoff [Sun, 1 Dec 2002 12:56:41 +0000 (12:56 +0000)]
MFbed: track changes in the english version and general cleanup
early-adopter/article.sgml: 1.5   -> 1.7
hardware/common/dev.sgml: 1.115 -> 1.116
installation/common/upgrade.sgml: 1.9   -> 1.10
relnotes/common/new.sgml: 1.455 -> 1.460

Approved by: re (blanket)

21 years agoMajor update to the ahd driver to fix many bugs found in the previous
Scott Long [Sun, 1 Dec 2002 08:13:37 +0000 (08:13 +0000)]
Major update to the ahd driver to fix many bugs found in the previous
version, plus add support for the new features found in the Rev B
version of the chip.  The changelog is quite long and can be provided
on request.  Major features include vastly improved protocol violation
handling, full support for the 7902 Rev B, better parity error
handling, and better packetized overrun handling, to name a few.

Approved by: re (blanket)

21 years agoImplement workaround for broken busfree-rev in the A4.
Scott Long [Sun, 1 Dec 2002 07:59:01 +0000 (07:59 +0000)]
Implement workaround for broken busfree-rev in the A4.

HP -> CPQ

Rearrange IDs to better match which chips they use.

Convert to uniform product description strings.

Simplify 7901A setup function.

Add the NONPACKFIFO_BUG and PACED_NEGTABLE_BUG entries
for the A.

Add rev B bugs and features.

The double write workaround for CURRSCB is only required if
abort pending is set.  Remove this work around and set the
abort pending bug bit on the B at least until we have better
confirmation that the double write is always safe.

Add updated H2B identifiers

Move IOCell paramters into softc and add a hook for the
OSM to modify these as well as other settings prior to
committing them to the chip.

SLEW -> SLEWRATE

PREQDIS in DEVCONFIG1 went away after the A2.
Remove all code that references this bit.  This
is especially important since this bit was reused
in the B for a different HW fix workaround.

Properly set the AHD_NEW_IOCELL_OPTS and
AHD_NEW_DFCNTRL_OPTS features for the B.

Remove stray/random extra 7901A generic PCI
table entry.  Also switch the correct 7901A
generic entry to use ID_ALL_MASK since we
can only differentiate the 7901A from the
7902 by checking for a "type field" of 0xE.

Set AHD_INTCOLLISIONT_BUG for the Rev B.

Set the PREQDIS bit in DEVCONFIG1 for the B.  The
bit is misnamed, but seems to disable a work-around
that breaks on the B on PCI busses.

Add a routine for testing memory mapped register access.
This will hopefully detect things like buggy via chipsets
so that the OSM can fallback to using I/O mapped access
when memory mapped I/O simply will not work.

Approved by: re (blanket)

21 years agoAdd our controller name to the front of our
Scott Long [Sun, 1 Dec 2002 07:54:53 +0000 (07:54 +0000)]
Add our controller name to the front of our
diagnostic "Setting Mode" messages.

Use a read of HCNTRL to flush our write to CLRCMDCMPLT
on the RevB.  This allows us to check to see if the sequencer
is paused and to initiate the interrupt collision workaround
without incuring an extra read.

Approved by: re (blanket)

21 years agoRemove redundant check for chip type being PCI-X.
Scott Long [Sun, 1 Dec 2002 07:52:35 +0000 (07:52 +0000)]
Remove redundant check for chip type being PCI-X.
PCI-X only workarounds are automatically masked out
if we are operating in PCI mode.

Make use of ahd_pci_test_register_access()

Approved by: re (blanket)

21 years agoUpdate to include new ahd_scb_timer_reset API
Scott Long [Sun, 1 Dec 2002 07:50:56 +0000 (07:50 +0000)]
Update to include new ahd_scb_timer_reset API

Move IOCell paramters into softc and add a hook for the
OSM to modify these as well as other settings prior to
committing them to the chip.

Approved by: re (blanket)

21 years agoHold the page queues lock when calling pmap_protect(); it updates fields
Alan Cox [Sun, 1 Dec 2002 05:40:18 +0000 (05:40 +0000)]
Hold the page queues lock when calling pmap_protect(); it updates fields
of the vm_page structure.  Nearby, remove an unnecessary semicolon and
return statement.

Approved by: re (blanket)

21 years agoAdd a manpage. Nobody expects the spanish inquisition, but the
Marcel Moolenaar [Sun, 1 Dec 2002 02:10:34 +0000 (02:10 +0000)]
Add a manpage. Nobody expects the spanish inquisition, but the
mdoc(7) police... It's all yours ru :-)

Approved by: re (blanket)

21 years agoAssert that the page queues lock is held in pmap_changebit()
Alan Cox [Sun, 1 Dec 2002 00:08:14 +0000 (00:08 +0000)]
Assert that the page queues lock is held in pmap_changebit()
and pmap_ts_referenced().

Approved by: re (blanket)

21 years agoIncrease the scope of the page queue lock in vm_pageout_scan().
Alan Cox [Sun, 1 Dec 2002 00:02:39 +0000 (00:02 +0000)]
Increase the scope of the page queue lock in vm_pageout_scan().

Approved by: re (blanket)

21 years agoMark snapshots so that dump will dump them as zero length
Kirk McKusick [Sat, 30 Nov 2002 23:28:22 +0000 (23:28 +0000)]
Mark snapshots so that dump will dump them as zero length
regular files rather than trying to interpret the snapshot.

Sponsored by:   DARPA & NAI Labs.

21 years agoIf both the unlink and the open fail, return the errno from the
Bill Fenner [Sat, 30 Nov 2002 23:12:59 +0000 (23:12 +0000)]
If both the unlink and the open fail, return the errno from the
 unlink (very likely EPERM), since the errno from the open might
 be a confusing ETXTBSY.

Approved by: re
MFC After: 1 week

21 years agoAllow aliases to be used when specifying partition types. The use of
Marcel Moolenaar [Sat, 30 Nov 2002 22:51:46 +0000 (22:51 +0000)]
Allow aliases to be used when specifying partition types. The use of
UUIDs can then be limited to those cases when an alias doesn't exist.
This greatly increases the likelyhood that a sysadmin finishes the
partitioning without intermittent mental breakdowns. Current aliases
are "efi", "swap" and "ufs".

While here, staticize global variables and expand the usage message.

Approved by: re (blanket)

21 years agoFix typo in comment. It's SYSINIT, not SYSINT.
Giorgos Keramidas [Sat, 30 Nov 2002 22:15:30 +0000 (22:15 +0000)]
Fix typo in comment.  It's SYSINIT, not SYSINT.

Approved by: re (murray)

21 years agoAdd Bluetooth section and a cross-reference from the USB section for
Bruce A. Mah [Sat, 30 Nov 2002 22:13:38 +0000 (22:13 +0000)]
Add Bluetooth section and a cross-reference from the USB section for
USB Bluetooth adapters.

While here, make some USB text conditional on building for an
architecture for which we actually have USB support.

Approved by: re (implicitly)

21 years agoNew release note: Bluetooth support.
Bruce A. Mah [Sat, 30 Nov 2002 22:11:45 +0000 (22:11 +0000)]
New release note:  Bluetooth support.

Approved by: re (implicitly)

21 years agoMove the check for the MD_SHUTDOWN flag to before the tsleep() call
Ian Dowse [Sat, 30 Nov 2002 22:03:53 +0000 (22:03 +0000)]
Move the check for the MD_SHUTDOWN flag to before the tsleep() call
in the per-device kthread. This ensures that synchronisation with
mddestroy() succeeds even if the kthread was not waiting in tsleep()
at the time of the wakeup(). Among other things, this fixes the
problem of mdconfig getting stuck when an attempt is made to use a
zero-length file as a vnode-type backing store.

Approved by: re

21 years agoSynchronize perforce Id tags
Scott Long [Sat, 30 Nov 2002 20:04:10 +0000 (20:04 +0000)]
Synchronize perforce Id tags

Approved by: re (blanket)

21 years agoia64: make link /boot->efi/boot relative.
Marcel Moolenaar [Sat, 30 Nov 2002 19:54:19 +0000 (19:54 +0000)]
ia64: make link /boot->efi/boot relative.

Approved by: re (murray)

21 years agoBring in many bugfixes and changes obtained from formal testing:
Scott Long [Sat, 30 Nov 2002 19:30:09 +0000 (19:30 +0000)]
Bring in many bugfixes and changes obtained from formal testing:

        aic7xxx.c:
        aic7xxx.h:
        aic7xxx.reg:
        aic7xxx.seq:
                Bring in the protocol violation handler from the U320
                driver and replace the NO_IDENT sequencer interrupt code
                with the PROTO_VIOLATION code.  Support for this code
                required the following changes:

                SEQ_FLAGS:
                        IDENTIFY_SEEN -> NOT_IDENTIFIED
                        Added NO_CDB_SENT

                SCB_CONTROL:
                        TARGET_SCB == STATUS_RCVD for initiator mode

                scb->flags:
                        Added SCB_TARGET_SCB since we cannot rely on
                        TARGET_SCB as a target/initiator differentiator
                        due to it being overloaded in initiator mode to
                        indicate that status has been received.

        aic7xxx.seq:
                Move data fifo CLRCHN to mesgin_rdptrs which is a safer
                location for doing this operation.  This also saves a
                sequencer instruction.

        aic7xxx.c:
        aic7xxx.h:
                Change ahc/ahd_upate_neg_request() to take a "negotiation
                type" enum that allows us to negotiate:
                        o only if the goal and current parameters differ.
                        o only if the goal is non-async
                        o always - even if the negotiation will be for async.
        aic7xxx.seq:
                Reset the FIFO whenever a short CDB transfer occurs
                so that the FIFO contents do not corrupt a future CDB
                transfer retry.

                Add support for catching the various protocol violations
                handled by ahc_handle_protocol_violation.

                Reformat some comments.

        aic7xxx.c:
        aic7xxx.h:
                Just for safety, have the aic7xxx driver probe
                the stack depth.

        aic7xxx.c:
        aic7xxx.h:
                Save and restore stack contents during diagnostics.
                Some chip variants overwrite stale entries on a
                stack "pop".

                Don't use 0 to probe the stack depth.  0 is the typical
                value used to backfill the stack if entries are overwritten
                on a "pop".

        aic7xxx.h:
                Add a missing typedef.

                Collapse SCB flag entries so they are bit contiguous.

                Add AHD_ULTRA2_XFER_PERIOD for narrow fallback calculations

        aic7xxx.c:
                Don't panic (as a diagnostic to catch bugs) if we decided to
                force the renegotiation of async even if we believe we are
                already async.  This should allow us to negotiate async instead
                of the full user goal rate during startup if bus resets are
                disabled.

                Add a space to the end of the ahc/ahd_print_devinfo routines
                so that it behaves as expected by the code that uses it.

                Only force a renegotiation on a selection timeout
                if the SCB was valid.  Doing otherwise may be dangerous
                as the connection was not valid for an unknown reason.

                Add additional diagnostic output to ahc_dump_card_state(),
                and have it use the register pretty printing functions.

                Update ahc_reg_print() to handle a NULL cur_col.

                Add a newline to ahc_dump_card_state() output.

                Bring back "use_ppr".  We need to use_ppr anytime
                doppr is true or we have non-zero protocol options.
                The later case was not handled in the recent removal
                of use_ppr.

                Move a comment and remove a useless clearing of use_ppr.

                Don't disable ENBUSFREE when single stepping on
                a DT capable controller.  We cannot re-enable unexpected
                busfree detection, so we must clear BUSFREE on each
                step instead.

                Correct the lookup of the SCB ID in ahc_handle_proto_error.

                Remove a diagnostic printf.
                Remove unecessary restoration of the STACK for older
                chips.

Approved by: re (blanket)

21 years agoAdd a routine for testing memory mapped register access.
Scott Long [Sat, 30 Nov 2002 19:13:55 +0000 (19:13 +0000)]
Add a routine for testing memory mapped register access.
This will hopefully detect things like buggy via chipsets
so that the OSM can fallback to using I/O mapped access
when memory mapped I/O simply will not work.

Approved by: re (blanket)

21 years agoAlways compile in the AHC_ALLOW_MEMIO code, but only try MEMIO if
Scott Long [Sat, 30 Nov 2002 19:11:44 +0000 (19:11 +0000)]
Always compile in the AHC_ALLOW_MEMIO code, but only try MEMIO if
AHC_ALLOW_MEMIO is set, or the hint hint.ahc.N.allow_memio=1 is
set in the bootloader.

Make use of ah?_pci_test_register_access().

Approved by: re (blanket)

21 years agoHandle changes to SCB_CONTROL, scb->flags and SEQ_FLAGS
Scott Long [Sat, 30 Nov 2002 19:08:58 +0000 (19:08 +0000)]
Handle changes to SCB_CONTROL, scb->flags and SEQ_FLAGS

Approved by: re (blanket)

21 years agoUse the new ahc_scb_timer_reset API
Scott Long [Sat, 30 Nov 2002 19:06:24 +0000 (19:06 +0000)]
Use the new ahc_scb_timer_reset API

Remove AHC_ALLOW_MEMIO

Approved by: re (blanket)

21 years agoAdd a check to disable the previous patch so that future filesystems
Kirk McKusick [Sat, 30 Nov 2002 19:04:57 +0000 (19:04 +0000)]
Add a check to disable the previous patch so that future filesystems
that choose to place their superblocks in non-standard locations will
not get them smashed.

Sponsored by:   DARPA & NAI Labs.

21 years agoRemove a race condition / deadlock from snapshots. When
Kirk McKusick [Sat, 30 Nov 2002 19:00:51 +0000 (19:00 +0000)]
Remove a race condition / deadlock from snapshots. When
converting from individual vnode locks to the snapshot
lock, be sure to pass any waiting processes along to the
new lock as well. This transfer is done by a new function
in the lock manager, transferlockers(from_lock, to_lock);
Thanks to Lamont Granquist <lamont@scriptkiddie.org> for
his help in pounding on snapshots beyond all reason and
finding this deadlock.

Sponsored by:   DARPA & NAI Labs.

21 years ago#include <sys/time.h> instead of depending on namespace pollution in
Bruce Evans [Sat, 30 Nov 2002 18:59:31 +0000 (18:59 +0000)]
#include <sys/time.h> instead of depending on namespace pollution in
<sys/stat.h> for the declaration of struct timeval.

Approved by: re (rwatson)

21 years agoAdd some more checks to newfs so that it will not build filesystems
Kirk McKusick [Sat, 30 Nov 2002 18:28:26 +0000 (18:28 +0000)]
Add some more checks to newfs so that it will not build filesystems
that the kernel will refuse to mount. Specifically it now enforces
the MAXBSIZE blocksize limit. This update also fixes a problem where
newfs could segment fault if the selected fragment size was too large.

PR: bin/30959
Submitted by: Ceri Davies <setantae@submonkey.net>
Sponsored by:   DARPA & NAI Labs.

21 years agoAdd support for the Olivetti branded EISA boards.
Scott Long [Sat, 30 Nov 2002 18:00:43 +0000 (18:00 +0000)]
Add support for the Olivetti branded EISA boards.

Approved by: re (blanket)

21 years agoAssert that the page queues lock is held in pmap_page_exists_quick().
Alan Cox [Sat, 30 Nov 2002 17:46:59 +0000 (17:46 +0000)]
Assert that the page queues lock is held in pmap_page_exists_quick().

Approved by: re (blanket)

21 years agoMerge the changes in the English version.
Jun Kuriyama [Sat, 30 Nov 2002 09:52:19 +0000 (09:52 +0000)]
Merge the changes in the English version.

1.324 -> 1.334  relnotes/common/new.sgml

Submitted by: Hideyuki KURASHINA <rushani@jp.FreeBSD.org>
Approved by: re (blanket)

21 years agoFix two deadlocks in snapshots:
Kirk McKusick [Sat, 30 Nov 2002 07:27:12 +0000 (07:27 +0000)]
Fix two deadlocks in snapshots:

1) Release the snapshot file lock while suspending the system. Otherwise
   a process trying to read the lock may block on its containing directory
   preventing the suspension from completing. Thanks to Sean Kelly
   <smkelly@zombie.org> for finding this deadlock.

2) Replace some bdwrite's with bawrite's so as not to fill all the
   buffers with dirty data. The buffers could not be cleaned as the
   snapshot vnode was locked hence the system could deadlock when
   making snapshots of really massive filesystems. Thanks to
   Hidetoshi Shimokawa <simokawa@sat.t.u-tokyo.ac.jp> for figuring
   this out.

Sponsored by:   DARPA & NAI Labs.

21 years agoWhitespace cleanup only.
Jun Kuriyama [Sat, 30 Nov 2002 04:39:28 +0000 (04:39 +0000)]
Whitespace cleanup only.

Approved by: re (blanket)

21 years agoFix typos.
Jun Kuriyama [Sat, 30 Nov 2002 04:23:18 +0000 (04:23 +0000)]
Fix typos.

Submitted by: Hideyuki KURASHINA <rushani@jp.FreeBSD.org>
Approved by: re (blanket)

21 years agoMerge the following from the English version:
Hiroki Sato [Sat, 30 Nov 2002 03:36:37 +0000 (03:36 +0000)]
Merge the following from the English version:

1.313 -> 1.324 relnotes/common/new.sgml

Submitted by: Hideyuki KURASHINA <rushani@jp.FreeBSD.org>
Approved by: re (blanket)

21 years agoChange punctuation style.
Hiroki Sato [Sat, 30 Nov 2002 02:56:52 +0000 (02:56 +0000)]
Change punctuation style.

Submitted by: Hideyuki KURASHINA <rushani@jp.FreeBSD.org>
Approved by: re (blanket)

21 years ago- Catch up with the English version:
Hiroki Sato [Sat, 30 Nov 2002 01:27:00 +0000 (01:27 +0000)]
- Catch up with the English version:

1.5   -> 1.6   early-adopter/article.sgml

- Change punctuation style:

relnotes/common/intro.sgml
relnotes/common/new.sgml
relnotes/common/upgrading.sgml

Submitted by: Hideyuki KURASHINA <rushani@jp.FreeBSD.org>
Approved by: re (blanket)
References: [doc-jp-work 521]

21 years agodevd kernel improvements:
Warner Losh [Sat, 30 Nov 2002 00:49:43 +0000 (00:49 +0000)]
devd kernel improvements:
1) Record all device events when devctl is enabled, rather than just when
   devd has devctl open.  This is necessary to prevent races between when
   a device arrives, and when devd starts.
2) Add hw.bus.devctl_disable to disable devctl, this can also be set as a
   tunable.
3) Fix async support. Reset nonblocking and async_td in open.  remove
   async flags.
4) Free all memory when devctl is disabled.

Approved by: re (blanket)

21 years agoTwo fixes for devd:
Warner Losh [Sat, 30 Nov 2002 00:40:53 +0000 (00:40 +0000)]
Two fixes for devd:
1) Move devd to the start of the head of the boot process rather than the end.
2) Disable devctl if devd is disabled.

21 years agoConditionalize a GCCism.
Mike Barcroft [Fri, 29 Nov 2002 23:49:27 +0000 (23:49 +0000)]
Conditionalize a GCCism.

Approved by: re

21 years agoImplement bus_space_subregion(). Identical to i386.
Marcel Moolenaar [Fri, 29 Nov 2002 20:14:03 +0000 (20:14 +0000)]
Implement bus_space_subregion(). Identical to i386.

Approved by: re (carte blanc)

21 years agoBetter handle sparse physical memory: Don't use the address range
Marcel Moolenaar [Fri, 29 Nov 2002 20:10:21 +0000 (20:10 +0000)]
Better handle sparse physical memory: Don't use the address range
as a measure for available memory to scale the VHPT. Instead, use
the previously determined Maxmem.

Approved by: re (carte blanc)

21 years agoCheck to make sure that the fs_sblockloc field was properly updated
Kirk McKusick [Fri, 29 Nov 2002 19:20:15 +0000 (19:20 +0000)]
Check to make sure that the fs_sblockloc field was properly updated
before using it to write the superblock. This is to guard against
accidentally trashing the disklabel if the superblock format missed
being upgraded by the new kernel.

Reported by: Sam Leffler <sam@errno.com>
Sponsored by:   DARPA & NAI Labs.
Approved by: Murray Stokely <murray@FreeBSD.org>

21 years agomdoc(7) police: sweep.
Ruslan Ermilov [Fri, 29 Nov 2002 17:35:09 +0000 (17:35 +0000)]
mdoc(7) police: sweep.

21 years agomdoc(7) police: sweep.
Ruslan Ermilov [Fri, 29 Nov 2002 16:42:23 +0000 (16:42 +0000)]
mdoc(7) police: sweep.

21 years agoPut back a test for binaries with no PT_LOAD entries I over-jealosly
Alexander Kabaev [Fri, 29 Nov 2002 16:41:31 +0000 (16:41 +0000)]
Put back a test for binaries with no PT_LOAD entries I over-jealosly
removed in r1.69.

Apploved by: re (rwatson)

21 years agomdoc(7) police: sweep.
Ruslan Ermilov [Fri, 29 Nov 2002 16:21:33 +0000 (16:21 +0000)]
mdoc(7) police: sweep.

21 years agomdoc(7) police: formatting nits.
Ruslan Ermilov [Fri, 29 Nov 2002 15:57:50 +0000 (15:57 +0000)]
mdoc(7) police: formatting nits.

Approved by: re

21 years agoRelease the reference to the peripheral if returning a error.
Yaroslav Tykhiy [Fri, 29 Nov 2002 15:40:10 +0000 (15:40 +0000)]
Release the reference to the peripheral if returning a error.
That reference is to be held only if daopen() has been successful
and until daclose() releases it.  daclose() won't be called if
daopen() has failed, though.

Approved by: re, njl
MFC after: 1 week

21 years agoBack out part of last delta that was committed by accident.
Ruslan Ermilov [Fri, 29 Nov 2002 11:50:08 +0000 (11:50 +0000)]
Back out part of last delta that was committed by accident.
(.Xr is not yet that smart.)

21 years agomdoc(7) police: scheduled sweep.
Ruslan Ermilov [Fri, 29 Nov 2002 11:39:20 +0000 (11:39 +0000)]
mdoc(7) police: scheduled sweep.

Approved by: re

21 years agomdoc(7) police: catch up to the code changes.
Ruslan Ermilov [Fri, 29 Nov 2002 11:36:32 +0000 (11:36 +0000)]
mdoc(7) police: catch up to the code changes.

Approved by: re

21 years agoWhitespace nits.
Dag-Erling Smørgrav [Thu, 28 Nov 2002 20:11:31 +0000 (20:11 +0000)]
Whitespace nits.

Approved by: re (bmah)

21 years agoAdd a PAM_MODULE_ENTRY to this module so it'll actually do something.
Dag-Erling Smørgrav [Thu, 28 Nov 2002 20:05:42 +0000 (20:05 +0000)]
Add a PAM_MODULE_ENTRY to this module so it'll actually do something.

Approved by: re (bmah)

21 years agoMisc. edits and cleanup.
Bruce A. Mah [Thu, 28 Nov 2002 19:50:56 +0000 (19:50 +0000)]
Misc. edits and cleanup.

21 years agotry SIOCGIFINDEX 1st to be able to use network aliasing.
Hajimu UMEMOTO [Thu, 28 Nov 2002 17:46:40 +0000 (17:46 +0000)]
try SIOCGIFINDEX 1st to be able to use network aliasing.

Submitted by: jlemon
Approved by: re

21 years agoFix the standards case by referring to __fds_bits instead of fds_bits
Mike Barcroft [Thu, 28 Nov 2002 15:34:32 +0000 (15:34 +0000)]
Fix the standards case by referring to __fds_bits instead of fds_bits
in FD_*() macros.

Submitted by: Marc Recht <marc@informatik.uni-bremen.de>
Approved by: re

21 years agoRemove the <sys/types.h> prerequisite for <ucontext.h> by including
Mike Barcroft [Thu, 28 Nov 2002 15:32:49 +0000 (15:32 +0000)]
Remove the <sys/types.h> prerequisite for <ucontext.h> by including
<sys/_types.h> in <sys/signal.h>.

Approved by: re

21 years agoAdd a test for what was fixed in revisions 1.39 and 1.50 of
Ruslan Ermilov [Thu, 28 Nov 2002 13:21:32 +0000 (13:21 +0000)]
Add a test for what was fixed in revisions 1.39 and 1.50 of
make/parse.c (allow embedded `:' and `!' in target names).

Approved by: re

21 years agoMoved make(1) regression tests from src/Makefile to where they
Ruslan Ermilov [Thu, 28 Nov 2002 13:08:09 +0000 (13:08 +0000)]
Moved make(1) regression tests from src/Makefile to where they
belong (src/tools/regression/usr.bin/make), and use the latter
to test if make(1) is adequate for building the world.

Approved by: re

21 years agoFinish the fix in revision 1.39 -- make(1)'s behavior is now
Ruslan Ermilov [Thu, 28 Nov 2002 12:47:56 +0000 (12:47 +0000)]
Finish the fix in revision 1.39 -- make(1)'s behavior is now
"greedy" with respect to finding the dependency operators.

Approved by: re

21 years agoImplement and document support for an HTTP_REFERER environment variable.
Dag-Erling Smørgrav [Thu, 28 Nov 2002 12:07:15 +0000 (12:07 +0000)]
Implement and document support for an HTTP_REFERER environment variable.

PR: 28171
Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
Approved by: re (bmah)
MFC after: 1 week

21 years agoUse pmap_remove_all() instead of pmap_remove() before freeing the page
Alan Cox [Thu, 28 Nov 2002 08:44:26 +0000 (08:44 +0000)]
Use pmap_remove_all() instead of pmap_remove() before freeing the page
in vm_pgmoveco(); the page may have more than one mapping.  Hold the page
queues lock when calling pmap_remove_all().

Approved by: re (blanket)

21 years agoLock page field accesses in mincore().
Alan Cox [Thu, 28 Nov 2002 08:01:39 +0000 (08:01 +0000)]
Lock page field accesses in mincore().

Approved by: re (blanket)

21 years agoReclaim major number 182 which I allocated for mdsio. It was decided
Robert Watson [Thu, 28 Nov 2002 04:29:52 +0000 (04:29 +0000)]
Reclaim major number 182 which I allocated for mdsio.  It was decided
that this should become a line discipline for sio rather than a
separate driver.