]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
26 years agoFTP_INTERNAL_LS -> FTPD_INTERNAL_LS
steve [Thu, 4 Dec 1997 03:58:02 +0000 (03:58 +0000)]
FTP_INTERNAL_LS -> FTPD_INTERNAL_LS

Pointed out by: Jaye Mathisen <mrcpu@cdsnet.net>

26 years agoUse 'proto <tcp|udp|all>' to select the protocol to display
steve [Thu, 4 Dec 1997 03:44:46 +0000 (03:44 +0000)]
Use 'proto <tcp|udp|all>' to select the protocol to display
in netstat-mode to avoid a conflict with tcp-mode.  Also
while documenting this new feature in the manpage, fix a
minor display nit.

PR: 5159
Submitted by: Sergei Chechetkin <csl@whale.sunbay.crimea.ua>

26 years agoBuffer overflow from bitblt's commit to OpenBSD. Committed here for lack
imp [Thu, 4 Dec 1997 02:54:05 +0000 (02:54 +0000)]
Buffer overflow from bitblt's commit to OpenBSD.  Committed here for lack
of a better place to do it.

Reviewed by: joerge
Obtained from: OpenBSD

26 years agoThe Libretto's BIOS doesn't set edx on the APM_GETPWSTATUS call, so
imp [Thu, 4 Dec 1997 02:40:00 +0000 (02:40 +0000)]
The Libretto's BIOS doesn't set edx on the APM_GETPWSTATUS call, so
the barrery time remaining is reported as a random number.  Initialize
edx to 0xffff in this case, and to 0 in all other cases.  This change
should be benign on other machines.
Reviewed by: jdp

26 years agoFix the CCP Type field value for DEFLATE.
brian [Wed, 3 Dec 1997 23:28:02 +0000 (23:28 +0000)]
Fix the CCP Type field value for DEFLATE.
(I *really* meant to do this  *before* committing the
deflate changes in the first place - oops).

Pppd is horribly broken in this respect - refer to the
ppp man page for details.  Ppp *WON'T* negotiate deflate
with pppd by default - you must ``enable'' and ``accept''
``pppd-deflate'' in your config.

While I'm in there, update the cftypes in ccp.c so that
we recognise some more protocols (we don't actually do
anything with them - just send a REJ).

26 years agoFixed corruption of the per-group used directories count. It wasn't
bde [Wed, 3 Dec 1997 16:46:21 +0000 (16:46 +0000)]
Fixed corruption of the per-group used directories count.  It wasn't
decremented when directories were removed because rev.1.12 broke the
fixup of the i_mode of the inode being removed.

26 years agoMake has_f00f_bug extern, and get rid of some unused code in the f00f
sef [Wed, 3 Dec 1997 16:27:33 +0000 (16:27 +0000)]
Make has_f00f_bug extern, and get rid of some unused code in the f00f
code.

Submitted by: Mikael Karpberg & Cy Schubert

26 years agoAbstract the CCP layer a level.
brian [Wed, 3 Dec 1997 10:23:54 +0000 (10:23 +0000)]
Abstract the CCP layer a level.
Add DEFLATE support.

26 years agoSync with sys/i386/i386/machdep.c and trap.c revisions 1.275 and
kato [Wed, 3 Dec 1997 09:46:34 +0000 (09:46 +0000)]
Sync with sys/i386/i386/machdep.c and trap.c revisions 1.275 and
1.116, respectively.

26 years agoUse full path in synopsis. Syslog will add trailing \n.
charnier [Wed, 3 Dec 1997 07:19:58 +0000 (07:19 +0000)]
Use full path in synopsis. Syslog will add trailing \n.

26 years agoUse err(3). Remove progname and trailing \n in syslog strings.
charnier [Wed, 3 Dec 1997 07:16:08 +0000 (07:16 +0000)]
Use err(3). Remove progname and trailing \n in syslog strings.

26 years agoWork around for the Intel Pentium F00F bug; this is Intel's recommended
sef [Wed, 3 Dec 1997 02:45:50 +0000 (02:45 +0000)]
Work around for the Intel Pentium F00F bug; this is Intel's recommended
workaround.  Note that this currently eats up two pages extra in the system;
this could be alleviated by aligning idt correctly, and then only dealing with
that (as opposed to the current method of allocated two pages and copying the
IDT table to that, and then setting that to be the IDT table).

26 years agoExpand default datasize to 22M, perl5 & pine are usual memory eaters
ache [Wed, 3 Dec 1997 01:12:48 +0000 (01:12 +0000)]
Expand default datasize to 22M, perl5 & pine are usual memory eaters

26 years ago- Remove the code that cleared out the registers (previously enabled by the
nate [Tue, 2 Dec 1997 23:23:14 +0000 (23:23 +0000)]
- Remove the code that cleared out the registers (previously enabled by the
  option PCIC_NOCLRREGS).  This is now the default behavior since it's
  apparently required for the CLPD6832, and doesn't negatively affect
  any of my test machines.

Requested by: Ted Faber <faber@ISI.EDU>

26 years agoFix size of start queue to 32 entries, independent of the default
se [Tue, 2 Dec 1997 22:37:58 +0000 (22:37 +0000)]
Fix size of start queue to 32 entries, independent of the default
number of tags (NCR_SCSI_DFLT_TAGS), which is 0 in the FAILSAFE case.
This should fix the incompatibility between kernel and ncrcontrol,
which is the result of FAILSAFE being defined in the kernel config
file, invisible to the build of ncrcontrol. (See kern/5133, which
should be fixed by this change.)

26 years ago- Add necessary include files and fix bugs in last.
nate [Tue, 2 Dec 1997 22:27:58 +0000 (22:27 +0000)]
- Add necessary include files and fix bugs in last.

26 years ago- Bring in code removed from /sys/pccard/pcic.c, including DEVICE IDs, and
nate [Tue, 2 Dec 1997 22:13:59 +0000 (22:13 +0000)]
- Bring in code removed from /sys/pccard/pcic.c, including DEVICE IDs, and
  more bootverbose code.
- Style nits.

No significant functional changes.

26 years agoNow that it's built at the same time as its source, disable building 07.lpr
wollman [Tue, 2 Dec 1997 21:46:58 +0000 (21:46 +0000)]
Now that it's built at the same time as its source, disable building 07.lpr
here.  Hopefully this will set a trend.  (What was I thinking when I set
this up the first time?)

26 years agoDocument the new long names for printcap capabilities.
wollman [Tue, 2 Dec 1997 21:41:40 +0000 (21:41 +0000)]
Document the new long names for printcap capabilities.

26 years ago- Remove PCI code from here, now that the PCI framework lives in /sys/pci.
nate [Tue, 2 Dec 1997 21:31:35 +0000 (21:31 +0000)]
- Remove PCI code from here, now that the PCI framework lives in /sys/pci.

26 years agoDefine MS_SYNC for compatibility.
dyson [Tue, 2 Dec 1997 21:30:03 +0000 (21:30 +0000)]
Define MS_SYNC for compatibility.

26 years ago- Framework for PCI/CardBus controllers running in PCMCIA emulation
nate [Tue, 2 Dec 1997 21:26:41 +0000 (21:26 +0000)]
- Framework for PCI/CardBus controllers running in PCMCIA emulation
  mode.  Currently, the only supported controller is the Cirrus Logic
  PD6832, but others can be supported with docs on them.

Submitted by: Ted Faber <faber@ISI.EDU>

26 years agoFix the copyright and attribution on this file. I forgot this
phk [Tue, 2 Dec 1997 21:20:06 +0000 (21:20 +0000)]
Fix the copyright and attribution on this file.  I forgot this
when the file was cloned.

26 years agoFix a bug that caused cdboot to stop reading the root directory at the
joerg [Tue, 2 Dec 1997 21:13:59 +0000 (21:13 +0000)]
Fix a bug that caused cdboot to stop reading the root directory at the
end of the first block.

Problem found by: Kenneth Merry <ken@plutotech.com>

26 years agoIn all such uses of struct buf: 's/b_un.b_addr/b_data/g'
phk [Tue, 2 Dec 1997 21:07:20 +0000 (21:07 +0000)]
In all such uses of struct buf: 's/b_un.b_addr/b_data/g'

26 years agoMega lpd/lpd upgrade, part I:
wollman [Tue, 2 Dec 1997 20:46:22 +0000 (20:46 +0000)]
Mega lpd/lpd upgrade, part I:

- Get rid of a lot of the static variables which were shared by
  many routines and programs in the suite.
- Create an abstract interface to the printcap database, so that
  other retrieval and iteration mechanisms could be developed
  (e.g., YP, Hesiod, or automatic retrieval from a trusted server).
- Give each capability a human-readable name in addition to the historic
  two-character one.
- Otherwise generally clean up a lot of dark corners.  Many still remain.
- When submitting jobs, use the official login name record (from getlogin())
  if there is one, rather than reverse-mapping the uid.

More to come...

26 years agoUse err(3). Add protos for -Wall.
charnier [Tue, 2 Dec 1997 12:33:42 +0000 (12:33 +0000)]
Use err(3). Add protos for -Wall.

26 years agoNo \n in syslog() strings. Add man page to Xrefs. Change null byte to NUL byte.
charnier [Tue, 2 Dec 1997 12:30:04 +0000 (12:30 +0000)]
No \n in syslog() strings. Add man page to Xrefs. Change null byte to NUL byte.

26 years agoDo not terminate syslog() messages with a dot, as others daemons do.
charnier [Tue, 2 Dec 1997 12:25:39 +0000 (12:25 +0000)]
Do not terminate syslog() messages with a dot, as others daemons do.

26 years agoDocument -n flag. Use err(3). Add usage.
charnier [Tue, 2 Dec 1997 12:20:17 +0000 (12:20 +0000)]
Document -n flag. Use err(3). Add usage.
Add syslog capability.

26 years ago"un-bump" the major number for libtermcap.so. This brings -current back
peter [Tue, 2 Dec 1997 11:56:36 +0000 (11:56 +0000)]
"un-bump" the major number for libtermcap.so.  This brings -current back
to the same version numbers as 2.2.x.
The problem with the way things were was:
 - if you took a 2.2.x binary, it either wouldn't run on -current or
   if you had the old -current version of libtermcap.so.2.1 then it could
   potentially be a security problem.
 - the alternative is to start a compat22 tree dist for -current with a
   uuencoded binary.  This makefile hack is less cost.
libtermcap.so.3.0 is provided via /usr/lib/compat to avoid transition
problems.

26 years ago`nextgennumber' can go away now that is no longer (ab)used by foreign
bde [Tue, 2 Dec 1997 11:43:45 +0000 (11:43 +0000)]
`nextgennumber' can go away now that is no longer (ab)used by foreign
fs's.

26 years agoUse the same algorithm as ffs for generation numbers.
bde [Tue, 2 Dec 1997 11:42:28 +0000 (11:42 +0000)]
Use the same algorithm as ffs for generation numbers.

26 years agoFix a small style bug in the generation number change (rev.1.33) before
bde [Tue, 2 Dec 1997 11:21:16 +0000 (11:21 +0000)]
Fix a small style bug in the generation number change (rev.1.33) before
copying the change to other fs's.

26 years agoRemoved __FreeBSD__ ifdefs.
bde [Tue, 2 Dec 1997 10:39:42 +0000 (10:39 +0000)]
Removed __FreeBSD__ ifdefs.

26 years agoCleaned up __getcwd(). This should be cosmetic except disabled calls
bde [Tue, 2 Dec 1997 10:32:21 +0000 (10:32 +0000)]
Cleaned up __getcwd().  This should be cosmetic except disabled calls
are now counted.

Reviewed by: phk

26 years agoSync with sys/i386/conf/files.i386 revision 1.180.
kato [Tue, 2 Dec 1997 08:20:34 +0000 (08:20 +0000)]
Sync with sys/i386/conf/files.i386 revision 1.180.

26 years agoRevisions 1.14 and 1.9.2.3 fixed PR conf/5127.
obrien [Mon, 1 Dec 1997 21:18:39 +0000 (21:18 +0000)]
Revisions 1.14 and 1.9.2.3 fixed PR conf/5127.

26 years agoFix a serious problem during resizing buffers where old buffers
dyson [Mon, 1 Dec 1997 19:04:00 +0000 (19:04 +0000)]
Fix a serious problem during resizing buffers where old buffers
address space wasn't being properly reclaimed.
Submitted by: Bruce Evans <bde@freebsd.org>

26 years agoFix a problem when creating a new kernel thread. In some cases, aio_read
dyson [Mon, 1 Dec 1997 18:41:08 +0000 (18:41 +0000)]
Fix a problem when creating a new kernel thread.  In some cases, aio_read
or aio_write can return the pid of the new thread.  This is due to the
way that return values from system calls being passed by side-effect in
the proc structure now.  This commit fixes the problem with aio_read and
aio_write.

26 years agoCleanup my last patch here
julian [Mon, 1 Dec 1997 11:34:41 +0000 (11:34 +0000)]
Cleanup my last patch here
Reviewed by: sef@kthrup.com and phk@freebsd.org

26 years agoSync with sys/i386/isa/isa.c revision 1.108.
kato [Mon, 1 Dec 1997 09:37:26 +0000 (09:37 +0000)]
Sync with sys/i386/isa/isa.c revision 1.108.

26 years agoInclude sound_timer.c for mss device and added
ahasty [Mon, 1 Dec 1997 09:29:50 +0000 (09:29 +0000)]
Include sound_timer.c for mss device and added
sound_timer.c, opl3.c, ad1848.c, adlib_card.c to trix device.
trix is a driver for an AudioTrix Pro.

26 years agoFix error handling for VCHR type I/O. Also, fix another spl problem, and
dyson [Mon, 1 Dec 1997 07:01:45 +0000 (07:01 +0000)]
Fix error handling for VCHR type I/O.  Also, fix another spl problem, and
remove alot of overly verbose debugging statements.
ioproclist {
int aioprocflags; /* AIO proc flags */
TAILQ_ENTRY(aioproclist) list; /* List of processes */
struct proc *aioproc; /* The AIO thread */
TAILQ_HEAD (,aiocblist) jobtorun; /* suggested job to run */
};

/*
 * data-structure for lio signal management
 */
struct aio_liojob {
int lioj_flags;
int lioj_buffer_count;
int lioj_buffer_finished_count;
int lioj_queue_count;
int lioj_queue_finished_count;
struct sigevent lioj_signal; /* signal on all I/O done */
TAILQ_ENTRY (aio_liojob) lioj_list;
struct kaioinfo *lioj_ki;
};
#define LIOJ_SIGNAL 0x1 /* signal on all done (lio) */
#define LIOJ_SIGNAL_POSTED 0x2 /* signal has been posted */

/*
 * per process aio data structure
 */
struct kaioinfo {
int kaio_flags; /* per process kaio flags */
int kaio_maxactive_count; /* maximum number of AIOs */
int kaio_active_count; /* number of currently used AIOs */
int kaio_qallowed_count; /* maxiumu size of AIO queue */
int kaio_queue_count; /* size of AIO queue */
int kaio_ballowed_count; /* maximum number of buffers */
int kaio_queue_finished_count; /* number of daemon jobs finished */
int kaio_buffer_count; /* number of physio buffers */
int kaio_buffer_finished_count; /* count of I/O done */
struct proc *kaio_p; /* process that uses this kaio block */
TAILQ_HEAD (,aio_liojob) kaio_liojoblist; /* list of lio jobs */
TAILQ_HEAD (,aiocblist) kaio_jobqueue; /* job queue for process */
TAILQ_HEAD (,aiocblist) kaio_jobdone; /* done queue for process */
TAILQ_HEAD (,aiocblist) kaio_bufqueue; /* buffer job queue for process */
TAILQ_HEAD (,aiocblist) kaio_bufdone; /* buffer done queue for process */
};

#define KAIO_RUNDOWN 0x1 /* process is being run down */
#define KAIO_WAKEUP 0x2 /* wakeup process when there is a significant
   event */

TAILQ_HEAD (,aioproclist) aio_freeproc, aio_activeproc;
TAILQ_HEAD(,aiocblist) aio_jobs; /* Async job list */
TAILQ_HEAD(,aiocblist) aio_bufjobs; /* Phys I/O job list */
TAILQ_HEAD(,aiocblist) aio_freejobs; /* Pool of free jobs */

static void aio_init_aioinfo(struct proc *p) ;
static void aio_onceonly(void *) ;
static int aio_free_entry(struct aiocblist *aiocbe);
static void aio_process(struct aiocblist *aiocbe);
static int aio_newproc(void) ;
static int aio_aqueue(struct proc *p, struct aiocb *job, int type) ;
static void aio_physwakeup(struct buf *bp);
static int aio_fphysio(struct proc *p, struct aiocblist *aiocbe, int type);
static int aio_qphysio(struct proc *p, struct aiocblist *iocb);
static void aio_daemon(void *uproc);

SYSINIT(aio, SI_SUB_VFS, SI_ORDER_ANY, aio_onceonly, NULL);

static vm_zone_t kaio_zone=0, aiop_zone=0,
aiocb_zone=0, aiol_zone=0, aiolio_zone=0;

/*
 * Single AIOD vmspace shared amongst all of them
 */
static struct vmspace *aiovmspace = NULL;

/*
 * Startup initialization
 */
void
aio_onceonly(void *na)
{
TAILQ_INIT(&aio_freeproc);
TAILQ_INIT(&aio_activeproc);
TAILQ_INIT(&aio_jobs);
TAILQ_INIT(&aio_bufjobs);
TAILQ_INIT(&aio_freejobs);
kaio_zone = zinit("AIO", sizeof (struct kaioinfo), 0, 0, 1);
aiop_zone = zinit("AIOP", sizeof (struct aioproclist), 0, 0, 1);
aiocb_zone = zinit("AIOCB", sizeof (struct aiocblist), 0, 0, 1);
aiol_zone = zinit("AIOL", AIO_LISTIO_MAX * sizeof (int), 0, 0, 1);
aiolio_zone = zinit("AIOLIO",
AIO_LISTIO_MAX * sizeof (struct aio_liojob), 0, 0, 1);
aiod_timeout = AIOD_TIMEOUT_DEFAULT;
aiod_lifetime = AIOD_LIFETIME_DEFAULT;
jobrefid = 1;
}

/*
 * Init the per-process aioinfo structure.
 * The aioinfo limits are set per-process for user limit (resource) management.
 */
void
aio_init_aioinfo(struct proc *p)
{
struct kaioinfo *ki;
if (p->p_aioinfo == NULL) {
ki = zalloc(kaio_zone);
p->p_aioinfo = ki

26 years agoMF 22s
obrien [Mon, 1 Dec 1997 06:11:34 +0000 (06:11 +0000)]
MF 22s

26 years agoReorder struct callout for better cacheline behavior.
dg [Mon, 1 Dec 1997 05:45:15 +0000 (05:45 +0000)]
Reorder struct callout for better cacheline behavior.

26 years agoAdd mount_devfs(8) to the SEE ALSO section.
steve [Mon, 1 Dec 1997 00:44:16 +0000 (00:44 +0000)]
Add mount_devfs(8) to the SEE ALSO section.

PR: 5125

26 years ago improve anti-spam rulesets: check_relay, check_mail, check_rcpt.
jmb [Mon, 1 Dec 1997 00:15:38 +0000 (00:15 +0000)]
    improve anti-spam rulesets: check_relay, check_mail, check_rcpt.
        break each ruleset into identified sections. (called groups).
        note which groups can be reordered.
        each group accepts and returns the same strings,
                as much as possible.
        reactivate Paul Vixie's RBL (in check_mail)
        add rules to limit mail relaying to a list of hosts and domains
                in the R class (check_rcpt, not active on hub.freebsd.org)
Submitted by:  jmb

26 years ago improve anti-spam rulesets: check_relay, check_mail, check_rcpt.
jmb [Sun, 30 Nov 1997 23:28:26 +0000 (23:28 +0000)]
improve anti-spam rulesets: check_relay, check_mail, check_rcpt.
break each ruleset into identified sections. (called groups).
note which groups can be reordered.
each group accepts and returns the same strings,
as much as possible.
reactivate Paul Vixie's RBL (in check_mail)
add rules to limit mail relaying to a list of hosts and domains
in the R class (check_rcpt, not active on hub.freebsd.org)

Submitted by:  jmb

26 years agoCorrect a last minute code change. Would have been an infinite loop under
dyson [Sun, 30 Nov 1997 23:21:08 +0000 (23:21 +0000)]
Correct a last minute code change.  Would have been an infinite loop under
certain error conditions.
Submitted by: pst@shockwave.com

26 years agoFix an spl nit.
dyson [Sun, 30 Nov 1997 21:47:36 +0000 (21:47 +0000)]
Fix an spl nit.

26 years agoRemove manpath.[ch] here and use the ones in ../manpath instead.
steve [Sun, 30 Nov 1997 20:08:27 +0000 (20:08 +0000)]
Remove manpath.[ch] here and use the ones in ../manpath instead.

26 years agoResurrect manpath.c here.
steve [Sun, 30 Nov 1997 19:55:16 +0000 (19:55 +0000)]
Resurrect manpath.c here.

26 years agoSleep 10secs after kill -HUP instead of 3secs, some daemons like Apache
ache [Sun, 30 Nov 1997 18:58:18 +0000 (18:58 +0000)]
Sleep 10secs after kill -HUP instead of 3secs, some daemons like Apache
may have log time to re-configure

26 years agoRemove this copy of manpath.c and add the .PATH directive to get to
steve [Sun, 30 Nov 1997 17:17:29 +0000 (17:17 +0000)]
Remove this copy of manpath.c and add the .PATH directive to get to
the same file in ${.CURDIR}/../man.

Suggested by: Bruce Evans

26 years agofix a few style nites...
jmg [Sun, 30 Nov 1997 09:44:28 +0000 (09:44 +0000)]
fix a few style nites...

make isa_dmacascade, isa_dmastart, isa_dmadone, and find_isadev MUCH
easier to be found by starting them at the beginging of the line...

remove braces inside of ifdef RESOURCE_CHECK... found by % in vi...

26 years agoFinish up the vast majority of the AIO/LIO functionality. Proper signal
dyson [Sun, 30 Nov 1997 04:36:31 +0000 (04:36 +0000)]
Finish up the vast majority of the AIO/LIO functionality.  Proper signal
support was missing in the previous version of the AIO code.  More
tunables added, and very efficient support for VCHR files has been added.
Kernel threads are not used for VCHR files, all work for such files is
done for the requesting process directly.  Some attempt has been made to
charge the requesting process for resource utilization, but more work
is needed.  aio_fsync is still missing (but the original fsync system
call can be used for now.)  aio_cancel is essentially a noop, but that
is okay per POSIX.  More aio_cancel functionality can be added later,
if it is found to be needed.

The functions implemented include:
aio_read, aio_write, lio_listio, aio_error, aio_return,
aio_cancel, aio_suspend.

The code has been implemented to support the POSIX spec 1003.1b
(formerly known as POSIX 1003.4 spec) features of the above.  The
async I/O features are truly async, with the VCHR mode of operation
being essentially the same as physio (for appropriate files) for
maximum efficiency.  This code also supports the signal capability,
is highly tunable, allowing management of resource usage, and
has been written to allow a per process usage quota.

Both the O'Reilly POSIX.4 book and the actual POSIX 1003.1b document
were the reference specs used.  Any filedescriptor can be used with
these new system calls.  I know of no exceptions where these
system calls will not work.  (TTY's will also probably work.)

26 years agoAlways call gripe_reading_mp_config with the required arguments.
steve [Sun, 30 Nov 1997 01:01:47 +0000 (01:01 +0000)]
Always call gripe_reading_mp_config with the required arguments.

PR: 3894
Submitted by: Stephen Clawson <sclawson@marker.cs.utah.edu>

26 years agos/geteid/geteuid/ - it's lucky I have a large supply of left-over pointy
peter [Sat, 29 Nov 1997 11:39:31 +0000 (11:39 +0000)]
s/geteid/geteuid/ - it's lucky I have a large supply of left-over pointy
hats from Tristan's last birthday party. :-]

26 years agoWork around the problems caused by calling issetugid() in libtermcap in
peter [Sat, 29 Nov 1997 11:30:57 +0000 (11:30 +0000)]
Work around the problems caused by calling issetugid() in libtermcap in
a similar way to libc. Sigh.  This is not pretty but seems to work.
Somthing like this was needed in preference to bogusly bumping the major
library number here.

The syscall(SYS_issetugid) idea is originally Bruce's.

26 years agoShuffle things a bit for better cacheline behavior.
dg [Sat, 29 Nov 1997 08:11:01 +0000 (08:11 +0000)]
Shuffle things a bit for better cacheline behavior.

26 years agoGet rid of the dynamic linker's internal malloc package, and arrange
jdp [Sat, 29 Nov 1997 03:32:48 +0000 (03:32 +0000)]
Get rid of the dynamic linker's internal malloc package, and arrange
things so that it uses the same malloc as is used by the program
being executed.  This has several advantages, the big one being
that you can now debug core dumps from dynamically linked programs
and get useful information out of them.  Until now, that didn't
work.  The internal malloc package placed the tables describing
the loaded shared libraries in a mapped region of high memory that
was not written to core files.  Thus the debugger had no way of
determining what was loaded where in memory.  Now that the dynamic
linker uses the application's malloc package (normally, but not
necessarily, the system malloc), its tables end up in the regular
heap area where they will be included in core dumps.  The debugger
now works very well indeed, thank you very much.

Also ...

Bring the program a little closer to conformance with style(9).
There is still a long way to go.

Add minimal const correctness changes to get rid of compiler warnings
caused by the recent const changes in <dlfcn.h> and <link.h>.

Improve performance by eliminating redundant calculations of symbols'
hash values.

26 years agoDisable the VCHR optimization for AIO until I have implemented it. Just in
dyson [Sat, 29 Nov 1997 02:57:46 +0000 (02:57 +0000)]
Disable the VCHR optimization for AIO until I have implemented it.  Just in
case anyone wants to play with the POSIX AIO/LIO stuff.  (As it is, it should
work with ANY vnode, on UP systems only, for now.)

26 years agoFix and complete the AIO syscalls. There are some performance enhancements
dyson [Sat, 29 Nov 1997 01:33:10 +0000 (01:33 +0000)]
Fix and complete the AIO syscalls.  There are some performance enhancements
coming up soon, but the code is functional.  Docs will be forthcoming.

26 years agoIn the "ldd -v" output, display the N_AUX information for each
jdp [Fri, 28 Nov 1997 19:34:27 +0000 (19:34 +0000)]
In the "ldd -v" output, display the N_AUX information for each
symbol.  It indicates whether the symbol refers to a function or a
data object.

26 years agoRemove the prototypes for dlopen and related functions. They don't
jdp [Fri, 28 Nov 1997 19:05:11 +0000 (19:05 +0000)]
Remove the prototypes for dlopen and related functions.  They don't
belong here.  The standard place for them is <dlfcn.h>, and there is
already another copy of the prototypes there.

26 years agoSort option list so that -amin works.
steve [Fri, 28 Nov 1997 15:48:08 +0000 (15:48 +0000)]
Sort option list so that -amin works.

PR: 5171
Submitted by: Dmitrij Tejblum <tejblum@arc.hq.cti.ru>

26 years agoThe previous revision (objfiles.c:1.2) fixed backtracing in the
jdp [Thu, 27 Nov 1997 21:26:08 +0000 (21:26 +0000)]
The previous revision (objfiles.c:1.2) fixed backtracing in the
absence of full debugging symbols for the kernel, but broke it for
application programs.  This commit disables that change except when
kernel debugging mode is in effect.

This needs to go into -2.2 as well, after a suitable burn-in period.

26 years agoUpgrade minor version
pst [Thu, 27 Nov 1997 20:52:28 +0000 (20:52 +0000)]
Upgrade minor version

26 years agoResolve merge conflicts.
pst [Thu, 27 Nov 1997 19:52:45 +0000 (19:52 +0000)]
Resolve merge conflicts.

26 years agoThis commit was generated by cvs2svn to compensate for changes in r31434,
pst [Thu, 27 Nov 1997 19:49:05 +0000 (19:49 +0000)]
This commit was generated by cvs2svn to compensate for changes in r31434,
which included commits to RCS files with non-trunk default branches.

26 years agoImport TCL v8.0 PL2.
pst [Thu, 27 Nov 1997 19:49:05 +0000 (19:49 +0000)]
Import TCL v8.0 PL2.

26 years agoMake this manpage reveal its true identity. A diff of these
steve [Thu, 27 Nov 1997 18:10:42 +0000 (18:10 +0000)]
Make this manpage reveal its true identity.  A diff of these
changes has also been sent to opie-bugs@inner.net.

PR: 5129
Submitted by: jkh

26 years agodatasize-curr -> datasize-cur, typo fix.
steve [Thu, 27 Nov 1997 17:52:52 +0000 (17:52 +0000)]
datasize-curr -> datasize-cur, typo fix.

PR: 5152
Submitted by: owaki@st.rim.or.jp

26 years agoRecognize 3C569 network interface when `PC98' is defined.
kato [Thu, 27 Nov 1997 12:59:46 +0000 (12:59 +0000)]
Recognize 3C569 network interface when `PC98' is defined.

Submitted by: HAMADA Naoki <hamada@astec.co.jp>

26 years agoSync with sys/i386/i386/userconfig.c revision 1.97.
kato [Thu, 27 Nov 1997 10:15:48 +0000 (10:15 +0000)]
Sync with sys/i386/i386/userconfig.c revision 1.97.

26 years agoSync with sys/i386/conf/files.i386 revision 1.179.
kato [Thu, 27 Nov 1997 10:14:54 +0000 (10:14 +0000)]
Sync with sys/i386/conf/files.i386 revision 1.179.

26 years agoSync with sys/i386/isa/syscons.c revision 1.240.
kato [Thu, 27 Nov 1997 10:14:11 +0000 (10:14 +0000)]
Sync with sys/i386/isa/syscons.c revision 1.240.

26 years agoRemove an extraneous ".kbd" that crept in for the Swiss German
joerg [Wed, 26 Nov 1997 22:48:24 +0000 (22:48 +0000)]
Remove an extraneous ".kbd" that crept in for the Swiss German
keyboard map.

Submitted by: georg.wagner@ubs.com

26 years agoRemoved an unused line of code, that caused an ``maybe used uninitialized''
joerg [Wed, 26 Nov 1997 22:45:47 +0000 (22:45 +0000)]
Removed an unused line of code, that caused an ``maybe used uninitialized''
warning.

Found by: Simon Shapiro

26 years agoUse err(3). -Wall cleaning. cosmetics in man page.
charnier [Wed, 26 Nov 1997 07:36:51 +0000 (07:36 +0000)]
Use err(3). -Wall cleaning. cosmetics in man page.

26 years agoRemove \n at end of syslog string. -Wall cleaning. Cosmetics in man page.
charnier [Wed, 26 Nov 1997 07:34:56 +0000 (07:34 +0000)]
Remove \n at end of syslog string. -Wall cleaning. Cosmetics in man page.

26 years agoCosmetics in man page. Add rcsid and sort #includes.
charnier [Wed, 26 Nov 1997 07:31:14 +0000 (07:31 +0000)]
Cosmetics in man page. Add rcsid and sort #includes.

26 years agoUse err(3). -Wall cleaning. Use Pa for file names and add section in Xrefs.
charnier [Wed, 26 Nov 1997 07:29:04 +0000 (07:29 +0000)]
Use err(3). -Wall cleaning. Use Pa for file names and add section in Xrefs.

26 years agoDisplay connect time as 0 if we've never been
brian [Wed, 26 Nov 1997 02:16:10 +0000 (02:16 +0000)]
Display connect time as 0 if we've never been
connected.
Pointed out by: Jeff Kletsky <Jeff@wagsky.com>

26 years agokinit(1) and its man page do not agre on what is reported with -v. Fix this.
markm [Tue, 25 Nov 1997 21:12:37 +0000 (21:12 +0000)]
kinit(1) and its man page do not agre on what is reported with -v. Fix this.
Submitted by: Sheldon Hearn.

26 years agoFrom the author:
markm [Tue, 25 Nov 1997 19:30:38 +0000 (19:30 +0000)]
From the author:

Here are the remanding changes required to support the Ensoniq
Soundscape using FreeBSD 3.0-current.

Notes:

  1) ad1848_init already has code to detect if DMA_DUPLEX should
     be set so it is not necessary (and is in fact a mistake) to
     hard code setting it.  Not all soundcards (i.e. the current
     sscape driver) are capable of using DMA_DUPLEX.

  2) The other changes are hopefully self explanatory.  Feel free
     to let me know if you need additional information.

Submitted by: john@feith.com (John Wehle)

26 years ago- Don't worry about the previous state of the card (insert/removed), and
nate [Tue, 25 Nov 1997 19:15:59 +0000 (19:15 +0000)]
- Don't worry about the previous state of the card (insert/removed), and
  instead Do The Right Thing when the kernel states a card was
  inserted/removed.  This isn't a complete fix, but better than nothing.

Reviewed by: Guido van Rooij <guido@gvr.org>

26 years agoMove the hardware currsor off screen when getting back from an
sos [Tue, 25 Nov 1997 12:44:44 +0000 (12:44 +0000)]
Move the hardware currsor off screen when getting back from an
unknown mode.
Fix warning on uninitialized var.

26 years agoSync with sys/i386/i386/machdep.c revision up to 1.274.
kato [Tue, 25 Nov 1997 09:54:36 +0000 (09:54 +0000)]
Sync with sys/i386/i386/machdep.c revision up to 1.274.

26 years agoSync with sys/i386/i386/trap.c revision 1.115.
kato [Tue, 25 Nov 1997 09:53:44 +0000 (09:53 +0000)]
Sync with sys/i386/i386/trap.c revision 1.115.

26 years agoSync with sys/i386/isa/isa.c revision 1.107.
kato [Tue, 25 Nov 1997 09:52:29 +0000 (09:52 +0000)]
Sync with sys/i386/isa/isa.c revision 1.107.

26 years agoSync with sys/i386/isa/syscons.c and syscons.h revisions 1.229 and
kato [Tue, 25 Nov 1997 09:51:46 +0000 (09:51 +0000)]
Sync with sys/i386/isa/syscons.c and syscons.h revisions 1.229 and
1.36, respectively.

26 years agoSync with sys/i386/isa/if_ed.c, if_fe.c and sio.c revisions 1.129,
kato [Tue, 25 Nov 1997 09:42:26 +0000 (09:42 +0000)]
Sync with sys/i386/isa/if_ed.c, if_fe.c and sio.c revisions 1.129,
1.35 and 1.189, respectively.

26 years agoSort #includes. Add rcsid. Add man page section in .Xrefs.
charnier [Tue, 25 Nov 1997 07:17:15 +0000 (07:17 +0000)]
Sort #includes. Add rcsid. Add man page section in .Xrefs.

26 years agoUse warn(3). Hardcode progname instead of using argv[0]. Use Pa for file
charnier [Tue, 25 Nov 1997 07:14:34 +0000 (07:14 +0000)]
Use warn(3). Hardcode progname instead of using argv[0]. Use Pa for file
name.

26 years agoShift a few SYSINT() calls around.
julian [Tue, 25 Nov 1997 07:07:48 +0000 (07:07 +0000)]
Shift a few SYSINT() calls around.
this results in a few functions becoming static, and
the SYSINITs being close to the code they are related to.
setting up the dump device is with dumpsys() and
kicking off the scheduler is with the scheduler.
Mounting root is with the code that does it.

Reviewed by: phk

26 years agoModify the return values to comply with POSIX. Previously these
alex [Tue, 25 Nov 1997 01:29:16 +0000 (01:29 +0000)]
Modify the return values to comply with POSIX.  Previously these
functions would return -1 and set errno to indicate the specific error.
POSIX requires that the functions return the error code as the return
value of the function instead.

26 years agoAdded missing source file uthread_sigwait.c.
alex [Mon, 24 Nov 1997 23:04:29 +0000 (23:04 +0000)]
Added missing source file uthread_sigwait.c.

Submitted by: Daniel M. Eischen <deischen@iworks.InterWorks.org>

26 years agoFixed missing #include of "opt_quota.h".
bde [Mon, 24 Nov 1997 19:25:24 +0000 (19:25 +0000)]
Fixed missing #include of "opt_quota.h".

Sorted the functions into the same order as in ufs_vnops.c so that this
can be compared with the latter without getting 2627 lines of diffs.
Now we get only 1920 lines of diffs.