]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoConvert interrupt count from signed to unsigned and the total
n_hibma [Mon, 26 Jul 1999 09:18:26 +0000 (09:18 +0000)]
Convert interrupt count from signed to unsigned and the total
from signed long to unsigned lon long.

PR: 12808
Submitted by: Kevin Day toasty@dragondata.com
Reviewed by: bde

25 years agoPull on my asbestos undies and claim ownership of inetd to prevent further
des [Mon, 26 Jul 1999 08:43:03 +0000 (08:43 +0000)]
Pull on my asbestos undies and claim ownership of inetd to prevent further
flamage between our beloved messrs Hearn and Feldman. Further commits go
through me. I urge the contestants to direct their energies at cleaning
up main() in inetd.c, which has over time become a crawling horror.

25 years agoHere goes, the "clear up any possible confusion" commit.
green [Mon, 26 Jul 1999 07:57:35 +0000 (07:57 +0000)]
Here goes, the "clear up any possible confusion" commit.
I've taken time to write up comments for the ident code tonight,
so there should no longer be any confusion about the purpouse of
whatever is in there. Wow, me commenting code... who'd have thought
that would happen?

Reviewed by: DES

25 years agoPave the way for the fla driver.
phk [Mon, 26 Jul 1999 07:43:26 +0000 (07:43 +0000)]
Pave the way for the fla driver.

25 years agoBring two wayward memory allocation failure messages in line with
sheldonh [Mon, 26 Jul 1999 06:39:46 +0000 (06:39 +0000)]
Bring two wayward memory allocation failure messages in line with
those featured in the rest of the code.

25 years agoAdd sysctl and support code to allow directories to be VMIO'd. The default
alc [Mon, 26 Jul 1999 06:25:53 +0000 (06:25 +0000)]
Add sysctl and support code to allow directories to be VMIO'd.  The default
setting for the sysctl is OFF, which is the historical operation.

Submitted by: dillon

25 years agoremove <ctype.h> - not needed
ache [Mon, 26 Jul 1999 05:50:43 +0000 (05:50 +0000)]
remove <ctype.h> - not needed

25 years agoVarious formatting fixes on my FPE trapcode commit.
cracauer [Mon, 26 Jul 1999 05:47:31 +0000 (05:47 +0000)]
Various formatting fixes on my FPE trapcode commit.

Submitted by: BDE

25 years agoTypo: assember -> assembler
kris [Mon, 26 Jul 1999 03:47:41 +0000 (03:47 +0000)]
Typo: assember -> assembler

25 years agoAdd recognition that Microsoft NT does not exist.
grog [Mon, 26 Jul 1999 03:35:08 +0000 (03:35 +0000)]
Add recognition that Microsoft NT does not exist.

Requested-by: John Baldwin <jobaldwi@vt.edu>

25 years agounsigned char cleanup
ache [Mon, 26 Jul 1999 01:33:38 +0000 (01:33 +0000)]
unsigned char cleanup
fix wrong index from p_simp_re()

PR: 8790
Submitted by: Alexander Viro <viro@math.psu.edu> (partially)

25 years agoDon't background pccardd, it nullfies the "-z" flag. pccardd itself
obrien [Mon, 26 Jul 1999 01:08:16 +0000 (01:08 +0000)]
Don't background pccardd, it nullfies the "-z" flag.  pccardd itself
daemonizes, lets just let things progress normally.

25 years agoMore cleanups to ident_stream. Variables moved around, changed.
green [Sun, 25 Jul 1999 23:15:03 +0000 (23:15 +0000)]
More cleanups to ident_stream.  Variables moved around, changed.
Got rid of an extra variable or two, while making corrections to
problems (that would probably not be a problem anyway, and worked.)

Partially Obtained from: David Malone <dwmalone@maths.tcd.ie>

25 years agoRemove paths that are listed as compulsory (sorta). /usr/local is not
markm [Sun, 25 Jul 1999 19:33:06 +0000 (19:33 +0000)]
Remove paths that are listed as compulsory (sorta). /usr/local is not
guaranteed to be there on a new system. This fixes whereis' "Could not
stat file..." errors.

25 years agoAdd comments, change variable names to make them consistent (r -> err,
n_hibma [Sun, 25 Jul 1999 18:54:22 +0000 (18:54 +0000)]
Add comments, change variable names to make them consistent (r -> err,
timo_handle -> timeout_handle, p -> pipe, *pipe -> *rpipe, etc.)

25 years agoasprintf() does use realloc() internally, but saying so in the manpage can be
hoek [Sun, 25 Jul 1999 17:38:59 +0000 (17:38 +0000)]
asprintf() does use realloc() internally, but saying so in the manpage can be
misinterpreted to mean that the pointer passed to asprintf() must be suitable
for passing to realloc() as-is (ie. either a NULL pointer or a valid pointer).

25 years agoOn FPU exceptions, pass a useful error code (one of the FPE_...
cracauer [Sun, 25 Jul 1999 13:16:09 +0000 (13:16 +0000)]
On FPU exceptions, pass a useful error code (one of the FPE_...
macros) to the signal handler, for old-style BSD signal handlers as
the second (int) argument, for SA_SIGINFO signal handlers as
siginfo_t->si_code. This is source-compatible with Solaris, except
that we have no <siginfo.h> (which isn't even mentioned in POSIX
1003.1b).

An rather complete example program is at
  http://www3.cons.org/cracauer/freebsd-signal.c
This will be added to the regression tests in src/.

This commit also adds code to disable the (hardware) FPU from
userconfig, so that you can use a software FP emulator on a machine
that has hardware floating point. See LINT.

25 years agoAdd a sensible message if an invalid ID was entered
n_hibma [Sun, 25 Jul 1999 11:00:05 +0000 (11:00 +0000)]
Add a sensible message if an invalid ID was entered

PR: 8739
Submitted by: Love lha@stacken.kth.se
Reviewed by: Doug Rabson <doug@freebsd.org>

25 years agoMake this compile on the Alpha. I'm not 100% sure about this but I
peter [Sun, 25 Jul 1999 06:46:19 +0000 (06:46 +0000)]
Make this compile on the Alpha.  I'm not 100% sure about this but I
think it's ok.  ti_bhandle is fetched from newbus on both the Alpha
and x86, the Alpha-only ti_vhandle is gone.

25 years agoRemember to clear the IFF_RUNNING and IFF_OACTIVE flags in sf_stop() and
wpaul [Sun, 25 Jul 1999 05:16:05 +0000 (05:16 +0000)]
Remember to clear the IFF_RUNNING and IFF_OACTIVE flags in sf_stop() and
sk_stop().

25 years agoThis commit adds device driver support for Adaptec Duralink PCI fast
wpaul [Sun, 25 Jul 1999 04:32:50 +0000 (04:32 +0000)]
This commit adds device driver support for Adaptec Duralink PCI fast
ethernet controllers based on the AIC-6915 "Starfire" controller chip.
There are single port, dual port and quad port cards, plus one 100baseFX
card. All are 64-bit PCI devices, except one single port model.

The Starfire would be a very nice chip were it not for the fact that
receive buffers have to be longword aligned. This requires buffer
copying in order to achieve proper payload alignment on the alpha.
Payload alignment is enforced on both the alpha and x86 platforms.
The Starfire has several different DMA descriptor formats and transfer
mechanisms. This driver uses frame descriptors for transmission which
can address up to 14 packet fragments, and a single fragment descriptor
for receive. It also uses the producer/consumer model and completion
queues for both transmit and receive. The transmit ring has 128
descriptors and the receive ring has 256.

This driver supports both FreeBSD/i386 and FreeBSD/alpha, and uses newbus
so that it can be compiled as a loadable kernel module. Support for BPF
and hardware multicast filtering is included.

25 years agoDon't set DE_ACCESS for unsuccessful reads.
bde [Sun, 25 Jul 1999 04:01:32 +0000 (04:01 +0000)]
Don't set DE_ACCESS for unsuccessful reads.
Translated from: a similar fix in ufs_readwrite.c rev.1.61.

Don't forget to set DE_ACCESS for short reads.

Check for invalid (negative) offsets before checking for reads of
0 bytes, as in ufs, although checking for invalid offsets at all
is probably a bug.

25 years agoDon't set IN_ACCESS for requests to read 0 bytes or for unsuccessful reads.
bde [Sun, 25 Jul 1999 02:56:17 +0000 (02:56 +0000)]
Don't set IN_ACCESS for requests to read 0 bytes or for unsuccessful reads.

Translated from: similar fixes in ufs_readwrite.c rev.1.61.  Things
are simpler (but annoyingly different) here because there are no
vm optimisations.

25 years agoFixed access timestamp bugs:
bde [Sun, 25 Jul 1999 02:07:16 +0000 (02:07 +0000)]
Fixed access timestamp bugs:

Set IN_ACCESS for successful reads of 0 bytes (except for requests to
read 0 bytes).  This was broken in rev.1.42.
PR: misc/10148

Don't set IN_ACCESS for requests to read 0 bytes.

Don't set IN_ACCESS for unsuccessful reads.

25 years agoAdded CyQ've ELA-010 Ethernet Card
hosokawa [Sun, 25 Jul 1999 01:56:16 +0000 (01:56 +0000)]
Added CyQ've ELA-010 Ethernet Card

25 years ago3C574TX Fast Etherlink PC-card support.
hosokawa [Sun, 25 Jul 1999 01:21:44 +0000 (01:21 +0000)]
3C574TX Fast Etherlink PC-card support.

25 years ago3C574TX 16bit FastEtherlink PC-card support.
hosokawa [Sun, 25 Jul 1999 01:20:37 +0000 (01:20 +0000)]
3C574TX 16bit FastEtherlink PC-card support.
Reviewed by: HAMADA Naoki <nao@tom-yam.or.jp>
Submitted by: Osamu MIHARA <mihara@prd.fc.nec.co.jp>

25 years agoIncreased max kmem to 200MB. This should fix some out-of-kmem panics on
dg [Sat, 24 Jul 1999 22:26:42 +0000 (22:26 +0000)]
Increased max kmem to 200MB. This should fix some out-of-kmem panics on
large systems.

25 years agoAdd mx and ax modules for the Macronix and ASIX drivers and update the
wpaul [Sat, 24 Jul 1999 20:55:05 +0000 (20:55 +0000)]
Add mx and ax modules for the Macronix and ASIX drivers and update the
modules Makefile so they get built.

25 years agoConvert the ASIX and Macronix drivers to newbus.
wpaul [Sat, 24 Jul 1999 20:52:57 +0000 (20:52 +0000)]
Convert the ASIX and Macronix drivers to newbus.

25 years agoOops, the previous commit only worked in the one case it was tested for.
bde [Sat, 24 Jul 1999 20:21:10 +0000 (20:21 +0000)]
Oops, the previous commit only worked in the one case it was tested for.

25 years agoUpdate intpm driver.
nsouch [Sat, 24 Jul 1999 19:13:54 +0000 (19:13 +0000)]
Update intpm driver.

PR:             kern/12631
Submitted by:   Takanori Watanabe <takawata@shidahara1.planet.sci.kobe-u.ac.jp>

25 years agoApply suggested patch- seems reasonable.
mjacob [Sat, 24 Jul 1999 17:49:17 +0000 (17:49 +0000)]
Apply suggested patch- seems reasonable.
PR: 12020
Submitted by: Matthew D. Fuller <fullermd@futuresouth.com>

25 years agoRecognise NetBSD slices.
rnordier [Sat, 24 Jul 1999 17:40:40 +0000 (17:40 +0000)]
Recognise NetBSD slices.

Submitted by : Lars Koeller <lkoeller@cc.fh-lippe.de>
PR           : 12772

25 years agoAdd -n to the example and explanation of the internal auth service.
green [Sat, 24 Jul 1999 17:19:54 +0000 (17:19 +0000)]
Add -n to the example and explanation of the internal auth service.

25 years agoCorrect a groff error in macro usage ("foo : bar" becomes "``foo: bar''").
green [Sat, 24 Jul 1999 17:11:50 +0000 (17:11 +0000)]
Correct a groff error in macro usage ("foo : bar" becomes "``foo: bar''").
Document the auth -n flag.

25 years agoMore cleanups, asprintf() usage (proper, as opposed to using snprintf()),
green [Sat, 24 Jul 1999 17:06:05 +0000 (17:06 +0000)]
More cleanups, asprintf() usage (proper, as opposed to using snprintf()),
and addition of a -n .noident-checking flag.

25 years agoUpdate the manpage for the number of symlinks in ELOOP
n_hibma [Sat, 24 Jul 1999 16:45:57 +0000 (16:45 +0000)]
Update the manpage for the number of symlinks in ELOOP

PR: 12634 (partial)
Submitted by: Julian H. Stacey jhs@FreeBSD.ORG

25 years agoClean up to match style(9) more closely. This should fix the problem of
green [Sat, 24 Jul 1999 16:24:03 +0000 (16:24 +0000)]
Clean up to match style(9) more closely. This should fix the problem of
people having ants in their pants ;)

25 years agoCorrect a typo (22th -> 22nd)
n_hibma [Sat, 24 Jul 1999 16:23:52 +0000 (16:23 +0000)]
Correct a typo (22th -> 22nd)
Remove some whitespace
Fix a problem where any event on the Last whatever of the month
was duplicated after the last day of the month (e.g. 32oct.)

PR: 4907
Submitted by: Mikhail Teterin mi@aldan.algebra.com

25 years agoUse comments to group functions by service more clearly. I've used the
sheldonh [Sat, 24 Jul 1999 13:02:09 +0000 (13:02 +0000)]
Use comments to group functions by service more clearly. I've used the
excuse of providing the RFC numbers for the associated services.

25 years agoStyle nits:
sheldonh [Sat, 24 Jul 1999 12:35:50 +0000 (12:35 +0000)]
Style nits:
      * Bring memory allocation failure handling in line with that of
the rest of the code.
      * Nestle block curlies between case statements correctly.

I've left the in-block declarations alone, since style(9) says we should
conform to the existing style within the code, and inetd already does
this. I've left the asprintf()'s in there because that's how Brian wants
it.

25 years agoFix typo.
n_hibma [Sat, 24 Jul 1999 10:44:50 +0000 (10:44 +0000)]
Fix typo.

PR: 12784
Submitted by: Doug <Doug@gorean.org>

25 years ago- Correctly initialize cn_dev_t and cn_udev_t.
yokota [Sat, 24 Jul 1999 09:41:06 +0000 (09:41 +0000)]
- Correctly initialize cn_dev_t and cn_udev_t.
- Add D_TTY for alpha.

Reviewed by: bde, dfr

25 years agoDo not print resource values which are not set.
yokota [Sat, 24 Jul 1999 09:35:21 +0000 (09:35 +0000)]
Do not print resource values which are not set.

Reviewed by: dfr

25 years agoThis makes the in kernel printf routines conform to the documented
dfr [Sat, 24 Jul 1999 09:34:12 +0000 (09:34 +0000)]
This makes the in kernel printf routines conform to the documented
behavior of their userland counterparts with respect to return values.

Submitted by: Matthew N. Dodd <winter@jurai.net>

25 years agoWhen we fetch previously retrieved IP fragments from the alias
brian [Sat, 24 Jul 1999 02:53:39 +0000 (02:53 +0000)]
When we fetch previously retrieved IP fragments from the alias
tables, copy them correctly back into our mbuf rather giving a
bzero'd count to memcpy() and ending up with a 0 byte fragment.

The old code resulted in a 0 byte write to the tun device which
tickled a bug that resulted in a panic :-(

25 years agoBacked out previous commit. devname.3 and devname.c were broken in Lite1
bde [Sat, 24 Jul 1999 00:33:27 +0000 (00:33 +0000)]
Backed out previous commit.  devname.3 and devname.c were broken in Lite1
(devname() returned "??" when the lookup failed, but callers expected it
to return NULL).  This was fixed in Lite2, but until recently the changes
were only merged into devname.3.  A day or two after devname.c was fixed,
devname.3 was made inconsistent again by backing out most of the Lite2
changes.

25 years agoatomic.h:
alc [Fri, 23 Jul 1999 23:45:50 +0000 (23:45 +0000)]
atomic.h:
Change "void *" to "volatile TYPE *", improving type safety
and eliminating some warnings (e.g., mp_machdep.c rev 1.106).

cpufunc.h:
Eliminate setbits.  As defined, it's not precisely correct;
and it's redundant.  (Use atomic_set_int instead.)

ipl_funcs.c:
Use atomic_set_int instead of setbits.

systm.h:
Include atomic.h.

Reviewed by: bde

25 years agoRegenerate to match changes made in the embedded yacc parser.
hoek [Fri, 23 Jul 1999 23:13:44 +0000 (23:13 +0000)]
Regenerate to match changes made in the embedded yacc parser.

25 years agoRegenerate this file. This fixes a y2k bogon. As an unintentional side-effect,
hoek [Fri, 23 Jul 1999 23:10:34 +0000 (23:10 +0000)]
Regenerate this file.  This fixes a y2k bogon.  As an unintentional side-effect,
it also fixes that fact that this file badly needed to be regenerated due
to changes in yacc.

Not done by: pst (in misc/1380)
Almost done by: danny (in ftp.y)

25 years ago(1) Avoid ambigous if/else
billf [Fri, 23 Jul 1999 21:31:08 +0000 (21:31 +0000)]
(1) Avoid ambigous if/else

PR: bin/12768
Submitted By: Stephen J. Roznowski <sjr@home.net>

(2) Follow KNF

25 years agofix a problem w/ zero byte writes to the tunnel device. It would bypass
jmg [Fri, 23 Jul 1999 20:08:27 +0000 (20:08 +0000)]
fix a problem w/ zero byte writes to the tunnel device.  It would bypass
the loop and not set an error, so we would then try to access an invalid
mbuf...

PR: 12780
Submitted by: bright@rush.net aka zb^3

a new record in length a pr was open... only about a half hour...

25 years agoClean up the buffer allocation code a bit. Make sure to initialize certain
wpaul [Fri, 23 Jul 1999 18:46:24 +0000 (18:46 +0000)]
Clean up the buffer allocation code a bit. Make sure to initialize certain
critical mbuf fields to sane values. Simplify the use of ETHER_ALIGN to
enforce payload alignment, and turn it on on the x86 as well as alpha
since it helps with NFS which wants the payload to be longword aligned
even though the hardware doesn't require it.

This fixes a problem with the ti driver causing an unaligned access trap
on the Alpha due to m_adj() sometimes not setting the alignment correctly
because of incomplete mbuf initialization.

25 years agoGrrr. Return the rman_get_bustag()/rman_get_bushandle() lines to their
wpaul [Fri, 23 Jul 1999 16:21:43 +0000 (16:21 +0000)]
Grrr. Return the rman_get_bustag()/rman_get_bushandle() lines to their
proper place in ti_attach(). I'm positive I typed them in there, but
they must have fallen victim to a drive-by cut & pasting.

25 years agoDocument the -o and -t options to the internal auth service and give an
sheldonh [Fri, 23 Jul 1999 15:49:34 +0000 (15:49 +0000)]
Document the -o and -t options to the internal auth service and give an
example of their usage in the sample config. Merge the two examples
for the green internal auth service.

This commit failed the first time around because Brian beat me to the
punch on inetd.8 . I like my descriptions better and I'm pretty sure
Brian won't mind.

25 years agoAhem. Put things back a bit. I declare variables in the scope they're
green [Fri, 23 Jul 1999 15:49:14 +0000 (15:49 +0000)]
Ahem. Put things back a bit. I declare variables in the scope they're
used! I don't declare every variable at the top of a function because
that wastes stack space. I've clarified the error a bit (for if asprintf()
filas.)

25 years agoDisable more PC98isms on the Alpha
jkh [Fri, 23 Jul 1999 15:39:51 +0000 (15:39 +0000)]
Disable more PC98isms on the Alpha

25 years agoAs per DES's prodding, document _all_ the arguments to inetd's auth
green [Fri, 23 Jul 1999 15:37:39 +0000 (15:37 +0000)]
As per DES's prodding, document _all_ the arguments to inetd's auth
service. This includes the -o "operating system" argument and the -t
"timeout" argument.

25 years agoStyle cleanups for iderror() and ident_stream(). Looks like c++ hang-over.
sheldonh [Fri, 23 Jul 1999 15:26:42 +0000 (15:26 +0000)]
Style cleanups for iderror() and ident_stream(). Looks like c++ hang-over.
;-)

25 years agoFix auth -t argument handling. It was broken for the "sec.usec" case.
sheldonh [Fri, 23 Jul 1999 15:00:07 +0000 (15:00 +0000)]
Fix auth -t argument handling. It was broken for the "sec.usec" case.
Add a warning for bogus -t arguments for the (debug) case.

25 years agoSet correct "flags" value when pccardd allocates/initializes driver.
hosokawa [Fri, 23 Jul 1999 14:58:33 +0000 (14:58 +0000)]
Set correct "flags" value when pccardd allocates/initializes driver.

25 years agoDon't match up TCP services with UDP sockets.
des [Fri, 23 Jul 1999 14:45:21 +0000 (14:45 +0000)]
Don't match up TCP services with UDP sockets.

25 years agoRemoved a very very old hack (I disabled it in PAO long time ago)
hosokawa [Fri, 23 Jul 1999 08:53:20 +0000 (08:53 +0000)]
Removed a very very old hack (I disabled it in PAO long time ago)
that existed in original PC-card driver for FreeBSD 2.0 (maybe).
This prevents from utilizing flags for drivers.

25 years agoFix a non-critical memory leak.
ru [Fri, 23 Jul 1999 08:28:46 +0000 (08:28 +0000)]
Fix a non-critical memory leak.

PR: 12769
Submitted by: Peter Jeremy <peter.jeremy@alcatel.com.au>

25 years agoOne last tweak before I turn in for the evening: the driver name in
wpaul [Fri, 23 Jul 1999 05:50:35 +0000 (05:50 +0000)]
One last tweak before I turn in for the evening: the driver name in
the driver_t declaration should be "skc" not "sk". Technically, "skc"
is the parent PCI device (the SysKonnect GEnesis controller) and "sk0"
and "sk1" are the network interfaces that get attached to it.

25 years agoCreate module directories for the xl, ti, tl and sk drivers and add
wpaul [Fri, 23 Jul 1999 05:48:01 +0000 (05:48 +0000)]
Create module directories for the xl, ti, tl and sk drivers and add
them to the Makefile so that modules will be generated for /modules.

25 years agoFixed a braino: lack of spaces in sscanf caused ident parsing to fail.
green [Fri, 23 Jul 1999 03:51:52 +0000 (03:51 +0000)]
Fixed a braino: lack of spaces in sscanf caused ident parsing to fail.
Sorry, guys.

25 years agoMore changes for the Alpha X support - don't list PC98 servers since they make
jkh [Fri, 23 Jul 1999 03:42:23 +0000 (03:42 +0000)]
More changes for the Alpha X support - don't list PC98 servers since they make
no sense here but list the TGA server, since it does.

25 years agoDangit. Somehow the pmap_kextract hack for alpha snuck back into these
wpaul [Fri, 23 Jul 1999 02:18:01 +0000 (02:18 +0000)]
Dangit. Somehow the pmap_kextract hack for alpha snuck back into these
files. Change them back to alpha_XXX_dmamap().

Pointed out by: Andrew Gallatin

25 years agoConvert the Alteon Tigon gigabit ethernet driver to newbus. Also upgrade
wpaul [Fri, 23 Jul 1999 02:10:11 +0000 (02:10 +0000)]
Convert the Alteon Tigon gigabit ethernet driver to newbus. Also upgrade
to the latest firmware release from Alteon (12.3.12).

25 years agoSome more small newbus cleanups. Remember to free all resources in case
wpaul [Fri, 23 Jul 1999 02:06:57 +0000 (02:06 +0000)]
Some more small newbus cleanups. Remember to free all resources in case
of failures in foo_attach(), simplify iospace/memspace things a little.

25 years agoSorry, I forgot to test "iosize auto" before merging "iosize" patch.
hosokawa [Fri, 23 Jul 1999 02:01:19 +0000 (02:01 +0000)]
Sorry, I forgot to test "iosize auto" before merging "iosize" patch.

25 years agoInappropriate commment for -i option.
hosokawa [Fri, 23 Jul 1999 01:46:47 +0000 (01:46 +0000)]
Inappropriate commment for -i option.

25 years agoAdded "iosize" directive in /etc/pccard.conf
hosokawa [Fri, 23 Jul 1999 01:33:34 +0000 (01:33 +0000)]
Added "iosize" directive in /etc/pccard.conf
Allocate free I/O window with given size to card.

(example)
# IBM PCMCIA Ethernet I/II
card "IBM Corp." "Ethernet"
config 0x1 "ed0" ?
iosize 32
ether 0xff0

(it's currently only useful for externalizing hacks for broken CIS cards,
 but it will play an important role with "function" directive I'm planninng
 to merge)

Reviewed by: freebsd-mobile list
Obtained from: PAO3's "cardio" directive

25 years agoRelease IRQ allocated dynamically.
hosokawa [Fri, 23 Jul 1999 00:12:22 +0000 (00:12 +0000)]
Release IRQ allocated dynamically.

25 years ago"knobs are cheap". Here's a -t timeout option for the internal ident
green [Thu, 22 Jul 1999 21:42:49 +0000 (21:42 +0000)]
"knobs are cheap". Here's a -t timeout option for the internal ident
service. It takes a number (w/ or w/out .usec) as an argument.

25 years agoThis commit encompasses the following changes to inetd:
green [Thu, 22 Jul 1999 21:11:40 +0000 (21:11 +0000)]
This commit encompasses the following changes to inetd:
1. Cleanups of ident_stream. "Evil" stdio is less used.
2. The BSD Copyright was added to the top of builtins.c.
3. As suggested, a timeout is now implemented in the ident
   service. It defaults to 10 seconds. If enough people want
   it, I'll make it configurable.

Suggested by: msmith

25 years agoRevert out part of the last patch that can be done better elsewhere.
julian [Thu, 22 Jul 1999 21:08:02 +0000 (21:08 +0000)]
Revert out part of the last patch that can be done better elsewhere.
The complexity added isn't worth it.

25 years agoRemove the definitions for the SiS 900 chip. This is not a RealTek
wpaul [Thu, 22 Jul 1999 20:56:49 +0000 (20:56 +0000)]
Remove the definitions for the SiS 900 chip. This is not a RealTek
clone after all. I have the datasheets for this part; hopefully I can
write a proper driver soon.

25 years agofix braino..
julian [Thu, 22 Jul 1999 19:45:33 +0000 (19:45 +0000)]
fix braino..
accidentally replaced PAGE_MASK with PAGE_SIZE.
(PAGE_MASK is PAGE_SIZE - 1)
bug does not manifest itself on our hardware.....

25 years agoWell, it seems that loading a PCI driver module after the system has
wpaul [Thu, 22 Jul 1999 18:10:20 +0000 (18:10 +0000)]
Well, it seems that loading a PCI driver module after the system has
been booted works too -- very neat. However I don't want the system to
stop for 5 seconds when the MII autoprobe is triggered in the xl and
tl drivers since that's lame. Instead, only use the hard delay when
we've been cold booted. If not, use the timeout mechanism instead.
(The SysKonnect driver doesn't use the same autonegotiation scheme, so
no change is required there.)

25 years agoDon't dump core for a known, documented bug.
sheldonh [Thu, 22 Jul 1999 17:33:11 +0000 (17:33 +0000)]
Don't dump core for a known, documented bug.

PR: 12611
Reviewed by: markm

25 years agoImprove printf(3) conversion specifier parsing so that silly formats
sheldonh [Thu, 22 Jul 1999 17:11:59 +0000 (17:11 +0000)]
Improve printf(3) conversion specifier parsing so that silly formats
aren't allowed and the right casts can be used for printf() statements.

Document the conversion specifier limitations and the fact that
arithmetic overflow causes a fatal error.

PR: 12611
Reported by: Frode Vatvedt Fjeld <frodef@acm.org>
Reviewed by: bde

25 years agoConvert the ThunderLAN driver to newbus. Also add splimp() protection to
wpaul [Thu, 22 Jul 1999 17:00:38 +0000 (17:00 +0000)]
Convert the ThunderLAN driver to newbus. Also add splimp() protection to
tl_stats_update().

25 years agoRelegate the diagnostic descriptor counter to the -DSANITY_CHECK case.
sheldonh [Thu, 22 Jul 1999 16:29:48 +0000 (16:29 +0000)]
Relegate the diagnostic descriptor counter to the -DSANITY_CHECK case.

25 years agoRemove unnecessary macro introduced in previous commit.
sheldonh [Thu, 22 Jul 1999 16:10:40 +0000 (16:10 +0000)]
Remove unnecessary macro introduced in previous commit.

Also, the previous commit failed to reference:

PR: 12731
Submitted by: dwmalone@maths.tcd.ie (David "Inetd" Malone)

25 years agoDon't leak pipe descriptor to daemons on execv().
sheldonh [Thu, 22 Jul 1999 15:57:37 +0000 (15:57 +0000)]
Don't leak pipe descriptor to daemons on execv().

25 years agoFix a small mind-o: one instance of SYS_RES_IOPORT should have been
wpaul [Thu, 22 Jul 1999 14:58:54 +0000 (14:58 +0000)]
Fix a small mind-o: one instance of SYS_RES_IOPORT should have been
SYS_RES_MEMOTY in sk_detatch().

25 years agoSignal handlers should use _exit(2) and not exit(3).
sheldonh [Thu, 22 Jul 1999 14:47:29 +0000 (14:47 +0000)]
Signal handlers should use _exit(2) and not exit(3).

25 years agoAdd braces to make if-else statement clearer.
iwasaki [Thu, 22 Jul 1999 14:45:22 +0000 (14:45 +0000)]
Add braces to make if-else statement clearer.

PR: 12663
Submitted by: Adam Wight <adamw@holonet.net>

25 years agoCorrect some style issues in my previous commit.
kris [Thu, 22 Jul 1999 14:26:21 +0000 (14:26 +0000)]
Correct some style issues in my previous commit.

Submitted by: bde

25 years ago- Add new argument `off' to the `-b' option. This will turn off
yokota [Thu, 22 Jul 1999 14:19:14 +0000 (14:19 +0000)]
- Add new argument `off' to the `-b' option.  This will turn off
  the bell.
- Document it in the man page.
- Fix a couple of typo in the man page.

Submitted by: cpiazza

25 years agoMove code for all builtin services from inetd.c to builtins.c, including
sheldonh [Thu, 22 Jul 1999 14:11:26 +0000 (14:11 +0000)]
Move code for all builtin services from inetd.c to builtins.c, including
the Green Piece. :-)

In future, new builtin services are less likely to need to touch the
already tangled inetd.c .

25 years agoDeal with new linux compat package naming.
jkh [Thu, 22 Jul 1999 09:18:12 +0000 (09:18 +0000)]
Deal with new linux compat package naming.

Submitted by: Marcel Moolenaar <marcel@scc.nl>

25 years agoIntentionally do the wrong thing in using the initial DHCP values for
jkh [Thu, 22 Jul 1999 08:51:42 +0000 (08:51 +0000)]
Intentionally do the wrong thing in using the initial DHCP values for
ifconfig, essentially stealing the lease until the user goes and changes
it.  The alternative, sadly, is total dysfunction since bpf isn't in
GENERIC and network connectivity would otherwise fail completely on first
bootup when DHCP configuration was attempted again.

The ultimate answer here is to make either bpf a loadable kernel module
(which security conscious admins will be able to simply remove from /modules)
or come up with a lighter weight mechanism just for dhcp and other apps that
need to see broadcast packets but not otherwise sniff the wire in full
bpf glory.

25 years agoModify device numbering method to work with latest -CURRENT. Briefly,
grog [Thu, 22 Jul 1999 08:40:37 +0000 (08:40 +0000)]
Modify device numbering method to work with latest -CURRENT.  Briefly,
the device numbers are now minor number only, so that we can still
compare them after dev_t has turned into a blob.

Broken-by: dev_t changes
Reported-by: Vallo Kallaste <vallo@matti.ee>
"Niels Chr. Bank-Pedersen" <ncbp@bank-pedersen.dk>

25 years agoReduce the number of "magic constants" used for page coloring
alc [Thu, 22 Jul 1999 06:04:17 +0000 (06:04 +0000)]
Reduce the number of "magic constants" used for page coloring
by one: PQ_PRIME2 and PQ_PRIME3 are used to accomplish the same
thing at different places in the kernel.  Drop PQ_PRIME3.

25 years agoConvert the SysKonnect gigabit ethernet driver to newbus.
wpaul [Thu, 22 Jul 1999 04:04:12 +0000 (04:04 +0000)]
Convert the SysKonnect gigabit ethernet driver to newbus.

25 years agoSmall tweak to newbus changes: return error status on failure correctly
wpaul [Thu, 22 Jul 1999 03:59:22 +0000 (03:59 +0000)]
Small tweak to newbus changes: return error status on failure correctly
in xl_attach() (not a problem if the attach never fails, but if it does
the function would still return 0, which is wrong).