]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
24 years agoMerged NetBSD version, as they have done improvements:
Semen Ustimenko [Fri, 3 Dec 1999 20:37:40 +0000 (20:37 +0000)]
Merged NetBSD version, as they have done improvements:
1. ntfs_read*attr*() functions now accept
uio structure to eliminate one data copying.
2. found and removed deadlock caused
by 6 concurent ls -lR.
3. started implementation of nromal
Unicode<->unix recodeing.

Obtained from: NetBSD

24 years agoFix bug parsing 32 bit integers on machines where sizeof(long) == 4.
Archie Cobbs [Fri, 3 Dec 1999 20:27:33 +0000 (20:27 +0000)]
Fix bug parsing 32 bit integers on machines where sizeof(long) == 4.

24 years agoAdded ntfs filesystem to be exported.
Semen Ustimenko [Fri, 3 Dec 1999 20:23:53 +0000 (20:23 +0000)]
Added ntfs filesystem to be exported.

24 years agoAdded ntfs subdir to be filled.
Semen Ustimenko [Fri, 3 Dec 1999 20:22:45 +0000 (20:22 +0000)]
Added ntfs subdir to be filled.

24 years agoAdded ntfs subdir to allow mountd.c include <ntfs/ntfsmount.h>
Semen Ustimenko [Fri, 3 Dec 1999 20:17:16 +0000 (20:17 +0000)]
Added ntfs subdir to allow mountd.c include <ntfs/ntfsmount.h>
and others.

24 years agoDon't use %E* in old fields for compatibility reasons
Andrey A. Chernov [Fri, 3 Dec 1999 19:52:06 +0000 (19:52 +0000)]
Don't use %E* in old fields for compatibility reasons

24 years agoFix for the new usage of mkioctls
Marcel Moolenaar [Fri, 3 Dec 1999 17:35:34 +0000 (17:35 +0000)]
Fix for the new usage of mkioctls

24 years agoChange -i host to whois.networksolutions.com
Andrey A. Chernov [Fri, 3 Dec 1999 16:56:57 +0000 (16:56 +0000)]
Change -i host to whois.networksolutions.com

24 years agoSupported to sound beep.
Yoshihiro Takahashi [Fri, 3 Dec 1999 13:20:55 +0000 (13:20 +0000)]
Supported to sound beep.

Submitted by: chi@bd.mbn.or.jp (Chiharu Shibata)

24 years agopc98/pc98/atapi.c
Yoshihiro Takahashi [Fri, 3 Dec 1999 12:56:21 +0000 (12:56 +0000)]
pc98/pc98/atapi.c
Copied from i386/isa/atapi.c.
Fixed to support slave devices.
Ignore the device that has strange model strings.

i386/isa/atapi.c
Removed pc98 codes.

Submitted by: chi@bd.mbn.or.jp (Chiharu Shibata)

24 years agoAvoid hardcoding any paths and variables. The include directory must
Marcel Moolenaar [Fri, 3 Dec 1999 12:50:02 +0000 (12:50 +0000)]
Avoid hardcoding any paths and variables. The include directory must
now be specified on the command line. Accept a '-s' option which
controls whether a switch-statement is to be used instead of a series
of if-statements.

Replace cpp with gcc -E.

Discussed with: bde

24 years agoChange splbio() to splcam(). bs is cam driver.
Yoshihiro Takahashi [Fri, 3 Dec 1999 11:58:12 +0000 (11:58 +0000)]
Change splbio() to splcam(). bs is cam driver.

24 years agoSync with sys/i386/conf/GENERIC revision 1.208.
Yoshihiro Takahashi [Fri, 3 Dec 1999 11:40:52 +0000 (11:40 +0000)]
Sync with sys/i386/conf/GENERIC revision 1.208.

24 years agoRemove mention of 'ivars'.
Matthew N. Dodd [Fri, 3 Dec 1999 11:22:28 +0000 (11:22 +0000)]
Remove mention of 'ivars'.

24 years agoMake the stub routines have the same prototypes as the real bpf
Julian Elischer [Fri, 3 Dec 1999 09:59:02 +0000 (09:59 +0000)]
Make the stub routines have the same prototypes as the real bpf
routines.

24 years agocorrect incomplete last change
Matt Jacob [Fri, 3 Dec 1999 09:10:04 +0000 (09:10 +0000)]
correct incomplete last change

24 years agoRemove the 'ivars' arguement to device_add_child() and
Matthew N. Dodd [Fri, 3 Dec 1999 08:41:24 +0000 (08:41 +0000)]
Remove the 'ivars' arguement to device_add_child() and
device_add_child_ordered().  'ivars' may now be set using the
device_set_ivars() function.

This makes it easier for us to change how arbitrary data structures are
associated with a device_t.  Eventually we won't be modifying device_t
to add additional pointers for ivars, softc data etc.

Despite my best efforts I've probably forgotten something so let me know
if this breaks anything.  I've been running with this change for months
and its been quite involved actually isolating all the changes from
the rest of the local changes in my tree.

Reviewed by: peter, dfr

24 years agoSet RB_VERBOSE into boothowto (provide proper layering from boot loader
Matt Jacob [Fri, 3 Dec 1999 07:20:22 +0000 (07:20 +0000)]
Set RB_VERBOSE into boothowto (provide proper layering from boot loader
(eventually) settings to kernel settings).

Prototype read_random to quiesce a warning.

24 years agoroll platform minor
Matt Jacob [Fri, 3 Dec 1999 06:56:43 +0000 (06:56 +0000)]
roll platform minor

24 years agoclean up sprintf and have buffer that won't overflow
Matt Jacob [Fri, 3 Dec 1999 06:56:23 +0000 (06:56 +0000)]
clean up sprintf and have buffer that won't overflow

24 years agoturn some messages into CFGPRINT messages
Matt Jacob [Fri, 3 Dec 1999 06:55:39 +0000 (06:55 +0000)]
turn some messages into CFGPRINT messages

24 years agoBe careful not to re-initialise `struct stat' while it still has
Brian Somers [Fri, 3 Dec 1999 06:33:10 +0000 (06:33 +0000)]
Be careful not to re-initialise `struct stat' while it still has
a running timer.  This fixes a problem where a dial is manually
aborted, the hangup script kicks in and the chat timer ends up
on the timer queue twice (tick tick tick tick *boom*)

24 years agoAdd '-l' which will match a certain status code.
Bill Fumerola [Fri, 3 Dec 1999 06:08:11 +0000 (06:08 +0000)]
Add '-l' which will match a certain status code.
Add the submitter as a contributor in the man page
freebsd -> FreeBSD, while I'm poking around.

PR: bin/15162
Submitted by: Dominic Mitchell <dom@palmerharvey.co.uk>

24 years agoCorrect co-MAINTAINER's e-mail address. I don't know how many
Bill Fumerola [Fri, 3 Dec 1999 03:20:59 +0000 (03:20 +0000)]
Correct co-MAINTAINER's e-mail address. I don't know how many
more ways I can screw his address up. Suggestions welcome.

24 years agoDefine name length differently for alpha in order to preserve
Jordan K. Hubbard [Fri, 3 Dec 1999 02:23:34 +0000 (02:23 +0000)]
Define name length differently for alpha in order to preserve
backwards compatibility.

Submitted by: Andrew Gallatin <gallatin@cs.duke.edu>
Reviewed by: mckusick

24 years agoFix the 'usb0: USB revision unknown, not supported' people have been seeing
Nick Hibma [Fri, 3 Dec 1999 01:34:42 +0000 (01:34 +0000)]
Fix the 'usb0: USB revision unknown, not supported' people have been seeing
by identifying the version in the PCI drivers.

The OHCI driver just presets this to 1.0 as it is not specified in the
PCI registers anywhere. This should be revisited once USB 2.0 is in
wide spread use.

24 years agoRemove the 'alpha, use at your own risk' death-statement.
Bill Fumerola [Fri, 3 Dec 1999 00:40:31 +0000 (00:40 +0000)]
Remove the 'alpha, use at your own risk' death-statement.

Reviewed by: mckusick (verbally at FreeBSDcon)

24 years agoFix typo, add $FreeBSD$
Bill Fumerola [Fri, 3 Dec 1999 00:34:26 +0000 (00:34 +0000)]
Fix typo, add $FreeBSD$

24 years agoA better version of the previous checkin. If the user specifies
Archie Cobbs [Fri, 3 Dec 1999 00:28:06 +0000 (00:28 +0000)]
A better version of the previous checkin.  If the user specifies
a custom file that could override a FreeBSD file under a different
configuration, but doesn't under this one, give a different warning.

24 years agoRemove the 'gzip' image activator. We're not using a.out anymore, so save
Mike Smith [Fri, 3 Dec 1999 00:16:20 +0000 (00:16 +0000)]
Remove the 'gzip' image activator.  We're not using a.out anymore, so save
ourselves just over 8k.

24 years agoAdd Matt Fuller's comment about long-standing x86 bug that allows you
Wes Peters [Thu, 2 Dec 1999 23:46:49 +0000 (23:46 +0000)]
Add Matt Fuller's comment about long-standing x86 bug that allows you
to install Windows.

24 years agoWhen specifying additiona user-specified kernel compilation source
Archie Cobbs [Thu, 2 Dec 1999 23:43:08 +0000 (23:43 +0000)]
When specifying additiona user-specified kernel compilation source
files in a 'files.XXX' file, config allows non-FreeBSD source files
with the same name as a FreeBSD source file to override the latter,
and in this situation it issues a warning.

However, if one of the user-specified files is actually a FreeBSD
source file (perhaps your kernel has some custom option that requires
that file), config mistakenly thinks it's a completely new file
and goes ahead and overrides all previous information for that file
(and issues the warning).

Fix this.

With help from: julian

24 years agoPrint device names for children that fail to detach.
Warner Losh [Thu, 2 Dec 1999 19:51:34 +0000 (19:51 +0000)]
Print device names for children that fail to detach.
Free child array when we're done with it.

Forgotten by: imp
Submitted by: Nick Hibma <hibma@skylink.it>

24 years agoNow that pccardc beep actually works, add knob for it in rc.conf/rc.pccard
Warner Losh [Thu, 2 Dec 1999 19:48:16 +0000 (19:48 +0000)]
Now that pccardc beep actually works, add knob for it in rc.conf/rc.pccard

Submitted by: sanpei@sanpei.org (MIHIRA-san Yoshiro)

24 years agoMake pccardc beep actually work. The kernel was doing the wrong thing
Warner Losh [Thu, 2 Dec 1999 19:46:41 +0000 (19:46 +0000)]
Make pccardc beep actually work.  The kernel was doing the wrong thing
with the beep information it had (like ignoring it).

Submitted by: sanpei@sanpei.org (MIHIRA-san Yoshiro)

Add $FreeBSD$ to pccard_beep.c while I'm here.

24 years agoReserve 138 for usio, the USB sio driver by Doug Ambrisko, Whistle.
Nick Hibma [Thu, 2 Dec 1999 19:45:19 +0000 (19:45 +0000)]
Reserve 138 for usio, the USB sio driver by Doug Ambrisko, Whistle.

Also, change mail address in all case to n_hibma@freebsd.org

24 years agoAdd 'const' to the bpf_filter() and bpf_validate() prototypes.
Archie Cobbs [Thu, 2 Dec 1999 19:36:05 +0000 (19:36 +0000)]
Add 'const' to the bpf_filter() and bpf_validate() prototypes.
Remove a stale comment from bpf_validate().

24 years agoFix typo from last commit.
Warner Losh [Thu, 2 Dec 1999 19:31:28 +0000 (19:31 +0000)]
Fix typo from last commit.

Noticed by: Maxim Sobolev <sobomax@altavista.net>

24 years agoReserve #137 for nCipher's nFast PCI hardware cryptographic
Sheldon Hearn [Thu, 2 Dec 1999 17:46:17 +0000 (17:46 +0000)]
Reserve #137 for nCipher's nFast PCI hardware cryptographic
accelerator, for which open source drivers are under development,
for John Hartley <john@ncipher.com>.

24 years agoIn one queue all the TDs (transfer descriptor, packets) for one transfer
Nick Hibma [Thu, 2 Dec 1999 16:43:18 +0000 (16:43 +0000)]
In one queue all the TDs (transfer descriptor, packets) for one transfer
are queued.  Traverse the queues vertically and then horizontally.
This means that TDs for one xfer are transmitted back to back until the
first NAK or error condition. Up to now we transmitted a TD per frame
and transmitted the next TD in the next frame.

The old approach is more fair if you have the end of the queue point at
the beginning of the control transfer queue, but also a lot more overhead
due to the fact that the QHs have to be read more often.

The new approach squirts the packets down the line as fast as possible
for one transfer and then does the next one. In the current situation,
with fairly empty USB buses, this is a more sensible approach. We might
have to revisit the scheduler later however.

It speeds up large transfers (Zip drive, Host-To-Host adapters) on UHCI
by a factor of 5 and makes it as fast as OHCI on the bus.

The next problem to solve is the question why the limit is 300kb/s and
not 1000/kb/s (kb == kilobyte).

24 years agoReplace the -q option to pwd_mkdb with a test for PW_SCAN_BIG_IDS in
Sheldon Hearn [Thu, 2 Dec 1999 16:39:15 +0000 (16:39 +0000)]
Replace the -q option to pwd_mkdb with a test for PW_SCAN_BIG_IDS in
the environment.  This allows big ID warnings to be suppressed for
vipw and chpass as well.

Since the environment variable test is only performed for callers
of pw_scan() that do not set pw_big_ids_warning, the test can still
be overriden.  Currently, chpass and pwd_mkdb are the only users
of pw_scan() and neither of them overrides the environment variable
test.

24 years agoAdd the device_get_children(9) man page.
Nick Hibma [Thu, 2 Dec 1999 16:34:09 +0000 (16:34 +0000)]
Add the device_get_children(9) man page.

24 years agoIt's device_add_child*, not bus_add_child*.
Nick Hibma [Thu, 2 Dec 1999 16:31:12 +0000 (16:31 +0000)]
It's device_add_child*, not bus_add_child*.

24 years agoRemove check for attached state.
Nick Hibma [Thu, 2 Dec 1999 16:30:21 +0000 (16:30 +0000)]
Remove check for attached state.

   sc = devclass_get_softc(devclass, unit);

doesn't return NULL during attach anymore, and produces the sc,
identical to (for devclass_get_unit(devclass, unit) != NULL that is):

   sc = device_get_softc(devclass_get_unit(devclass, unit));

Reviewed-by: dfr
24 years agoSync with sys/i386/conf/GENERIC revision up to 1.207.
Yoshihiro Takahashi [Thu, 2 Dec 1999 14:00:28 +0000 (14:00 +0000)]
Sync with sys/i386/conf/GENERIC revision up to 1.207.

24 years agoMake sure DESTDIR is set in the environment of mkioctls. This fixes
Marcel Moolenaar [Thu, 2 Dec 1999 13:47:17 +0000 (13:47 +0000)]
Make sure DESTDIR is set in the environment of mkioctls. This fixes
the breakage people have encountered at certain times (for example
when the altq_*.h files were removed).

$FreeBSD$ tag added.

24 years agomention AlphaServer 1000 and AlphaServer DS20
Andrew Gallatin [Thu, 2 Dec 1999 13:39:05 +0000 (13:39 +0000)]
mention AlphaServer 1000 and AlphaServer DS20

24 years agoRemove the "MCA not supported" from RELNOTES.TXT and add
Matthew N. Dodd [Thu, 2 Dec 1999 10:01:06 +0000 (10:01 +0000)]
Remove the "MCA not supported" from RELNOTES.TXT and add
mention of the various devices that are supported.

Add some text and entry to LINT for 'controller mca0'.

I'd like to turn this option on in GENERIC as well as it
isn't impacting and has a small footprint.

24 years agoUpdates for 4.x:
Mike Smith [Thu, 2 Dec 1999 09:16:02 +0000 (09:16 +0000)]
Updates for 4.x:

 - Add AMI and Mylex RAID controllers
 - Reflect the demise of the 'eg' and 'ft' drivers
 - Various minor cleanups
 - Add some initial Microchannel information (this could do with some
   fleshing out)

24 years agoCorrect outdated aic entries.
Jordan K. Hubbard [Thu, 2 Dec 1999 08:25:53 +0000 (08:25 +0000)]
Correct outdated aic entries.

Submitted by: Greg Lewis <glewis@trc.adelaide.edu.au>
PR: 15186

24 years ago[[ First attempt used -m /tmp/msg rather than -F /tmp/msg ]]
Warner Losh [Thu, 2 Dec 1999 05:57:50 +0000 (05:57 +0000)]
[[ First attempt used -m /tmp/msg rather than -F /tmp/msg ]]

Kernel portion of pccardc power virtual power interface (it helps me
test).  Patch was originally from mihira-san in message to
freebsd-mobile.  I reworked it a little and tested it here.  He ported
the code originally from PAO.

Submitted by: MIHIRA Sanpei Yoshiro <sanpei@sanpei.org>

24 years agoForgotten part of pccardc power command
Warner Losh [Thu, 2 Dec 1999 05:56:36 +0000 (05:56 +0000)]
Forgotten part of pccardc power command

24 years agoImplement pccardc power command. This allows one to power off
Warner Losh [Thu, 2 Dec 1999 05:52:58 +0000 (05:52 +0000)]
Implement pccardc power command.  This allows one to power off
individual slots at one's whim.  Useful for turning the slots into
card carrying cases, etc.  Patch was originally from mihira-san in
message to freebsd-mobile.  He ported the code originally from PAO.

Submitted by: MIHIRA Sanpei Yoshiro <sanpei@sanpei.org>

24 years ago/tmp/msg
Warner Losh [Thu, 2 Dec 1999 05:46:22 +0000 (05:46 +0000)]
/tmp/msg

24 years agoAdd bus suffix to mii.
Daniel C. Sobral [Thu, 2 Dec 1999 03:48:50 +0000 (03:48 +0000)]
Add bus suffix to mii.

24 years agoAdd if_ prefix to network drivers.
Daniel C. Sobral [Thu, 2 Dec 1999 03:47:46 +0000 (03:47 +0000)]
Add if_ prefix to network drivers.

24 years agoChange the delayed ack time from 200ms to 100ms.
Jonathan Lemon [Thu, 2 Dec 1999 03:25:19 +0000 (03:25 +0000)]
Change the delayed ack time from 200ms to 100ms.

This results in closer behavior to earlier versions, where the fixed
200ms timer actually resulted in a delay anywhere from 1..200ms, with
the average delay being 100ms.

Pointed out by:  dg

24 years agoRemove code to select APM version with flags to the apm0 device. This
Jonathan Lemon [Thu, 2 Dec 1999 03:13:11 +0000 (03:13 +0000)]
Remove code to select APM version with flags to the apm0 device.  This
code has been disabled for the last 4 months.

Prodded into action by:  n_hibma

24 years agoGrammar nit.
Bill Fumerola [Thu, 2 Dec 1999 03:01:54 +0000 (03:01 +0000)]
Grammar nit.

24 years agoAllow 2 more characters for Mike's long device names.
Jordan K. Hubbard [Thu, 2 Dec 1999 02:55:57 +0000 (02:55 +0000)]
Allow 2 more characters for Mike's long device names.

24 years agoAdd the AMI MegaRAID and Mylex DAC960 drivers. Installation to arrays
Mike Smith [Thu, 2 Dec 1999 02:53:08 +0000 (02:53 +0000)]
Add the AMI MegaRAID and Mylex DAC960 drivers.  Installation to arrays
on these controllers is now no different to the process for any other
supported disk controller.

24 years agoWe need COMPAT3X for both i386 and alpha, not just i386.
Jordan K. Hubbard [Thu, 2 Dec 1999 01:54:15 +0000 (01:54 +0000)]
We need COMPAT3X for both i386 and alpha, not just i386.

24 years agoAdd the hostname command for dhcp.
Jordan K. Hubbard [Thu, 2 Dec 1999 01:52:45 +0000 (01:52 +0000)]
Add the hostname command for dhcp.

24 years agoUse m_dup() instead of m_copypacket() to duplicate the mbuf chain.
Archie Cobbs [Wed, 1 Dec 1999 23:11:58 +0000 (23:11 +0000)]
Use m_dup() instead of m_copypacket() to duplicate the mbuf chain.
This fixes a bug where if the original packet was modified downstream
of the tee node, then the copy could get modified too.

24 years agoEliminate compiler warning.
Archie Cobbs [Wed, 1 Dec 1999 22:56:51 +0000 (22:56 +0000)]
Eliminate compiler warning.

24 years agoThe functions m_copym() and m_copypacket() return read-only copies,
Archie Cobbs [Wed, 1 Dec 1999 22:31:32 +0000 (22:31 +0000)]
The functions m_copym() and m_copypacket() return read-only copies,
because in the case of mbuf clusters they only increment the reference
count rather than actually copying the data.

Add comments to this effect, and add a new routine called m_dup() that
returns a real, writable copy of an mbuf chain.

This is preliminary work required for implementing 'ipfw tee'.

Reviewed by: julian

24 years agoAdd more comments describing how to use parse types and how they work.
Archie Cobbs [Wed, 1 Dec 1999 19:41:15 +0000 (19:41 +0000)]
Add more comments describing how to use parse types and how they work.

24 years agoShow how to supply a struct ng_cmdlist for (de)asciification
Archie Cobbs [Wed, 1 Dec 1999 19:40:37 +0000 (19:40 +0000)]
Show how to supply a struct ng_cmdlist for (de)asciification
of control messages.

Suggested by: julian

24 years agoPreferentially allocate the first indirect block in the same
Kirk McKusick [Wed, 1 Dec 1999 19:33:12 +0000 (19:33 +0000)]
Preferentially allocate the first indirect block in the same
cylinder group as the inode. This makes a 15% difference in
read speed for files in the 96K to 500K size range.

24 years agoAdd fairings. Do not depend on user actually supplying the arguments
Daniel C. Sobral [Wed, 1 Dec 1999 18:30:26 +0000 (18:30 +0000)]
Add fairings. Do not depend on user actually supplying the arguments
he is supposed to supply.

24 years agoCosmetic changes to comments:
Sheldon Hearn [Wed, 1 Dec 1999 16:25:24 +0000 (16:25 +0000)]
Cosmetic changes to comments:

* Bring source file references in line with the style used in
  GENERIC (i.e. src/sys/...).
* Update outdated source file references.
* Use proper URL syntax for URLs.
* Update outdated URLs.

PR: 15194
Submitted by: jedgar@fxp.org (Chris D. Faulhaber)

24 years agoForgot to commit this in last commit.
Warner Losh [Wed, 1 Dec 1999 15:45:39 +0000 (15:45 +0000)]
Forgot to commit this in last commit.

24 years agoA port of NetBSD's AlphaServer 1000 and 1000A support. Thanks to Cristian
Andrew Gallatin [Wed, 1 Dec 1999 15:25:04 +0000 (15:25 +0000)]
A port of NetBSD's AlphaServer 1000 and 1000A support.  Thanks to Cristian
Angelini for allowing me to use his AS1000 to do the port.

Note that this is untested on AlphaServer 1000A hardware.

Reviewed by: dfr
Tested by:    Cristian Angelini <chr.ang@biella.alpcom.it>
Obtained From:  NetBSD

24 years agoThe 6th bit of configuration port is not defined on pc98.
Yoshihiro Takahashi [Wed, 1 Dec 1999 14:15:30 +0000 (14:15 +0000)]
The 6th bit of configuration port is not defined on pc98.

Submitted by: chi@bd.mbn.or.jp (Chiharu Shibata)

24 years agoSync with sys/isa/sio.c revision 1.276.
Yoshihiro Takahashi [Wed, 1 Dec 1999 13:40:03 +0000 (13:40 +0000)]
Sync with sys/isa/sio.c revision 1.276.

24 years agoInitialize type correctly.
Poul-Henning Kamp [Wed, 1 Dec 1999 10:20:21 +0000 (10:20 +0000)]
Initialize type correctly.

24 years agoAdd NULL for new entrypoint.
Julian Elischer [Wed, 1 Dec 1999 09:37:28 +0000 (09:37 +0000)]
Add NULL for new entrypoint.

24 years agoAdd explicit null entries for new entrypoints.
Julian Elischer [Wed, 1 Dec 1999 09:34:14 +0000 (09:34 +0000)]
Add explicit null entries for new entrypoints.
Forgotten by: Archie

24 years agochange intial timeout for session negotiation from 1 to 2 seconds.
Julian Elischer [Wed, 1 Dec 1999 08:05:56 +0000 (08:05 +0000)]
change intial timeout for session negotiation from 1 to 2 seconds.
One second was being hit too many times in normal situations.

24 years agoFix the hang on card eject problem and maybe the hang on suspend
Warner Losh [Wed, 1 Dec 1999 07:38:54 +0000 (07:38 +0000)]
Fix the hang on card eject problem and maybe the hang on suspend
problem.

o Create new timeout routine so we don't detach the card inside a ISR
  but instead drop back to spl0 via a timeout of 0.
o Actually delete the child of the pccard device rather than just faking
  it badly.
o Fix sio, ed and ep to have pccard detach routines that are int rather
  than void.
o Fix ep and ed pccard detach routines to use if_detach rather than just
  if_down.  if_detach destroys the device, while if_down just marks it
  down.  In this incarnation of the pccard things, we map the disable
  the slot action to detach the driver, which removes the driver from the
  device tree.  When that is done, a panic would soon follow as the
  ifconfig tried to down the device.

Didn't fix:
o Should cache the pccard dev child's pointer in struct slot
o remove now unused parts of struct slot
o Any driver using softc after detach has been called.  sio's softc used
  to be statically allocated, so you could check sc->gone, but that is
  now gone.
o Didn't remove gone from softc of drivers that use the old pccard method.

Didn't test:
o ed driver changes
o sio driver changes on pccards
o suspend (no laptop or apm support on my desktop)

24 years agoSeparate some common sysctl code into sysctl_find_oid() and calling
Brian Feldman [Wed, 1 Dec 1999 02:25:19 +0000 (02:25 +0000)]
Separate some common sysctl code into sysctl_find_oid() and calling
thereof.  Also, make the errno returns  _correct_, and add a new one
which is more appropriate.

24 years agoPrint out the filesystem read counts now collected by the kernel.
Kirk McKusick [Wed, 1 Dec 1999 02:16:51 +0000 (02:16 +0000)]
Print out the filesystem read counts now collected by the kernel.

Submitted by: Craig A Soules <soules+@andrew.cmu.edu>
Reviewed by: Kirk McKusick <mckusick@mckusick.com>

24 years agoCollect read and write counts for filesystems. This new code
Kirk McKusick [Wed, 1 Dec 1999 02:09:30 +0000 (02:09 +0000)]
Collect read and write counts for filesystems. This new code
drops the counting in bwrite and puts it all in spec_strategy.
I did some tests and verified that the counts collected for writes
in spec_strategy is identical to the counts that we previously
collected in bwrite. We now also get read counts (async reads
come from requests for read-ahead blocks). Note that you need
to compile a new version of mount to get the read counts printed
out. The old mount binary is completely compatible, the only
reason to install a new mount is to get the read counts printed.

Submitted by: Craig A Soules <soules+@andrew.cmu.edu>
Reviewed by: Kirk McKusick <mckusick@mckusick.com>

24 years agoDon't make the failure of a LOCAL dist a reported error.
Jordan K. Hubbard [Wed, 1 Dec 1999 01:34:33 +0000 (01:34 +0000)]
Don't make the failure of a LOCAL dist a reported error.
If we're running 4.x and install X, auto-select COMPAT3X.

24 years agoChange the way we transfer links (again). The previous
Brian Somers [Tue, 30 Nov 1999 23:52:37 +0000 (23:52 +0000)]
Change the way we transfer links (again).  The previous
method avoided all race conditions, but suffered from
sometimes running out of buffer space if enough clients
were piled up at the same time.

Now, the client pushes the link descriptor, one end of a
socketpair() and the ppp version via sendmsg() at the
server.  The server replies with a pid.  The client then
transfers any link lock with uu_lock_txfr() and writev()s
the actual link contents.  The socketpair is now the only
place we need to have large socket buffers and the bind()ed
socket can keep the default 4k buffer while still handling
around 90 racing clients.

24 years agodummy version of read_random() until the real random driver is ported.
Dima Ruban [Tue, 30 Nov 1999 23:49:54 +0000 (23:49 +0000)]
dummy version of read_random() until the real random driver is ported.

24 years agoA (harmless) forgotten MFS_ROOT -> MD_ROOT.
Pierre Beyssac [Tue, 30 Nov 1999 21:41:11 +0000 (21:41 +0000)]
A (harmless) forgotten MFS_ROOT -> MD_ROOT.

24 years agoA forgotten "disk" -> "device".
Pierre Beyssac [Tue, 30 Nov 1999 21:24:31 +0000 (21:24 +0000)]
A forgotten "disk" -> "device".

24 years agoIf SRCS is not defined assume ${PROG}.c, not when PROG isn't defined.
Michael Haro [Tue, 30 Nov 1999 20:24:44 +0000 (20:24 +0000)]
If SRCS is not defined assume ${PROG}.c, not when PROG isn't defined.

24 years agoAdd arc4random.c.
Dima Ruban [Tue, 30 Nov 1999 20:18:20 +0000 (20:18 +0000)]
Add arc4random.c.
Please note that kernel is still broken on alpha.

24 years agoUse %EF instead of hardcoded order for each country
Andrey A. Chernov [Tue, 30 Nov 1999 20:08:50 +0000 (20:08 +0000)]
Use %EF instead of hardcoded order for each country

24 years ago%Ex -> %Ef
Andrey A. Chernov [Tue, 30 Nov 1999 19:44:33 +0000 (19:44 +0000)]
%Ex -> %Ef

24 years agoConvert after %Ex->%Ef and %EF adding
Andrey A. Chernov [Tue, 30 Nov 1999 19:43:32 +0000 (19:43 +0000)]
Convert after %Ex->%Ef and %EF adding

24 years ago%Ex -> %Ef to not conflict with POSIX
Andrey A. Chernov [Tue, 30 Nov 1999 19:24:07 +0000 (19:24 +0000)]
%Ex -> %Ef to not conflict with POSIX
Add %EF (long months name / day order)
Check that O and E not intermixed
Add missing POSIX extension to example

24 years agoDocument %Ex and %OB
Andrey A. Chernov [Tue, 30 Nov 1999 18:37:36 +0000 (18:37 +0000)]
Document %Ex and %OB

24 years agoAccept both types of devices to work with older kernels too
Andrey A. Chernov [Tue, 30 Nov 1999 16:27:30 +0000 (16:27 +0000)]
Accept both types of devices to work with older kernels too
Fix confusing message

24 years agoJust to avoid warning message about trigraph.
Yoshinobu Inoue [Tue, 30 Nov 1999 16:24:36 +0000 (16:24 +0000)]
Just to avoid warning message about trigraph.

Commented by: green

24 years agoAdd support for pr's locale
Andrey A. Chernov [Tue, 30 Nov 1999 16:15:22 +0000 (16:15 +0000)]
Add support for pr's locale

24 years agoChange meaning of newly introduced -L option - it helps lpr later
Andrey A. Chernov [Tue, 30 Nov 1999 15:09:26 +0000 (15:09 +0000)]
Change meaning of newly introduced -L option - it helps lpr later

24 years agoImplement linux_sigaltstack.
Marcel Moolenaar [Tue, 30 Nov 1999 15:02:28 +0000 (15:02 +0000)]
Implement linux_sigaltstack.