]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoput the null modem driver (nmdm) in the right place with the right name.
Julian Elischer [Tue, 27 Feb 2001 16:41:28 +0000 (16:41 +0000)]
put the null modem driver (nmdm) in the right place with the right name.

23 years agoDo it right this time. Give it a better name and place
Julian Elischer [Tue, 27 Feb 2001 16:39:54 +0000 (16:39 +0000)]
Do it right this time. Give it a better name and place
CVSrepo deletion of the previous attempt will be requested:

--original message--
Add the 'virtual nulmodem driver'
Particularly useful for debuging kernels using vmware.

If your name is Bruce evans and you are a WIZ at tty interfaces,
then you should probably rip this to shreds and offer lots of suggestions and
patches. I've been using this since 4.0-CURRENT and it's never caused
problems but I'm sure I got something wrong. This is similar to the pty/cty
device driver except that both sides are ttys. Even minor numbers
are side A and odd minor numbers are side B.
Work needs to be done regarding what happens to the other side when you
close a node.

to use with vmware, configure vmware to redirect COM2 out to side A of one
of these and boot a kernel with teh gdb remote port set to sio1.
AFTER dropping into the gdb kernel debugger in your test kernel,
fire up gdb with it's remote port pointing at the appropriate side B.

To catch all console output, you can boot the vmware kernel with a serial
console, (COM1) similarly redirected to a nulmodem, and use 'tip' to observe it.

This is practically unaltered since pre 4.0 days except for
changes made along the way needed to make it compile, so any suggestions
or offers of total rewrites will be listenned to :-)

23 years agoTell the world about the nulmodem device.
Julian Elischer [Tue, 27 Feb 2001 16:20:57 +0000 (16:20 +0000)]
Tell the world about the nulmodem device.

23 years agoCorrectly declare variables as u_int rather than doing typecasts.
Jonathan Lemon [Tue, 27 Feb 2001 15:11:31 +0000 (15:11 +0000)]
Correctly declare variables as u_int rather than doing typecasts.
Kill some register declarations while I'm here.

Submitted by:  bde (1)

23 years agoNew release notes: TCP delayed ACK fixes, ICMP bugfix, find(1)
Bruce A. Mah [Tue, 27 Feb 2001 15:04:50 +0000 (15:04 +0000)]
New release notes:  TCP delayed ACK fixes, ICMP bugfix, find(1)
case-insensitive and regexp additions, pkg_delete(1) globbing,
tar(1) TAR_SSH, disklabel(8) fixes, rand(3) improvements.

New MFCs noted:  pkg_info(1) globbing, burncd(8) -m and related changes.

23 years agoUse formula with better random distribution for rand()
Andrey A. Chernov [Tue, 27 Feb 2001 14:42:19 +0000 (14:42 +0000)]
Use formula with better random distribution for rand()

Even better formula from random() could not be intetgrated because rand_r()
supposed to store its state in the single variable (but table needed for
random() algorithm integration).

23 years agoI accidently deleted an include when I added the $FreeBSD$ so I could
David E. O'Brien [Tue, 27 Feb 2001 14:30:42 +0000 (14:30 +0000)]
I accidently deleted an include when I added the $FreeBSD$ so I could
check in my changes.

23 years agoIn soshutdown(), use SHUT_{RD,WR,RDWR} instead of FREAD and FWRITE.
Ruslan Ermilov [Tue, 27 Feb 2001 13:48:07 +0000 (13:48 +0000)]
In soshutdown(), use SHUT_{RD,WR,RDWR} instead of FREAD and FWRITE.
Also, return EINVAL if `how' is invalid, as required by POSIX spec.

23 years agoImpliment the ISO-C99 strto[u]ll()
David E. O'Brien [Tue, 27 Feb 2001 13:33:07 +0000 (13:33 +0000)]
Impliment the ISO-C99 strto[u]ll()
and rewrite strto[u]q() in terms of it.

23 years agoAdd speaker volume adjusting support
Cameron Grant [Tue, 27 Feb 2001 12:44:31 +0000 (12:44 +0000)]
Add speaker volume adjusting support

Submitted by: Tai-hwa Liang <avatar@mmlab.cse.yzu.edu.tw>
PR: i386/21452

23 years agoAdded another wd33c93 based SCSI card driver which replaces the bs driver.
Yoshihiro Takahashi [Tue, 27 Feb 2001 12:34:01 +0000 (12:34 +0000)]
Added another wd33c93 based SCSI card driver which replaces the bs driver.
Now, default is still bs.

Submitted by: nyan and non.
Obtained from: NetBSD/pc98

23 years agoMake critical toolchain binaries staticly linked in this development
David E. O'Brien [Tue, 27 Feb 2001 11:25:43 +0000 (11:25 +0000)]
Make critical toolchain binaries staticly linked in this development
version of the OS.

23 years agoAdd definitions and support for the AMD k6-2, Pentium MMX (i586/MMX),
Kris Kennaway [Tue, 27 Feb 2001 11:21:47 +0000 (11:21 +0000)]
Add definitions and support for the AMD k6-2, Pentium MMX (i586/MMX),
and Pentium II, III and IV processors (p2, p3, p4), as well as 'mmx' and
'3dnow' MACHINE_CPU tags as appropriate.  In the near future this will
be used to control various ports which have MMX/3dNow optimizations,
instead of the ad-hoc methods currently used.

Reviewed by:    peter

23 years agoProperly detect and report malloc(3) failures.
Maxim Sobolev [Tue, 27 Feb 2001 11:15:04 +0000 (11:15 +0000)]
Properly detect and report malloc(3) failures.

23 years agoAdd rcsid.
David E. O'Brien [Tue, 27 Feb 2001 10:50:06 +0000 (10:50 +0000)]
Add rcsid.

23 years agoThe TCP header-specific section suffered a little bit of bitrot recently:
Bill Fumerola [Tue, 27 Feb 2001 10:20:44 +0000 (10:20 +0000)]
The TCP header-specific section suffered a little bit of bitrot recently:

When we recieve a fragmented TCP packet (other than the first) we can't
extract header information (we don't have state to reference). In a rather
unelegant fashion we just move on and assume a non-match.

Recent additions to the TCP header-specific section of the code neglected
to add the logic to the fragment code so in those cases the match was
assumed to be positive and those parts of the rule (which should have
resulted in a non-match/continue) were instead skipped (which means
the processing of the rule continued even though it had already not
matched).

Fault can be spread out over Rich Steenbergen (tcpoptions) and myself
(tcp{seq,ack,win}).

rwatson sent me a patch that got me thinking about this whole situation
(but what I'm committing / this description is mine so don't blame him).

23 years agoShort lived fame for -Wundef.
Jeroen Ruigrok van der Werven [Tue, 27 Feb 2001 10:16:56 +0000 (10:16 +0000)]
Short lived fame for -Wundef.

Second thoughts by: bde

23 years agoAdd -Wundef to BDECFLAGS:
Jeroen Ruigrok van der Werven [Tue, 27 Feb 2001 09:03:55 +0000 (09:03 +0000)]
Add -Wundef to BDECFLAGS:

Warn if an undefined identifier is evaluated in an `#if' directive.

Not objected to by: bde

23 years agoPrint a diagnostic message if no interface could be determined
Ruslan Ermilov [Tue, 27 Feb 2001 09:02:10 +0000 (09:02 +0000)]
Print a diagnostic message if no interface could be determined
when attempting to create a published ARP entry.  For example:

# arp -s 1.1.1.1 auto pub

PR: bin/7753
Submitted by: Jonathan Hanna <pangolin@home.com>

23 years ago- Merge recently added into pkg_info(1) regex/glob matching functionality into
Maxim Sobolev [Tue, 27 Feb 2001 09:00:18 +0000 (09:00 +0000)]
- Merge recently added into pkg_info(1) regex/glob matching functionality into
  pkg_delete(1) as well;
- add a new `-a' option for pkg_delete(1) to delete all installed packages;
- add a new `-i' option for pkg_delete(1) to request simple rm(1)-like
  interactive confirmation before attempting to delete each package.

Silently approved by: jkh, -ports

23 years agoAdd a 'clobber' target.. Like 'clean' but it takes out even the config
Peter Wemm [Tue, 27 Feb 2001 08:13:32 +0000 (08:13 +0000)]
Add a 'clobber' target.. Like 'clean' but it takes out even the config
generated stuff, leaving only 'version' in the build dir.

23 years agoAdd and document the LINPROCFS option, so that we can build linprocfs
Peter Wemm [Tue, 27 Feb 2001 08:11:28 +0000 (08:11 +0000)]
Add and document the LINPROCFS option, so that we can build linprocfs
(either as a module or in the kernel) after sys/modules/* dies.

23 years agoadd functions for sound drivers to use for locking and setting up interrupt
Cameron Grant [Tue, 27 Feb 2001 07:45:09 +0000 (07:45 +0000)]
add functions for sound drivers to use for locking and setting up interrupt
handlers.  these are not yet used, but will allow compatibility for driver
modules from 5.x to 4.x.

23 years agoRemove some cruft
Peter Wemm [Tue, 27 Feb 2001 07:40:09 +0000 (07:40 +0000)]
Remove some cruft

23 years ago"Document" the COMPAT_LINUX and IBCS2 ABI emulation support together
Peter Wemm [Tue, 27 Feb 2001 07:39:12 +0000 (07:39 +0000)]
"Document" the COMPAT_LINUX and IBCS2 ABI emulation support together
rather than in silly places like "VFS Cluster debugging".  People
should really be using COMPAT_LINUX instead of the linux module on
dynamic systems like -current.

23 years agoMFS: 4.x/5.x compatibility #ifdefs
Cameron Grant [Tue, 27 Feb 2001 07:01:49 +0000 (07:01 +0000)]
MFS: 4.x/5.x compatibility #ifdefs

23 years agoMFS: 4.x/5.x compatibility #defines
Cameron Grant [Tue, 27 Feb 2001 06:58:55 +0000 (06:58 +0000)]
MFS: 4.x/5.x compatibility #defines

23 years agoAdd pci/agp_if.m to the MFILES list so that we can auto depend on agp_if.h
Peter Wemm [Tue, 27 Feb 2001 01:23:34 +0000 (01:23 +0000)]
Add pci/agp_if.m to the MFILES list so that we can auto depend on agp_if.h

23 years agoDisable the mutex locking calls. These do not work in their present form
Peter Wemm [Tue, 27 Feb 2001 01:05:25 +0000 (01:05 +0000)]
Disable the mutex locking calls.  These do not work in their present form
as the code calls the usb stack (which can sleep) while holding the driver
lock.  This leads to a deadlock.

23 years agoCast nfds to u_int before range checking it in order to catch negative
Jonathan Lemon [Tue, 27 Feb 2001 00:50:20 +0000 (00:50 +0000)]
Cast nfds to u_int before range checking it in order to catch negative
values.

PR: 25393

23 years agoFix at2_entry_t to reflect what the firmware actually writes (instead
Matt Jacob [Tue, 27 Feb 2001 00:14:39 +0000 (00:14 +0000)]
Fix at2_entry_t to reflect what the firmware actually writes (instead
of just deriving from SCSI at_entry_t). In this case, there is no
'suggested sense' for FC cards.

23 years agoInitialize native priority to PRI_MAX. It was usually 0 which made a
Jake Burkholder [Mon, 26 Feb 2001 23:27:35 +0000 (23:27 +0000)]
Initialize native priority to PRI_MAX.  It was usually 0 which made a
process's priority go through the roof when it released a (contested)
mutex.  Only set the native priority in mtx_lock if hasn't already
been set.

Reviewed by: jhb

23 years agoAdd crashdump support.
Jonathan Lemon [Mon, 26 Feb 2001 22:25:30 +0000 (22:25 +0000)]
Add crashdump support.

Tested by: ps

23 years agoWorkaround to prevent VMware from melting down. The pseudo PCnet interface
Bill Paul [Mon, 26 Feb 2001 22:23:55 +0000 (22:23 +0000)]
Workaround to prevent VMware from melting down. The pseudo PCnet interface
in VMware reports 0x00000000 in the PCI subsystem ID register, but
0x10001000 when you read the mirror registers in I/O space. This causes
pcn_probe() to think it's found a card in 32-bit mode, and performing
a 32-bit I/O access makes on a 16-bit port makes VMware go boom. Special
case the 0x10001000 value until somebody at VMware grows a clue.

Finally discovered by: Andrew Gallatin

23 years agoUse more aggressive retransmit timeouts for the initial SYN packet.
Jonathan Lemon [Mon, 26 Feb 2001 21:33:55 +0000 (21:33 +0000)]
Use more aggressive retransmit timeouts for the initial SYN packet.
As we currently drop the connection after 4 retransmits + 2 ICMP errors,
this allows initial connection attempts to be dropped much faster.

23 years agoRemove in_pcbnotify and use in_pcblookup_hash to find the cb directly.
Jonathan Lemon [Mon, 26 Feb 2001 21:19:47 +0000 (21:19 +0000)]
Remove in_pcbnotify and use in_pcblookup_hash to find the cb directly.

For TCP, verify that the sequence number in the ICMP packet falls within
the tcp receive window before performing any actions indicated by the
icmp packet.

Clean up some layering violations (access to tcp internals from in_pcb)

23 years agoProperly protect the parameters to the EC_{GET,SET}_{DATA,CSR} macros with
John Baldwin [Mon, 26 Feb 2001 20:39:28 +0000 (20:39 +0000)]
Properly protect the parameters to the EC_{GET,SET}_{DATA,CSR} macros with
parens.

23 years ago- Use a loop to read consecutive bytes from the embedded controller to
John Baldwin [Mon, 26 Feb 2001 20:36:56 +0000 (20:36 +0000)]
- Use a loop to read consecutive bytes from the embedded controller to
  handle read and write requests for widths of multiple bytes.  This
  can be used to read 16-bit battery status registers for example.
- Remove some unused variables and #if 0'd debugging cruft.
- Don't complain about a GPE query that fails due to AE_NOT_FOUND if the
  query method was _Q00.

23 years agoWhen ensuring the destination buffer is truncated for a string obtained
John Baldwin [Mon, 26 Feb 2001 20:32:18 +0000 (20:32 +0000)]
When ensuring the destination buffer is truncated for a string obtained
from a BIF, use the size of the destinatino buffer, not the length of the
string to determine where to put the nul char.  As a side effect, the
old code would truncate the string by one character while it was possibly
overflowing the buffer.

23 years agoTypo fix; use & to test for bits set in the status register.
Mike Smith [Mon, 26 Feb 2001 20:13:19 +0000 (20:13 +0000)]
Typo fix; use & to test for bits set in the status register.

Submitted by: Joel Jacobson <jake@3ware.com>

23 years agoRemove struct full_tcpiphdr{}.
Jeroen Ruigrok van der Werven [Mon, 26 Feb 2001 20:10:16 +0000 (20:10 +0000)]
Remove struct full_tcpiphdr{}.

This piece of code has not been referenced since it was put there
in 1995.  Also done a codebased search on popular networking libraries
and third-party applications.  This is an orphan.

Reviewed by: jesper

23 years agoRemove conditionals for vax support.
Jeroen Ruigrok van der Werven [Mon, 26 Feb 2001 20:05:32 +0000 (20:05 +0000)]
Remove conditionals for vax support.
People who care much about this are welcomed to try 2.11BSD. :)

Noticed by: luigi
Reviewed by: jesper

23 years agoFix clock selection for X.21 interfaces.
John Hay [Mon, 26 Feb 2001 16:30:02 +0000 (16:30 +0000)]
Fix clock selection for X.21 interfaces.

23 years agoUse ``.St -p1003.[12]''.
Ruslan Ermilov [Mon, 26 Feb 2001 16:12:39 +0000 (16:12 +0000)]
Use ``.St -p1003.[12]''.

23 years agoUse ``.St -p1003.1g''.
Ruslan Ermilov [Mon, 26 Feb 2001 16:08:15 +0000 (16:08 +0000)]
Use ``.St -p1003.1g''.

23 years agomdoc(7) police: utilize .St macro.
Ruslan Ermilov [Mon, 26 Feb 2001 16:02:53 +0000 (16:02 +0000)]
mdoc(7) police: utilize .St macro.

23 years agomdoc(7) police: utilize .St.
Ruslan Ermilov [Mon, 26 Feb 2001 15:57:37 +0000 (15:57 +0000)]
mdoc(7) police: utilize .St.

23 years agomdoc(7) police: utilize .St macro.
Ruslan Ermilov [Mon, 26 Feb 2001 15:55:52 +0000 (15:55 +0000)]
mdoc(7) police: utilize .St macro.

23 years agoRemove "autounit" from settable options, it's the default unless you
Poul-Henning Kamp [Mon, 26 Feb 2001 15:31:47 +0000 (15:31 +0000)]
Remove "autounit" from settable options, it's the default unless you
specify -u.

Spotted by: dcs

23 years agoUse ``.St -p1003.1-96''.
Ruslan Ermilov [Mon, 26 Feb 2001 15:16:43 +0000 (15:16 +0000)]
Use ``.St -p1003.1-96''.

23 years ago/^\.St/ s/-iso9945-1/-p1003.1-96/
Ruslan Ermilov [Mon, 26 Feb 2001 14:48:38 +0000 (14:48 +0000)]
/^\.St/ s/-iso9945-1/-p1003.1-96/

23 years agoSynch with NetBSD and OpenBSD: replace -iso9945-1 with -p1003.1-96.
Ruslan Ermilov [Mon, 26 Feb 2001 14:41:05 +0000 (14:41 +0000)]
Synch with NetBSD and OpenBSD: replace -iso9945-1 with -p1003.1-96.

23 years ago``.St -p1003.1b'' -> ``.St -p1003.1b-93''.
Ruslan Ermilov [Mon, 26 Feb 2001 14:33:54 +0000 (14:33 +0000)]
``.St -p1003.1b'' -> ``.St -p1003.1b-93''.

23 years agoRevert part of rev 1.2: make -ansiC equivalent to -ansiC-89.
Ruslan Ermilov [Mon, 26 Feb 2001 13:29:46 +0000 (13:29 +0000)]
Revert part of rev 1.2: make -ansiC equivalent to -ansiC-89.

23 years ago.St -ansiC -> .St -isoC
Ruslan Ermilov [Mon, 26 Feb 2001 13:23:47 +0000 (13:23 +0000)]
.St -ansiC -> .St -isoC

23 years agoo Check the size of I/O window handed by parent bus.
Noriaki Mitsunaga [Mon, 26 Feb 2001 12:26:29 +0000 (12:26 +0000)]
o Check the size of I/O window handed by parent bus.
o Allocate memory mapped by pcic even when not used for ncv.
  This is for PC-Cards which needs offset, because I/O space should not be
  used by other devices.

Pointed-out-by: YAMAMOTO Shigeru <shigeru@iij.ad.jp>
23 years agotake major number 18 for the nmdm "nullmodem" back-to-back tty device.
Julian Elischer [Mon, 26 Feb 2001 09:55:02 +0000 (09:55 +0000)]
take major number 18 for the nmdm "nullmodem" back-to-back tty device.
Incredibally useful for debugging kernels using vmware.
Vmware com1 is diverted to one side, and gdb listens to the other side.
viola.. instant debugging sandbox on one system.

23 years agoProtect against negative numbers as well
Alfred Perlstein [Mon, 26 Feb 2001 09:52:43 +0000 (09:52 +0000)]
Protect against negative numbers as well

23 years agoDocument the EINTR error.
Jason Evans [Mon, 26 Feb 2001 09:38:01 +0000 (09:38 +0000)]
Document the EINTR error.

23 years agoAllow a changed MAC address to show up in ifconfig by changing it
Julian Elischer [Mon, 26 Feb 2001 09:31:54 +0000 (09:31 +0000)]
Allow a changed MAC address to show up in ifconfig by changing it
in the ifaddr list as well. Also change an error return in the base system.

23 years agoDo not output extra space after .Ef call.
Ruslan Ermilov [Mon, 26 Feb 2001 09:24:14 +0000 (09:24 +0000)]
Do not output extra space after .Ef call.

Ported from: mdocNG

23 years agomdoc(7) police: use .Vt macro.
Ruslan Ermilov [Mon, 26 Feb 2001 09:15:17 +0000 (09:15 +0000)]
mdoc(7) police: use .Vt macro.

23 years agofix typo in comment
Alfred Perlstein [Mon, 26 Feb 2001 09:13:42 +0000 (09:13 +0000)]
fix typo in comment

23 years agoSantize a size variable passed to kernel malloc.
Alfred Perlstein [Mon, 26 Feb 2001 09:07:55 +0000 (09:07 +0000)]
Santize a size variable passed to kernel malloc.

Since we know there's always an upper bound we force that bound,
otherwise users can cause a panic via malloc getting hit with a
odd (huge or negative) amount of memory to allocate.

Tested by: kris
Pointed out by: Andrey Valyaev <dron@infosec.ru>

23 years agoPrepare for mdocNG.
Ruslan Ermilov [Mon, 26 Feb 2001 09:05:48 +0000 (09:05 +0000)]
Prepare for mdocNG.

23 years agoBackout -a restriction hack.
Ruslan Ermilov [Mon, 26 Feb 2001 08:09:51 +0000 (08:09 +0000)]
Backout -a restriction hack.

Requested by: rwatson

23 years ago- Mutexify midi(4). The driver runs under the giant lock by default.
Seigo Tanimura [Mon, 26 Feb 2001 07:36:24 +0000 (07:36 +0000)]
- Mutexify midi(4). The driver runs under the giant lock by default.
If you ever want to run midi(4) out of the giant lock, uncomment
MIDI_OUTOFGIANT in midi.h. Confirmed to work for csamidi with WITNESS
and INVARIANTS.

- midi_info, midi_open and seq_info are now tailqs, allowing arbitrary
numbers of devices to be configured.

- Do not send an active sensing message to reset midi modules.

- Clone /dev/sequencer*. /dev/sequencer0 and /dev/sequencer are generated
upon initialization.

23 years agoUpdate for bc 1.06
Kris Kennaway [Mon, 26 Feb 2001 07:23:27 +0000 (07:23 +0000)]
Update for bc 1.06

23 years agoResolve conflicts
Kris Kennaway [Mon, 26 Feb 2001 07:17:03 +0000 (07:17 +0000)]
Resolve conflicts

23 years agoThis commit was generated by cvs2svn to compensate for changes in r73064,
Kris Kennaway [Mon, 26 Feb 2001 07:13:00 +0000 (07:13 +0000)]
This commit was generated by cvs2svn to compensate for changes in r73064,
which included commits to RCS files with non-trunk default branches.

23 years agoInitial import of bc 1.0.6
Kris Kennaway [Mon, 26 Feb 2001 07:13:00 +0000 (07:13 +0000)]
Initial import of bc 1.0.6

23 years agoDocument various changes to kq:
Jonathan Lemon [Mon, 26 Feb 2001 04:16:19 +0000 (04:16 +0000)]
Document various changes to kq:
    - new EV_SET macro,
    - NOTE_LOWAT option for low water marks on read/write filters,
    - NOTE_REVOKE for filesystem unmounting (and revoke() calls)
    - improved API for EVFILT_AIO

23 years agoMore IP option length validation.
Kris Kennaway [Mon, 26 Feb 2001 03:41:13 +0000 (03:41 +0000)]
More IP option length validation.
Includes the following revisions from KAME (two of these were actually
committed previously but the CVS revisions weren't documented):

1.40      kame/kame/sys/netinet6/ah_core.c (committed in previous rev)
1.41      kame/kame/sys/netinet6/ah_core.c
1.28      kame/kame/sys/netinet6/ah_output.c (committed in previous rev)
1.29      kame/kame/sys/netinet6/ah_output.c
1.30      kame/kame/sys/netinet6/ah_output.c
1.129     kame/kame/sys/netinet6/nd6.c
1.130     kame/kame/sys/netinet6/nd6.c
1.24      kame/kame/sys/netinet6/dest6.c
1.25      kame/kame/sys/netinet6/dest6.c

Obtained from: KAME

23 years agoReally set the flags for a private mutex (used by libc/libc_r).
Daniel Eischen [Mon, 26 Feb 2001 01:06:52 +0000 (01:06 +0000)]
Really set the flags for a private mutex (used by libc/libc_r).

23 years agoLimit threads clock resolution to no less than 1000usec (1000Hz).
Daniel Eischen [Mon, 26 Feb 2001 01:05:33 +0000 (01:05 +0000)]
Limit threads clock resolution to no less than 1000usec (1000Hz).

PR: 25300
Submitted by: Tom Pavel <pavel@alum.mit.edu> (in part)

23 years agoFix my ambiguous message about ECONNABORTED.
Jimmy Olgeni [Sun, 25 Feb 2001 23:56:41 +0000 (23:56 +0000)]
Fix my ambiguous message about ECONNABORTED.

Submitted by: Ian Dowse <iedowse@maths.tcd.ie>

23 years agoAdd the 'mly' device nodes.
Mike Smith [Sun, 25 Feb 2001 22:52:55 +0000 (22:52 +0000)]
Add the 'mly' device nodes.

23 years agoRemove the 'gdt' and 'gdtd' majors; the ICP driver is taking a
Mike Smith [Sun, 25 Feb 2001 22:51:36 +0000 (22:51 +0000)]
Remove the 'gdt' and 'gdtd' majors; the ICP driver is taking a
different direction.

Add 'mly' for the newer Mylex driver's control interface.

23 years agoMajor update and bugfix for the 'mly' driver.
Mike Smith [Sun, 25 Feb 2001 22:48:34 +0000 (22:48 +0000)]
Major update and bugfix for the 'mly' driver.

 - Convert to a more efficient queueing implementation.
 - Don't allocate command buffers on the fly; simply work from a
   static pool.
 - Add a control device interface, for later use.
 - Handle controller overload better as a consequence of the
   improved queue implementation.
 - Add support for the XPT_GET_TRAN_SETTINGS ccb, and correctly
   set the virtual SCSI channels up for multiple outstanding I/Os.
 - Update copyrights for 2001.
 - Some whitespace fixes to improve readability.

Due to a misunderstanding on my part, previous versions of the
driver were limited to a single outstanding I/O per virtual drive.
Needless to say, this update improves performance substantially.

23 years agoSpelling: Signalling -> Signaling.
Jimmy Olgeni [Sun, 25 Feb 2001 22:26:37 +0000 (22:26 +0000)]
Spelling: Signalling -> Signaling.
Removed whitespaces at end of lines.

23 years agoAdd ECONNABORTED to the ERRORS section.
Jimmy Olgeni [Sun, 25 Feb 2001 22:12:40 +0000 (22:12 +0000)]
Add ECONNABORTED to the ERRORS section.

23 years agoUpdate the list of OpenSSL manpages (now contains many more describing
Kris Kennaway [Sun, 25 Feb 2001 21:42:12 +0000 (21:42 +0000)]
Update the list of OpenSSL manpages (now contains many more describing
libssl, for example), and hide it behind a make.conf option,
WANT_OPENSSL_MANPAGES, instead of having it commented out.  We still can't
install these by default because of clobbering of a number of system
manpages with the same name, but they're there for people who want them.

23 years agoUse vsnprintf in logmsg() to avoid overflowing the array on the stack.
Gary Jennejohn [Sun, 25 Feb 2001 19:52:42 +0000 (19:52 +0000)]
Use vsnprintf in logmsg() to avoid overflowing the array on the stack.
The problem was noted with an older model 3Com 3C589 which seems
to return more than 256 bytes of data.

23 years agoRemove tcp_drop_all_states, which is unneeded after jlemon removed it
Jesper Skriver [Sun, 25 Feb 2001 17:20:19 +0000 (17:20 +0000)]
Remove tcp_drop_all_states, which is unneeded after jlemon removed it
from tcp_subr.c in rev 1.92

23 years agoslight cleanups during testing.
Julian Elischer [Sun, 25 Feb 2001 16:49:04 +0000 (16:49 +0000)]
slight cleanups during testing.

23 years ago Allow for easier configuration when using disklabel. A sample
John W. De Boskey [Sun, 25 Feb 2001 16:47:44 +0000 (16:47 +0000)]
   Allow for easier configuration when using disklabel. A sample
being:

#        size   offset    fstype   [fsize bsize bps/cpg]
  a:   400M        0    4.2BSD     4096 16384    75     # (Cyl.    0 - 812*)
  b:     1G        *      swap
  c:      *        *    unused
  e: 204800        *    4.2BSD
  f:     5g        *    4.2BSD
  g:      *        *    4.2BSD

   These patches are the original work of Randell Jesup, and
I believe Matt Dillon, with additional work by Warner Losh.
Please let me know if I've left someone out.

   Incorporated into this is the fix for PR bin/22727.

   This patchset still has style issues and a possible problem on
large disks. However, it was a agreed to get these committed before
performing major surgery on them.

PR: bin/22727
Submitted by: Randell Jesup <rjesup@wgate.com>

23 years agoRemove brackets around variables in a function that used to be
Jake Burkholder [Sun, 25 Feb 2001 16:18:13 +0000 (16:18 +0000)]
Remove brackets around variables in a function that used to be
a macro.

23 years agoDo not delay a new ack if there already is a delayed ack pending on the
Jonathan Lemon [Sun, 25 Feb 2001 15:17:24 +0000 (15:17 +0000)]
Do not delay a new ack if there already is a delayed ack pending on the
connection, but send it immediately.  Prior to this change, it was possible
to delay a delayed-ack for multiple times, resulting in degraded TCP
behavior in certain corner cases.

23 years agoo Check the size of I/O window handed by parent bus.
Noriaki Mitsunaga [Sun, 25 Feb 2001 14:01:46 +0000 (14:01 +0000)]
o Check the size of I/O window handed by parent bus.

23 years agoo Check the size of I/O window handed by parent bus.
Noriaki Mitsunaga [Sun, 25 Feb 2001 14:01:05 +0000 (14:01 +0000)]
o Check the size of I/O window handed by parent bus.
o Check if it is in PIO_MODE when memory window is not handed.

23 years agoStop pac dereferencing a null pointer if accounting is not enabeled.
David Malone [Sun, 25 Feb 2001 13:50:29 +0000 (13:50 +0000)]
Stop pac dereferencing a null pointer if accounting is not enabeled.

PR: 24798 18191
Submitted by: Nick Hilliard <nick@netability.ie>
Reviewed by: gad

23 years agoForgot to remove unneeded "intcmp" function.
Poul-Henning Kamp [Sun, 25 Feb 2001 13:14:41 +0000 (13:14 +0000)]
Forgot to remove unneeded "intcmp" function.

23 years agoMake "md" and "mdctl" macroized parameters.
Poul-Henning Kamp [Sun, 25 Feb 2001 13:12:57 +0000 (13:12 +0000)]
Make "md" and "mdctl" macroized parameters.

Implement "-l" option to mdconfig which can list one or all md devices.

Submitted by:   Dima Dorfman <dima@unixfreak.org>

23 years agoo Support AUTO SENSE correctly.
Noriaki Mitsunaga [Sun, 25 Feb 2001 12:40:30 +0000 (12:40 +0000)]
o Support AUTO SENSE correctly.
o Offset and period in synch messages and width negotiation should be
  done for per target not per lun. Move these from *lun_info to
  *targ_info.
o Change in handling XPT_RESET_DEV and XPT_GET_TRAN_SETTINGS .
o Change CAM_* xpt_done return values.
o Busy loop did not timeout. Change this to timeout as original NetBSD/pc98.

Reviewed by: bsd-nomads ML

23 years agoFix references to Chapman & Zwicky and Cheswick & Bellowin.
Dag-Erling Smørgrav [Sun, 25 Feb 2001 11:44:51 +0000 (11:44 +0000)]
Fix references to Chapman & Zwicky and Cheswick & Bellowin.

PR: 24652
Submitted by: jjreynold@home.com

23 years agoSupported pcmcia modem card.
Yoshihiro Takahashi [Sun, 25 Feb 2001 08:55:07 +0000 (08:55 +0000)]
Supported pcmcia modem card.

Submitted by: MURAMATSU Atsushi <amura@ma3.seikyou.ne.jp>

23 years agoUpdate unused __dtoa prototypes to match reality.
Tor Egge [Sun, 25 Feb 2001 08:51:41 +0000 (08:51 +0000)]
Update unused __dtoa prototypes to match reality.

23 years agoAdd fsck_4.2bsd to boot crunch target. This returns the ability to
Jordan K. Hubbard [Sun, 25 Feb 2001 08:02:46 +0000 (08:02 +0000)]
Add fsck_4.2bsd to boot crunch target.  This returns the ability to
actually fsck (and upgrade) a box from boot media again.

23 years agoMerged from sys/i386/i386/machdep.c revision 1.443.
KATO Takenori [Sun, 25 Feb 2001 08:00:35 +0000 (08:00 +0000)]
Merged from sys/i386/i386/machdep.c revision 1.443.

23 years agogenassym.sh does not work with a.out because the sizes are rounded up
Peter Wemm [Sun, 25 Feb 2001 07:51:19 +0000 (07:51 +0000)]
genassym.sh does not work with a.out because the sizes are rounded up
by the compiler.  ie: char foo[0] comes out as 4 bytes on a.out, and
we depended on it coming out as 0 for the script version. :-(

Make double sure that genassym.o is built and nm'ed in elf mode.

(ia64 skipped since it is stuck on the linux toolchain and doesn't
 understand the -elf switches)