]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoMy 5520 turned out to be a 5510 in disguise.. correct the probe message.
Julian Elischer [Fri, 7 May 1999 04:04:42 +0000 (04:04 +0000)]
My 5520 turned out to be a 5510 in disguise.. correct the probe message.

25 years agoAdd support to wicontrol(8) and wi(4) for enabling and configuring
Bill Paul [Fri, 7 May 1999 03:28:54 +0000 (03:28 +0000)]
Add support to wicontrol(8) and wi(4) for enabling and configuring
power management. This will only work on newer firmware revisions; older
firmware will silently ignore the attempts to turn power management on.

Patches supplied by: Brad Karp <karp@eecs.harvard.edu>

25 years agoRemove the last trace of disk.h
Greg Lehey [Fri, 7 May 1999 03:28:46 +0000 (03:28 +0000)]
Remove the last trace of disk.h

Put-up-to-it-by: phk

25 years agoSevere slowdowns have been reported when creating or removing many
Kirk McKusick [Fri, 7 May 1999 02:26:47 +0000 (02:26 +0000)]
Severe slowdowns have been reported when creating or removing many
files at once on a filesystem running soft updates. The root of
the problem is that soft updates limits the amount of memory that
may be allocated to dependency structures so as to avoid hogging
kernel memory. The original algorithm just waited for the disk I/O
to catch up and reduce the number of dependencies. This new code
takes a much more aggressive approach. Basically there are two
resources that routinely hit the limit. Inode dependencies during
periods with a high file creation rate and file and block removal
dependencies during periods with a high file removal rate. I have
attacked these problems from two fronts. When the inode dependency
limits are reached, I pick a random inode dependency, UFS_UPDATE
it together with all the other dirty inodes contained within its
disk block and then write that disk block. This trick usually
clears 5-50 inode dependencies in a single disk I/O. For block and
file removal dependencies, I pick a random directory page that has
at least one remove pending and VOP_FSYNC its directory. That
releases all its removal dependencies to the work queue. To further
hasten things along, I also immediately start the work queue process
rather than waiting for its next one second scheduled run.

25 years agoRemove old grunge, including references to sys/disk.h (which is going
Greg Lehey [Fri, 7 May 1999 01:35:29 +0000 (01:35 +0000)]
Remove old grunge, including references to sys/disk.h (which is going
away).

Submitted-by: phk

25 years agoRemove old grunge
Greg Lehey [Fri, 7 May 1999 01:25:54 +0000 (01:25 +0000)]
Remove old grunge

Submitted-by: phk

25 years agoUse id_ointr for setting the pnp interrupt handler, it uses the oldstyle
Peter Wemm [Thu, 6 May 1999 22:26:28 +0000 (22:26 +0000)]
Use id_ointr for setting the pnp interrupt handler, it uses the oldstyle
int unit argument still.

25 years ago``fix'' the devfs_add_devswf() calls, the printf string wasn't factoring
Peter Wemm [Thu, 6 May 1999 22:21:31 +0000 (22:21 +0000)]
``fix'' the devfs_add_devswf() calls, the printf string wasn't factoring
in the unit number.  I'm not so sure about this at all, the SVR4 systems
I have access to have a mixture of names and often %03d format units.

25 years agoUndo excess staticization - these two are meant to be callable from
Peter Wemm [Thu, 6 May 1999 22:19:38 +0000 (22:19 +0000)]
Undo excess staticization - these two are meant to be callable from
DDB.

25 years agoahatimeout is static..
Peter Wemm [Thu, 6 May 1999 22:18:21 +0000 (22:18 +0000)]
ahatimeout is static..

25 years agoUse consistant function definitions which also silences a warning.
Peter Wemm [Thu, 6 May 1999 22:17:26 +0000 (22:17 +0000)]
Use consistant function definitions which also silences a warning.

25 years agoEnsure prototype for pnp_configure() is visible.
Peter Wemm [Thu, 6 May 1999 22:16:19 +0000 (22:16 +0000)]
Ensure prototype for pnp_configure() is visible.

25 years agoGC unused variable
Peter Wemm [Thu, 6 May 1999 22:14:46 +0000 (22:14 +0000)]
GC unused variable

25 years agoPull in prototype for splq().
Peter Wemm [Thu, 6 May 1999 22:13:53 +0000 (22:13 +0000)]
Pull in prototype for splq().

25 years agoFix a static/extern conflict. Put extra brackets to ensure the
Peter Wemm [Thu, 6 May 1999 22:13:04 +0000 (22:13 +0000)]
Fix a static/extern conflict.  Put extra brackets to ensure the
tsleep() priority is clear and not parsed incorrectly.

25 years agoPut the old-style isa interrupt handlers in id_ointr to avoid warnings.
Peter Wemm [Thu, 6 May 1999 22:11:19 +0000 (22:11 +0000)]
Put the old-style isa interrupt handlers in id_ointr to avoid warnings.

25 years agoFix a warning - make sure the register is read regardless of the debugging
Peter Wemm [Thu, 6 May 1999 22:09:55 +0000 (22:09 +0000)]
Fix a warning - make sure the register is read regardless of the debugging
options.

25 years agoFix two warnings; and note a problem where a pointer is stored in an
Peter Wemm [Thu, 6 May 1999 22:08:57 +0000 (22:08 +0000)]
Fix two warnings; and note a problem where a pointer is stored in an
int variable - this can't work on an Alpha.

25 years agoOne too many vfsops..
Peter Wemm [Thu, 6 May 1999 22:07:34 +0000 (22:07 +0000)]
One too many vfsops..

25 years agoAdd brackets to silence egcs and help clarity.
Peter Wemm [Thu, 6 May 1999 22:06:45 +0000 (22:06 +0000)]
Add brackets to silence egcs and help clarity.

25 years agoFix a goof on my part; s/struct moduledata */struct module */
Peter Wemm [Thu, 6 May 1999 22:05:39 +0000 (22:05 +0000)]
Fix a goof on my part; s/struct moduledata */struct module */

25 years agoFix some variable naming confusion
Peter Wemm [Thu, 6 May 1999 22:04:21 +0000 (22:04 +0000)]
Fix some variable naming confusion

25 years agoGC unused pps_drvinit() declaration
Peter Wemm [Thu, 6 May 1999 22:03:14 +0000 (22:03 +0000)]
GC unused pps_drvinit() declaration

25 years agoAdd Cyrix (NatSemi) 5520 and 5530 PCI-ISA bridges.
Julian Elischer [Thu, 6 May 1999 21:21:30 +0000 (21:21 +0000)]
Add Cyrix (NatSemi) 5520 and 5530 PCI-ISA bridges.

25 years agoThe base transfer speed for the parallel port bus is 93K/sec, not 3.3MB/sec.
Kenneth D. Merry [Thu, 6 May 1999 21:14:49 +0000 (21:14 +0000)]
The base transfer speed for the parallel port bus is 93K/sec, not 3.3MB/sec.

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

25 years agoTake out calls to cam_sim_set_basexfer_speed(), the base transfer speed is
Kenneth D. Merry [Thu, 6 May 1999 21:12:58 +0000 (21:12 +0000)]
Take out calls to cam_sim_set_basexfer_speed(), the base transfer speed is
now returned in the path inquiry CCB.

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

25 years agoAdd new member for XPT_PATH_INQ, follows recent changes in
Nick Hibma [Thu, 6 May 1999 20:58:37 +0000 (20:58 +0000)]
Add new member for XPT_PATH_INQ, follows recent changes in
version v1.2 of cam_sim.h.

25 years agoFix confusing sentence, the PR noticed the sentence, I rewrote the sentence.
Bill Fumerola [Thu, 6 May 1999 20:51:31 +0000 (20:51 +0000)]
Fix confusing sentence, the PR noticed the sentence, I rewrote the sentence.

PR: docs/11257
Submitted by: Motomichi Matsuzaki <mzaki@e-mail.ne.jp>

25 years agoAdd a number of interrelated CAM feature enhancements and bug fixes.
Kenneth D. Merry [Thu, 6 May 1999 20:16:39 +0000 (20:16 +0000)]
Add a number of interrelated CAM feature enhancements and bug fixes.

NOTE:  These changes will require recompilation of any userland
applications, like cdrecord, xmcd, etc., that use the CAM passthrough
interface.  A make world is recommended.

camcontrol.[c8]:
 - We now support two new commands, "tags" and "negotiate".

- The tags commands allows users to view the number of tagged
  openings for a device as well as a number of other related
  parameters, and it allows users to set tagged openings for
  a device.

- The negotiate command allows users to enable and disable
  disconnection and tagged queueing, set sync rates, offsets
  and bus width.  Note that not all of those features are
  available for all controllers.  Only the adv, ahc, and ncr
  drivers fully support all of the features at this point.
  Some cards do not allow the setting of sync rates, offsets and
  the like, and some of the drivers don't have any facilities to
  do so.  Some drivers, like the adw driver, only support enabling
  or disabling sync negotiation, but do not support setting sync
  rates.

 - new description in the camcontrol man page of how to format a disk
 - cleanup of the camcontrol inquiry command
 - add support in the 'devlist' command for skipping unconfigured devices if
   -v was not specified on the command line.
 - make use of the new base_transfer_speed in the path inquiry CCB.
 - fix CCB bzero cases

cam_xpt.c, cam_sim.[ch], cam_ccb.h:

 - new flags on many CCB function codes to designate whether they're
   non-immediate, use a user-supplied CCB, and can only be passed from
   userland programs via the xpt device.  Use these flags in the transport
   layer and pass driver to categorize CCBs.

 - new flag in the transport layer device matching code for device nodes
   that indicates whether a device is unconfigured

 - bump the CAM version from 0x10 to 0x11

 - Change the CAM ioctls to use the version as their group code, so we can
   force users to recompile code even when the CCB size doesn't change.

 - add + fill in a new value in the path inquiry CCB, base_transfer_speed.
   Remove a corresponding field from the cam_sim structure, and add code to
   every SIM to set this field to the proper value.

 - Fix the set transfer settings code in the transport layer.

scsi_cd.c:

 - make some variables volatile instead of just casting them in various
   places
 - fix a race condition in the changer code
 - attach unless we get a "logical unit not supported" error.  This should
   fix all of the cases where people have devices that return weird errors
   when they don't have media in the drive.

scsi_da.c:

 - attach unless we get a "logical unit not supported" error

scsi_pass.c:

 - for immediate CCBs, just malloc a CCB to send the user request in.  This
   gets rid of the 'held' count problem in camcontrol tags.

scsi_pass.h:

 - change the CAM ioctls to use the CAM version as their group code.

adv driver:

 - Allow changing the sync rate and offset separately.

adw driver

 - Allow changing the sync rate and offset separately.

aha driver:

 - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs.

ahc driver:

 - Allow setting offset and sync rate separately

bt driver:

 - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs.

NCR driver:

 - Fix the ultra/ultra 2 negotiation bug
 - allow setting both the sync rate and offset separately

Other HBA drivers:
 - Put code in to set the base_transfer_speed field for
   XPT_GET_TRAN_SETTINGS CCBs.

Reviewed by: gibbs, mjacob (isp), imp (aha)

25 years agoremove b_proc from struct buf, it's (now) unused.
Poul-Henning Kamp [Thu, 6 May 1999 20:00:34 +0000 (20:00 +0000)]
remove b_proc from struct buf, it's (now) unused.

Reviewed by: dillon, bde

25 years agoOops, forgot this bit: don't use <sys/disk.h>
Poul-Henning Kamp [Thu, 6 May 1999 19:35:55 +0000 (19:35 +0000)]
Oops, forgot this bit: don't use <sys/disk.h>

25 years agoAdded details of PCI network cards which work by emulating the NE2000.
Roger Hardiman [Thu, 6 May 1999 19:24:45 +0000 (19:24 +0000)]
Added details of PCI network cards which work by emulating the NE2000.
  RealTek 8029, NetVin 5000, Winbond W89C940, Surecom NE-34, VIA VT86C926.
(checked with Bill Paul)

Mention the Brooktree Bt878 is supported by the Bt848 driver.

25 years agodon't use <sys/disk.h>
Poul-Henning Kamp [Thu, 6 May 1999 19:20:34 +0000 (19:20 +0000)]
don't use <sys/disk.h>

25 years agoDon't use <sys/disk.h>
Poul-Henning Kamp [Thu, 6 May 1999 19:19:30 +0000 (19:19 +0000)]
Don't use <sys/disk.h>

25 years agoNuke bogus prototypes which have bogotified ccd and vinum
Poul-Henning Kamp [Thu, 6 May 1999 19:07:06 +0000 (19:07 +0000)]
Nuke bogus prototypes which have bogotified ccd and vinum
(who shouldn't really use this file in the first place!)

25 years agoFix 'signed char as array index' warnings and an unused variable.
Peter Wemm [Thu, 6 May 1999 18:58:05 +0000 (18:58 +0000)]
Fix 'signed char as array index' warnings and an unused variable.

25 years agoMissing 'int' in declaration of variables.
Peter Wemm [Thu, 6 May 1999 18:54:19 +0000 (18:54 +0000)]
Missing 'int' in declaration of variables.

25 years agoFix a precedence bug in the atapi tape driver. I think it could either
Peter Wemm [Thu, 6 May 1999 18:50:10 +0000 (18:50 +0000)]
Fix a precedence bug in the atapi tape driver.  I think it could either
write a filemark where it wasn't needed, or neglect to write one at all,
depending on how the boolean converted to an int value for the &.

25 years agoFix up a few easy 'assignment used as truth value' and 'suggest parens
Peter Wemm [Thu, 6 May 1999 18:44:42 +0000 (18:44 +0000)]
Fix up a few easy 'assignment used as truth value' and 'suggest parens
around && within ||' type warnings.  I'm pretty sure I have not masked
any problems here, I've committed real problem fixes seperately.

25 years agoThe joypart() macro had a precedence bug. Add seatbelts for UNIT() too.
Peter Wemm [Thu, 6 May 1999 18:39:32 +0000 (18:39 +0000)]
The joypart() macro had a precedence bug.  Add seatbelts for UNIT() too.

25 years agoMissing 'int' declaration in devsw_installed variable.
Peter Wemm [Thu, 6 May 1999 18:33:48 +0000 (18:33 +0000)]
Missing 'int' declaration in devsw_installed variable.

25 years agoAdd sufficient braces to keep egcs happy about potentially ambiguous
Peter Wemm [Thu, 6 May 1999 18:13:11 +0000 (18:13 +0000)]
Add sufficient braces to keep egcs happy about potentially ambiguous
if/else nesting.

25 years agoAdd missing comment characters from wi driver description.
Peter Wemm [Thu, 6 May 1999 18:08:23 +0000 (18:08 +0000)]
Add missing comment characters from wi driver description.

25 years agoRemove unused fields from struct buf:
Poul-Henning Kamp [Thu, 6 May 1999 17:06:41 +0000 (17:06 +0000)]
Remove unused fields from struct buf:
b_savekva
b_validoff
b_validend

Reviewed by: dillon, bde

25 years agoModify wicontrol(8) and wi(4) to allow setting the frequency of the
Bill Paul [Thu, 6 May 1999 16:32:45 +0000 (16:32 +0000)]
Modify wicontrol(8) and wi(4) to allow setting the frequency of the
WaveLAN's radio modem. The default is whatever the NIC uses since NICs
sold in different countries may default to different frequencies. (The
Lose95/LoseNT software doesn't let you select the channel so it's probably
not really meant to be changed.)

25 years agoTweak the Macronix driver to hopefully make it more reliable:
Bill Paul [Thu, 6 May 1999 15:32:52 +0000 (15:32 +0000)]
Tweak the Macronix driver to hopefully make it more reliable:

- Change to the same transmit scheme as the PNIC driver.
- Dynamically set the cache alignment, and set burst size the same as
  the PNIC driver in mx_init().
- Enable 'store and forward' mode by default. This is the slowest option
  and it does reduce 100Mbps performance somewhat, but it's the most
  reliable setting I can find. I'm more interested in having the driver
  work reliably than trying to squeeze the best performance out of it.
  The reason I'm doing this is that on *some* systems you may see a lot
  of transmit underruns (which I can't explain: these are *fast* test
  systems) and these errors seem to cause unusual and decidedly
  non-tulip-like behavior. In normal 10Mbps mode, performance is fine
  (you can easily saturate a 10Mbps link).

Also tweak some of the other drivers:

- Increase the size of the TX ring for the Winbond, ASIX, VIA Rhine
  and PNIC drivers.
- Set a larger value for ifq_maxlen in the ThunderLAN driver. The setting
  of TL_TX_LIST_CNT - 1 is too low (the ThunderLAN driver only allocates
  20 transmit descriptors, and I don't want to fiddle with that now
  because the ThunderLAN's descriptor structure is an oddball size
  compared to the others).

25 years agoreadklog(): rename variable 'l' to 'len', to avoid possible confusion with 'i'
Dmitrij Tejblum [Thu, 6 May 1999 13:57:57 +0000 (13:57 +0000)]
readklog(): rename variable 'l' to 'len', to avoid possible confusion with 'i'
and '1'.

Requested by: mckay

25 years agoAdded USB HID devices major number.
Nick Hibma [Thu, 6 May 1999 13:51:26 +0000 (13:51 +0000)]
Added USB HID devices major number.

Submitted by: MAEKAWA Masahide

25 years agoMove the proc0 init before the driver probe/attach etc since machdep.c
Peter Wemm [Thu, 6 May 1999 13:42:25 +0000 (13:42 +0000)]
Move the proc0 init before the driver probe/attach etc since machdep.c
doesn't set curproc anymore, and certain drivers like to tsleep() during
probes, usb for example.

25 years agoI'm not sure why the #ifdef SMP became #if 1 (this overrode the npx probe
Peter Wemm [Thu, 6 May 1999 12:47:21 +0000 (12:47 +0000)]
I'm not sure why the #ifdef SMP became #if 1 (this overrode the npx probe
and always succeeded as is required on SMP).  Anyway, reverting this
still compiles and appears ok.

25 years agoFix spelling and typos.
Joseph Koshy [Thu, 6 May 1999 12:17:58 +0000 (12:17 +0000)]
Fix spelling and typos.

PR: docs/11506
Submitted by: Norihiro Kumagai <kuma@jp.freebsd.org>

25 years agoFix typos and -mdoc usage.
Joseph Koshy [Thu, 6 May 1999 11:28:01 +0000 (11:28 +0000)]
Fix typos and -mdoc usage.

PR: docs/11537
Submitted by: Motomichi Matsuzaki <mzaki@e-mail.ne.jp>

25 years agorevert premature DHCP commit
Jordan K. Hubbard [Thu, 6 May 1999 11:15:33 +0000 (11:15 +0000)]
revert premature DHCP commit

25 years agoArgh, those last two commits snuck in with the other one by mistake.
Jordan K. Hubbard [Thu, 6 May 1999 11:05:19 +0000 (11:05 +0000)]
Argh, those last two commits snuck in with the other one by mistake.
Disable the non-working dhcp client code I just committed by mistake.

25 years agoFix includes; I had a slightly different Makefile than everyone else
Jordan K. Hubbard [Thu, 6 May 1999 11:03:21 +0000 (11:03 +0000)]
Fix includes; I had a slightly different Makefile than everyone else
which is why I didn't see this. :)

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

25 years agoFixed profiling of elf kernels. Made high resolution profiling compile
Bruce Evans [Thu, 6 May 1999 09:44:57 +0000 (09:44 +0000)]
Fixed profiling of elf kernels.  Made high resolution profiling compile
for elf kernels (it is broken for all kernels due to lack of egcs support).

Renaming of many assembler labels is avoided by declaring by declaring
the labels that need to be visible to gprof as having type "function"
and depending on the elf version of gprof being zealous about discarding
the others.  A few type declarations are still missing, mainly for SMP.

PR: 9413
Submitted by: Assar Westerlund <assar@sics.se> (initial parts)

25 years agoSync with sys/i386/i386/userconfig.c revision 1.138.
KATO Takenori [Thu, 6 May 1999 09:15:20 +0000 (09:15 +0000)]
Sync with sys/i386/i386/userconfig.c revision 1.138.

25 years agoSync with sys/i386/conf/files.i386 revision 1.237.
KATO Takenori [Thu, 6 May 1999 09:13:56 +0000 (09:13 +0000)]
Sync with sys/i386/conf/files.i386 revision 1.237.

25 years agoFixed bitrot in comments.
Bruce Evans [Thu, 6 May 1999 03:35:44 +0000 (03:35 +0000)]
Fixed bitrot in comments.

25 years agoFix some byte ordering problems; I was storing string lengths wrong,
Bill Paul [Thu, 6 May 1999 03:34:02 +0000 (03:34 +0000)]
Fix some byte ordering problems; I was storing string lengths wrong,
which was causing wicontrol to crash when reading string parameters
from the WaveLAN.

Patches submitted by: Brad Karp <karp@eecs.harvard.edu>

25 years agoAdd the capability for traceroute(8) to send packets of any IP protocol
Archie Cobbs [Thu, 6 May 1999 03:23:24 +0000 (03:23 +0000)]
Add the capability for traceroute(8) to send packets of any IP protocol
instead of just UDP; an alternate protocol is specified by '-P proto'.
This is useful for finding routers that are blocking packets based on
IP protocol.  New handlers can be added fairly easily to do protocol-
specific things.

25 years ago"Fixed" missing dependency of ${PROG} on ${LIBC} in the elf case.
Bruce Evans [Thu, 6 May 1999 02:58:30 +0000 (02:58 +0000)]
"Fixed" missing dependency of ${PROG} on ${LIBC} in the elf case.

25 years agoNow that each cpu has its own gdt table, we need to setup apm gdt entries in
Luoqi Chen [Thu, 6 May 1999 01:07:03 +0000 (01:07 +0000)]
Now that each cpu has its own gdt table, we need to setup apm gdt entries in
all the tables.

25 years agoInitialize dblfault_tss.tss_fs to the per-cpu private data segment selector.
Luoqi Chen [Thu, 6 May 1999 00:54:54 +0000 (00:54 +0000)]
Initialize dblfault_tss.tss_fs to the per-cpu private data segment selector.

25 years agoDon't ignore mmap() address hint below the text section.
Luoqi Chen [Thu, 6 May 1999 00:46:19 +0000 (00:46 +0000)]
Don't ignore mmap() address hint below the text section.

25 years agoDo not set curproc until proc0 is fully initialized (in proc0_init()).
Luoqi Chen [Thu, 6 May 1999 00:39:00 +0000 (00:39 +0000)]
Do not set curproc until proc0 is fully initialized (in proc0_init()).

25 years agoDon't forget to install crash.8 after moving it.
Bruce Evans [Thu, 6 May 1999 00:01:48 +0000 (00:01 +0000)]
Don't forget to install crash.8 after moving it.

25 years agoMFS: don't build the pcic.ko module.
David E. O'Brien [Wed, 5 May 1999 20:58:15 +0000 (20:58 +0000)]
MFS: don't build the pcic.ko module.

25 years agoIncomplete. Back out until I can revisit.
Mark Murray [Wed, 5 May 1999 20:28:54 +0000 (20:28 +0000)]
Incomplete. Back out until I can revisit.

25 years agoAdd extra dirs needed by install. Do the include/* by reusing
Mark Murray [Wed, 5 May 1999 19:44:14 +0000 (19:44 +0000)]
Add extra dirs needed by install. Do the include/* by reusing
BSD.include.dist to make tracking this easier.

Pointed out by: Rod Grimes

25 years agoThis has bitrotted to the state that it is no longer functional.
Mark Murray [Wed, 5 May 1999 19:34:29 +0000 (19:34 +0000)]
This has bitrotted to the state that it is no longer functional.
Make Mike Smith very happy by nuking the PCIC auto-load code. :-)

NOTE - You will have to have PCIC in your kernel again!!
You have been warned!!

25 years agoVarious cosmetics.
Poul-Henning Kamp [Wed, 5 May 1999 19:23:45 +0000 (19:23 +0000)]
Various cosmetics.

Submitted by: Rudolf Cejka <cejkar@dcse.fee.vutbr.cz>
Reviewed by: phk

25 years agoRemove $Log garbage from these; it was causing failure when expanded.
Mark Murray [Wed, 5 May 1999 19:09:48 +0000 (19:09 +0000)]
Remove $Log garbage from these; it was causing failure when expanded.

Found by: Udo Schweigert <ust@cert.siemens.de>

25 years agoDynamically increase TX start threshold if TX underruns are detected.
Bill Paul [Wed, 5 May 1999 17:05:07 +0000 (17:05 +0000)]
Dynamically increase TX start threshold if TX underruns are detected.

25 years agoFix a typo (wi. -> wi.4)
Chris Timmons [Wed, 5 May 1999 12:54:50 +0000 (12:54 +0000)]
Fix a typo (wi. -> wi.4)

Submitted by: Zach Heilig <zach@uffdaonline.net>

25 years agoAdd the INIT_PATH option for embedded systems.
John Birrell [Wed, 5 May 1999 12:22:31 +0000 (12:22 +0000)]
Add the INIT_PATH option for embedded systems.

25 years agoAllow the init_path to be customised in an embedded system using the
John Birrell [Wed, 5 May 1999 12:20:23 +0000 (12:20 +0000)]
Allow the init_path to be customised in an embedded system using the
INIT_PATH config option.

Also fix two bugs which caused an infinite loop in none of the programs
in the init_path were found. That code was obviously not tested!

25 years agoAnother round of updates for -current and 3.2
Jordan K. Hubbard [Wed, 5 May 1999 11:34:03 +0000 (11:34 +0000)]
Another round of updates for -current and 3.2

25 years agoAdd an INIT_PATH option to allow the init_path (for sys/kern/init_main.c)
John Birrell [Wed, 5 May 1999 11:24:15 +0000 (11:24 +0000)]
Add an INIT_PATH option to allow the init_path (for sys/kern/init_main.c)
to be customised in an embedded system which doesn't want to
run either init of sysinstall.

25 years agoTypo fix.
Jordan K. Hubbard [Wed, 5 May 1999 10:18:55 +0000 (10:18 +0000)]
Typo fix.

25 years agoFix a few bogons and make this safer for humanity.
Jordan K. Hubbard [Wed, 5 May 1999 10:14:06 +0000 (10:14 +0000)]
Fix a few bogons and make this safer for humanity.

25 years agoAdd information strings for a number of devices which have suddenly appeared
Jordan K. Hubbard [Wed, 5 May 1999 09:37:22 +0000 (09:37 +0000)]
Add information strings for a number of devices which have suddenly appeared
in the configuration name space.

25 years agoDeal with MAKEDEV's braindedness.
Andrzej Bialecki [Wed, 5 May 1999 08:37:46 +0000 (08:37 +0000)]
Deal with MAKEDEV's braindedness.

Submitted by: Patrick Powell <papowell@astart4.astart.com>
Reviewed by: abial

25 years agoAdd device driver support for the Lucent WaveLAN/IEEE 802.11 PCMCIA
Bill Paul [Wed, 5 May 1999 07:37:11 +0000 (07:37 +0000)]
Add device driver support for the Lucent WaveLAN/IEEE 802.11 PCMCIA
adapter (and some workalikes). Also add man pages and a wicontrol
utility to manipulate some of the card parameters.

This driver was written using information gleaned from the Lucent HCF Light
library, though it does not use any of the HCF Light code itself, mainly
because it's contaminated by the GPL (but also because it's pretty gross).
The HCF Light lacks certain featurs from the full (but proprietary) HCF
library, including 802.11 frame encapsulation support, however it has
just enough register information about the Hermes chip to allow someone
with enough spare time and energy to implement a proper driver. (I would
have prefered getting my hands on the Hermes manual, but that's proprietary
too. For those who are wondering, the Linux driver uses the proprietary
HCF library, but it's provided in object code form only.)

Note that I do not have access to a WavePOINT access point, so I have
only been able to test ad-hoc mode. The wicontrol utility can turn on
BSS mode, but I don't know for certain that the NIC will associate with
an access point correctly. Testers are encouraged to send their results
to me so that I can find out if I screwed up or not.

25 years agoSimple_HTTPd on steroids - major rework of the original program, by
Andrzej Bialecki [Wed, 5 May 1999 07:22:34 +0000 (07:22 +0000)]
Simple_HTTPd on steroids - major rework of the original program, by
William Lloyd. New features include:

* many additional command line options
* "fetch" mode
* less bugs :-)
* better README.

Submitted by: William Lloyd <wlloyd@lap.net>
Reviewed by: abial

25 years agoAdd device driver support for the Lucent WaveLAN/IEEE 802.11 wireless
Bill Paul [Wed, 5 May 1999 07:11:38 +0000 (07:11 +0000)]
Add device driver support for the Lucent WaveLAN/IEEE 802.11 wireless
network adapters. These are all PCMCIA devices (the ISA version is a
PCMCIA to ISA bridge with a PCMCIA card plugged into it). Also add a
wicontrol utility to read and write some of the card's parameters.

Note: I do not have access to a WavePOINT access point, so I have only
been able to test this driver in ad-hoc (point to point) mode. The
wicontrol utility allows programming the desired service set name (SSID)
and enabling BSS mode, but I can't tell for sure if it works (I know the
card switches modes, but I can't verify that it joins a service set
correctly).

This driver was written using information gleaned from the Lucent HCF Light
library, which is an API library designed to simplify driver development
for devices based on the Lucent Hermes chip. Unfortunately, the HCF Light
is missing certain features (like 802.11 frame encapsulation!) which are
available only in the proprietary complete HCF code, which is not available
to the public. This driver uses none of the HCF Light code: it's very ugly
and contaminated by the GPL. IP and ARP packets are encapsulated as 802.11
frames, everything else is encapsulated as 802.3.

(It would be easier to just get the Hermes programming manual, but that's
not publically available either. For those who are wondering, the Linux
WaveLAN/IEEE driver uses the proprietary HCF code, which is provided in
object code form only. So much for supporting open source sofware.)

Multicast filter support is implemented, however it appears that the
filter doesn't work: programming in one IP mutlicast group enables them
all.

25 years agoInstall the right man page.
Jordan K. Hubbard [Wed, 5 May 1999 07:09:21 +0000 (07:09 +0000)]
Install the right man page.

25 years agoRemove bt_softcs. It is no longer used.
Warner Losh [Wed, 5 May 1999 06:45:10 +0000 (06:45 +0000)]
Remove bt_softcs.  It is no longer used.

25 years agoRestore perror --> err changes.
Mark Murray [Wed, 5 May 1999 06:15:30 +0000 (06:15 +0000)]
Restore perror --> err changes.

25 years agoCorrect `sd', `st' -> `da', `sa'; add xref to scsi(4); remove xref
Joseph Koshy [Wed, 5 May 1999 04:01:40 +0000 (04:01 +0000)]
Correct `sd', `st' -> `da', `sa'; add xref to scsi(4); remove xref
to nca(4).

PR: docs/11493
Submitted by: "Steven G. Kargl" <kargl@troutmask.apl.washington.edu>

25 years agoSync with sys/i386/isa/fd.c revision 1.137.
KATO Takenori [Wed, 5 May 1999 03:43:00 +0000 (03:43 +0000)]
Sync with sys/i386/isa/fd.c revision 1.137.

25 years agoSync with sys/i386/conf/majors.i386 revision 1.71.
KATO Takenori [Wed, 5 May 1999 03:41:39 +0000 (03:41 +0000)]
Sync with sys/i386/conf/majors.i386 revision 1.71.

25 years agoCorrect prototype for tigetstr().
Joseph Koshy [Wed, 5 May 1999 02:51:45 +0000 (02:51 +0000)]
Correct prototype for tigetstr().

PR: docs/11489
Submitted by: Andrew <andrew@ugh.net.au>

25 years agoFixed missing parenthesis.
KATO Takenori [Wed, 5 May 1999 01:53:43 +0000 (01:53 +0000)]
Fixed missing parenthesis.

Submitted by: Takahashi Yoshihiro <nyan@dd.catv.ne.jp>

25 years agoTypo fix and a very minor wording adjustment.
Masafumi Max NAKANE [Wed, 5 May 1999 00:20:15 +0000 (00:20 +0000)]
Typo fix and a very minor wording adjustment.

25 years agoSubstantially update this to match reality. Also do a typo/facto pass over
Jordan K. Hubbard [Tue, 4 May 1999 23:27:07 +0000 (23:27 +0000)]
Substantially update this to match reality.  Also do a typo/facto pass over
Max's serial console instructions (this is good stuff, thanks Max!).

25 years agoClean up after another half-Danish.
Mike Smith [Tue, 4 May 1999 22:44:48 +0000 (22:44 +0000)]
Clean up after another half-Danish.

Add 'ad' as another known disk driver of the same kind as 'wd'.

25 years agoUse unit, not device_id as an argument to an old-style ISA interrupt
Doug Rabson [Tue, 4 May 1999 21:18:20 +0000 (21:18 +0000)]
Use unit, not device_id as an argument to an old-style ISA interrupt
handler. This fixes pnp interrupts and would have fixed pccard interrupts
but a workaround has been applied there.

This the sound driver problems which people have reported with new-bus.

25 years agoAdd procfs_rlimit.c so the kernel actually compiles.
Bill Fumerola [Tue, 4 May 1999 21:08:58 +0000 (21:08 +0000)]
Add procfs_rlimit.c so the kernel actually compiles.

PR: kern/11497
Submitted by: Jos Backus <jos.backus@nl.origin-it.com> (misordered patch)
Pointy Hat: phk, adrian

25 years agoMake `ed0' example match reality.
David E. O'Brien [Tue, 4 May 1999 21:04:01 +0000 (21:04 +0000)]
Make `ed0' example match reality.
Merge `ed1' and `ed2' examples as we don't ship with an `ed1' any more we
yet, the example of extra flags could be useful to someone.