]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoMove makectx() after kdb_cpu_trap(), so the PCB will have possible MD
Marcel Moolenaar [Fri, 17 Sep 2004 22:27:23 +0000 (22:27 +0000)]
Move makectx() after kdb_cpu_trap(), so the PCB will have possible MD
corrections made to the trapframe. This is more logical.

20 years agoFix an issue with ng_tty which (ab)used the tty->t_sc field which is
Poul-Henning Kamp [Fri, 17 Sep 2004 22:26:27 +0000 (22:26 +0000)]
Fix an issue with ng_tty which (ab)used the tty->t_sc field which is
reserved for the device drivers:

Add a t_lsc field for line discipline private use.

20 years agoProvide our own FPSWA definitions, instead of depending on the Intel
Marcel Moolenaar [Fri, 17 Sep 2004 22:19:41 +0000 (22:19 +0000)]
Provide our own FPSWA definitions, instead of depending on the Intel
EFI headers and put them all in <machine/fpu.h>. The Intel EFI headers
conflict with the Intel ACPI headers (duplicate type definitions), so
are being phased out in the kernel.

20 years agoDon't attempt to install the (obsolete and disconnected) Early Adopters
Bruce A. Mah [Fri, 17 Sep 2004 21:46:17 +0000 (21:46 +0000)]
Don't attempt to install the (obsolete and disconnected) Early Adopters
Guide on release media.

20 years agoRemove useless inclusion of <machine/fpu.h>
Marcel Moolenaar [Fri, 17 Sep 2004 20:42:45 +0000 (20:42 +0000)]
Remove useless inclusion of <machine/fpu.h>

20 years ago- properly capitalize NetFlow
Gleb Smirnoff [Fri, 17 Sep 2004 20:09:59 +0000 (20:09 +0000)]
- properly capitalize NetFlow
- small grammar fix

Submitted by: ru

20 years ago- Remove advertising clause from copyright [1]
Gleb Smirnoff [Fri, 17 Sep 2004 19:58:03 +0000 (19:58 +0000)]
- Remove advertising clause from copyright [1]
- Change my email to glebius@FreeBSD.org

Requested by: ru [1]

20 years agoFix compiler warnings, when __stdcall is #defined, by adding explicit casts.
Bruce M Simpson [Fri, 17 Sep 2004 19:54:26 +0000 (19:54 +0000)]
Fix compiler warnings, when __stdcall is #defined, by adding explicit casts.
These normally only manifest if the ndis compat module is statically
compiled into a kernel image by way of 'options NDISAPI'.

Submitted by: Dmitri Nikulin
Approved by: wpaul
PR: kern/71449
MFC after: 1 week

20 years agoA man page for Netflow node.
Gleb Smirnoff [Fri, 17 Sep 2004 19:30:39 +0000 (19:30 +0000)]
A man page for Netflow node.

English lessons by: julian
mdoc(7) lessons by: ru
Approved by: julian (mentor)

20 years agoThe early-adopter doesn't exist in HEAD anymore. The source is there,
Giorgos Keramidas [Fri, 17 Sep 2004 19:28:46 +0000 (19:28 +0000)]
The early-adopter doesn't exist in HEAD anymore.  The source is there,
but it's disconnected from the build.  Remove it from the architecture
independent release documentation set, because it breaks "make release"
when NODOC is undefined.

Reviewed by: hrs

20 years agoFix compilation again.
Poul-Henning Kamp [Fri, 17 Sep 2004 19:24:58 +0000 (19:24 +0000)]
Fix compilation again.

20 years agoThe long dead hand of the past has pushed forward useless bits in this
Warner Losh [Fri, 17 Sep 2004 18:50:38 +0000 (18:50 +0000)]
The long dead hand of the past has pushed forward useless bits in this
driver.  Trim its fingernails by removing some useless bits before
fixing the 'thread not terminated on detach' problem.

o dmacnt is no longer used now that we allocate at attach time.  Remove
  it from struct fdc_data.
o ISPNP was only ever set, but never tested.  It used to be used for the
  allocation routines to change how it allocated resources.  Since that's
  no longer necessary, retire the flag.
o ISPCMICA was only ever tested, but never set.  GC it.  This removes
  a special case in determining the drive type.  The drive type is
  now set in fdc_pcmcia.c, so the hack isn't needed anymore.  Sadly,
  this isn't tested with a Y-E Data pcmcia floppy drive because there
  are a number of other issues that preclude it from working.
o Fix ifdef for reading from the rtc.  I'm of the opinion that this ifdef
  should be moved into fdc_isa.c, but not today as ideally there'd be
  other fixes to the probing of children.  So now we just read it on
  i386 ! pc98 (there's no #define for MACHINE_ARCH, just MACHINE, hence
  this slightly inelegant kludge) and amd64.  The PC98 exclusion likely
  isn't meaningful since pc98 uses a different driver, but will be when
  merging of the pc98 floppy code into this driver is complete (this is the
  other reason I think this block of code belongs outside fdc.c).

All of these changes are safe to MT5.

20 years agoTeach swap1 script how to remove added swap devices on system shutdown.
Pawel Jakub Dawidek [Fri, 17 Sep 2004 17:58:19 +0000 (17:58 +0000)]
Teach swap1 script how to remove added swap devices on system shutdown.
Without this change, if one had a swap-on-mirror configuration, gmirror
will rebuild mirror component(s) on boot, because they are dirty (they
were open on shutdown).

20 years agoUse correct format string.
Lukas Ertl [Fri, 17 Sep 2004 17:08:44 +0000 (17:08 +0000)]
Use correct format string.

Submitted by:   ssouhlal

20 years agoDon't refer to the (now disconnected) early adopter's guide anymore.
Bruce A. Mah [Fri, 17 Sep 2004 16:29:46 +0000 (16:29 +0000)]
Don't refer to the (now disconnected) early adopter's guide anymore.

20 years agoSince early-adopter has been disconnected from the build, don't try
Bruce A. Mah [Fri, 17 Sep 2004 16:16:31 +0000 (16:16 +0000)]
Since early-adopter has been disconnected from the build, don't try
to install it to a RELNOTESng snapshot.

20 years agoFix typo.
Diomidis Spinellis [Fri, 17 Sep 2004 14:53:46 +0000 (14:53 +0000)]
Fix typo.

MFC after: 2 days

20 years agoUserland control utility for ng_netflow.
Gleb Smirnoff [Fri, 17 Sep 2004 12:02:22 +0000 (12:02 +0000)]
Userland control utility for ng_netflow.

Approved by: julian (mentor)

20 years agoOverhaul ucom serial driver by using generic stuff instead of homerolled
Poul-Henning Kamp [Fri, 17 Sep 2004 11:53:45 +0000 (11:53 +0000)]
Overhaul ucom serial driver by using generic stuff instead of homerolled
all over the place.

20 years agoAdd ttyopen and ttyclose functions which will do the right stuff for
Poul-Henning Kamp [Fri, 17 Sep 2004 11:43:35 +0000 (11:43 +0000)]
Add ttyopen and ttyclose functions which will do the right stuff for
most if not all of our tty drivers in the future.

Centralizing this stuff enables us to remove about 100 lines of
almost but not quite perfectly copy&paste code from each tty driver.

20 years agoforward declare struct cdev, not cdevsw;
Poul-Henning Kamp [Fri, 17 Sep 2004 11:39:56 +0000 (11:39 +0000)]
forward declare struct cdev, not cdevsw;

20 years agoAllocate tty at attach time instead of open time.
Poul-Henning Kamp [Fri, 17 Sep 2004 11:04:57 +0000 (11:04 +0000)]
Allocate tty at attach time instead of open time.

20 years agoBe slightly less bogus in struct tty allocation.
Poul-Henning Kamp [Fri, 17 Sep 2004 11:02:53 +0000 (11:02 +0000)]
Be slightly less bogus in struct tty allocation.

20 years agoUse tty->t_sc, ttyalloc() and lock/init termios from struct tty.
Poul-Henning Kamp [Fri, 17 Sep 2004 10:59:17 +0000 (10:59 +0000)]
Use tty->t_sc, ttyalloc() and lock/init termios from struct tty.

20 years agoInclude <sys/malloc.h> to satisfy new isa_dma stuff.
Poul-Henning Kamp [Fri, 17 Sep 2004 10:55:01 +0000 (10:55 +0000)]
Include <sys/malloc.h> to satisfy new isa_dma stuff.

20 years agoPass the idea of the make(1) binary to use down to newvers.sh.
Ruslan Ermilov [Fri, 17 Sep 2004 09:17:33 +0000 (09:17 +0000)]
Pass the idea of the make(1) binary to use down to newvers.sh.
This is necessary so source upgrades use the correct binary.

MFC after: 3 days

For the record: Problem spotted by Scott Long, who mentioned
that source upgrades from 4.7 to recent 5.x and 6.0 are broken.
Detailed analysis shows that 4.7 has a broken make(1) binary.
A breakage was fixed in RELENG_4 in make/main.c,v 1.35.2.7 by
imp@, though the commit log erroneously stated "MFC 1.68"
while in fact it should have been spelled as "MFC 1.67".

20 years agoDo not exit after printing usage, give geom(8) a chance to show standard
Pawel Jakub Dawidek [Fri, 17 Sep 2004 08:41:15 +0000 (08:41 +0000)]
Do not exit after printing usage, give geom(8) a chance to show standard
commands.

20 years ago- Show all commands in usage, even if those commands are not available.
Pawel Jakub Dawidek [Fri, 17 Sep 2004 08:38:53 +0000 (08:38 +0000)]
- Show all commands in usage, even if those commands are not available.
- When command is not available, just say it, instead of says, that command
  is unknown. Old behaviour seems to be too confusing.

20 years agoAdd various stuff to struct tty and surounding areas in preparation
Poul-Henning Kamp [Fri, 17 Sep 2004 08:34:43 +0000 (08:34 +0000)]
Add various stuff to struct tty and surounding areas in preparation
for getting stuff from P4::phk_tty into -current.

20 years agoUse ttyalloc() instead of ttymalloc(NULL)
Poul-Henning Kamp [Fri, 17 Sep 2004 07:28:07 +0000 (07:28 +0000)]
Use ttyalloc() instead of ttymalloc(NULL)

20 years agoBring kernel time variable and access function docs up to date by
Brooks Davis [Fri, 17 Sep 2004 07:24:18 +0000 (07:24 +0000)]
Bring kernel time variable and access function docs up to date by
updating the list of variables and mentioning the bin(up)time functions.

20 years agoDocument the struct if_data members ifi_datalen, ifi_epoch, and
Brooks Davis [Fri, 17 Sep 2004 06:49:43 +0000 (06:49 +0000)]
Document the struct if_data members ifi_datalen, ifi_epoch, and
ifi_link_state.

Reminded by: ru

20 years agoAdd my birth date to our calendar.
Xin LI [Fri, 17 Sep 2004 06:36:01 +0000 (06:36 +0000)]
Add my birth date to our calendar.

Approved by: murray (mentor)

20 years agoAdd ttyalloc() which in due time will be the successor to ttymalloc(),
Poul-Henning Kamp [Fri, 17 Sep 2004 06:13:47 +0000 (06:13 +0000)]
Add ttyalloc() which in due time will be the successor to ttymalloc(),
but without the "struct tty *" argument.

20 years agoFurther refine some #ifs:
David Schultz [Fri, 17 Sep 2004 05:15:33 +0000 (05:15 +0000)]
Further refine some #ifs:
- Simplify the logic by using __GNUC_PREREQ__.
  Suggested by stefanf.
- Make math.h compile with old (pre-8.0) versions of icc.
  Submitted by sf [sic].

20 years agoCommit patch to supress spurious link change events. Apparently, with
Bill Paul [Fri, 17 Sep 2004 04:58:17 +0000 (04:58 +0000)]
Commit patch to supress spurious link change events. Apparently, with
copper NICs, a link change event is posted whenever MII autopolling is
toggled off and on, which happens whenever someone calls
bge_miibus_readreg() or bge_miibus_writereg() to access the PHY
registers. This means anytime someone called the SIOCGIFMEDIA ioctl
on a bge interface, the link would reset. Even a simple "ifconfig bge0"
would do it, though other apps like dhclient or the PPPoE daemon could
trigger it as well. An obvious symptom of this problem is lots of
"bgeX: gigabit link up" messages appearing on the console for no
apparent reason.

Through experimentation, I determined that when a real link change
event occurs, the BGE_MIMODE_AUTOPOLL in the BGE_MI_MODE register
is always set, so now if we have a copper NIC and an link change
event occurs and the BGE_MIMODE_AUTOPOLL bit is clear, we ignore
the event.

Note that this does not apply to the original BCM5700 chip since we
use a different method for sensing link changes with that chip (the
status block method was broken), nor to fiber optic NICs since they
don't use the GMII PHY access registers.

20 years agoFix two ugly errors:
Tim Kientzle [Fri, 17 Sep 2004 04:39:07 +0000 (04:39 +0000)]
Fix two ugly errors:
1. The correct cutoff for large uid/gid handling is 1<<18, not 1<<20.
2. Limit the uid/gid in the 'x' extension header (where numeric extensions
are not permitted) to 1<<18, but use the correct value in the regular
header (where numeric extensions are permitted).
Thanks to: Dan Nelson
MFC after: 3 days

20 years agoHandle _FDE results of 5 bytes (vs. 5 uint32_t's). BIOS vendors find yet
Nate Lawson [Fri, 17 Sep 2004 04:14:38 +0000 (04:14 +0000)]
Handle _FDE results of 5 bytes (vs. 5 uint32_t's).  BIOS vendors find yet
another way to misinterpret the spec.  Also, always fall back to the hints
probe on any attach failure, not just when _FDE fails.

Thanks to imp and scottl for finding this.

Tested by: rwatson (minimally)
MFC after: 5 days

20 years agoDestroy global tapmtx when the if_tap module is unloaded.
Robert Watson [Fri, 17 Sep 2004 03:55:50 +0000 (03:55 +0000)]
Destroy global tapmtx when the if_tap module is unloaded.

RELENG_5 candidated.

20 years agoBreak out altq_enable/disable from DIOC{START,STOP}ALTQ into seprate
Max Laier [Fri, 17 Sep 2004 02:15:05 +0000 (02:15 +0000)]
Break out altq_enable/disable from DIOC{START,STOP}ALTQ into seprate
functions that can be called from enable/disable pf as well. This improves
switching from non-altq ruleset to altq ruleset (and the other way 'round)
by a great deal and makes pfctl act like the user would except it to.

PR: kern/71746
Tested by: Aurilien "beorn" Rougemont (PR submitter)
MFC after: 3 days

20 years agoBump document date forgotten in last commit.
Simon L. B. Nielsen [Thu, 16 Sep 2004 22:26:45 +0000 (22:26 +0000)]
Bump document date forgotten in last commit.

20 years agoChange tabs to whitespaces.
Gleb Smirnoff [Thu, 16 Sep 2004 21:33:56 +0000 (21:33 +0000)]
Change tabs to whitespaces.

Noticed by: ru
Pointy hat to: glebius

20 years ago- Make md(4) 64-bit clean.
Pawel Jakub Dawidek [Thu, 16 Sep 2004 21:32:13 +0000 (21:32 +0000)]
- Make md(4) 64-bit clean.
  After this change it should be possible to use very big md(4) devices.
- Clean up and simplify the code a bit.
- Use humanize_number(3) to print size of md(4) devices.
- Add 't' suffix which stands for terabyte.
- Make '-S' to really work with all types of devices.
- Other minor changes.

20 years agoInstall netflow includes.
Gleb Smirnoff [Thu, 16 Sep 2004 20:42:03 +0000 (20:42 +0000)]
Install netflow includes.

Approved by: julian (mentor)

20 years agoAutogenerate device listings for ed(4)
Christian Brueffer [Thu, 16 Sep 2004 20:38:28 +0000 (20:38 +0000)]
Autogenerate device listings for ed(4)

MFC after: 3 days

20 years agoThere is no need to keep 'npage' value inside our softc structure,
Pawel Jakub Dawidek [Thu, 16 Sep 2004 20:38:11 +0000 (20:38 +0000)]
There is no need to keep 'npage' value inside our softc structure,
it is only used in one function. While doing so, change its type to
vm_ooffset_t.
We are still limited for swap-backed devices to 16TB on 32-bit architectures
where PAGE_SIZE is 4096 bytes.

20 years agoAdd ed(4)
Christian Brueffer [Thu, 16 Sep 2004 20:35:51 +0000 (20:35 +0000)]
Add ed(4)

MFC after: 3 days

20 years agoAttach ng_netflow to kernel build.
Gleb Smirnoff [Thu, 16 Sep 2004 20:35:28 +0000 (20:35 +0000)]
Attach ng_netflow to kernel build.

Approved by: julian (mentor)

20 years ago- add a HARDWARE section based on the hardware notes
Christian Brueffer [Thu, 16 Sep 2004 20:33:32 +0000 (20:33 +0000)]
- add a HARDWARE section based on the hardware notes
- capitalize Ethernet
- minor fixes

MFC after: 3 days

20 years agoAdaptec ANA-6944/TX is supported
Christian Brueffer [Thu, 16 Sep 2004 20:29:19 +0000 (20:29 +0000)]
Adaptec ANA-6944/TX is supported

MFC after: 3 days

20 years agoA netgraph node implementing Netflow version 5.
Gleb Smirnoff [Thu, 16 Sep 2004 20:24:23 +0000 (20:24 +0000)]
A netgraph node implementing Netflow version 5.

Supported by: Bestcom ISP, Rinet ISP
Approved by: julian (mentor)

20 years agoAllocate tty at attach time rather than open time.
Poul-Henning Kamp [Thu, 16 Sep 2004 19:46:23 +0000 (19:46 +0000)]
Allocate tty at attach time rather than open time.

Use the init/lock*in/out fields in struct tty and remove them from our softc.

Use tty->t_sc to find out softc when convenient.

Mostly OK'ed by: bde

20 years ago- Do not use bio_pblkno as it is going away anyway.
Pawel Jakub Dawidek [Thu, 16 Sep 2004 19:42:17 +0000 (19:42 +0000)]
- Do not use bio_pblkno as it is going away anyway.
- Prefer bio_length than bio_bcount.

20 years agoFirst wakeup, then unlock.
Pawel Jakub Dawidek [Thu, 16 Sep 2004 18:59:19 +0000 (18:59 +0000)]
First wakeup, then unlock.

20 years agoType 'int' is too small for 'i' and 'lastp' variables. Use proper type,
Pawel Jakub Dawidek [Thu, 16 Sep 2004 18:56:20 +0000 (18:56 +0000)]
Type 'int' is too small for 'i' and 'lastp' variables. Use proper type,
which is vm_pindex_t (unsigned 64bit on i386).

20 years agoAdd missing comma
Christian Brueffer [Thu, 16 Sep 2004 18:46:21 +0000 (18:46 +0000)]
Add missing comma

Submitted by: ceri
MFC after: 3 days

20 years agoFix an out of bounds write during the initialization of the PF_INET protocol
Andre Oppermann [Thu, 16 Sep 2004 18:33:39 +0000 (18:33 +0000)]
Fix an out of bounds write during the initialization of the PF_INET protocol
family to the ip_protox[] array.  The protocol number of IPPROTO_DIVERT is
larger than IPPROTO_MAX and was initializing memory beyond the array.
Catch all these kinds of errors by ignoring protocols that are higher than
IPPROTO_MAX or 0 (zero).

Add more comments ip_init().

20 years agoThere is no such manual page in FreeBSD.
Pawel Jakub Dawidek [Thu, 16 Sep 2004 18:32:58 +0000 (18:32 +0000)]
There is no such manual page in FreeBSD.

20 years agoDo not traverse list of snapshots if there isn't one.
Poul-Henning Kamp [Thu, 16 Sep 2004 17:28:56 +0000 (17:28 +0000)]
Do not traverse list of snapshots if there isn't one.

Found by: scottl

20 years agoDon't print a warning message if the _CRS value is empty. This is already
Nate Lawson [Thu, 16 Sep 2004 17:19:16 +0000 (17:19 +0000)]
Don't print a warning message if the _CRS value is empty.  This is already
covered by other printfs under ACPI_DEBUG and is not a failure case.

MFC after: 3 days

20 years agoNew release notes:
Hiroki Sato [Thu, 16 Sep 2004 17:18:55 +0000 (17:18 +0000)]
New release notes:
crontab(1) now properly prompts when an entry is written
in the incorrect format.

20 years agoWe don't have any providers of `beforenetlkm' in FreeBSD. Remove the
Giorgos Keramidas [Thu, 16 Sep 2004 17:04:20 +0000 (17:04 +0000)]
We don't have any providers of `beforenetlkm' in FreeBSD.  Remove the
dependency to it from our rc.d scripts.

Approved by: mtm

20 years agoFix requirement of `network' to `NETWORK' because the former isn't
Giorgos Keramidas [Thu, 16 Sep 2004 17:03:12 +0000 (17:03 +0000)]
Fix requirement of `network' to `NETWORK' because the former isn't
provided by any rc.d script.

Approved by: mtm

20 years agoDisconnect early-adopter/ from the build.
Hiroki Sato [Thu, 16 Sep 2004 16:46:37 +0000 (16:46 +0000)]
Disconnect early-adopter/ from the build.

20 years agoMissed a place where snapshots were allocated in my last commit to
Poul-Henning Kamp [Thu, 16 Sep 2004 15:58:18 +0000 (15:58 +0000)]
Missed a place where snapshots were allocated in my last commit to
this file.

20 years agog_nop_create: destroy newly created provider in case of errors.
Max Khon [Thu, 16 Sep 2004 15:28:48 +0000 (15:28 +0000)]
g_nop_create: destroy newly created provider in case of errors.

20 years agoImplement cancellation points in libc interfaces, as specified by POSIX.
Mike Makonnen [Thu, 16 Sep 2004 13:55:46 +0000 (13:55 +0000)]
Implement cancellation points in libc interfaces, as specified by POSIX.

20 years agoUse the tty->t_sc field to find our softc.
Poul-Henning Kamp [Thu, 16 Sep 2004 12:07:25 +0000 (12:07 +0000)]
Use the tty->t_sc field to find our softc.

20 years agoAdd fields to struct tty for the initial/lock * tty/cua states.
Poul-Henning Kamp [Thu, 16 Sep 2004 10:43:47 +0000 (10:43 +0000)]
Add fields to struct tty for the initial/lock * tty/cua states.

20 years agoIgnore leading '_' in model name returned by devices.
Søren Schmidt [Thu, 16 Sep 2004 09:35:01 +0000 (09:35 +0000)]
Ignore leading '_' in model name returned by devices.

This make "_NEC" devices appear as "NEC" which is more corrent.
The reason is tha NEC originally screwed up on the byteorder in the
model string, so now that they have realized that they prefixed the '_'
so that not every ATA driver on the planet would call them "EN C" :)

20 years agoclean up thread runq accounting a bit.
Julian Elischer [Thu, 16 Sep 2004 07:12:59 +0000 (07:12 +0000)]
clean up thread runq accounting a bit.

MFC after: 3 days

20 years agodefault: case shouldn't set an error yet either.
Warner Losh [Thu, 16 Sep 2004 06:37:05 +0000 (06:37 +0000)]
default: case shouldn't set an error yet either.

20 years agoReserve the DMA memory at attach time, rather than at use. While
Warner Losh [Thu, 16 Sep 2004 06:34:00 +0000 (06:34 +0000)]
Reserve the DMA memory at attach time, rather than at use.  While
reserving it at use time is more miserly, low memory (< 16MB)
evaporates quickly on many systems, so there may not be any suitable
buffers available.  This specifically doesn't use the newer, fancier
isa_dma_init to ease merging to 5.

Reviewed by: tegge, phk

20 years agowhitespace change
Poul-Henning Kamp [Thu, 16 Sep 2004 06:00:52 +0000 (06:00 +0000)]
whitespace change

20 years agoe specific code to revert a partial add ot teh run queue, not
Julian Elischer [Thu, 16 Sep 2004 05:37:40 +0000 (05:37 +0000)]
e specific code to revert a partial add ot teh run queue, not
remrunqueue() which can't handle a partially added thread.

MFC after: 1 week

20 years agoIf the timeout handler runs and notices that commands are timed out, check
Scott Long [Thu, 16 Sep 2004 02:37:40 +0000 (02:37 +0000)]
If the timeout handler runs and notices that commands are timed out, check
the firmware status register on the card to see if the firmware is still
running.  There is no way to recover from this, but at least it can give
a hint as whether the car has crashed (which happens all too often).

MFC after: 3 days

20 years agoFor the moment, treat failures to attach floppy drives as non-fatal
Warner Losh [Thu, 16 Sep 2004 01:51:21 +0000 (01:51 +0000)]
For the moment, treat failures to attach floppy drives as non-fatal
errors for the attachment process for the floppy controller.  This is
a band-aide because it doesn't try any of the fallback methods when
_FDE isn't long enough, but should be sufficient for people
experiencing the dreaded mutex not initialized panic.

20 years agoUpdate path to c++ internals includes for the 3.4 location.
Peter Wemm [Thu, 16 Sep 2004 01:41:11 +0000 (01:41 +0000)]
Update path to c++ internals includes for the 3.4 location.
No longer use make -k since the libraries target should build now.
Remove hacks for gnu/lib/libregex.

20 years agoBe explicit about which regex.h file we are including. This reduces the
Peter Wemm [Thu, 16 Sep 2004 01:39:13 +0000 (01:39 +0000)]
Be explicit about which regex.h file we are including.  This reduces the
need for hack workarounds in tools/lib32/build32.sh.

20 years agoAdd libmagic to the explicit build-tools list.
Peter Wemm [Thu, 16 Sep 2004 00:22:59 +0000 (00:22 +0000)]
Add libmagic to the explicit build-tools list.

20 years agoBackout the code which tries to use undocumented way to determine if
Maxim Sobolev [Wed, 15 Sep 2004 23:47:17 +0000 (23:47 +0000)]
Backout the code which tries to use undocumented way to determine if
fm801 has sound capabilities or not. Unfortunately this code doesn't
work as expected.

Submitted by: many
MFC after: 3 days

20 years agoFix breakpoint handling for i386.
Julian Elischer [Wed, 15 Sep 2004 23:26:49 +0000 (23:26 +0000)]
Fix breakpoint handling for i386.
not sure yet about 5.x... MFC if needed.
Also fixes small problems with examining some registers and
some specific gdb transfer problems.

As the patch says:
This is not a pretty patch and only meant as a temporary
fix until a better solution is committed.

PR: i386/71715
Submitted by: Stephan Uphoff <ups@tree.com>
MFC after: 1 week

20 years agoMake libcrypto/libssh cross compile
Peter Wemm [Wed, 15 Sep 2004 23:15:53 +0000 (23:15 +0000)]
Make libcrypto/libssh cross compile

20 years agoFix typo in rev 1.2 "-DNOINFOall" should be "-DNOINFO all"
Peter Wemm [Wed, 15 Sep 2004 23:01:28 +0000 (23:01 +0000)]
Fix typo in rev 1.2 "-DNOINFOall" should be "-DNOINFO all"

20 years agoRemove unused B_WRITEINPROG flag
Poul-Henning Kamp [Wed, 15 Sep 2004 21:49:22 +0000 (21:49 +0000)]
Remove unused B_WRITEINPROG flag

20 years agoSimplify initialization of va_null a little bit.
Poul-Henning Kamp [Wed, 15 Sep 2004 21:42:03 +0000 (21:42 +0000)]
Simplify initialization of va_null a little bit.

20 years agoComment lines may have leading whitespace.
Dag-Erling Smørgrav [Wed, 15 Sep 2004 21:10:46 +0000 (21:10 +0000)]
Comment lines may have leading whitespace.

PR: 71773
Submitted by: Antoine Brodin <antoine.brodin@laposte.net>

20 years agoundent some functions a bit.
Poul-Henning Kamp [Wed, 15 Sep 2004 21:08:58 +0000 (21:08 +0000)]
undent some functions a bit.

20 years agostylistic polishing.
Poul-Henning Kamp [Wed, 15 Sep 2004 20:54:23 +0000 (20:54 +0000)]
stylistic polishing.

20 years agoClarify some comments for the M_FASTFWD_OURS case in ip_input().
Andre Oppermann [Wed, 15 Sep 2004 20:17:03 +0000 (20:17 +0000)]
Clarify some comments for the M_FASTFWD_OURS case in ip_input().

20 years agoRemove the last two global variables that are used to store packet state while
Andre Oppermann [Wed, 15 Sep 2004 20:13:26 +0000 (20:13 +0000)]
Remove the last two global variables that are used to store packet state while
it travels through the IP stack.  This wasn't much of a problem because IP
source routing is disabled by default but when enabled together with SMP and
preemption it would have very likely cross-corrupted the IP options in transit.

The IP source route options of a packet are now stored in a mtag instead of the
global variable.

20 years agoMy fingers keep typing 'burncd data foo fixate eject' instead of 'burncd -e
Dag-Erling Smørgrav [Wed, 15 Sep 2004 19:03:35 +0000 (19:03 +0000)]
My fingers keep typing 'burncd data foo fixate eject' instead of 'burncd -e
data foo fixate'.  Humor them.

MFC after: 4 weeks

20 years agoPark my birthday truck in this free space.
Koop Mast [Wed, 15 Sep 2004 19:03:27 +0000 (19:03 +0000)]
Park my birthday truck in this free space.

Approved by: pav (mentor)

20 years agoMechanical whitespace cleanup.
Dag-Erling Smørgrav [Wed, 15 Sep 2004 19:01:08 +0000 (19:01 +0000)]
Mechanical whitespace cleanup.

20 years agoUpdate to reflect recent changes such as the fields for O(log n) address
Alan Cox [Wed, 15 Sep 2004 18:41:10 +0000 (18:41 +0000)]
Update to reflect recent changes such as the fields for O(log n) address
space allocation.

Submitted by: Mark W. Krentel

20 years agoTry harder to get back to being a non threaded process.
Julian Elischer [Wed, 15 Sep 2004 18:39:09 +0000 (18:39 +0000)]
Try harder to get back to being a non threaded process.

Submitted by: DavidXu
MFC after: 3 days

20 years agoFlush the queue of minor fixes to pst.
Søren Schmidt [Wed, 15 Sep 2004 15:39:28 +0000 (15:39 +0000)]
Flush the queue of minor fixes to pst.

20 years agoAdd new a function isa_dma_init() which returns an errno when it fails
Poul-Henning Kamp [Wed, 15 Sep 2004 12:09:50 +0000 (12:09 +0000)]
Add new a function isa_dma_init() which returns an errno when it fails
and which takes a M_WAITOK/M_NOWAIT flag argument.

Add compatibility isa_dmainit() macro which whines loudly if
isa_dma_init() fails.

Problem uncovered by: tegge

20 years agoRemove now unused #include files.
Poul-Henning Kamp [Wed, 15 Sep 2004 12:02:35 +0000 (12:02 +0000)]
Remove now unused #include files.

20 years ago#include <isa/isavar.h> instead of <i386/isa/isa_dma.h>
Poul-Henning Kamp [Wed, 15 Sep 2004 11:58:34 +0000 (11:58 +0000)]
#include <isa/isavar.h> instead of <i386/isa/isa_dma.h>