]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoModify the transmit packet queuing strategy a bit to be a little less
Bill Paul [Mon, 19 Oct 1998 22:08:56 +0000 (22:08 +0000)]
Modify the transmit packet queuing strategy a bit to be a little less
agressive. With the old code, if a descriptor chain was already on its
way to the chip, xl_start() would try to splice new chains onto the end
of the current chain by stopping the transmitter, modifying the tail
pointer of the current chain to point to the head of the new chain, then
restart the transmitter. The manual says you're allowed to do this and
it works, but I'm not too keen on it anymore.

The new code waits until the eixsting chain has been sent and then
queues the next waiting chain in the 'transmit ok' handler.

Performance still looks good one way or the other.

25 years agoUse sysctl() rather than kvm_read() when getting information
Dima Ruban [Mon, 19 Oct 1998 20:48:08 +0000 (20:48 +0000)]
Use sysctl() rather than kvm_read() when getting information
about clock stuff from kernel.
This fixes xntpd on alpha.

25 years agoDefault to aout only on i386. Before we defaulted to elf only on
Warner Losh [Mon, 19 Oct 1998 20:09:17 +0000 (20:09 +0000)]
Default to aout only on i386.  Before we defaulted to elf only on
Alpha.  This is a minor, but important distinction.  Should be a no-op
to the install base.  If OBJFORMAT is set elsewhere, things work
exactly as they did before.

25 years agoMake gcore work for ELF.
John Polstra [Mon, 19 Oct 1998 19:42:18 +0000 (19:42 +0000)]
Make gcore work for ELF.

25 years agoMake configurable (a) whether to use int 0x13 extensions; (b) timeout
Robert Nordier [Mon, 19 Oct 1998 19:13:53 +0000 (19:13 +0000)]
Make configurable (a) whether to use int 0x13 extensions; (b) timeout
value.  Ignore failure to update mbr sector.  Disable interrupts while
using extended registers.  Default to F1 not F2.

25 years agoAdd entries for DiskOnChip2000 Flash device.
Jordan K. Hubbard [Mon, 19 Oct 1998 14:58:38 +0000 (14:58 +0000)]
Add entries for DiskOnChip2000 Flash device.
Submitted by: phk

25 years agoReclaim the 48k used by the first bootstrap stage by adding it to the malloc
Doug Rabson [Mon, 19 Oct 1998 09:12:41 +0000 (09:12 +0000)]
Reclaim the 48k used by the first bootstrap stage by adding it to the malloc
pool.

25 years agoImplement a hook to allow us to reclaim the memory used by the first stage
Doug Rabson [Mon, 19 Oct 1998 09:08:40 +0000 (09:08 +0000)]
Implement a hook to allow us to reclaim the memory used by the first stage
of the bootstrap (the bit which loads /boot/loader).

25 years agoWe're now in post-release; make the docs indicate as much.
Jordan K. Hubbard [Mon, 19 Oct 1998 08:49:50 +0000 (08:49 +0000)]
We're now in post-release; make the docs indicate as much.

25 years agoRename boot2 to loader.
Doug Rabson [Sun, 18 Oct 1998 19:05:07 +0000 (19:05 +0000)]
Rename boot2 to loader.

25 years agoR_ALPHA_RELATIVE relocations need to add the value to the existing memory
Doug Rabson [Sun, 18 Oct 1998 19:04:13 +0000 (19:04 +0000)]
R_ALPHA_RELATIVE relocations need to add the value to the existing memory
contents.

25 years agoAdd driver support for PCI fast ethernet adapters based on the
Bill Paul [Sun, 18 Oct 1998 16:24:34 +0000 (16:24 +0000)]
Add driver support for PCI fast ethernet adapters based on the
RealTek 8129/8139 chipset like I've been threatening. Update kernel
configs, userconfig.c, relnotes and sysinstall. No man page yet;
comming soon.

I consider this driver stable enough that I want to give it some
exposure in -current.

25 years agoCommented out aic.
KATO Takenori [Sun, 18 Oct 1998 15:55:20 +0000 (15:55 +0000)]
Commented out aic.

25 years agoSome cleanups and optimizations:
Peter Wemm [Sun, 18 Oct 1998 15:55:12 +0000 (15:55 +0000)]
Some cleanups and optimizations:
- Use the system headers method for Elf32/Elf64 symbol compatability
- get rid of the UPRINTF debugging.
- check the ELF header for compatability much more completely
- optimize the section mapper.  Use the same direct VM interfaces that
  imgact_aout.c and kern_exec.c use.
- Check the return codes from the vm_* functions better.  Some return
  KERN_* results, not an errno.
- prefault the page tables to reduce startup faults on page tables like
  a.out does.
- reset the segment protection to zero for each loop, otherwise each
  segment could get progressively more privs. (eg: if the first was
  read/write/execute, and the second was meant to be read/execute, the
  bug would make the second r/w/x too.  In practice this was not a
  problem because executables are normally laid out with text first.)
- Don't impose arbitary limits.  Use the limits on headers imposed by
  the need to fit them into one page.
- Remove unused switch() cases now that the verbose debugging is gone.

I've been using an earlier version of this for a month or so.
This sped up ELF exec speed a bit for me but I found it hard to get
consistant benchmarks when I tested it last (a few weeks ago).
I'm still bothered by the page read out of order caused by the
transition from data to bss. This which requires either part filling the
transition page or clearing the remainder.

25 years agoAdd an ELF_MACHINE_OK() macro for compatability with the Alpha version.
Peter Wemm [Sun, 18 Oct 1998 15:31:23 +0000 (15:31 +0000)]
Add an ELF_MACHINE_OK() macro for compatability with the Alpha version.

25 years agoUpdate SMP-GENERIC98 to something a bit more recent based on GENERIC98.
KATO Takenori [Sun, 18 Oct 1998 14:02:43 +0000 (14:02 +0000)]
Update SMP-GENERIC98 to something a bit more recent based on GENERIC98.

25 years agoSync with sys/i386/conf/GENERIC revision 1.125.
KATO Takenori [Sun, 18 Oct 1998 13:58:16 +0000 (13:58 +0000)]
Sync with sys/i386/conf/GENERIC revision 1.125.

25 years agoadjust to match current times.
Jordan K. Hubbard [Sun, 18 Oct 1998 13:56:49 +0000 (13:56 +0000)]
adjust to match current times.

25 years agoAdd a missing ++.
Poul-Henning Kamp [Sun, 18 Oct 1998 11:58:57 +0000 (11:58 +0000)]
Add a missing ++.

Noticed by: gcc via phk
Submitted by: Mike Spengler <mks@networkcs.com>

25 years agoAllow building games w/o an existing /usr/games.
David E. O'Brien [Sun, 18 Oct 1998 11:14:58 +0000 (11:14 +0000)]
Allow building games w/o an existing /usr/games.

25 years agoAdded missing dependency for bus_if.o on bus_if.h
Poul-Henning Kamp [Sun, 18 Oct 1998 10:57:55 +0000 (10:57 +0000)]
Added missing dependency for bus_if.o on bus_if.h

25 years agoAllow ``make BINDIR=/foo/games'' to work.
David E. O'Brien [Sun, 18 Oct 1998 09:49:02 +0000 (09:49 +0000)]
Allow ``make BINDIR=/foo/games'' to work.

25 years agoAllow ``make BINDIR=/foo/games'' to work.
David E. O'Brien [Sun, 18 Oct 1998 09:35:17 +0000 (09:35 +0000)]
Allow ``make BINDIR=/foo/games'' to work.

25 years agoInitialize earlier (ie: before npx) in case it makes any difference.
Peter Wemm [Sun, 18 Oct 1998 07:44:33 +0000 (07:44 +0000)]
Initialize earlier (ie: before npx) in case it makes any difference.
Print a message when it attaches to pmath_emulate if bootverbose so that
we can be sure it's online.

25 years agoPrint a message if bootverbose that the emulator is present in the kernel.
Peter Wemm [Sun, 18 Oct 1998 07:40:29 +0000 (07:40 +0000)]
Print a message if bootverbose that the emulator is present in the kernel.

Move the initialization before isa_configure() and npx, in case npx does
something to initialize the state of the emulator somehow.

I do not have any machines without a FPU so that I can test this with -
except an old 386sx motherboard in a box somewhere that might work...

25 years agoPR: 7583
Peter Hawkins [Sun, 18 Oct 1998 04:46:24 +0000 (04:46 +0000)]
PR: 7583
Submitted by: Dave Glowacki <dglo@SSEC.WISC.EDU>
Support input via stdin if no input file name specified explicitly

25 years agoOOPS!
Peter Wemm [Sun, 18 Oct 1998 04:36:58 +0000 (04:36 +0000)]
OOPS!
Supply the moduledata handle rather than the event dispatcher function.
This should explain the panic on boot problem that's been discussed in
-current at the moment.  Both machines had GNU_MATH_EMULATE.

25 years agoOnly process the ``PWD'' environmental var if built with "WANT_ENV_PWD"
David E. O'Brien [Sun, 18 Oct 1998 00:51:46 +0000 (00:51 +0000)]
Only process the ``PWD'' environmental var if built with "WANT_ENV_PWD"
defined.  Bash v2 sets PWD and it creates major problems for those of us
with /usr/src being a symlink.  See the lists for examples of the problems.

25 years agoTo clarify last commit, msmith says:
David E. O'Brien [Sat, 17 Oct 1998 22:20:06 +0000 (22:20 +0000)]
To clarify last commit, msmith says:
The reason the Alpha has less room than the i386 is because the bootstrap
is given a 256K mapping by the firmware; to add more requires extra work.

25 years ago"Fixed" a printf format error. Use bogus casts to avoid using %p so that
Bruce Evans [Sat, 17 Oct 1998 16:14:00 +0000 (16:14 +0000)]
"Fixed" a printf format error.  Use bogus casts to avoid using %p so that
the output doesn't change (unless the old format caused runtime errors).

25 years agoFixed printf format errors.
Bruce Evans [Sat, 17 Oct 1998 16:04:28 +0000 (16:04 +0000)]
Fixed printf format errors.

25 years agoDon't even think about using a sysctl to build osreldate.h, since this
Bruce Evans [Sat, 17 Oct 1998 15:52:48 +0000 (15:52 +0000)]
Don't even think about using a sysctl to build osreldate.h, since this
breaks cross-builds.  Just depend on ${.CURDIR}/../sys/conf/newvers.sh
existing.

Don't override the (correct) defaults for the depend, lint or tags target.

In LDIRS: fixed order-rot.  Don't comment out dead networking directories;
remove them.

25 years agoRestored all lost user targets that are supported by bsd.subdir.mk:
Bruce Evans [Sat, 17 Oct 1998 15:25:26 +0000 (15:25 +0000)]
Restored all lost user targets that are supported by bsd.subdir.mk:
checkdpadd, lint, maninstall, objlink, regress and tags.

Removed bogus user target cleanobj.  It is the non-recursive base of
the cleandir target, so it is not useful (or usable) here.

25 years agoFixed the type of a signal handler.
Bruce Evans [Sat, 17 Oct 1998 14:47:00 +0000 (14:47 +0000)]
Fixed the type of a signal handler.

25 years agoSet BINMODE to 444.
Robert Nordier [Sat, 17 Oct 1998 14:20:25 +0000 (14:20 +0000)]
Set BINMODE to 444.

25 years agoFixed printf format errors. Assume that uid_t's are representable as
Bruce Evans [Sat, 17 Oct 1998 14:08:12 +0000 (14:08 +0000)]
Fixed printf format errors.  Assume that uid_t's are representable as
unsigned longs.

25 years agoFixed printf format errors. Assume that time_t's are representable as
Bruce Evans [Sat, 17 Oct 1998 14:06:20 +0000 (14:06 +0000)]
Fixed printf format errors.  Assume that time_t's are representable as
longs.

25 years agoFixed missing libraries in DPADD.
Bruce Evans [Sat, 17 Oct 1998 13:55:22 +0000 (13:55 +0000)]
Fixed missing libraries in DPADD.

25 years agoHonour the default CFLAGS. This fixes a spew of warnings for compiling
Bruce Evans [Sat, 17 Oct 1998 13:48:37 +0000 (13:48 +0000)]
Honour the default CFLAGS.  This fixes a spew of warnings for compiling
unused static inlines in headers without -O.

Fixed missing libraries in DPADD.

Use .PATH instead of a symlink farm.

Removed bogus dependency of util.c on statetext.h (only util.o depends
on it).  Removed rule for building statetext.h.  The dependency was used
to get statetext.h built, but statetext.h is a non-generated source file
so it doesn't need building.  This fixes an annoying message for the null
build and use of `make' instead of ${MAKE} to do the null build.

Fixed some style bugs.

25 years agoDon't assume that time_t is long.
Bruce Evans [Sat, 17 Oct 1998 13:02:46 +0000 (13:02 +0000)]
Don't assume that time_t is long.

25 years agoDon't assume that time_t is long.
Bruce Evans [Sat, 17 Oct 1998 12:44:55 +0000 (12:44 +0000)]
Don't assume that time_t is long.

25 years agoWarn about (but process anyway) CBCP messages with an incorrect id.
Brian Somers [Sat, 17 Oct 1998 12:28:11 +0000 (12:28 +0000)]
Warn about (but process anyway) CBCP messages with an incorrect id.

25 years agoAdjust our CBCP RESP id so that it's the same as the REQ.
Brian Somers [Sat, 17 Oct 1998 12:28:09 +0000 (12:28 +0000)]
Adjust our CBCP RESP id so that it's the same as the REQ.
Submitted by: Tom <root@majestix.cmr.no>

25 years agoIf the client asks for CBCP, then replies with CBCP_NONUM
Brian Somers [Sat, 17 Oct 1998 12:28:08 +0000 (12:28 +0000)]
If the client asks for CBCP, then replies with CBCP_NONUM
to our REQ, allow no dialback - but only if we've
``set cbcp *''..
Suggested by: Andrzej Tobola <san@koziolek.lublin.top.pl>

25 years agoExplain what the ``*'' means for a CBCP phone number
Brian Somers [Sat, 17 Oct 1998 12:28:06 +0000 (12:28 +0000)]
Explain what the ``*'' means for a CBCP phone number
in ``show link''.

25 years agoMove help displays left one column to avoid problems with
Brian Somers [Sat, 17 Oct 1998 12:28:05 +0000 (12:28 +0000)]
Move help displays left one column to avoid problems with
broken terminals that can't handle 80 columns followed by
a linefeed.
Pointed out by: bde@FreeBSD.org

25 years agoWarn if we encounter an unindented line that isn't
Brian Somers [Sat, 17 Oct 1998 12:28:03 +0000 (12:28 +0000)]
Warn if we encounter an unindented line that isn't
terminated with a ``:''.

25 years agoWarn that the specific chap protocol isn't supported when
Brian Somers [Sat, 17 Oct 1998 12:28:02 +0000 (12:28 +0000)]
Warn that the specific chap protocol isn't supported when
rejecting CHAP because of byte 5.
Suggested by: Daniel O'Callaghan <danny@hilink.com.au>

25 years agoDon't add /usr/games to $PATH. Adding it here can only work for the
Bruce Evans [Sat, 17 Oct 1998 11:56:20 +0000 (11:56 +0000)]
Don't add /usr/games to $PATH.  Adding it here can only work for the
`make world' case, and only made a difference for the evil NOTOOLS case
of `make world' since games tools are installed in ${TOOLROOT}/usr/bin
if they are built, but the PR was for normal builds.  This is fixed
in rev.1.14 of src/games/fortune/datfiles/Makefile.

PR: 7936

25 years agoFixed building without /usr/games in $PATH. Just add /usr/games
Bruce Evans [Sat, 17 Oct 1998 11:44:28 +0000 (11:44 +0000)]
Fixed building without /usr/games in $PATH.  Just add /usr/games
to the _end_ of $PATH before using tools that are normally in
/usr/games.  I broke this in rev.1.9 by assuming that these tools
are in $PATH (as they are for `make world' unless NOTOOLS is set).
Revs.1.10-1.13 of this file had various wrong fixes.  Rev.1.18
of src/Makefile.inc1 has an incomplete fix.

PR: 7936

Fixed comments about what to comment out to [not] install the
potentially offensive fortunes.  This should be configured using
an ifdef.

Fixed missing dependencies of fortunes.dat on fortunes, etc.

Removed bogus dependency of `all' on source files.

Fixed some style bugs.

25 years agoFix some glitches in the input routine:
Robert Nordier [Sat, 17 Oct 1998 11:25:05 +0000 (11:25 +0000)]
Fix some glitches in the input routine:
Don't display a \b if not acting on it.
Don't process binary zero chars (which result from pressing function
keys, etc. on the PC).

25 years agoHaving probed the keyboard, turn off the -P flag.
Robert Nordier [Sat, 17 Oct 1998 10:35:15 +0000 (10:35 +0000)]
Having probed the keyboard, turn off the -P flag.

25 years agoUpdated the error message for EXDEV to match recent kernel fixes.
Bruce Evans [Sat, 17 Oct 1998 09:56:32 +0000 (09:56 +0000)]
Updated the error message for EXDEV to match recent kernel fixes.
Fixed nearby indentation.

Use RAW_PART instead of RAWPARTITION.

25 years agoTreat all options as toggles (ie. -c -c is the same no -c). Since
Robert Nordier [Sat, 17 Oct 1998 09:50:09 +0000 (09:50 +0000)]
Treat all options as toggles (ie. -c -c is the same no -c).  Since
the boot.config settings are persistent, this seems to provide a
useful override capability, and should break only on broken
boot.config's.

Output a cosmetic newline if booting with no input.

25 years agoOverload the correct errno for attempts to set an in-core label with
Bruce Evans [Sat, 17 Oct 1998 09:46:42 +0000 (09:46 +0000)]
Overload the correct errno for attempts to set an in-core label with
a raw partition at a nonzero offset (EINVAL should have been EXDEV;
DIOCSDINFO was broken, and DIOCWDINFO was broken because it depended
on DIOCSDINFO).

A zero offset for the raw partition should probably be enforced in
setdisklabel(), and DIOCWDINFO should probably always be handled by
first calling setdisklabel() so that writedisklabel() doesn't need to
enforce it, but this has never been done; dsioctl() has a special
check.  Changes in this commit are limited to dsioctl() to preserve
bug for bug compatibility in drivers that don't use the slice code
(notably the ccd driver, which allows setting a bogus label in
DIOCWDINFO and doesn't undo the setting when writedisklabel() fails).

25 years agobiosboot compatibility fix:
Robert Nordier [Sat, 17 Oct 1998 09:16:01 +0000 (09:16 +0000)]
biosboot compatibility fix:
If we have a boot.config command, display it.

25 years agobiosboot compatibility fix:
Robert Nordier [Sat, 17 Oct 1998 09:01:13 +0000 (09:01 +0000)]
biosboot compatibility fix:
Read boot.help before parsing boot.config.  We were parsing
boot.config first, which could result in boot.help being read
from a different location (or not found), which would probably
just cause surprise, without being useful.

25 years agoBacked out previous commit. It broke fsck again. See rev.1.22 and the
Bruce Evans [Sat, 17 Oct 1998 08:03:52 +0000 (08:03 +0000)]
Backed out previous commit.  It broke fsck again.  See rev.1.22 and the
references there, and rev.1.38 of sys/ufs/ufs/ufs_disksubr.c.

25 years agoUse only the correct raw partition for writing labels. Don't use the
Bruce Evans [Sat, 17 Oct 1998 07:49:04 +0000 (07:49 +0000)]
Use only the correct raw partition for writing labels.  Don't use the
partition that the label ioctl is being done on just because it has
offset 0, since there is no guarantee that such a partition is large
enough to contain the label.  Don't use the wrong raw partition (0
instead of RAW_PART).

This fixes problems rewriting bizarre labels (with a nonzero offset
for the 'a' partition) in newfs(8).  Such labels shouldn't normally
be used, but creating them was allowed if the ioctl was done on the
raw partition, and sysinstall creates them if the root partition isn't
allocated first.

Note that allowing write access to a partition other than the one that
has been checked for write access doesn't increase security holes
significantly, since write access to any partition already allows
changing the in-core label.

This fix should be in 3.0R.  Rev.1.26 of newfs/newfs.c shouldn't be
in 3.0R.

25 years agoFix paths. This sorts out the remaining p5-* port-creation hassles.
Mark Murray [Sat, 17 Oct 1998 07:10:43 +0000 (07:10 +0000)]
Fix paths. This sorts out the remaining p5-* port-creation hassles.

25 years ago3.0-RELEASE and now the post-release 3.0-CURRENT were being declared
Peter Wemm [Sat, 17 Oct 1998 05:40:46 +0000 (05:40 +0000)]
3.0-RELEASE and now the post-release 3.0-CURRENT were being declared
'too old' to run ELF.  Check the kernel version properly.

25 years agoDon't rewrite the disk label. The type field is already set correctly
Jordan K. Hubbard [Sat, 17 Oct 1998 04:19:29 +0000 (04:19 +0000)]
Don't rewrite the disk label.  The type field is already set correctly
and we don't use the frags info, so why bother?  More to the point, it
seems to result in an EXDEV error when the label is written out and we
lose because of it (don't know why though).  This is a work-around and
is marked as such.

25 years agoLoad the first page of the file and use the headers in it. This should
Peter Wemm [Sat, 17 Oct 1998 03:06:38 +0000 (03:06 +0000)]
Load the first page of the file and use the headers in it.  This should
avoid the need to seek back to offset zero which is causing trouble on
the Alpha with a gzipped kernel.

25 years agoSome directories would like to install things into /modules
Peter Wemm [Sat, 17 Oct 1998 01:21:35 +0000 (01:21 +0000)]
Some directories would like to install things into /modules

25 years agoComplete the description of sio flag 0x40.
Alexander Langer [Fri, 16 Oct 1998 22:26:19 +0000 (22:26 +0000)]
Complete the description of sio flag 0x40.

Reviewed by: bde

25 years agoAdd the dpt(4) man page.
Kenneth D. Merry [Fri, 16 Oct 1998 22:00:55 +0000 (22:00 +0000)]
Add the dpt(4) man page.

Reviewed by: ken
Submitted by: gibbs

25 years agoUpdate the bt(4) man page for CAM.
Kenneth D. Merry [Fri, 16 Oct 1998 21:33:34 +0000 (21:33 +0000)]
Update the bt(4) man page for CAM.

Reviewed by: ken
Submitted by: gibbs

25 years agoAdd in the adv(4) and adw(4) man pages for the AdvanSys narrow and wide
Kenneth D. Merry [Fri, 16 Oct 1998 21:12:07 +0000 (21:12 +0000)]
Add in the adv(4) and adw(4) man pages for the AdvanSys narrow and wide
SCSI controllers, respectively.

Once these drivers are tested on the alpha, these man pages can probably be
moved up a directory to reflect the fact that they're architecture
independent.

An mdoc guru should probably look at the AUTHORS sections in both of these
pages -- the .An macro seems to cause strange spacing problems.

Reviewed by: ken
Submitted by: gibbs

25 years agoTurn of disk debugging by default (suppress annoying "partition marked as
Mike Smith [Fri, 16 Oct 1998 19:26:11 +0000 (19:26 +0000)]
Turn of disk debugging by default (suppress annoying "partition marked as
unused" diagnostic on floppies).

25 years agoReduce the sbrk() increment from 64K to 4K. There's not much room on the
Mike Smith [Fri, 16 Oct 1998 19:23:37 +0000 (19:23 +0000)]
Reduce the sbrk() increment from 64K to 4K.  There's not much room on the
Alpha, and wasting potentially 64K-4 bytes of RAM just isn't an option.

25 years agoSplit up rule for ${COMMONHDRS} so that `make -j N' doesn't create the
Bruce Evans [Fri, 16 Oct 1998 16:53:15 +0000 (16:53 +0000)]
Split up rule for ${COMMONHDRS} so that `make -j N' doesn't create the
6 headers up to 6 times concurrently.  Debugged by Don Lewis
(Don.Lewis@tsc.tdk.com).

Fixed some disorder.

25 years agoCommented out bs driver.
KATO Takenori [Fri, 16 Oct 1998 16:22:35 +0000 (16:22 +0000)]
Commented out bs driver.

25 years agoGenerate code to interpret ioctl numbers for all ioctls defined in
Bruce Evans [Fri, 16 Oct 1998 15:33:17 +0000 (15:33 +0000)]
Generate code to interpret ioctl numbers for all ioctls defined in
headers under /usr/include, not just for the ones in <sys/ioctl.h>.
The generated file includes all headers that seem to define ioctls,
so build errors will probably occur if headers become less self-
sufficient than they are already.  This is a feature.  Build errors
shall not be fixed by adding more includes here.

Optionally generate a case statement instead of a list of if
statements.  This source must be edited to change this.  The case
statement should be non-optional.  It currently can't be, because
many ioctl numbers are not unique.

25 years agoRemoved unused/unusable "ha" keyword. Device classes should never have
Bruce Evans [Fri, 16 Oct 1998 15:08:30 +0000 (15:08 +0000)]
Removed unused/unusable "ha" keyword.  Device classes should never have
been visible here.

Fixed disordering of keyword table by "cam".

25 years agoUpdate to include all supported chips.
Justin T. Gibbs [Fri, 16 Oct 1998 15:06:26 +0000 (15:06 +0000)]
Update to include all supported chips.

Add a table comparing the features of all aic7xxx chips.

Clean up the discussion of SCB paging.

Reviewed by: Kenneth Merry <ken@FreeBSD.org>

25 years agoDon't generate unused file ioconf.h.
Bruce Evans [Fri, 16 Oct 1998 15:00:18 +0000 (15:00 +0000)]
Don't generate unused file ioconf.h.

Didn't bump CONFIGVERS, since ioconf.h was already unused when
CONFIGVERS was last bumped (although essentially the same (CAM)
commit batch that bumped CONFIGVERS also added bogus includes of
ioconf.h).

25 years agoFixed missing directories in the libraries target:
Bruce Evans [Fri, 16 Oct 1998 14:22:54 +0000 (14:22 +0000)]
Fixed missing directories in the libraries target:
- in the elf case, csu must be built and installed before any shared
  libraries.  It wasn't, but we usually used a stale version that
  happened to work.  E.g., in the !NOTOOLS case we used the version
  built and installed by the bootstrap-libraries target.  Only cross
  building was completely broken.
- the shared libmd must be built and installed before any shared
  libraries that link to it.  It wasn't, but we sometimes used a stale
  version that happened to work, as above.  For elf, this caused
  bogus linkage of the target shared libatm and libopie with the host
  static libmd.  It isn't clear what this actually breaks, except for
  cross compiling.  For aout, the shared libmd is not built at all, so
  all shared libraries linked to libmd may be broken.  The linker
  reports them by spewing RRS warnings.

Note that building src/lib early and building subdirs of src/lib in
the correct order in src/lib/Makefile doesn't help, since the subdirs
are all built before any are installed.

Fixed bitrot in the comments about the ordering requirements.

25 years agoUse the same boot.conf file for alpha, now that it works.
Jordan K. Hubbard [Fri, 16 Oct 1998 10:17:29 +0000 (10:17 +0000)]
Use the same boot.conf file for alpha, now that it works.

25 years agofixup for alpha.
Jordan K. Hubbard [Fri, 16 Oct 1998 10:14:21 +0000 (10:14 +0000)]
fixup for alpha.

25 years agofixup for the alpha. tag slid forward, since it's benign for the x86.
Jordan K. Hubbard [Fri, 16 Oct 1998 10:13:09 +0000 (10:13 +0000)]
fixup for the alpha.  tag slid forward, since it's benign for the x86.

25 years agoBack to -current again.
Jordan K. Hubbard [Fri, 16 Oct 1998 06:55:36 +0000 (06:55 +0000)]
Back to -current again.

25 years agopost-tag version bump. We're now back at 3.0-current!
Jordan K. Hubbard [Fri, 16 Oct 1998 06:55:07 +0000 (06:55 +0000)]
post-tag version bump.  We're now back at 3.0-current!

25 years agoadd note about CD-220E drives.
Jordan K. Hubbard [Fri, 16 Oct 1998 04:47:12 +0000 (04:47 +0000)]
add note about CD-220E drives.

25 years agoUpdate SMP-GENERIC to something a bit more recent based on GENERIC.
Peter Wemm [Fri, 16 Oct 1998 04:44:05 +0000 (04:44 +0000)]
Update SMP-GENERIC to something a bit more recent based on GENERIC.

Approved by: jkh

25 years agoUpdate the aha(4) and ahb(4) man pages for CAM.
Kenneth D. Merry [Fri, 16 Oct 1998 04:41:38 +0000 (04:41 +0000)]
Update the aha(4) and ahb(4) man pages for CAM.

Reviewed by: ken
Submitted by: gibbs

25 years agoMissed this file.
Peter Wemm [Fri, 16 Oct 1998 04:31:35 +0000 (04:31 +0000)]
Missed this file.

25 years agoSample initial set of kld-ified modules. Not all have been completely
Peter Wemm [Fri, 16 Oct 1998 04:30:52 +0000 (04:30 +0000)]
Sample initial set of kld-ified modules.  Not all have been completely
converted yet.  These are more of a starting point.  This is NOT connected
to the parent Makefile.

OK'ed by jkh (who is ever so patiently waiting)

25 years agoUpdate the scsi(4) man page for CAM, and add in links so that it appears
Kenneth D. Merry [Fri, 16 Oct 1998 04:28:39 +0000 (04:28 +0000)]
Update the scsi(4) man page for CAM, and add in links so that it appears
as cam(4) as well.

This includes a description of all the generic CAM kernel options, as
well as a description of some of the CAM debugging printf options.

25 years agoBump freebsd version #.
Jordan K. Hubbard [Fri, 16 Oct 1998 04:28:04 +0000 (04:28 +0000)]
Bump freebsd version #.

25 years agoAdd acknowledgements.
Jordan K. Hubbard [Fri, 16 Oct 1998 04:26:00 +0000 (04:26 +0000)]
Add acknowledgements.

25 years agoUpdate the sa(4) man page for CAM. This was repository-copied from the
Kenneth D. Merry [Fri, 16 Oct 1998 04:06:21 +0000 (04:06 +0000)]
Update the sa(4) man page for CAM.  This was repository-copied from the
st(4) man page.

Take out most of the sd(4) and st(4) man pages and point to the new
da(4) and sa(4) man pages.

Add sa.4 to the makefile.

Reviewed by: ken
Submitted by: gibbs

25 years ago*gulp*. Jordan specifically OK'ed this..
Peter Wemm [Fri, 16 Oct 1998 03:55:01 +0000 (03:55 +0000)]
*gulp*.  Jordan specifically OK'ed this..

This is the bulk of the support for doing kld modules.  Two linker_sets
were replaced by SYSINIT()'s.  VFS's and exec handlers are self registered.
kld is now a superset of lkm.  I have converted most of them, they will
follow as a seperate commit as samples.
This all still works as a static a.out kernel using LKM's.

25 years agoFix some bugs in link_aout.c caused by using uninitialized malloc space.
Peter Wemm [Fri, 16 Oct 1998 03:45:35 +0000 (03:45 +0000)]
Fix some bugs in link_aout.c caused by using uninitialized malloc space.

Pre-Approved by:  jkh

25 years agoUpdate the kld glue
Peter Wemm [Fri, 16 Oct 1998 03:32:03 +0000 (03:32 +0000)]
Update the kld glue

Pre-Approved by: jkh

25 years agoAdd the da(4) man page.
Kenneth D. Merry [Fri, 16 Oct 1998 03:29:49 +0000 (03:29 +0000)]
Add the da(4) man page.

25 years agoUpdate the da(4) man page for CAM. The da.4 man page was repository-copied
Kenneth D. Merry [Fri, 16 Oct 1998 03:28:12 +0000 (03:28 +0000)]
Update the da(4) man page for CAM.  The da.4 man page was repository-copied
from the sd.4 man page.

This includes a discussion of write caching and its effects.

Reviewed by: ken
Submitted by: gibbs

25 years agoCommented out example of changing the default kernel format with warning.
Peter Wemm [Fri, 16 Oct 1998 03:26:54 +0000 (03:26 +0000)]
Commented out example of changing the default kernel format with warning.

25 years ago"fix" the gzipped kernel load problem by having the loader check that it
Peter Wemm [Fri, 16 Oct 1998 03:04:15 +0000 (03:04 +0000)]
"fix" the gzipped kernel load problem by having the loader check that it
can seek back to the first PT_LOAD and doing a close/reopen if it cannot.
This is because the first PT_LOAD section includes the ELF headers.
This fixes gzipped kernels on the i386, it should solve mike's problem
for the Alpha.

25 years agoAdjust for -RELEASE
Jordan K. Hubbard [Fri, 16 Oct 1998 02:27:07 +0000 (02:27 +0000)]
Adjust for -RELEASE

25 years agoChange to 3.0-RELEASE in anticipation of the tag.
Jordan K. Hubbard [Fri, 16 Oct 1998 02:08:20 +0000 (02:08 +0000)]
Change to 3.0-RELEASE in anticipation of the tag.