]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoThis commit was generated by cvs2svn to compensate for changes in r40767,
joerg [Fri, 30 Oct 1998 18:20:54 +0000 (18:20 +0000)]
This commit was generated by cvs2svn to compensate for changes in r40767,
which included commits to RCS files with non-trunk default branches.

25 years agoThis is enigma, aka. crypt(1). It has suppsedly been taken from
joerg [Fri, 30 Oct 1998 18:20:54 +0000 (18:20 +0000)]
This is enigma, aka. crypt(1).  It has suppsedly been taken from
Cryptbreakers Workbench.  While arguably a rather weak encryption,
it's in some use in the Internet still, and provided by a bunch of
other Unix systesms, so we include it here for compatibility.

Silently agreed by: core

25 years agoRename a function name so that it doesn't conflict with a future system call.
dg [Fri, 30 Oct 1998 16:17:50 +0000 (16:17 +0000)]
Rename a function name so that it doesn't conflict with a future system call.

25 years agoUnwire everything to the inactive queue in order to preserve LRU ordering.
dg [Fri, 30 Oct 1998 14:53:54 +0000 (14:53 +0000)]
Unwire everything to the inactive queue in order to preserve LRU ordering.

25 years agoFix a typo, and remove verbose error reporting when in
luigi [Fri, 30 Oct 1998 11:53:58 +0000 (11:53 +0000)]
Fix a typo, and remove verbose error reporting when in
promisc mode (they are annoying in normal mode
as well so i am really tempted to remove them unconditionally...)

25 years agoUse "KB" instead of "Kb" for KiloBytes, consistently with other
luigi [Fri, 30 Oct 1998 10:57:09 +0000 (10:57 +0000)]
Use "KB" instead of "Kb" for KiloBytes, consistently with other
drivers and common practices.

25 years ago(1) Add new variables LOOP_VAR and LOOP_OPTIONS to implement building several
asami [Fri, 30 Oct 1998 08:28:02 +0000 (08:28 +0000)]
(1) Add new variables LOOP_VAR and LOOP_OPTIONS to implement building several
    packages from a single port.  LOOP_VAR is the name of the variable
    and LOOP_OPTIONS is a space-separated list of values it should
    take.  When these are set, the target "package-loop" will go
    through a clean and package loop for all the options.  The
    "package-loop" target is defined as "package" when LOOP_VAR is not
    defined, so if you are in the business for building packages, you
    should use "package-loop" all the time.  (This target is added to
    bsd.port.subdir.mk too.)

    Also, the "describe" target prints out multiple lines so that all
    options will go into the INDEX.  (In other words, if you define
    these variables, INDEX is going to look real silly if you don't
    put ${${LOOP_VAR}} in PKGNAME.)

Seconded by: obrien ("ANYTHING")

(2) Turn off regexp support for LIB_DEPENDS.  It is a fixed string of
    the form <NAME>.<VER> now.
Tested by: several rounds of complete package builds

(3) Check checksum even if NO_EXTRACT is defined.

(4) Cosmetic fix for message in MANUAL_PACKAGE_BUILD case.

25 years agoFix an uninitialized variable in the `dangerously dedicated disk' case.
luoqi [Fri, 30 Oct 1998 07:15:52 +0000 (07:15 +0000)]
Fix an uninitialized variable in the `dangerously dedicated disk' case.

25 years ago``MAKEDEV ccd3'' is now consistant with many of the other devices in that
obrien [Fri, 30 Oct 1998 06:02:48 +0000 (06:02 +0000)]
``MAKEDEV ccd3'' is now consistant with many of the other devices in that
*ccd{0,1,2}* will be created.

25 years agoAdd the ability to specify where on the at_shutdown queue a handler is
msmith [Fri, 30 Oct 1998 05:41:15 +0000 (05:41 +0000)]
Add the ability to specify where on the at_shutdown queue a handler is
installed.

Remove cpu_power_down, and replace it with an entry at the end of the
SHUTDOWN_FINAL queue in the only place it's used (APM).

Submitted by: Some ideas from Bruce Walter <walter@fortean.com>

25 years agoDo not disable the ISA compatibility window if it is the same value as
gibbs [Fri, 30 Oct 1998 02:06:44 +0000 (02:06 +0000)]
Do not disable the ISA compatibility window if it is the same value as
that set for our PCI IO address space.  This can happen on the BT-946.

25 years agoPut back MFS_ROOT now that it's been fixed correctly.
jkh [Fri, 30 Oct 1998 01:37:33 +0000 (01:37 +0000)]
Put back MFS_ROOT now that it's been fixed correctly.

25 years agoFix this correctly - check if mfs_getimage() succeeds before settings
jkh [Fri, 30 Oct 1998 01:36:40 +0000 (01:36 +0000)]
Fix this correctly - check if mfs_getimage() succeeds before settings
the MFS root unconditionally, just as on the x86.
Prompted by: msmith

25 years agoTake MFS_ROOT out for now; for some reason, the alpha GENERIC chokes
jkh [Fri, 30 Oct 1998 01:17:42 +0000 (01:17 +0000)]
Take MFS_ROOT out for now; for some reason, the alpha GENERIC chokes
and dies if it can't find the MFS root whereas the x86 one seems to sail
past.  Looking at the code, I can't see how either one works, so I'm
confused. :)

25 years agoMake the BTX loader much more quiet about what it's doing. This removes
msmith [Fri, 30 Oct 1998 00:11:35 +0000 (00:11 +0000)]
Make the BTX loader much more quiet about what it's doing.  This removes
most of the "what the (*^&%(*^ was that?" stuff that normally flies by.

25 years agoDon't call PacketAliasIn() when we turn around packets destined
brian [Thu, 29 Oct 1998 23:47:10 +0000 (23:47 +0000)]
Don't call PacketAliasIn() when we turn around packets destined
for our interface address.  We're about to call ip_Input()
anyway, and ip_Input() does the PacketAliasIn().
Stack trace provided by: Cameron Grant <gandalf@vilnya.demon.co.uk>

25 years agoPrevent buffer overflow in getpwnam()
msmith [Thu, 29 Oct 1998 23:16:24 +0000 (23:16 +0000)]
Prevent buffer overflow in getpwnam()

PR: bin/8176
Submitted by: Archie Cobbs <archie@whistle.com>

25 years agotx(4) - misspelled word
nate [Thu, 29 Oct 1998 22:54:14 +0000 (22:54 +0000)]
tx(4) - misspelled word

PR:  misc/8491
Submitted by: billf@jade.chc-chimes.com

25 years agoAdd support for -Wformat consistency checking between format strings
jdp [Thu, 29 Oct 1998 22:18:47 +0000 (22:18 +0000)]
Add support for -Wformat consistency checking between format strings
and their argument lists for the err(3) family of functions.  Note,
I intentionally used __printflike instead of __printf0like for
warnx.  Although a NULL format string is legal for that function, it
doesn't make any sense.

25 years agoClarify processing of the string argument by psignal().
rnordier [Thu, 29 Oct 1998 22:17:46 +0000 (22:17 +0000)]
Clarify processing of the string argument by psignal().

25 years ago``MAKEDEV bpf3'' is now consistant with many of the other devices in that
obrien [Thu, 29 Oct 1998 22:16:29 +0000 (22:16 +0000)]
``MAKEDEV bpf3'' is now consistant with many of the other devices in that
bpf{0,1,2} will be created.

25 years agoPer request from ache, change the numbering scheme of __FreeBSD_cc_version.
jdp [Thu, 29 Oct 1998 22:11:08 +0000 (22:11 +0000)]
Per request from ache, change the numbering scheme of __FreeBSD_cc_version.
Its form is now like __FreeBSD_version, with the FreeBSD revision in
the top digits.

25 years ago(Whoops: make it better instead of worse this time). Clarify
rnordier [Thu, 29 Oct 1998 22:08:59 +0000 (22:08 +0000)]
(Whoops: make it better instead of worse this time).  Clarify
processing of the string argument by perror().

25 years agoClarify processing of the string argument by perror().
rnordier [Thu, 29 Oct 1998 21:59:38 +0000 (21:59 +0000)]
Clarify processing of the string argument by perror().

25 years agoCorrect the reporting of the queue full condition so that the XPT layer
gibbs [Thu, 29 Oct 1998 17:41:34 +0000 (17:41 +0000)]
Correct the reporting of the queue full condition so that the XPT layer
can properly throttle tags.

Add diagnostic printfs for firmware result codes that we encounter but
don't know anything about.

25 years agoBump VERSREQ to make ``config'' happy.
dima [Thu, 29 Oct 1998 17:09:36 +0000 (17:09 +0000)]
Bump VERSREQ to make ``config'' happy.

Obtained from: Makefile.i386

25 years agoCheck for a zero-length as well as a NULL string argument.
rnordier [Thu, 29 Oct 1998 14:40:20 +0000 (14:40 +0000)]
Check for a zero-length as well as a NULL string argument.

25 years agoUpdate help files to reflect changes in vm utility.
abial [Thu, 29 Oct 1998 12:46:31 +0000 (12:46 +0000)]
Update help files to reflect changes in vm utility.

25 years agoFix conversion from pages to kB's.
abial [Thu, 29 Oct 1998 12:38:06 +0000 (12:38 +0000)]
Fix conversion from pages to kB's.

Noticed by: luigi
Obtained from: sources for vmstat(8)

25 years agoOops, rev.1.167 made the device number checking in bdevvp() too strict
bde [Thu, 29 Oct 1998 11:50:32 +0000 (11:50 +0000)]
Oops, rev.1.167 made the device number checking in bdevvp() too strict
for mfs root mounts.  Don't require major 255 to be in bdevsw[].

25 years agoThe man page implies that the string argument to psignal() may be
rnordier [Thu, 29 Oct 1998 11:39:39 +0000 (11:39 +0000)]
The man page implies that the string argument to psignal() may be
NULL, in line with perror(3).  However, the code presently checks only
for a zero-length string.  Check for both.

25 years agoFixed editing error. Pointed out by bde.
dg [Thu, 29 Oct 1998 11:04:22 +0000 (11:04 +0000)]
Fixed editing error. Pointed out by bde.

25 years agoIncrease the granularity of the debugging messages.
obrien [Thu, 29 Oct 1998 10:38:57 +0000 (10:38 +0000)]
Increase the granularity of the debugging messages.

25 years agoAdd some debugging statments (only if DEBUG opt is set), to try pin down
obrien [Thu, 29 Oct 1998 10:18:49 +0000 (10:18 +0000)]
Add some debugging statments (only if DEBUG opt is set), to try pin down
the hangs many of us are experiencing.

25 years agoDon't install the lp.4 -> plip.4 links, since the important one of them
bde [Thu, 29 Oct 1998 09:57:40 +0000 (09:57 +0000)]
Don't install the lp.4 -> plip.4 links, since the important one of them
clobbers the new plip.4 non-link.

25 years agoRemove the V_SAVEMETA flag, nothing uses it any more now that msdosfs and
peter [Thu, 29 Oct 1998 09:51:28 +0000 (09:51 +0000)]
Remove the V_SAVEMETA flag, nothing uses it any more now that msdosfs and
ext2fs call vtruncbuf() directly.  This simplifies and cleans up
vinvalbuf() a little.

25 years agoerror return assignment was less than ideal. Fix the part that caused
peter [Thu, 29 Oct 1998 09:44:12 +0000 (09:44 +0000)]
error return assignment was less than ideal.  Fix the part that caused
warnings to be the same as the ffs code.  Previously, any error from
the UFS_UPDATE() call was lost (I think).

25 years ago`vector xxxintr' should cause a syntax error.
bde [Thu, 29 Oct 1998 09:42:05 +0000 (09:42 +0000)]
`vector xxxintr' should cause a syntax error.

25 years agoFixed missing link and disordering in previous commit.
bde [Thu, 29 Oct 1998 09:35:10 +0000 (09:35 +0000)]
Fixed missing link and disordering in previous commit.

25 years agoUse vtruncbuf() to clean out cached blocks on a file shorten rather than
peter [Thu, 29 Oct 1998 09:30:52 +0000 (09:30 +0000)]
Use vtruncbuf() to clean out cached blocks on a file shorten rather than
the more expensive vinvalbuf(), based on the FFS version of the same
routine.  I don't have any ext2fs filesystems to test this on.

25 years agoUse vtruncbuf() rather than vinvalbuf() when shortening files.
peter [Thu, 29 Oct 1998 09:29:17 +0000 (09:29 +0000)]
Use vtruncbuf() rather than vinvalbuf() when shortening files.

25 years agoMake the definition of __printf0like() vary according to the value
jdp [Thu, 29 Oct 1998 04:41:24 +0000 (04:41 +0000)]
Make the definition of __printf0like() vary according to the value
of __FreeBSD_cc_version, so that its use is safe with older compilers
and with non-FreeBSD compilers.

25 years agoAdd a new predefined preprocessor symbol "__FreeBSD_cc_version",
jdp [Thu, 29 Oct 1998 04:27:06 +0000 (04:27 +0000)]
Add a new predefined preprocessor symbol "__FreeBSD_cc_version",
with a numeric value that describes the feature level of the
compiler.  This can be used to check for the presence/absence of
FreeBSD-specific compiler features.  The value is a decimal number
whose digits have the form VRRRRFF, where:

V    = Compiler vendor.  0 (elided) means gcc.
RRRR = Vendor's version number, e.g., 2721 for the current
       gcc version (2.7.2.1).
FF   = FreeBSD-specific revision level.  00 means the stock
       compiler from the vendor.

The value of "__FreeBSD_cc_version" is hard-coded in
"src/contrib/gcc/config/i386/freebsd.h" and must be incremented
when new FreeBSD-specific compiler features are added.  I considered
simply picking up the value of FreeBSD_version from <osreldate.h>.
But that would break cross compiles of gcc.

PR: Part of the fix for gnu/8452
Suggested by: bde

25 years agoRemove some redundant (#if 0'd) code.
brian [Thu, 29 Oct 1998 02:12:54 +0000 (02:12 +0000)]
Remove some redundant (#if 0'd) code.

25 years agoA small fragment of new ISA framework: manifest constants for the resources
wollman [Thu, 29 Oct 1998 01:56:23 +0000 (01:56 +0000)]
A small fragment of new ISA framework: manifest constants for the resources
implemented by the i386 root nexus.

25 years agoA Perl5 version of makedevops.sh. Fixes a bug in the sh/awk version
wollman [Thu, 29 Oct 1998 01:52:59 +0000 (01:52 +0000)]
A Perl5 version of makedevops.sh.  Fixes a bug in the sh/awk version
which really annoyed me at the time but I don't think bites me any
more.

25 years agoThe new resource manager, hopefully in a reasonably stable form.
wollman [Thu, 29 Oct 1998 01:48:36 +0000 (01:48 +0000)]
The new resource manager, hopefully in a reasonably stable form.

25 years agoUnlink old contents of package when extracting new.
jkh [Wed, 28 Oct 1998 22:44:24 +0000 (22:44 +0000)]
Unlink old contents of package when extracting new.
Submitted by: pst

25 years agoChange the way unmounting happens to guarantee that the
rvb [Wed, 28 Oct 1998 20:31:13 +0000 (20:31 +0000)]
Change the way unmounting happens to guarantee that the
client programs are allowed to finish up (coda_call is
forced to complete) and release their locks.  Thus there
is a reasonable chance that the vflush implicit in the
unmount will not get hung on held locks.

25 years agoUpdate
rvb [Wed, 28 Oct 1998 19:39:12 +0000 (19:39 +0000)]
Update

25 years agoVenus must be passed O_CREAT flag on VOP_OPEN iff this is
rvb [Wed, 28 Oct 1998 19:33:50 +0000 (19:33 +0000)]
Venus must be passed O_CREAT flag on VOP_OPEN iff this is
a creat so that we can will allow a mode 444 file to be
written into.  Sync with the latest coda.h and deal with
collateral damage.

25 years agoSimplify the "is a character ready" test, so that we don't return a false
msmith [Wed, 28 Oct 1998 19:24:15 +0000 (19:24 +0000)]
Simplify the "is a character ready" test, so that we don't return a false
negative for keypresses with zero ascii values.  This is in line with the
comconsole test, rather than being more ambitious.

Submitted by: rnordier

25 years agoCorect info about help files being stored in ar archive.
abial [Wed, 28 Oct 1998 16:42:51 +0000 (16:42 +0000)]
Corect info about help files being stored in ar archive.

Submitted by: Eric P. Scott <eps@sirius.com>

25 years agoAdd missing psm0 dev entry.
abial [Wed, 28 Oct 1998 16:26:29 +0000 (16:26 +0000)]
Add missing psm0 dev entry.

Submitted by: Eric P. Scott <eps@sirius.com>

25 years agoAdded TODO list. Automate changing of the "Last updated" field.
abial [Wed, 28 Oct 1998 16:16:31 +0000 (16:16 +0000)]
Added TODO list. Automate changing of the "Last updated" field.

25 years agoFixed wrong comments in and about vm_page_deactivate().
dg [Wed, 28 Oct 1998 13:41:43 +0000 (13:41 +0000)]
Fixed wrong comments in and about vm_page_deactivate().

25 years agoAdded a second argument, "activate" to the vm_page_unwire() call so that
dg [Wed, 28 Oct 1998 13:37:02 +0000 (13:37 +0000)]
Added a second argument, "activate" to the vm_page_unwire() call so that
the caller can select either inactive or active queue to put the page on.

25 years agoSync with sys/i386/conf/Makefile.i386 revision 1.127.
kato [Wed, 28 Oct 1998 10:48:37 +0000 (10:48 +0000)]
Sync with sys/i386/conf/Makefile.i386 revision 1.127.

25 years agoClarify a rather ambiguous debugging message.
jkh [Wed, 28 Oct 1998 10:37:54 +0000 (10:37 +0000)]
Clarify a rather ambiguous debugging message.

25 years agoAdd BUS_DEBUG option in opt_bus.h
dfr [Wed, 28 Oct 1998 08:37:10 +0000 (08:37 +0000)]
Add BUS_DEBUG option in opt_bus.h

25 years agoBack out previous work-around for "vmstat -i" failing on ELF
jdp [Wed, 28 Oct 1998 06:41:24 +0000 (06:41 +0000)]
Back out previous work-around for "vmstat -i" failing on ELF
kernels.  A better fix is now committed to "src/lib/libc/gen/nlist.c"
and "src/usr.sbin/kvm_mkdb/nlist.c".

25 years agoHandle ELF symbols better. This fixes "vmstat -i" for the case
jdp [Wed, 28 Oct 1998 06:39:41 +0000 (06:39 +0000)]
Handle ELF symbols better.  This fixes "vmstat -i" for the case
where "/var/db/kvm_kernel.db" exists.

Note, kvm_mkdb tries to be clever, and skips rebuilding the database
if it thinks it's already up to date.  To see the effects of this
fix, you may need to manually delete "/var/db/kvm_kernel.db" and
then run "kvm_mkdb".

25 years agoHandle ELF symbols better. This fixes "vmstat -i" for the case
jdp [Wed, 28 Oct 1998 06:37:49 +0000 (06:37 +0000)]
Handle ELF symbols better.  This fixes "vmstat -i" for the case
where "/var/db/kvm_kernel.db" doesn't exist.

25 years agoCheck for C:\FREEBSD also, for backwards compatibility
jkh [Wed, 28 Oct 1998 02:18:08 +0000 (02:18 +0000)]
Check for C:\FREEBSD also, for backwards compatibility

25 years agomicroseq.9: general purpose parallel microcode for ppbus(4)
nsouch [Wed, 28 Oct 1998 00:55:42 +0000 (00:55 +0000)]
microseq.9: general purpose parallel microcode for ppbus(4)
ppbconf.9: general info about ppbus(4) structures

25 years agoppc.4 moved from man4 to man4.i386: parallel port chipset driver manpage
nsouch [Wed, 28 Oct 1998 00:51:40 +0000 (00:51 +0000)]
ppc.4 moved from man4 to man4.i386: parallel port chipset driver manpage

pcf.4 added: Philips I2C/isa interface manpage

25 years agoFinally they arrive! ppbus (Parallel Port Bus framework) manpages.
nsouch [Wed, 28 Oct 1998 00:44:22 +0000 (00:44 +0000)]
Finally they arrive! ppbus (Parallel Port Bus framework) manpages.

25 years agoMore I2C framework docs. See iicbus(4) for more info.
nsouch [Wed, 28 Oct 1998 00:42:34 +0000 (00:42 +0000)]
More I2C framework docs. See iicbus(4) for more info.

25 years agoI2C framework manpages. See iicbus(4) for more info.
nsouch [Wed, 28 Oct 1998 00:40:53 +0000 (00:40 +0000)]
I2C framework manpages. See iicbus(4) for more info.

25 years agoUntimeout the chat expect timer as soon as we get a
brian [Tue, 27 Oct 1998 22:53:25 +0000 (22:53 +0000)]
Untimeout the chat expect timer as soon as we get a
match - otherwise, with a delayed (\\d) ``send'', the
timeout may happen during the send and cause a failure.
Problem reported by: David L. Vondrasek <dallas.tx@airmail.net>

25 years agoAdd ``set proctitle'' for changing argv[0]. All substitutions
brian [Tue, 27 Oct 1998 22:53:22 +0000 (22:53 +0000)]
Add ``set proctitle'' for changing argv[0].  All substitutions
are done in the same way as command execution.

For example, ``set proctitle USER INTERFACE PROCESSID'' would
be useful in a -direct profile for identifying who's connected.

25 years agoAdd ``PROCESSID'' as a constant expanded when running
brian [Tue, 27 Oct 1998 22:53:19 +0000 (22:53 +0000)]
Add ``PROCESSID'' as a constant expanded when running
commands.

25 years agoStress that ``none'' must be specified on the `set callback'
brian [Tue, 27 Oct 1998 22:53:18 +0000 (22:53 +0000)]
Stress that ``none'' must be specified on the `set callback'
line if callback is to be optional.
Requested by: Andrzej Tobola <san@koziolek.lublin.top.pl>

25 years agoFixes for handling 'wfd' (and any other disk with a non-2-character name)
msmith [Tue, 27 Oct 1998 21:14:03 +0000 (21:14 +0000)]
Fixes for handling 'wfd' (and any other disk with a non-2-character name)
Submitted by: Some from Satoh Junichi (junichi@astec.co.jp)

25 years agoAvoid interruptions while talking to keyboard controller.
rnordier [Tue, 27 Oct 1998 20:19:24 +0000 (20:19 +0000)]
Avoid interruptions while talking to keyboard controller.

25 years agobiosboot compatibility fix: change behavior of backspace in getstr().
rnordier [Tue, 27 Oct 1998 20:16:36 +0000 (20:16 +0000)]
biosboot compatibility fix: change behavior of backspace in getstr().
Noticed by: abial

Optimize away a few bytes to make space for the above.

25 years agoAdded needed splvm() protection around object page traversal in
dg [Tue, 27 Oct 1998 13:22:51 +0000 (13:22 +0000)]
Added needed splvm() protection around object page traversal in
vm_object_terminate().

25 years agoOops, the redundant tests for major numbers weren't redundant here.
bde [Tue, 27 Oct 1998 11:47:08 +0000 (11:47 +0000)]
Oops, the redundant tests for major numbers weren't redundant here.
They checked for the magic major number for the "device" behind mfs
mount points.  Use a more obvious check for this device.

Debugged by: Andrew Gallatin <gallatin@cs.duke.edu>

25 years agoLots of debugging code and some bugfixes.
dfr [Tue, 27 Oct 1998 09:21:43 +0000 (09:21 +0000)]
Lots of debugging code and some bugfixes.

Submitted by: Nick Hibma <nick.hibma@jrc.it>

25 years agoSome optimisations to the fragment reassembly code.
dfr [Tue, 27 Oct 1998 09:19:03 +0000 (09:19 +0000)]
Some optimisations to the fragment reassembly code.

Submitted by: Don Lewis <Don.Lewis@tsc.tdk.com>

25 years agoFix a bug in the new fragment reassembly code which was tickled by recieving
dfr [Tue, 27 Oct 1998 09:11:41 +0000 (09:11 +0000)]
Fix a bug in the new fragment reassembly code which was tickled by recieving
a fragment which wholly overlapped one or more existing fragments.

Submitted by: Don Lewis <Don.Lewis@tsc.tdk.com>

25 years ago64bit fixes.
dima [Tue, 27 Oct 1998 04:01:19 +0000 (04:01 +0000)]
64bit fixes.
(Note: ``dump'' doesn't work on alpha yet. Apparently there's a problem
somewhere is the physio() area)

Submitted by: myself && Matt Dillon.

25 years ago(`'): and (`'); -> (`:') and (`;')
fenner [Mon, 26 Oct 1998 19:36:38 +0000 (19:36 +0000)]
(`'): and (`'); -> (`:') and (`;')

25 years agoShuffle the iface-alias option so that's in alphabetical
brian [Mon, 26 Oct 1998 19:07:42 +0000 (19:07 +0000)]
Shuffle the iface-alias option so that's in alphabetical
order like the rest of the options.

25 years agoAdd ``enable proxyall'' support. This adds proxy ARP entries
brian [Mon, 26 Oct 1998 19:07:39 +0000 (19:07 +0000)]
Add ``enable proxyall'' support.  This adds proxy ARP entries
for every machine on every class C or smaller subnet that we
route to.
Add ``set {send,recv}pipe'' for controlling our socket buffer
sizes.
Mention the IP number with the problem in a few error messages.
All submitted by:     Craig Leres <leres@ee.lbl.gov>
Modified slightly by: me

25 years agoFix the interface alias code. Previously, I was expecting something
brian [Mon, 26 Oct 1998 19:07:36 +0000 (19:07 +0000)]
Fix the interface alias code.  Previously, I was expecting something
like

tun0: flags=blah
10.0.0.1 -> 10.0.0.100
10.0.0.2 -> 10.0.0.100
10.0.0.3 -> 10.0.0.100

to DTRT, despite the SIOCAIFADDR for each new alias returning
-1 & EEXIST while adding the alias anyway.  In real life, once
we have the second alias with the same destination, nothing will
route any more !  Also, because I was ignoring EEXIST, the
dynamic IP assignment code was assigning duplicate addresses
('cos it was being lied to by iface_inAdd()).

Now we have

tun0: flags=blah
10.0.0.1 -> 255.255.255.255
10.0.0.2 -> 10.0.0.100
10.0.0.3 -> 255.255.255.255

This works - stuff bound to 10.1 & 10.3 will be considered alive
by the kernel, and when they route back to the tun device, the
packets get aliased to 10.2 and go out to 10.100 (as with the
original plan).

We still see the EEXIST in SIOCAIFADDR, but ignore it when our
destination is 255.255.255.255, assuming that the alias *was*
actually added.

Additionally, ``iface add'' may now optionally be given only
the interface address.  The mask & destination default to
255.255.255.255.

25 years agoAnother instance where we default to elf only on alpha, but should
imp [Mon, 26 Oct 1998 18:27:32 +0000 (18:27 +0000)]
Another instance where we default to elf only on alpha, but should
default to aout only on i386.

25 years agoBogon I somehow inserted between compiling and commiting...
obrien [Mon, 26 Oct 1998 13:33:39 +0000 (13:33 +0000)]
Bogon I somehow inserted between compiling and commiting...

25 years agoQuiet many compiler warnings. Still fails -Wconversion in one case.
obrien [Mon, 26 Oct 1998 11:54:36 +0000 (11:54 +0000)]
Quiet many compiler warnings.  Still fails -Wconversion in one case.

Required because: -Werror is in Makefile

25 years agoRemoved redundant bitrotted checks for major numbers instead of updating
bde [Mon, 26 Oct 1998 08:53:13 +0000 (08:53 +0000)]
Removed redundant bitrotted checks for major numbers instead of updating
them.

25 years agoUpdated the major number check in vfs_object_create(). It's not
bde [Mon, 26 Oct 1998 08:07:00 +0000 (08:07 +0000)]
Updated the major number check in vfs_object_create().  It's not
clear if the check is necessary, but vfs_object_create() is called
for all vnodes and it was silly to create objects for VBLK vnodes
that don't even have a driver.

25 years agoCheck the major number of the boot device more carefully. There was only
bde [Mon, 26 Oct 1998 07:05:34 +0000 (07:05 +0000)]
Check the major number of the boot device more carefully.  There was only
a problem if the boot blocks passed bad data.

Check the major number of the dump device consistently.

25 years agoFixed breakage of the GPROF case of statclock() in the previous commit.
bde [Mon, 26 Oct 1998 06:13:18 +0000 (06:13 +0000)]
Fixed breakage of the GPROF case of statclock() in the previous commit.

25 years agoIf we know the content-length, only read that number of bytes from
fenner [Mon, 26 Oct 1998 02:39:21 +0000 (02:39 +0000)]
If we know the content-length, only read that number of bytes from
the server.  There exists a broken server which sends a few extra
garbage bytes in response to HTTP/1.1 requests.

25 years agoDon't send a DELAY value when sending CBCP_NONUM
brian [Sun, 25 Oct 1998 23:35:15 +0000 (23:35 +0000)]
Don't send a DELAY value when sending CBCP_NONUM
(as per the spec).
Spotted by: Andrzej Tobola <san@koziolek.lublin.top.pl>

25 years agoupdate to match reality
obrien [Sun, 25 Oct 1998 22:34:33 +0000 (22:34 +0000)]
update to match reality

25 years agoOops, back out some local changes that shouldn't have been in the
bde [Sun, 25 Oct 1998 20:11:36 +0000 (20:11 +0000)]
Oops, back out some local changes that shouldn't have been in the
previous commit.

25 years agoFixed breakage of the !SMP case of roundrobin() in the previous commit.
bde [Sun, 25 Oct 1998 19:57:23 +0000 (19:57 +0000)]
Fixed breakage of the !SMP case of roundrobin() in the previous commit.

25 years agoDon't follow null bdevsw pointers. The `major(dev) < nblkdev' test rotted
bde [Sun, 25 Oct 1998 19:26:18 +0000 (19:26 +0000)]
Don't follow null bdevsw pointers.  The `major(dev) < nblkdev' test rotted
when bdevsw[] became sparse.  We still depend on magic to avoid having to
check that (v_rdev) device numbers in vnodes are not NODEV.

25 years agoDon't follow null bdevsw pointers. The `major(dev) < nblkdev' test rotted
bde [Sun, 25 Oct 1998 19:24:04 +0000 (19:24 +0000)]
Don't follow null bdevsw pointers.  The `major(dev) < nblkdev' test rotted
when bdevsw[] became sparse.  We still depend on magic to avoid having to
check that (v_rdev) device numbers in vnodes are not NODEV.

Removed a redundant `major(dev) < nblkdev' test instead of updating it.

Don't follow a garbage bdevsw pointer for attempts to swap on empty
regular files.  This case currently can't happen.  Swapping on regular
files is ifdefed out in swapon() and isn't attempted for empty files
in nfs_mountroot().