]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
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().

25 years agoDon't follow null bdevsw pointers. The `major(dev) < nblkdev' test rotted
bde [Sun, 25 Oct 1998 19:02:48 +0000 (19:02 +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 redundant `major(dev) < nblkdev' tests instead of updating them.

25 years agoNitpicking and dusting performed on a train. Removes trivial warnings
phk [Sun, 25 Oct 1998 17:44:59 +0000 (17:44 +0000)]
Nitpicking and dusting performed on a train.  Removes trivial warnings
about unused variables, labels and other lint.

25 years agoFixed device number checking in bdevvp():
bde [Sun, 25 Oct 1998 16:11:49 +0000 (16:11 +0000)]
Fixed device number checking in bdevvp():
- dev != NODEV was checked for, but 0 was returned on failure.  This was
  fixed in Lite2 (except the return code was still slightly wrong (ENODEV
  instead of ENXIO)) but the changes were not merged.  This case probably
  doesn't actually occur under FreeBSD.
- major(dev) was not checked to have a valid non-NULL bdevsw entry.  This
  caused panics when the driver for the root device didn't exist.

Fixed minor misformattings in bdevvp().  Rev.1.14 consisted mainly of
gratuitous reformattings that seem to have caused many Lite2 merge
errors.

PR: 8417

25 years agoRemove redundant files (they are recreated during building from templates
abial [Sun, 25 Oct 1998 15:48:18 +0000 (15:48 +0000)]
Remove redundant files (they are recreated during building from templates
in ../../lang/).

25 years agoChanges to comply with new syntax.
abial [Sun, 25 Oct 1998 15:31:26 +0000 (15:31 +0000)]
Changes to comply with new syntax.

Noticed by: bde, the Hawk Eye :-)

25 years agoAdd msg.
abial [Sun, 25 Oct 1998 15:17:24 +0000 (15:17 +0000)]
Add msg.

25 years agoChanges to comply with new syntax.
abial [Sun, 25 Oct 1998 15:16:58 +0000 (15:16 +0000)]
Changes to comply with new syntax.

Noticed by: bde, the Hawk Eye :-)

25 years agoAllow to use other floppy sizes for "router" floppy.
abial [Sun, 25 Oct 1998 15:15:55 +0000 (15:15 +0000)]
Allow to use other floppy sizes for "router" floppy.

25 years agoSlight style police.
obrien [Sun, 25 Oct 1998 13:29:57 +0000 (13:29 +0000)]
Slight style police.
Add some content from objformat(1).

25 years agoAdd manpage for objformat. This needs major word smithing.
obrien [Sun, 25 Oct 1998 13:25:42 +0000 (13:25 +0000)]
Add manpage for objformat.  This needs major word smithing.

25 years agoBackout out previous commit. The bug was in the kernel.
bde [Sun, 25 Oct 1998 10:59:44 +0000 (10:59 +0000)]
Backout out previous commit.  The bug was in the kernel.

25 years agoFixed sysctl attachment for statically configured vfs's.
bde [Sun, 25 Oct 1998 10:52:34 +0000 (10:52 +0000)]
Fixed sysctl attachment for statically configured vfs's.

Broken in: previous commit

25 years agoAdd some missing dependencies.
phk [Sun, 25 Oct 1998 07:23:32 +0000 (07:23 +0000)]
Add some missing dependencies.

Grumbled about in principle: bde

25 years agofix unsigned overflow
ache [Sun, 25 Oct 1998 05:06:42 +0000 (05:06 +0000)]
fix unsigned overflow
PR: 8437

25 years agoRemove redundant copies, and give credit to proper person.
abial [Sun, 25 Oct 1998 03:16:03 +0000 (03:16 +0000)]
Remove redundant copies, and give credit to proper person.

Noticed by: jkh

25 years agoWith much regret, I had to remove these two programs - their license
abial [Sun, 25 Oct 1998 03:02:56 +0000 (03:02 +0000)]
With much regret, I had to remove these two programs - their license
doesn't permit us to distribute them. I think I should take some law
classes...

Explained by: bde

25 years agoFix the interrupt mask generated for enabling/disabling interrupts 8 to 15.
paul [Sun, 25 Oct 1998 01:30:16 +0000 (01:30 +0000)]
Fix the interrupt mask generated for enabling/disabling interrupts 8 to 15.

25 years agoWork around some variables having N_UNDF types but valid values; this
msmith [Sat, 24 Oct 1998 23:12:09 +0000 (23:12 +0000)]
Work around some variables having N_UNDF types but valid values; this
makes vmstat work on ELF kernels again.
Submitted by: Daniel Rock <rock@cs.uni-sb.de>

25 years agoUpdate to current reality. We can now track several LORSTA on separate
phk [Sat, 24 Oct 1998 19:55:09 +0000 (19:55 +0000)]
Update to current reality.  We can now track several LORSTA on separate
minor devices.

Improve PLL/OCXO DAC dithering.

General remodeling.

Performance is now 2.5e-11 in frequency and +/- 100 nsec in time, both
of which are actually the limits of the transmitted signal.

25 years agoUpdate and add timekeeping code.
phk [Sat, 24 Oct 1998 19:47:42 +0000 (19:47 +0000)]
Update and add timekeeping code.

25 years agoHeader file for XRPU devices.
phk [Sat, 24 Oct 1998 19:47:29 +0000 (19:47 +0000)]
Header file for XRPU devices.

25 years agoUse sysctlbyname() to locate the vfs.nfs.nfsstats node.
msmith [Sat, 24 Oct 1998 19:36:52 +0000 (19:36 +0000)]
Use sysctlbyname() to locate the vfs.nfs.nfsstats node.
Submitted by: Daniel Rock <rock@cs.uni-sb.de>

25 years agoDon't complain about SIOCGIFMEDIA; it's harmless and not something
msmith [Sat, 24 Oct 1998 18:55:53 +0000 (18:55 +0000)]
Don't complain about SIOCGIFMEDIA; it's harmless and not something
useful here yet.

Submitted by: markm and others

25 years agoDon't put 0x in front of %p, it does it already.
msmith [Sat, 24 Oct 1998 18:35:09 +0000 (18:35 +0000)]
Don't put 0x in front of %p, it does it already.
Submitted by:  Brian Feldman <green@janus.syracuse.net>

25 years agoPR: 8429
thepish [Sat, 24 Oct 1998 10:08:05 +0000 (10:08 +0000)]
PR: 8429
Submitted by: jkb@FreeBSD.ORG
correct typo in example

25 years agoWhen invoked as "objformat" to print out the current object format,
jdp [Sat, 24 Oct 1998 02:01:30 +0000 (02:01 +0000)]
When invoked as "objformat" to print out the current object format,
produce an error message if any arguments are given on the command
line.

25 years agoLoosen our restrictions on setting enddisc, mrru,
brian [Sat, 24 Oct 1998 01:08:45 +0000 (01:08 +0000)]
Loosen our restrictions on setting enddisc, mrru,
shortseq, authname and authkey.

o Auth{name,key} may additionally be set in PHASE_ESTABLISH.
o The others may be set in PHASE_ESTABLISH as long as no links
  have yet reached DATALINK_LCP.

25 years agoReserve some space for the stack in the region we recover from the initial
msmith [Sat, 24 Oct 1998 00:31:21 +0000 (00:31 +0000)]
Reserve some space for the stack in the region we recover from the initial
bootloader.
Submitted by: dfr

25 years agoClean the ${BASE}.sym file as well.
msmith [Fri, 23 Oct 1998 22:32:27 +0000 (22:32 +0000)]
Clean the ${BASE}.sym file as well.

25 years agoRemove a debugging copy of offsetof() that snuck in.
msmith [Fri, 23 Oct 1998 22:30:52 +0000 (22:30 +0000)]
Remove a debugging copy of offsetof() that snuck in.

25 years agoEnable the PCI BIOS PnP enumerator.
msmith [Fri, 23 Oct 1998 22:29:19 +0000 (22:29 +0000)]
Enable the PCI BIOS PnP enumerator.

25 years agoPnP enumerator using the PCI BIOS. This is needlessly complex due to the
msmith [Fri, 23 Oct 1998 22:29:08 +0000 (22:29 +0000)]
PnP enumerator using the PCI BIOS.  This is needlessly complex due to the
lack of a simple "enumerate all PCI devices" function.

25 years agoAdded ability to auto-run a presentation.
abial [Fri, 23 Oct 1998 19:34:50 +0000 (19:34 +0000)]
Added ability to auto-run a presentation.

25 years agoOops, don't quite use RAW_PART instead of RAWPARTITION. Use 'a' + RAW_PART
bde [Fri, 23 Oct 1998 18:57:39 +0000 (18:57 +0000)]
Oops, don't quite use RAW_PART instead of RAWPARTITION.  Use 'a' + RAW_PART
in both places.

25 years agoIgnore `vector xxxintr' specifications except for checking their syntax.
bde [Fri, 23 Oct 1998 17:05:06 +0000 (17:05 +0000)]
Ignore `vector xxxintr' specifications except for checking their syntax.
Interrupt handlers are now configured in drivers.

Didn't update config/SMM.doc.  It doesn't have any i386 examples (not
even `isa').

Bumped CONFIGVERS.  This is not necessary for -current yet, but using
the new config with old system sources gives null pointers for all
vectors.

25 years agoSync with sys/i386/isa/clock.c revision 1.128.
kato [Fri, 23 Oct 1998 13:13:43 +0000 (13:13 +0000)]
Sync with sys/i386/isa/clock.c revision 1.128.

25 years agoRefer to vinum driver sources in a single location (outside of the
peter [Fri, 23 Oct 1998 11:56:45 +0000 (11:56 +0000)]
Refer to vinum driver sources in a single location (outside of the
kernel src/sys tree, sigh - it should really be sys/dev/vinum/*).

25 years agoMove the kld stuff to here.
peter [Fri, 23 Oct 1998 11:49:47 +0000 (11:49 +0000)]
Move the kld stuff to here.

25 years agoUpdate timecounters to new interface.
phk [Fri, 23 Oct 1998 10:46:20 +0000 (10:46 +0000)]
Update timecounters to new interface.

25 years agoChange the way we simulate stable storage for timecounters.
phk [Fri, 23 Oct 1998 10:44:52 +0000 (10:44 +0000)]
Change the way we simulate stable storage for timecounters.

If you have problems with the "calcru" messages and processes being
killed for excessive cpu time, try to increase the NTIMECOUNTER
#define and report your findings.

25 years agoRemove the tc_cost field from timecounters.
phk [Fri, 23 Oct 1998 10:42:42 +0000 (10:42 +0000)]
Remove the tc_cost field from timecounters.

25 years agofix spelling error.
jkh [Fri, 23 Oct 1998 10:27:50 +0000 (10:27 +0000)]
fix spelling error.
Submitted by: Pomegranate <daver@flag.blackened.net>

25 years agoFix behaviour for 'mkdir -m 777 / /tmp/foo'. Play "guess the style bug"
msmith [Fri, 23 Oct 1998 06:28:40 +0000 (06:28 +0000)]
Fix behaviour for 'mkdir -m 777 / /tmp/foo'.  Play "guess the style bug"
with Bruce again.

Reported by: bde

25 years agoOops, revert part of last fix. vm_pager_dealloc() can't be called until
dg [Fri, 23 Oct 1998 05:43:13 +0000 (05:43 +0000)]
Oops, revert part of last fix. vm_pager_dealloc() can't be called until
after the pages are removed from the object...so fix the problem by
not printing the diagnostic for wired fictitious pages (which is normal).

25 years agoFixed two bugs in recent commit: in vm_object_terminate, vm_pager_dealloc
dg [Fri, 23 Oct 1998 05:25:49 +0000 (05:25 +0000)]
Fixed two bugs in recent commit: in vm_object_terminate, vm_pager_dealloc
needs to be called prior to freeing remaining pages in the object so that
the device pager has an opportunity to grab its "fake" pages. Also, in
the case of wired pages, the page must be made busy prior to calling
vm_page_remove. This is a difference from 2.2.x that I overlooked when
I brought these changes forward.

25 years agoFix a problem with the way we handled device invalidation when attaching
ken [Thu, 22 Oct 1998 22:16:56 +0000 (22:16 +0000)]
Fix a problem with the way we handled device invalidation when attaching
to a device failed.

In theory, the same steps that happen when we get an AC_LOST_DEVICE async
notification should have been taken when a driver fails to attach.  In
practice, that wasn't the case.

This only affected the da, cd and ch drivers, but the fix affects all
peripheral drivers.

There were several possible problems:
 - In the da driver, we didn't remove the peripheral's softc from the da
   driver's linked list of softcs.  Once the peripheral and softc got
   removed, we'd get a kernel panic the next time the timeout routine
   called dasendorderedtag().
 - In the da, cd and possibly ch drivers, we didn't remove the
   peripheral's devstat structure from the devstat queue.  Once the
   peripheral and softc were removed, this could cause a panic if anyone
   tried to access device statistics.  (one component of the linked list
   wouldn't exist anymore)
 - In the cd driver, we didn't take the peripheral off the changer run
   queue if it was scheduled to run.  In practice, it's highly unlikely,
   and maybe impossible that the peripheral would have been on the
   changer run queue at that stage of the probe process.

The fix is:
 - Add a new peripheral callback function (the "oninvalidate" function)
   that is called the first time cam_periph_invalidate() is called for a
   peripheral.

 - Create new foooninvalidate() routines for each peripheral driver.  This
   routine is always called at splsoftcam(), and contains all the stuff
   that used to be in the AC_LOST_DEVICE case of the async callback
   handler.

 - Move the devstat cleanup call to the destructor/cleanup routines, since
   some of the drivers do I/O in their close routines.

 - Make sure that when we're flushing the buffer queue, we traverse it at
   splbio().

 - Add a check for the invalid flag in the pt driver's open routine.

Reviewed by: gibbs

25 years agoHonour the spirit rather than the implementation of the previous changes;
msmith [Thu, 22 Oct 1998 21:42:20 +0000 (21:42 +0000)]
Honour the spirit rather than the implementation of the previous changes;
if we are invoked with -m, use chmod() on the final directory component
in order to ensure the mode is correctly set.

25 years agoPnP BIOS enumerator; mostly useful for hunting PnP devices that are supported
msmith [Thu, 22 Oct 1998 20:25:06 +0000 (20:25 +0000)]
PnP BIOS enumerator; mostly useful for hunting PnP devices that are supported
by the system BIOS rather than supporting ISA PnP.

25 years agoAdd a dependancy on the BTX crt0 object
msmith [Thu, 22 Oct 1998 20:23:58 +0000 (20:23 +0000)]
Add a dependancy on the BTX crt0 object
Add the biospnp handler to the pnp handler array
Drop some old debugging code

25 years agoAdd biospnp.c
msmith [Thu, 22 Oct 1998 20:23:15 +0000 (20:23 +0000)]
Add biospnp.c

25 years agoPop the return address off the stack before making a V86_CALLF call;
msmith [Thu, 22 Oct 1998 20:22:07 +0000 (20:22 +0000)]
Pop the return address off the stack before making a V86_CALLF call;
this allows us to implement what look like C function calls from user
space "directly" to v86 mode code.  (Used for calling the PnP BIOS)

25 years agoExport an interface for presetting the ISA PnP read port address.
msmith [Thu, 22 Oct 1998 20:20:51 +0000 (20:20 +0000)]
Export an interface for presetting the ISA PnP read port address.
Make the EISA ID formatter generally available

25 years agoFixed gratuitous unformatting in rev.1.8.
bde [Thu, 22 Oct 1998 18:17:30 +0000 (18:17 +0000)]
Fixed gratuitous unformatting in rev.1.8.

25 years agoClose PR #8409:
wpaul [Thu, 22 Oct 1998 16:46:26 +0000 (16:46 +0000)]
Close PR #8409:

Fix syntax errors inside #ifdef FORCE_AUTONEG_TFOUR.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 16:16:02 +0000 (16:16 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Don't quote port names that don't have a digit in them.

25 years agoFIxed a bogus comment.
bde [Thu, 22 Oct 1998 16:12:16 +0000 (16:12 +0000)]
FIxed a bogus comment.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 16:10:29 +0000 (16:10 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Attempted to update the generated interrupt handler attachment to the
current "temporary" method.  Not tested.  To test it, someone would first
have to fix the bitrot in the ioctl command arg type.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 15:53:06 +0000 (15:53 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

25 years agoClose PR #8384:
wpaul [Thu, 22 Oct 1998 15:52:25 +0000 (15:52 +0000)]
Close PR #8384:

Revert the transmission packet queueing strategy changes. Clearly I missed
something while debugging this, although I never encountered any problems
on my test machines.

Also make one other minor change: jack up the TX reclaim threshold for
3c90xB adapters in order to stave off 'transmission error: 82' errors.

Document the existence of the tx reclaim register (for inspecting the
current reclaim threshold) in register window 5 (if_xlreg.h).

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 15:39:35 +0000 (15:39 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 15:31:33 +0000 (15:31 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 14:51:19 +0000 (14:51 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Fixed some missing closing quotes.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 14:49:00 +0000 (14:49 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Quote the last few args to form 1 arg.  Quoting them in pairs almost
defeated the point of quoting them, which is to reduce the arg count
to <= 9.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 14:36:55 +0000 (14:36 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Don't quote IO_KBD.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 14:32:20 +0000 (14:32 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Format "irq ?" the same as in other man pages.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 14:29:03 +0000 (14:29 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Use .Cd instead of a a home made format.

Don't quote IO_KBD.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 14:22:13 +0000 (14:22 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Fixed the port name, and don't quote it.  IO_NPX0 never existed.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 14:17:44 +0000 (14:17 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Fixed quoting of IO_ASC1.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 14:12:55 +0000 (14:12 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

25 years agoQuote port names that have a digit in them. IO_TIMER1 was lexed as
bde [Thu, 22 Oct 1998 13:49:51 +0000 (13:49 +0000)]
Quote port names that have a digit in them.  IO_TIMER1 was lexed as
{ port_name = "IO_TIMER", port_number = 1 } and only worked because
it was reassembled to "IO_TIMER1".  Trailing digits always work, but
this is too magic to depend on.

Don't quote port names that don't have a digit in them.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 13:01:19 +0000 (13:01 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

Use the correct macro in configuration declarations.  .Cm somehow gave
the same results as .Cm.

25 years agoRemoved all `vector xxxintr' specifications. Interrupt handlers are now
bde [Thu, 22 Oct 1998 12:23:04 +0000 (12:23 +0000)]
Removed all `vector xxxintr' specifications.  Interrupt handlers are now
configured in drivers.

25 years agoFixed the ifdef for `uart' driver being present. The sound-driver-specific
bde [Thu, 22 Oct 1998 11:41:22 +0000 (11:41 +0000)]
Fixed the ifdef for `uart' driver being present.  The sound-driver-specific
ifdef used in uart6850.c is bogus.

Reported by: Paul Allenby <pallenby@mikom.csir.co.za>

25 years agoAdded rl driver.
kato [Thu, 22 Oct 1998 11:29:58 +0000 (11:29 +0000)]
Added rl driver.

25 years agoFix behaviour of the driver with small files -- due to a stupid
luigi [Thu, 22 Oct 1998 11:20:33 +0000 (11:20 +0000)]
Fix behaviour  of the driver with small files -- due to a stupid
bug sound was not played if the total amount of data written to
the device was less than one blocksize

Noticed by: NABETANI Masaki and FreeBSD-users-jp

25 years agoThis is a copyrighted software. I forgot to include the license here.
abial [Thu, 22 Oct 1998 11:16:41 +0000 (11:16 +0000)]
This is a copyrighted software. I forgot to include the license here.
Apologies to all parties involved -  it wasn't intended...

Reminded by: bde

25 years agoThe previous change was much too simple to actually compensate for
bde [Thu, 22 Oct 1998 10:34:11 +0000 (10:34 +0000)]
The previous change was much too simple to actually compensate for
braindamage in sound driver confiruration.  Attempt to only reference
interrupt handlers if their driver is configured.

25 years agooops, forgot to add his birthday
torstenb [Thu, 22 Oct 1998 07:56:22 +0000 (07:56 +0000)]
oops, forgot to add his birthday

25 years agoAdd Jon Postel's birth/death
torstenb [Thu, 22 Oct 1998 07:40:05 +0000 (07:40 +0000)]
Add Jon Postel's birth/death

PR: 8402

25 years agoInitialize isa_devtab entries for interrupt handlers in individual
bde [Thu, 22 Oct 1998 05:58:45 +0000 (05:58 +0000)]
Initialize isa_devtab entries for interrupt handlers in individual
device drivers, not in ioconf.c.  Use a different hack in isa_device.h
so that a new config(8) is not required yet.

pc98 parts approved by: kato

25 years agos/boot2/loader/ on the alpha
gpalmer [Thu, 22 Oct 1998 04:58:38 +0000 (04:58 +0000)]
s/boot2/loader/ on the alpha

25 years agoAdded Microsoft's PE formatted executables for mips, i386 and Alpha.
imp [Thu, 22 Oct 1998 04:41:34 +0000 (04:41 +0000)]
Added Microsoft's PE formatted executables for mips, i386 and Alpha.
Also added entry for the debugger files as well.  The PE format is
used for all Win32 platforms (Win95, Win98, WinNT and WinCE), so it is
nice to be able to tell what kind of thing the foo.dll....  Don't have
any data for powerpc formats, however...

25 years agoWhen the "-s" option is given, try to ensure that we restart the
jdp [Thu, 22 Oct 1998 04:02:37 +0000 (04:02 +0000)]
When the "-s" option is given, try to ensure that we restart the
target process even if we are killed or die due to an error.

25 years agoSolve the ``first connection'' problem that occurs on
brian [Thu, 22 Oct 1998 02:32:50 +0000 (02:32 +0000)]
Solve the ``first connection'' problem that occurs on
demand-dial links with dynamic IP numbers where the program
that causes the dial bind()s to an interface address that is
subsequently changed after ppp negotiation.

The problem is defeated by adding negotiated addresses to the
tun interface as additional alias addresses and providing a set
of ``iface'' commands for managing the interface.  Libalias is
also required (and what a name clash!) - it happily IP-aliases
the address so that the source is that of the primary (negotiated)
interface and un-IP-aliases it on the way back.

An ``enable iface-alias'' is done implicitly by the -alias command
line switch.  If -alias isn't given, iface-aliasing is disabled by
default and can't be enabled 'till an ``alias enable yes'' is done.
``alias enable no'' silently disables iface-alias.

So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah''
will work for the first connection, although existing bindings will
not survive a disconnect/connect as the TCP peer will be trying to
send to the old IP address - the packets won't route.

It's now a lot easier to add IPXCP to ppp with minor updates to
the new iface.[ch] (if anyone ever gets 'round to it).

It's also now possible to manually add interface aliases with
something like ``iface add 1.2.3.4/24 5.6.7.8''.  This allows
multi-homed ppp links :-)

25 years agoMake the VM system handle the case where a terminating object contains
dg [Thu, 22 Oct 1998 02:16:53 +0000 (02:16 +0000)]
Make the VM system handle the case where a terminating object contains
legitimately wired pages. Currently we print a diagnostic when this
happens, but this will be removed soon when it will be common for this
to occur with zero-copy TCP/IP buffers.

25 years agoDefer rundown (m_freem) of completed transmit buffers for no longer
dg [Thu, 22 Oct 1998 02:00:49 +0000 (02:00 +0000)]
Defer rundown (m_freem) of completed transmit buffers for no longer
than 1 second.

25 years agoConvert fake page allocs to use the zone allocator, thus eliminating the
dg [Thu, 22 Oct 1998 01:45:29 +0000 (01:45 +0000)]
Convert fake page allocs to use the zone allocator, thus eliminating the
private pool management code in here.

25 years agoSet m->object to NULL in dev_pager_getfake().
dg [Wed, 21 Oct 1998 23:06:50 +0000 (23:06 +0000)]
Set m->object to NULL in dev_pager_getfake().

25 years agoBye-bye setdefs.
msmith [Wed, 21 Oct 1998 20:11:03 +0000 (20:11 +0000)]
Bye-bye setdefs.

25 years ago - Enable PnP and ISA PnP code.
msmith [Wed, 21 Oct 1998 20:10:33 +0000 (20:10 +0000)]
 - Enable PnP and ISA PnP code.
 - Use the ISA PnP enumerator.
 - Use the new linker set code, throw out the gensetdefs stuff.
 - Produce an intermediate loader image that has symbols stripped, to aid
 - in debugging.
 - Supply ISA port access functions required for ISA PnP

25 years agoDon't depend on being part of the kernel tree. If we are, use the kernel's
msmith [Wed, 21 Oct 1998 20:07:57 +0000 (20:07 +0000)]
Don't depend on being part of the kernel tree.  If we are, use the kernel's
<machine> includes rather than the system's.

25 years agoMake ISA PnP work. This successfully enumerates as many ISA devices as I
msmith [Wed, 21 Oct 1998 20:07:05 +0000 (20:07 +0000)]
Make ISA PnP work.  This successfully enumerates as many ISA devices as I
can fit into my test machine.

 - Move to using STAILQs rather than ad-hoc singly-linked lists.
 - Use a mostly procedural interface to the PnP information.  This
   improves data-hiding.

Implement a new linker-set technique (currently on i386 only but should work
on Alpha as well).  This is a good candidate for replacing the current
gensetdefs cruft completely.

25 years agoAdded two new utilities, both ported from Minix: mined (editor), and
abial [Wed, 21 Oct 1998 17:36:28 +0000 (17:36 +0000)]
Added two new utilities, both ported from Minix: mined (editor), and
msh (largely compatible Bourne shell).

25 years agoGet things limping along again for the 80386 and friends. The
rnordier [Wed, 21 Oct 1998 17:10:49 +0000 (17:10 +0000)]
Get things limping along again for the 80386 and friends.  The
ELF assembler emits a redundant operand-size prefix for the
fnstsw %ax instruction, and this stops the show for 3.0-RELEASE.

25 years agoEliminate a superfluous comment.
jdp [Wed, 21 Oct 1998 16:31:38 +0000 (16:31 +0000)]
Eliminate a superfluous comment.

25 years agoRemove most of the code and replace it with a call to getobjformat().
jdp [Wed, 21 Oct 1998 15:13:16 +0000 (15:13 +0000)]
Remove most of the code and replace it with a call to getobjformat().

25 years agoNuked PG_TABLED flag. Replaced with m->object != NULL.
dg [Wed, 21 Oct 1998 14:46:42 +0000 (14:46 +0000)]
Nuked PG_TABLED flag. Replaced with m->object != NULL.

25 years agoFixed previous commit to actually work. The wrong variable was initialized
bde [Wed, 21 Oct 1998 11:48:39 +0000 (11:48 +0000)]
Fixed previous commit to actually work.  The wrong variable was initialized
earlier (with style bugs of course), so zcat was an alias for uncompress
instead of `uncompress -c'.