]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoDisconnect kzip from the build, as it's about to enter late retirement.
sheldonh [Thu, 2 Nov 2000 10:00:34 +0000 (10:00 +0000)]
Disconnect kzip from the build, as it's about to enter late retirement.

PR: 20643
Submitted by: Peter Pentchev <roam@orbitel.bg>

23 years agoFixed the bug I have introduced in icmp_error() in revision 1.44.
ru [Thu, 2 Nov 2000 09:46:23 +0000 (09:46 +0000)]
Fixed the bug I have introduced in icmp_error() in revision 1.44.
The amount of data we copy from the original IP datagram into the
ICMP message was computed incorrectly for IP packets with payload
less than 8 bytes.

23 years agoFix cross-references which referred to xntp* instead of ntp*.
sheldonh [Thu, 2 Nov 2000 09:32:24 +0000 (09:32 +0000)]
Fix cross-references which referred to xntp* instead of ntp*.

PR: 22483
Submitted by: Philipp Mergenthaler <un1i@rz.uni-karlsruhe.de>

23 years agouse ``config auto'' for COREGA Ether PCC-T
sanpei [Thu, 2 Nov 2000 06:52:37 +0000 (06:52 +0000)]
use ``config auto'' for COREGA Ether PCC-T

23 years agoadd support for 3Com 3c575TX Fast Etherlink XL.
sanpei [Thu, 2 Nov 2000 06:49:38 +0000 (06:49 +0000)]
add support for 3Com 3c575TX Fast Etherlink XL.

Device information for 3C575-TX is from NetBSD,
sys/dev/cardbus/if_ex_cardbus.c file.

Reviewed by: wpaul, imp

23 years agoDon't use a trivially predictable temporary filename and keep recreating
kris [Thu, 2 Nov 2000 06:33:57 +0000 (06:33 +0000)]
Don't use a trivially predictable temporary filename and keep recreating
it again and again, practically begging the Bad Man to insert his symlink
underneath it and send us down the path to oblivion.

Noticed by: David Lary <dlary@secureworks.net>

23 years agoFix linux_ustat syscall. We only have cdevs now, so looking
marcel [Thu, 2 Nov 2000 06:08:26 +0000 (06:08 +0000)]
Fix linux_ustat syscall. We only have cdevs now, so looking
for a block device isn't that useful anymore.

Reported by: Wesley Morgan <morganw@chemicals.tacorp.com>
Submitted by: gallatin
Acknowledged by: phk

23 years agoSort.
obrien [Thu, 2 Nov 2000 04:13:51 +0000 (04:13 +0000)]
Sort.

23 years agoCopy the built kernel from the right place. (Forgotten in the last
joe [Thu, 2 Nov 2000 01:57:23 +0000 (01:57 +0000)]
Copy the built kernel from the right place.  (Forgotten in the last
commit.)

23 years agoImprove the PCI interrupt routing code. Now the process is as follows:
msmith [Thu, 2 Nov 2000 00:37:45 +0000 (00:37 +0000)]
Improve the PCI interrupt routing code.  Now the process is as follows:

 - Look for a hardwired interrupt in the routing table for this
   bus/device/pin (we already did this).
 - Look for another device with the same link byte which has a hardwired
   interrupt.
 - Look for a PCI device matching an entry with the same link byte
   which has already been assigned an interrupt, and use that.
 - Look for a routable interrupt listed in the "PCI only" interrupts
   field and use that.
 - Pick the first interrupt that's marked as routable and use that.

23 years agoFix a couple of cases where I tried to release the I/O space resource twice
wpaul [Thu, 2 Nov 2000 00:00:30 +0000 (00:00 +0000)]
Fix a couple of cases where I tried to release the I/O space resource twice
(once as as an I/O space resource and once as an IRQ resource). There was
a problem with this in if_rl too, which is how I found it.

23 years agoFix 8-bit EEPROM breakage: compare against 0x8129, not 8129.
wpaul [Wed, 1 Nov 2000 23:56:46 +0000 (23:56 +0000)]
Fix 8-bit EEPROM breakage: compare against 0x8129, not 8129.

23 years agoSupport for the linux ipc syscalls on the alpha, where each one has
gallatin [Wed, 1 Nov 2000 23:17:31 +0000 (23:17 +0000)]
Support for the linux ipc syscalls on the alpha, where each one has
its own syscall rather than going through a demux function like
linux_ipc() on i386

23 years agouse DUMMY macro and support from linux_util.h
gallatin [Wed, 1 Nov 2000 23:13:49 +0000 (23:13 +0000)]
use DUMMY macro and support from linux_util.h

23 years agoMove the linux abi's alpha md files over to using David's build-time
gallatin [Wed, 1 Nov 2000 23:07:25 +0000 (23:07 +0000)]
Move the linux abi's alpha md files over to using David's build-time
syscall generation method

23 years agoFixup some wrong statements in the new mbuf(9) man page.
bmilekic [Wed, 1 Nov 2000 23:02:19 +0000 (23:02 +0000)]
Fixup some wrong statements in the new mbuf(9) man page.
Also introduce a bunch of (missed?) macros and functions.
This man page still needs a lot of work, most likely a re-ordering
of the macros/functions, and a more complete, more accurate, listing of
available routines.
A good and worthy start nonetheless.

23 years agofix linux_termio and linux_termios structs on alpha. alpha differences
gallatin [Wed, 1 Nov 2000 22:36:41 +0000 (22:36 +0000)]
fix linux_termio and linux_termios structs on alpha.  alpha differences
are in the termios struct (probably because linux wants to be compatible
with the osf/1 termios struct), not the termio struct.

23 years agoMake internal clock sourcing work right by tying the NCO to the
phk [Wed, 1 Nov 2000 22:04:15 +0000 (22:04 +0000)]
Make internal clock sourcing work right by tying the NCO to the
Xtal reference instead of the CLADI input.

In unframed E1 mode, tie SIGFRZ low so that the mysycc doesn't
get confused.

Don't mask errors with OOF.  Don't ignore OOF errors.

Stop the channel before freeing mbufs in disconnect.

I still have no T1 devices to test with, so the T1 code is non-existent.

23 years agoAt the beginning of pthread_mutex_lock(), call _thread_init() if
jdp [Wed, 1 Nov 2000 20:19:07 +0000 (20:19 +0000)]
At the beginning of pthread_mutex_lock(), call _thread_init() if
necessary.  This works around a bug in old versions of libgcc_r.a
which are statically linked into old executables.

23 years agoThe MI/MD split wasn't perfect and the MI files need hacks for the
obrien [Wed, 1 Nov 2000 19:48:35 +0000 (19:48 +0000)]
The MI/MD split wasn't perfect and the MI files need hacks for the
AlphaLinux compat bits.  This will be better cleaned up soon.

Agreed to what ever was necessary by: marcel

23 years agoFix overflow from jail hostname.
eivind [Wed, 1 Nov 2000 19:38:08 +0000 (19:38 +0000)]
Fix overflow from jail hostname.

Bug found by: Esa Etelavuori <eetelavu@cc.hut.fi>

23 years agoPass in the new-bus flags to alpha_setup_intr().
jhb [Wed, 1 Nov 2000 18:40:42 +0000 (18:40 +0000)]
Pass in the new-bus flags to alpha_setup_intr().

23 years ago- Modify alpha_setup_intr() to take interrupt handler flags as an additional
jhb [Wed, 1 Nov 2000 18:40:02 +0000 (18:40 +0000)]
- Modify alpha_setup_intr() to take interrupt handler flags as an additional
  argument.  These flags include INTR_FAST, INTR_MPSAFE, etc.
- Properly handle INTR_EXCL when it is passed in to allow an interrupt
  handler to claim exclusive ownership of an interrupt thread.
- Add support for psuedo-fast interrupts on the alpha.  For fast interrupts,
  we don't allocate an interrupt thread; instead, during dispatching of an
  interrupt, we run the handler directly instead of scheduling the thread
  to run.  Note that the handler is currently run without Giant and must be
  MP safe.  The only fast handler currently is for the sio driver.

Requested by: dfr

23 years agoClose PR #21078: the aue driver was not correctly programming the
wpaul [Wed, 1 Nov 2000 18:26:36 +0000 (18:26 +0000)]
Close PR #21078: the aue driver was not correctly programming the
multicast filter on the Pegasus chip. Since IPv6 depends a lot
on multicasting, this caused several failures for people trying to
use IPv6 with Pegasus USB ethernet devices.

Submitted by: Jun Kuriyama <kuriyama@FreeBSD.org>

23 years agoMetricom Ricochet GS wireless USB modems are now supported.
n_hibma [Wed, 1 Nov 2000 18:03:14 +0000 (18:03 +0000)]
Metricom Ricochet GS wireless USB modems are now supported.

23 years agoGive vop_mmap an untimely death. The opportunity to give it a timely
eivind [Wed, 1 Nov 2000 17:57:24 +0000 (17:57 +0000)]
Give vop_mmap an untimely death.  The opportunity to give it a timely
death timed out in 1996.

23 years agoMake the Ricochet modems work through the USB interface.
n_hibma [Wed, 1 Nov 2000 17:35:57 +0000 (17:35 +0000)]
Make the Ricochet modems work through the USB interface.

Tested by: jim
Obtained from: NetBSD

23 years agoUpdate the config gathering code for both Promise & HPT
sos [Wed, 1 Nov 2000 17:35:44 +0000 (17:35 +0000)]
Update the config gathering code for both Promise & HPT

23 years agoIPv6 was not work on FDDI.
ume [Wed, 1 Nov 2000 16:57:01 +0000 (16:57 +0000)]
IPv6 was not work on FDDI.

Reported by: Akihiro IIJIMA <aki@noc.titech.ac.jp>
Submitted by: JINMEI Tatuya <jinmei@isl.rdc.toshiba.co.jp>
Reviewed by: Akihiro IIJIMA <aki@noc.titech.ac.jp>

23 years agoWrong checksum may have been computed for certain UDP packets.
ru [Wed, 1 Nov 2000 16:56:33 +0000 (16:56 +0000)]
Wrong checksum may have been computed for certain UDP packets.

Reviewed by: jlemon

23 years agoApply style(9).
obrien [Wed, 1 Nov 2000 16:38:40 +0000 (16:38 +0000)]
Apply style(9).

23 years agoMinimal set of diffs from the i386 bits to the AlphaLinux.
obrien [Wed, 1 Nov 2000 16:34:16 +0000 (16:34 +0000)]
Minimal set of diffs from the i386 bits to the AlphaLinux.
(based on a patch to sys/i386/linux/linux_dummy.c by gallatin)

23 years ago- correct the trace file date/time format to match reality
hm [Wed, 1 Nov 2000 14:55:02 +0000 (14:55 +0000)]
- correct the trace file date/time format to match reality
- add mail address of Gary Jennejohn and change mine to @freebsd.org

23 years agoOoops, forgot to add the originator for the last FAQ entry
hm [Wed, 1 Nov 2000 14:08:58 +0000 (14:08 +0000)]
Ooops, forgot to add the originator for the last FAQ entry

23 years agoAdd entry to the FAQ
hm [Wed, 1 Nov 2000 14:06:46 +0000 (14:06 +0000)]
Add entry to the FAQ

23 years agofixes:
andreas [Wed, 1 Nov 2000 13:30:24 +0000 (13:30 +0000)]
fixes:
a) the configured default printer entry might turn out to become a security hole
   Although lpd isn't enabled by default in FreeBSD 4 and later versions
   bad things might happen because of a simple copy and paste failure:
- fill up root-fs, if /dev/lpt0 doesn't exist
- fill up the spool dir (either root-fs or var-fs) if machine hadn't been
  designed as print server
   Therefore best decision: if people want printing, then configure both
1. /etc/rc.conf
2. /etc/printcap
   the default entry wouldn't have served well all tastes and needs anyway...
   design of most of our config files is, to have commented out suggestions
   if a service is not active.
b) [Garance]
   fix old and wrong documentation:
input filters are possible even if you print to a remote printer
   this makes FreeBSDs implementation of lpd currently the best without
   having to switch to port monster LPRng ...
c) fix pointer to wrong handbook section for a longer time .. so I doesn't fix the number
   I'm only referring to the printing section, otherwise this would have to be fixed several
   times, if the chapters should be reordered again ...
d) typo: chose -> choose

Submitted by: me and some suggestions by Garance
Approved by: Eivind and Garance A Drosihn <drosih@rpi.edu> in private e-mail

23 years agoWrong checksum used for certain reassembled IP packets before diverting.
ru [Wed, 1 Nov 2000 11:21:45 +0000 (11:21 +0000)]
Wrong checksum used for certain reassembled IP packets before diverting.

23 years agoMdoc style fixes.
mpp [Wed, 1 Nov 2000 11:17:33 +0000 (11:17 +0000)]
Mdoc style fixes.

PR: docs/22445

23 years agoAdd BUFFALO LPC3-CLX
toshi [Wed, 1 Nov 2000 10:30:53 +0000 (10:30 +0000)]
Add BUFFALO LPC3-CLX
Submitted by:   Takashi Ikeda <ccikeda@moon.aichi-u.ac.jp>

Add TOSHIBA Ethernet Card LABCTA00A.
Submitted by:   "minoru kurata" <minoru.kurata@toshiba.co.jp>
Obtained from: PAO3

23 years agoDon't use the old URL to the GNU ftp location any more:
rse [Wed, 1 Nov 2000 10:24:08 +0000 (10:24 +0000)]
Don't use the old URL to the GNU ftp location any more:
ftp://prep.ai.mit.edu/pub/gnu/ -> ftp://ftp.gnu.org/gnu/

23 years agoAllow the building of the syscall bits at compile time.
obrien [Wed, 1 Nov 2000 09:23:12 +0000 (09:23 +0000)]
Allow the building of the syscall bits at compile time.

23 years agoNeed to dummy-out rt_sigreturn.
obrien [Wed, 1 Nov 2000 08:52:24 +0000 (08:52 +0000)]
Need to dummy-out rt_sigreturn.

submitted by: gallatin

23 years agoNeed to include sys/proc.h.
obrien [Wed, 1 Nov 2000 08:49:48 +0000 (08:49 +0000)]
Need to include sys/proc.h.

23 years agoI don't know why this used to compile for me, but now struct
obrien [Wed, 1 Nov 2000 08:48:53 +0000 (08:48 +0000)]
I don't know why this used to compile for me, but now struct
linux_sigcontext seems to have changed and some members are gone.

23 years agoMake the target a little bit more generic.
obrien [Wed, 1 Nov 2000 08:47:34 +0000 (08:47 +0000)]
Make the target a little bit more generic.

23 years agoWhitespace only: Correct poor line-breaking introduced in rev 1.17,
sheldonh [Wed, 1 Nov 2000 07:51:14 +0000 (07:51 +0000)]
Whitespace only:  Correct poor line-breaking introduced in rev 1.17,
which was limited to correcting mark-up.

23 years agoCorrect mark-up used in rev 1.16, as discussed with its contributor:
sheldonh [Wed, 1 Nov 2000 07:49:29 +0000 (07:49 +0000)]
Correct mark-up used in rev 1.16, as discussed with its contributor:

* Use a sub-section (Ss) instead of a section (Sh) for
  "Sysctl MIB Entries".

* Use a tagged list (Bl, El and It) instead of sub-sections (Ss) for
  the actual MIB entries.

* Mark paths up as such (Pa).

* Mark defined values up as such (Dv).

23 years agoReorder one of the new features based on experience. Since the number
dougb [Wed, 1 Nov 2000 07:21:56 +0000 (07:21 +0000)]
Reorder one of the new features based on experience. Since the number
of files auto-installed during an upgrade from a really old system
can get quite long, and it's piped to the PAGER already, print
that first, then print any of the 4 two-line messages that might
apply.

23 years agoUse test -z rather than -n.
obrien [Wed, 1 Nov 2000 07:09:38 +0000 (07:09 +0000)]
Use test -z rather than -n.

Submitted by: bde

23 years agoIt's no longer true that "nobody uses ia beyond here"; it's now
joe [Wed, 1 Nov 2000 01:59:28 +0000 (01:59 +0000)]
It's no longer true that "nobody uses ia beyond here"; it's now
used to keep address based if_data statistics in.

Submitted by: ru

23 years agoChange 'lpr' so -C (class) information is included in the control file
gad [Wed, 1 Nov 2000 01:48:06 +0000 (01:48 +0000)]
Change 'lpr' so -C (class) information is included in the control file
even for jobs or queues which will not print header (separator) pages.

Submitted by: "Steve O'Hara-Smith" <steveo@eircom.net>

23 years agoUpdate the entry for uscanner (only list devices that we know that work)
n_hibma [Wed, 1 Nov 2000 01:27:09 +0000 (01:27 +0000)]
Update the entry for uscanner (only list devices that we know that work)
Also add an entry for umodem (3Com 5605)

23 years agoTell libgcc_r how to initialize a mutex before it uses it. This
jdp [Wed, 1 Nov 2000 01:18:27 +0000 (01:18 +0000)]
Tell libgcc_r how to initialize a mutex before it uses it.  This
fixes the immediate segmentation violations from some multithreaded
programs built on very up-to-date systems.

23 years ago(null commit)
n_hibma [Wed, 1 Nov 2000 01:10:03 +0000 (01:10 +0000)]
(null commit)

Patch for previous commit submitted by Daniel O'Connor
<doconnor@gsoft.com.au>

23 years agokill old sb driver
cg [Wed, 1 Nov 2000 00:56:52 +0000 (00:56 +0000)]
kill old sb driver

23 years agoswitch over to new sb8/sb16 code
cg [Wed, 1 Nov 2000 00:53:16 +0000 (00:53 +0000)]
switch over to new sb8/sb16 code

23 years agoCatch up to mutexes becoming MI: machine/mutex.h -> sys/mutex.h
jhb [Wed, 1 Nov 2000 00:39:27 +0000 (00:39 +0000)]
Catch up to mutexes becoming MI: machine/mutex.h -> sys/mutex.h

23 years agofix type check in pccard_beep_select
sanpei [Wed, 1 Nov 2000 00:37:34 +0000 (00:37 +0000)]
fix type check in pccard_beep_select

23 years ago"it's" --> "its" (possessive).
mph [Wed, 1 Nov 2000 00:36:12 +0000 (00:36 +0000)]
"it's" --> "its" (possessive).

Correct a malformed sentence.

23 years agoCut&paste bug: Set USBD_SHORT_XFER_OK unconditionally
n_hibma [Wed, 1 Nov 2000 00:28:40 +0000 (00:28 +0000)]
Cut&paste bug: Set USBD_SHORT_XFER_OK unconditionally

23 years agoNote MFC of uscanner(4).
bmah [Wed, 1 Nov 2000 00:01:49 +0000 (00:01 +0000)]
Note MFC of uscanner(4).

23 years agoDon't ignore the return from tcsetattr() - moan about it instead.
brian [Tue, 31 Oct 2000 23:27:35 +0000 (23:27 +0000)]
Don't ignore the return from tcsetattr() - moan about it instead.

Suggested by: Andre Albsmeier <andre@akademie3000.de>

23 years ago1. Oops on last commit: note a few SCSI-3 items for data compression page.
mjacob [Tue, 31 Oct 2000 22:36:33 +0000 (22:36 +0000)]
1. Oops on last commit: note a few SCSI-3 items for data compression page.
2. Change copyright to me- per practive of scsi_sa.c that Justin had
me do a while back.

23 years agoscsi_sa.h
mjacob [Tue, 31 Oct 2000 22:34:51 +0000 (22:34 +0000)]
scsi_sa.h

23 years agoBuild with -DUSE_INET6 so that we can actually use the IPv6 support in
nectar [Tue, 31 Oct 2000 22:29:27 +0000 (22:29 +0000)]
Build with -DUSE_INET6 so that we can actually use the IPv6 support in
IPFilter 3.4.x.

Approved by: darrenr, guido

23 years agoTreat engine inquiry the same as path inquiry in xptioctl().
gibbs [Tue, 31 Oct 2000 22:07:02 +0000 (22:07 +0000)]
Treat engine inquiry the same as path inquiry in xptioctl().
Return ENOTSUP for any opcode that is not supported by the XPT
device.

Add back a missing local declaration that seems to have been deleted
by my last commit.

23 years agoNote bktr(4) update to 2.18, cvs(1) update MFC-ed, new C/C++
bmah [Tue, 31 Oct 2000 22:01:32 +0000 (22:01 +0000)]
Note bktr(4) update to 2.18, cvs(1) update MFC-ed, new C/C++
runtime initialization code, BIND update to 8.2.3-T6B.

23 years agoDon't needlessly indirect the APM softstate. It does nothing but
nsayer [Tue, 31 Oct 2000 21:00:15 +0000 (21:00 +0000)]
Don't needlessly indirect the APM softstate. It does nothing but
obfuscate the code.

23 years agoMake the standard 'lpq' output a little more informative when listing jobs
gad [Tue, 31 Oct 2000 20:11:30 +0000 (20:11 +0000)]
Make the standard 'lpq' output a little more informative when listing jobs
which have long names.  Instead of just listing '...', try to list some
reasonable subset of the name (with a "..." to indicate something missing).

Reviewed by: freebsd-print@bostonradio.org (only a little review)

23 years agoFix 'lpq' so it can correctly display jobs which come from hosts using
gad [Tue, 31 Oct 2000 19:57:07 +0000 (19:57 +0000)]
Fix 'lpq' so it can correctly display jobs which come from hosts using
'lprNG' (which writes control-lines in a different order than our lpr).

Reviewed by: freebsd-print@bostonradio.org

23 years agoGenerate bmov instructions for all move requests. The driver
gibbs [Tue, 31 Oct 2000 18:44:32 +0000 (18:44 +0000)]
Generate bmov instructions for all move requests.  The driver
converts these to an equivalent 'and' operation on chips that
don't support this opcode.

23 years agoaic7xxx.c:
gibbs [Tue, 31 Oct 2000 18:43:29 +0000 (18:43 +0000)]
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.

In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up.  When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.

When something strange happens, dump the card's transaction
state via ahc_dump_card_state().  This should aid in debugging.

Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly.  The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs.  I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.

Add additional sequencer interrupt codes to support firmware
diagnostics.  The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.

Make it possible to switch into and out of target mode on
the fly.  The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed.  As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.

In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.

Never issue a PPR as a target.  It is forbidden by the spec.

Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.

Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue.  This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.

When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips.  Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.

Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case.  The free list is only
used if we must page SCBs.

Correct the transmission of cdbs > 12 bytes in length.  When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location.  We now defer
the calculation of the cdb address until just before queing
it to the card.

When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.

Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.

Use the BMOV instruction for all single byte moves on
controllers that support it.  The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.

Correct a few bugs in ahc_dump_card_state().  If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue.  When enumerating
the untagged queue, it helps to use the correct index for
the queue.

aic7xxx.h:

Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.

Add the AHC_SEQUENCER_DEBUG flag.

Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.

Add new prototypes.

aic7xxx.reg:
Allow the SCSIBUSL register to be written to.  This is
required to fix a selection timeout problem on the 7892/99.

Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.

Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram.  This prevented the driver from properly honoring
these settings when no serial eeprom was available.

Remove an unused sequencer flag.

aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register.  Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.

Add sequencer diagnostic code to detect software and or
hardware bugs.  The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.

Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.

Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.

Cleanup the overrun case.  It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes.  This drasticaly simplifies
things.

Prior to leaving the data phase, make sure that the S/G
preload queue is empty.

Remove code to place a request sense request on the waiting
queue.  This is all handled by the kernel now.

Change the semantics of "findSCB".  In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list.  The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation.  We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.

Inline some busy target table operations.

Add a critical section to protect adding an SCB to
the disconnected list.

aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests.  No filtering is necessary
for "goal" requests from the XPT.

Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.

In ahc_timeout, only take action if the active SCB is
the timedout SCB.  This deals with the case of two
transactions to the same device with different timeout
values.

Use ahc_qinfifo_requeu_tail() instead of home grown
version.

aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.

aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.

23 years agoAdd the aic7xxx option AHC_DEBUG_SEQUENCER. This enables some diagnostic
gibbs [Tue, 31 Oct 2000 18:01:15 +0000 (18:01 +0000)]
Add the aic7xxx option AHC_DEBUG_SEQUENCER.  This enables some diagnostic
code in the firmware downloaded to the aic7xxx RISC engine.

23 years ago0x03F -> 0x3F for consistancy with the rest of the file.
gibbs [Tue, 31 Oct 2000 18:00:23 +0000 (18:00 +0000)]
0x03F -> 0x3F for consistancy with the rest of the file.

23 years agoAllow async transfer negotiation updates to modify tagging behavior.
gibbs [Tue, 31 Oct 2000 17:59:43 +0000 (17:59 +0000)]
Allow async transfer negotiation updates to modify tagging behavior.
The XPT uses this to prevent tags from being used on parallel SCSI
interfaces immediately after a bus reset or BDR so that controllers
have an oportunity to renegotiate without tag messages in the way.
Somehow this got disabled... the functionality has been here for
quite some time.

Noticed by: my SCSI bus analyzer

23 years agoUse do { ... } while (0) to wrap the body of mtx_assert().
jhb [Tue, 31 Oct 2000 17:20:55 +0000 (17:20 +0000)]
Use do { ... } while (0) to wrap the body of mtx_assert().

Reported by: rwatson

23 years agoSwap the order of two tags in the pppoe PADI and PADS packets
julian [Tue, 31 Oct 2000 14:40:23 +0000 (14:40 +0000)]
Swap the order of two tags in the pppoe PADI and PADS packets
as there are apparently some buggy switches that need them in that order.
(I hope there aren't any that require them in the old order!)

23 years agoFix typo in comments
roger [Tue, 31 Oct 2000 14:30:29 +0000 (14:30 +0000)]
Fix typo in comments

23 years agoResolve conflicts.
asmodai [Tue, 31 Oct 2000 14:19:55 +0000 (14:19 +0000)]
Resolve conflicts.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r68073,
asmodai [Tue, 31 Oct 2000 14:17:05 +0000 (14:17 +0000)]
This commit was generated by cvs2svn to compensate for changes in r68073,
which included commits to RCS files with non-trunk default branches.

23 years agoVirgin import of BIND v8.2.3-T6B
asmodai [Tue, 31 Oct 2000 14:17:05 +0000 (14:17 +0000)]
Virgin import of BIND v8.2.3-T6B

23 years agoFix a fundamental error. The device is not owned by the parent
asmodai [Tue, 31 Oct 2000 13:16:47 +0000 (13:16 +0000)]
Fix a fundamental error.  The device is not owned by the parent
bus before allocation.  The resource is.

Asked about by: julian on -current
Verified by: msmith

23 years agoUpgrade to driver 2.18
roger [Tue, 31 Oct 2000 13:09:56 +0000 (13:09 +0000)]
Upgrade to driver 2.18
Support for new Hauppauge Model 44xxx WinTV Cards
(the ones with no audio mux)

Submitted by: Christian Gusenbauer <Christian.Gusenbauer@netway.at>

23 years agoResolve conflicts.
asmodai [Tue, 31 Oct 2000 12:38:22 +0000 (12:38 +0000)]
Resolve conflicts.

23 years agoVirgin import of BIND v8.2.3-T6B
asmodai [Tue, 31 Oct 2000 12:35:45 +0000 (12:35 +0000)]
Virgin import of BIND v8.2.3-T6B

23 years agoThis commit was generated by cvs2svn to compensate for changes in r68068,
asmodai [Tue, 31 Oct 2000 12:35:45 +0000 (12:35 +0000)]
This commit was generated by cvs2svn to compensate for changes in r68068,
which included commits to RCS files with non-trunk default branches.

23 years agoIf acpica driver is loaded using kldload(8), warn and just ignore.
takawata [Tue, 31 Oct 2000 11:54:10 +0000 (11:54 +0000)]
If acpica driver is loaded using kldload(8), warn and just ignore.

23 years agoQuiet -Wall.
obrien [Tue, 31 Oct 2000 11:36:33 +0000 (11:36 +0000)]
Quiet -Wall.

23 years agoDeprecate devsw->d_bmaj entirely.
phk [Tue, 31 Oct 2000 10:58:14 +0000 (10:58 +0000)]
Deprecate devsw->d_bmaj entirely.

This removes support for booting current kernels with very old bootblocks.

Device driver writers: Please remove initializations for the d_bmaj
field in your cdevsw{}.

23 years agoAdd missing redirection to /dev/null to prevent disturbing
hm [Tue, 31 Oct 2000 10:00:51 +0000 (10:00 +0000)]
Add missing redirection to /dev/null to prevent disturbing
of fullscreen display.

23 years agoDon't issue a warning if isdncontroller is set to -1
hm [Tue, 31 Oct 2000 09:59:04 +0000 (09:59 +0000)]
Don't issue a warning if isdncontroller is set to -1

23 years agoDo not waste a time saving a copy of IP header if we are certainly
ru [Tue, 31 Oct 2000 09:13:02 +0000 (09:13 +0000)]
Do not waste a time saving a copy of IP header if we are certainly
not going to send an ICMP error message (net.inet.udp.blackhole=1).

23 years agoRemove unused #includes.
phk [Tue, 31 Oct 2000 08:22:26 +0000 (08:22 +0000)]
Remove unused #includes.

Approved by: nyan

23 years agoDisabled EGA/VGA 1bpp/4bpp modes support. This is not real fix, but this comes
nyan [Tue, 31 Oct 2000 07:58:34 +0000 (07:58 +0000)]
Disabled EGA/VGA 1bpp/4bpp modes support. This is not real fix, but this comes
back to support 8bpp mode.

23 years agoI forgot the driver configuration file for mfsroot.flp.
hosokawa [Tue, 31 Oct 2000 07:41:38 +0000 (07:41 +0000)]
I forgot the driver configuration file for mfsroot.flp.
http://people.freebsd.org/~hosokawa/driver-floppy/ for details.

Reviewed by: current@FreeBSD.org

23 years agoMoved driver modules for some PCI NICs and PCCARD-only NICs to mfsroot.flp.
hosokawa [Tue, 31 Oct 2000 07:39:07 +0000 (07:39 +0000)]
Moved driver modules for some PCI NICs and PCCARD-only NICs to mfsroot.flp.
http://people.freebsd.org/~hosokawa/driver-floppy/ for details.

Reviewed by: current@FreeBSD.org

23 years agoCorrect a bignum.
grog [Tue, 31 Oct 2000 07:20:46 +0000 (07:20 +0000)]
Correct a bignum.

Detected-by: lang/clisp
Approved-by: jkh

23 years agoAdd support for virgin disklabels
jkh [Tue, 31 Oct 2000 07:07:39 +0000 (07:07 +0000)]
Add support for virgin disklabels

Submitted by: dillon

23 years agoAdd a new ioctl for doing virgin disklabels.
jkh [Tue, 31 Oct 2000 07:05:40 +0000 (07:05 +0000)]
Add a new ioctl for doing virgin disklabels.

Submitted by: dillon

23 years agoremove duplicated item.
sanpei [Tue, 31 Oct 2000 06:19:58 +0000 (06:19 +0000)]
remove duplicated item.
"Laneed Wireless card" is "ELECOM *Laneed*
Air@Hawk/LD-WL11/PCC Wireless card" which
was already committed.