]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoInclude rdp(4).
joerg [Mon, 21 Dec 1998 18:04:20 +0000 (18:04 +0000)]
Include rdp(4).

Should i also include it into GENERIC?

25 years agoThis commit was generated by cvs2svn to compensate for changes in r41980,
joerg [Mon, 21 Dec 1998 18:01:15 +0000 (18:01 +0000)]
This commit was generated by cvs2svn to compensate for changes in r41980,
which included commits to RCS files with non-trunk default branches.

25 years agoThis is my first cut on a driver for the RealTek RTL 8002 pocket
joerg [Mon, 21 Dec 1998 18:01:15 +0000 (18:01 +0000)]
This is my first cut on a driver for the RealTek RTL 8002 pocket
ethernet driver.

The BUGS section is still impressive, but the driver seems to work for
me now.  Disclaimer: i haven't been able to test this under -current
so far (but it compiles, and the notebook it's intended for can now be
updated to -current more easy than before).  Don't be afraid of the
many #ifdefs on __FreeBSD_version in the imported file; i want them in
the repository on the vendor-branch so other people can also manually
integrate it into older systems.  I'll clean it up on the -current
branch in a followup commit.  The vendor-banch version right now
supports systems back to 2.2R.

This driver should be layered upon ppc(4), but i currently have no
idea how to do this.

Eventually i'll further develop the driver to also support the more
modern RTL 8012 success, which seems to be present in a number of
cheap pocket ethernet adapters these days.  Right now, i doubt it will
run with the 8012 without any changes.

Finally a big Thanks! to RealTek for promptly providing me with
documentation and with the source code for the 8012 pocket driver upon
request.  I wish all vendors were that cooperative!.

25 years agoJust fetch distfiles from a single location. The change to textproc/docproj
jkh [Mon, 21 Dec 1998 17:34:33 +0000 (17:34 +0000)]
Just fetch distfiles from a single location.  The change to textproc/docproj
broke my fetch heuristic anyway.

25 years ago update HISTORY section
dillon [Mon, 21 Dec 1998 10:34:53 +0000 (10:34 +0000)]
update HISTORY section

25 years ago Add manual page for experimental kernel asleep() and await() routines
dillon [Mon, 21 Dec 1998 10:29:28 +0000 (10:29 +0000)]
Add manual page for experimental kernel asleep() and await() routines

25 years agoRegrettably, you can't take setdef0.c out of order, or, surprise,
msmith [Mon, 21 Dec 1998 10:29:09 +0000 (10:29 +0000)]
Regrettably, you can't take setdef0.c out of order, or, surprise,
linker sets don't work and thus neither does the kernel.

This should stop being a problem with an improved linker set mechanism in
the pipeline (see the bootloader) but for now this has to stay like this.

25 years agoSync with sys/i386/conf/options.i386 revision 1.96.
kato [Mon, 21 Dec 1998 09:58:37 +0000 (09:58 +0000)]
Sync with sys/i386/conf/options.i386 revision 1.96.

25 years agoSync with sys/i386/conf/Makefile.i386 revision 1.131.
kato [Mon, 21 Dec 1998 09:57:42 +0000 (09:57 +0000)]
Sync with sys/i386/conf/Makefile.i386 revision 1.131.

25 years ago Elf (and a.out?) rebuild of bootstrap tools appeared to be broken due to
dillon [Mon, 21 Dec 1998 09:41:26 +0000 (09:41 +0000)]
Elf (and a.out?) rebuild of bootstrap tools appeared to be broken due to
    some tools requiring header files, such as errno.h, that are softlinks.
    The mini installation of include files in Makefile.inc1 wasn't doing
    the job, so it has been ripped out and replaced with a true make of the
    /usr/src/include/ directory (make all and make beforeinstall).  I think
    the original idea was to save time by not installing unnecessary header
    files, but it doesn't really save all that much time.

    Also, I have moved a NOCLEAN conditional to cover rebuilding the object
    tree 'Rebuilding the ${OBJFORMAT} obj tree' section.  This may or may not
    be correct but it appears to function properly.  If it is not correct we
    need to find another way to avoid scanning every single file in the
    entire source hierarchy for make restarts.

25 years agoCheck for DMA capbility is against unit,not controller.
msmith [Mon, 21 Dec 1998 08:55:56 +0000 (08:55 +0000)]
Check for DMA capbility is against unit,not controller.

Submitted by: Lee Cremeans <lee@st-lcremean.tidalwave.net>

25 years ago Add asleep() and await() support. Currently highly experimental. A
dillon [Mon, 21 Dec 1998 07:41:51 +0000 (07:41 +0000)]
Add asleep() and await() support.  Currently highly experimental.  A
    small support structure had to be added to the proc structure, and
    a few minor conditional panics no longer apply.

25 years agoFlush the output file before exiting; short-lived programs don't even fill
msmith [Mon, 21 Dec 1998 06:34:50 +0000 (06:34 +0000)]
Flush the output file before exiting; short-lived programs don't even fill
the stdio buffer.

25 years agoClarify that "/usr/share" is used for architecture independent files
jkoshy [Mon, 21 Dec 1998 05:57:57 +0000 (05:57 +0000)]
Clarify that "/usr/share" is used for architecture independent files
and need not be restricted to ASCII text files.

PR: docs/9124
Submitted by: Mattias Pantzare <pantzer@ludd.luth.se>

25 years agoBad Dog! No Biscuit! *Never* commit without testing- even if it was
mjacob [Sun, 20 Dec 1998 20:32:34 +0000 (20:32 +0000)]
Bad Dog! No Biscuit! *Never* commit without testing- even if it was
"just a printf formatting change"....

25 years ago Moved security page from man 1 to man 7
dillon [Sun, 20 Dec 1998 20:12:17 +0000 (20:12 +0000)]
Moved security page from man 1 to man 7

25 years ago Moving security page to section 7
dillon [Sun, 20 Dec 1998 20:11:25 +0000 (20:11 +0000)]
Moving security page to section 7

25 years ago Add quick overview of basic firewalling practices
dillon [Sun, 20 Dec 1998 20:05:44 +0000 (20:05 +0000)]
Add quick overview of basic firewalling practices

25 years ago Add information on how to deal with DOS attacks
dillon [Sun, 20 Dec 1998 19:49:43 +0000 (19:49 +0000)]
Add information on how to deal with DOS attacks

25 years agoAdd two fields for the lower layers convenience.
phk [Sun, 20 Dec 1998 19:06:22 +0000 (19:06 +0000)]
Add two fields for the lower layers convenience.

25 years agoadd Bus and Device Reset commands
mjacob [Sun, 20 Dec 1998 18:51:56 +0000 (18:51 +0000)]
add Bus and Device Reset commands

25 years agoRemove unused file. It seems to have been a vestige of when mfs did its
bde [Sun, 20 Dec 1998 17:05:54 +0000 (17:05 +0000)]
Remove unused file.  It seems to have been a vestige of when mfs did its
own memory allocation.

25 years agoBacked out rev.1.46. It had no effect for aout, was incomplete for elf,
bde [Sun, 20 Dec 1998 16:54:27 +0000 (16:54 +0000)]
Backed out rev.1.46.  It had no effect for aout, was incomplete for elf,
and had gratuitous gcc dependencies.  Rev.1.47 has a better fix.

25 years agoFixed building elf kernels without a `make depend' step. The magic
bde [Sun, 20 Dec 1998 16:25:21 +0000 (16:25 +0000)]
Fixed building elf kernels without a `make depend' step.  The magic
ordering in rev.1.38 was broken by putting setdef0.o before vnode_if.o
in SYSTEM_OBJS.

25 years agoRemove a vestige of the amd driver.
bde [Sun, 20 Dec 1998 15:26:02 +0000 (15:26 +0000)]
Remove a vestige of the amd driver.

Forgotten by: msmith

25 years agoSplit out afterdistribute and distribute from BKTGTS because they do
jhay [Sun, 20 Dec 1998 14:30:08 +0000 (14:30 +0000)]
Split out afterdistribute and distribute from BKTGTS because they do
need MK_ENV. This should make release work again. Maybe we should get
rid of the /usr/obj/elf at some stage?

25 years agoAdd a workaround to avoid 'defined but not used' warnings for linker
dfr [Sun, 20 Dec 1998 14:20:11 +0000 (14:20 +0000)]
Add a workaround to avoid 'defined but not used' warnings for linker
sets on i386 and alpha ELF kernels.

25 years agoAdd LDGP(pv) to suword, subyte, fuword, fubyte so that they can be called
dfr [Sun, 20 Dec 1998 13:21:55 +0000 (13:21 +0000)]
Add LDGP(pv) to suword, subyte, fuword, fubyte so that they can be called
from kernel modules.

25 years agoIn ufs_setattr(), if only one of va_atime or va_mtime are != VNOVAL, then
dfr [Sun, 20 Dec 1998 12:36:01 +0000 (12:36 +0000)]
In ufs_setattr(), if only one of va_atime or va_mtime are != VNOVAL, then
the code set the other field in the inode to VNOVAL.  This can happen
sometimes on an NFS server.

25 years agoFIxed unsorting of cross references in previous commit.
bde [Sun, 20 Dec 1998 06:30:05 +0000 (06:30 +0000)]
FIxed unsorting of cross references in previous commit.

25 years agoFixed unsorting of MAN1 in previous commit.
bde [Sun, 20 Dec 1998 06:27:00 +0000 (06:27 +0000)]
Fixed unsorting of MAN1 in previous commit.

25 years agoAdded documenation for the existing implementation of asctime_r,
wes [Sun, 20 Dec 1998 06:22:06 +0000 (06:22 +0000)]
Added documenation for the existing implementation of asctime_r,
ctime_r, gmtime_r, and localtime_r.

25 years agoFixed typo done, thanks!
foxfair [Sun, 20 Dec 1998 02:53:49 +0000 (02:53 +0000)]
Fixed typo done, thanks!

25 years agoFixed typo.
foxfair [Sun, 20 Dec 1998 02:52:30 +0000 (02:52 +0000)]
Fixed typo.

25 years ago1) Fix some serious bugs (1 botch on my part which caused a filemark to be
mjacob [Sat, 19 Dec 1998 23:33:21 +0000 (23:33 +0000)]
1) Fix some serious bugs (1 botch on my part which caused a filemark to be
written even it the tape was opened readonly- 2 botches in deferred error
handling for FIXED LENGTH mode which caused panic && hand resp.). Fixed
a memory leak in sa_mount.
2) Fixed an annoying bug when turning of compression to actually reflect
this for future status calls.
3) Implement the MTIOCERRSTAT call where latched control and I/O residuals
and sense data are returned to the application asking for them.

25 years agoMove the "generic" SMP options from options.i386 to conf/options
msmith [Sat, 19 Dec 1998 23:07:51 +0000 (23:07 +0000)]
Move the "generic" SMP options from options.i386 to conf/options

25 years agoMove the "generic" SMP options from options.i386 to conf/options.
msmith [Sat, 19 Dec 1998 23:07:16 +0000 (23:07 +0000)]
Move the "generic" SMP options from options.i386 to conf/options.
This has the useful side-effect of ensuring that opt_smp.h is
always generated, so we can assume it will always be available.

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

25 years agoadd in new errstat subcommand (MTIOCSERRSTAT)
mjacob [Sat, 19 Dec 1998 20:23:37 +0000 (20:23 +0000)]
add in new errstat subcommand (MTIOCSERRSTAT)

25 years agoAdd MTIOCERRSTAT ioctl and structures that define (at least for
mjacob [Sat, 19 Dec 1998 20:21:56 +0000 (20:21 +0000)]
Add MTIOCERRSTAT ioctl and structures that define (at least for
SCSI) the data returned for latched residual and error statistics.
Reserve space for LOG SENSE data which we'll get later. This area
in transition as the HP TapeAlert initiative is looked at.

25 years agoDisplay uptime in upper right corner.
obrien [Sat, 19 Dec 1998 20:09:20 +0000 (20:09 +0000)]
Display uptime in upper right corner.

Submitted by: Andy Farkas <andyf@speednet.com.au>
Obtained from: freebsd-current list

25 years agoMake it compile again.
sos [Sat, 19 Dec 1998 19:05:57 +0000 (19:05 +0000)]
Make it compile again.
I have no idea if the new stuff works though.

Broken by: Julian

25 years agoAdd "fla" to Disk_Names()
phk [Sat, 19 Dec 1998 18:48:33 +0000 (18:48 +0000)]
Add "fla" to Disk_Names()

25 years agoWait for channel commands to complete after issuing the commands.
bde [Sat, 19 Dec 1998 16:28:57 +0000 (16:28 +0000)]
Wait for channel commands to complete after issuing the commands.
The optimisation of only waiting before issuing new commands is
obviously invalid in general and it caused many errors in NIST-PCTS.
I think the errors were mostly for characters sent with the wrong
parity, etc., after a half complete tcsetattr().

Use microtime() instead of a magic loop count to limit the wait.
The wait is a busy-wait :-( and normally takes about 500 usec.

25 years agoAdd more non-Intel family ((new)) chipset, just like VIA technology MVP3
foxfair [Sat, 19 Dec 1998 16:05:19 +0000 (16:05 +0000)]
Add more non-Intel family ((new)) chipset, just like VIA technology MVP3
AcerLabs Aladdin-V. It makes the PCI probing work when system booting. I
will try to merge some additional funtion(i.e. wdc1 problem cause tons of
PR appear :<) ASAP if I could.

Remind me if something wrong after committing, thanks!

25 years ago Commit initial man page for 'security'... an introduction to security
dillon [Sat, 19 Dec 1998 09:33:03 +0000 (09:33 +0000)]
Commit initial man page for 'security'... an introduction to security
    mechanisms.  It is a little light on detail but still a pretty good
    overview.  I suggest that for specific detail (such as, for example,
    how to setup kerberos), that additional security-DETAILNAME man pages
    be written and refered to.

25 years agoRemove the bogus charracters "42" from the beginning of the first line.
julian [Sat, 19 Dec 1998 08:35:30 +0000 (08:35 +0000)]
Remove the bogus charracters "42" from the beginning of the first line.
looks like "editor turd".

25 years agoFix two bogons created by 'patch(1)' in my last commit.
julian [Sat, 19 Dec 1998 08:23:31 +0000 (08:23 +0000)]
Fix two bogons created by 'patch(1)' in my last commit.

25 years ago Take bind out of sandbox and run it as root again, but leave support
dillon [Sat, 19 Dec 1998 07:25:56 +0000 (07:25 +0000)]
Take bind out of sandbox and run it as root again, but leave support
    mechanisms ('bind' user and group) in place so the feature can be easily
    turned on.  There were too many complaints.  The security(1) man
    page will be created/updated to include the appropriate info.

25 years agoTrim the ethernet CRC from received frames manually in wb_rxeof().
wpaul [Sat, 19 Dec 1998 04:19:44 +0000 (04:19 +0000)]
Trim the ethernet CRC from received frames manually in wb_rxeof().
The Winbond chip always includes the CRC with every received frame,
and I can't find anything in the Winbond manual that indicates you can
program it not to do this.

25 years agoFix for bogus BIOS configuration of the 450NX PCI interface on some
msmith [Sat, 19 Dec 1998 02:58:29 +0000 (02:58 +0000)]
Fix for bogus BIOS configuration of the 450NX PCI interface on some
systems (eg. Dell 6300).

PR: kern/8928
Submitted by: David Malone <dwmalone@maths.tcd.ie>

25 years agoReviewed by: Luoqi Chen, Jordan Hubbard
julian [Sat, 19 Dec 1998 02:55:34 +0000 (02:55 +0000)]
Reviewed by:  Luoqi Chen, Jordan Hubbard
Submitted by:  "Richard Seaman, Jr." <lists@tar.com>
Obtained from: linux :-)

Code to allow Linux Threads to run under FreeBSD.

By default not enabled
This code is dependent on the conditional
COMPAT_LINUX_THREADS (suggested by Garret)
This is not yet a 'real' option but will be within some number of hours.

25 years agoSupport for Intel 450NX-based systems with more than one PCI bus (ie.
msmith [Sat, 19 Dec 1998 02:51:22 +0000 (02:51 +0000)]
Support for Intel 450NX-based systems with more than one PCI bus (ie.
most of them).

Many thanks to Kevin Van Maren for the work here, Intel for lending us
a 450NX system to work this out on, and several other folks for testing
the patches.  See the PR for an extensive discussion of the nature of
the problem and resolution.

PR: kern/8928
Submitted by: Kevin Van Maren <vanmaren@fast.cs.utah.edu>

25 years agoMake <sys/dirent.h> and <dirent.h> self-sufficient, using new __intN_t types.
dt [Sat, 19 Dec 1998 00:30:01 +0000 (00:30 +0000)]
Make <sys/dirent.h> and <dirent.h> self-sufficient, using new __intN_t types.

25 years agoInstall <sys/inttypes.h> as <inttypes.h>.
dt [Sat, 19 Dec 1998 00:21:13 +0000 (00:21 +0000)]
Install <sys/inttypes.h> as <inttypes.h>.

25 years agoLittle reorganization:
dt [Sat, 19 Dec 1998 00:02:34 +0000 (00:02 +0000)]
Little reorganization:
- created internal names for fixed-size integral types, like __int32_t. They
  will be used to make several headers self-sufficient.
- <stdlib.h> don't include <machine/types.h> anymore.
- created <sys/inttypes.h>, which can be used as <inttypes.h>.
- declaration of uoff_t and ufs_daddr_t moved to <sys/types.h>.

Reviewed by: bde

25 years agoCorrect the definition for PN_NETCFG_NO_RXCRC: it's 0x20000000, not
wpaul [Fri, 18 Dec 1998 18:31:34 +0000 (18:31 +0000)]
Correct the definition for PN_NETCFG_NO_RXCRC: it's 0x20000000, not
0x02000000. This error was causing the chip to always include the
ethernet CRC along with every received frame (the driver turns on
PN_NETCFG_NO_RXCRC, but it was frobbing the wrong bit).

25 years agoMade changes suggested by bruce, plus a couple of others which made -Wall work.
mjacob [Fri, 18 Dec 1998 18:16:35 +0000 (18:16 +0000)]
Made changes suggested by bruce, plus a couple of others which made -Wall work.
Obtained from:bde@freebsd.org

25 years agoRemoved unused (slightly wrong) stub functions.
bde [Fri, 18 Dec 1998 18:07:10 +0000 (18:07 +0000)]
Removed unused (slightly wrong) stub functions.

Fixed some staticization obfuscations.

25 years agoDon't specify a language to compile_et.
des [Fri, 18 Dec 1998 14:32:48 +0000 (14:32 +0000)]
Don't specify a language to compile_et.
Rename fetchConnect to _fetch_connect since it's internal.

25 years agoUnbreak K&R support, and generate code that works in K&R C as well as
des [Fri, 18 Dec 1998 14:29:51 +0000 (14:29 +0000)]
Unbreak K&R support, and generate code that works in K&R C as well as
ANSI C no matter what language is specified.

25 years agoAdd share/emacs/site-lisp. Too many things break without it.
asami [Fri, 18 Dec 1998 08:06:34 +0000 (08:06 +0000)]
Add share/emacs/site-lisp.  Too many things break without it.

Reviewed by: ports list (a long time ago)

25 years agoNow that textproc/docproj works properly, use it instead of the long
jkh [Fri, 18 Dec 1998 05:13:01 +0000 (05:13 +0000)]
Now that textproc/docproj works properly, use it instead of the long
list of ports.

25 years agoAdd in block position/block locate functions.
mjacob [Fri, 18 Dec 1998 04:31:43 +0000 (04:31 +0000)]
Add in block position/block locate functions.

25 years agoAdd structures and function definitions pertinent for hardware locate support.
mjacob [Fri, 18 Dec 1998 04:29:16 +0000 (04:29 +0000)]
Add structures and function definitions pertinent for hardware locate support.

25 years agoAdd references to test(1) and expr(1).
jkoshy [Fri, 18 Dec 1998 03:51:18 +0000 (03:51 +0000)]
Add references to test(1) and expr(1).

25 years agoAdd cross-references to test(1) and expr(1) respectively.
jkoshy [Fri, 18 Dec 1998 03:16:47 +0000 (03:16 +0000)]
Add cross-references to test(1) and expr(1) respectively.

PR:  docs/9111
Submitted by: Josh Gilliam <josh@quick.net>

25 years agoFix spelling nits.
jkoshy [Fri, 18 Dec 1998 03:08:57 +0000 (03:08 +0000)]
Fix spelling nits.

PR: docs/9115
Submitted by: Kris Kennaway <kkennawa@physics.adelaide.edu.au>

25 years ago1) Clarify that a weof count of zero is allowed (flush buffers).
mjacob [Fri, 18 Dec 1998 02:02:20 +0000 (02:02 +0000)]
1) Clarify that a weof count of zero is allowed (flush buffers).
2) Add in the set of {sethpos,setspos,rdhpos,rdspos) commands to implement the
Hardware && SCSI Logical Block position ioctls.

25 years agoEnable building libio on the alpha.
dfr [Thu, 17 Dec 1998 23:02:11 +0000 (23:02 +0000)]
Enable building libio on the alpha.

25 years agoMove some compile flags from the kernel makefile to bsd.kern.mk so that
dfr [Thu, 17 Dec 1998 22:36:21 +0000 (22:36 +0000)]
Move some compile flags from the kernel makefile to bsd.kern.mk so that
kernel modules are built with the right flags.

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

25 years agoCorrect the definition of the changer device capabilities page. The
gibbs [Thu, 17 Dec 1998 22:26:39 +0000 (22:26 +0000)]
Correct the definition of the changer device capabilities page.  The
previous definition confused some reserved bytes for exchange capabilities.

25 years agoimport logical/hardware block locate ioctls from NetBSD
mjacob [Thu, 17 Dec 1998 19:26:49 +0000 (19:26 +0000)]
import logical/hardware block locate ioctls from NetBSD

25 years agoFlush the tx fifo in cystop(). Now ttyflush() (and thus tcflush(3))
bde [Thu, 17 Dec 1998 19:23:09 +0000 (19:23 +0000)]
Flush the tx fifo in cystop().  Now ttyflush() (and thus tcflush(3))
almost works properly.  Unfortunately, there is no way to flush
the rx fifo without resetting the channel, which also flushes the
tx fifo.  We avoid resetting even when both fifos need to be flushed,
since resetting seems to cause the rx to lose sync if it is done
while data is arriving.

Reminded by: NIST-PCTS

25 years agoAdd some tape specific density codes- only enough of them here to recognize
mjacob [Thu, 17 Dec 1998 19:04:18 +0000 (19:04 +0000)]
Add some tape specific density codes- only enough of them here to recognize
some specific older units so we can choose 2FM@EOD or FIXED blocksize
quirks.

25 years agoSeveral changes having to do blocksize- mostly to force variable as the default.
mjacob [Thu, 17 Dec 1998 18:56:23 +0000 (18:56 +0000)]
Several changes having to do blocksize- mostly to force variable as the default.
Attempt to determine (at mount time if not done so already) via density code
whether a device should default to fixed mode or not. Attempts to set to
variable that fail will cause fixed to be selected.

Similarly, the '2 filemarks at EOM' quirk is now determined (or attempted to
be determined) via density code. Some as yet not entirely tested code for
coping with 2FM@EOD position is now also in place.

25 years agoFixed handling of BREAK in input. BREAK was not being converted into an
bde [Thu, 17 Dec 1998 18:43:08 +0000 (18:43 +0000)]
Fixed handling of BREAK in input.  BREAK was not being converted into an
escape sequence in the `-ignbrk -brkint parmrk' case.

Found by: NIST-PCTS

25 years agoOops, this should have been in the previous commit (ensure atomic update
bde [Thu, 17 Dec 1998 18:18:06 +0000 (18:18 +0000)]
Oops, this should have been in the previous commit (ensure atomic update
of com->cor[1]).

25 years agoImplemented sending of BREAKs. This is quite complicated because the
bde [Thu, 17 Dec 1998 17:40:13 +0000 (17:40 +0000)]
Implemented sending of BREAKs.  This is quite complicated because the
hardware is interrupt-driven to a fault and sending a BREAK requires
mode switching.  Always running in the BREAK-capable mode as in PR 8318
would double the overhead for sending \0's.

Reminded by: PR 8318

25 years agoNote that dying on NULL is an implementation detail.
eivind [Thu, 17 Dec 1998 17:13:47 +0000 (17:13 +0000)]
Note that dying on NULL is an implementation detail.

25 years agoLook for boot blocks in new default location.
jkh [Thu, 17 Dec 1998 16:50:10 +0000 (16:50 +0000)]
Look for boot blocks in new default location.

25 years agoRestore old semantics (broken in rev 1.47's buffer overflow fix).
eivind [Thu, 17 Dec 1998 16:31:02 +0000 (16:31 +0000)]
Restore old semantics (broken in rev 1.47's buffer overflow fix).

25 years agoActually document my last addition.
cracauer [Thu, 17 Dec 1998 12:21:20 +0000 (12:21 +0000)]
Actually document my last addition.
General cosmetics.
Add a note that scsi-defects has not been ported to CAM.
Submitted by: BDE

25 years agoAdd an option for insecure mode, in which rwhod does not discard packets
des [Thu, 17 Dec 1998 11:05:57 +0000 (11:05 +0000)]
Add an option for insecure mode, in which rwhod does not discard packets
from incorrect source ports.

25 years agoSync with sys/i386/isa/clock.c revision 1.129.
kato [Thu, 17 Dec 1998 08:54:47 +0000 (08:54 +0000)]
Sync with sys/i386/isa/clock.c revision 1.129.

25 years agoThe fe driver has been merged into sys/i386/isa/if_fe.c.
kato [Thu, 17 Dec 1998 08:45:07 +0000 (08:45 +0000)]
The fe driver has been merged into sys/i386/isa/if_fe.c.

Forgotten by: kato

25 years agoSync with sys/i386/isa/wd.c revision 1.181.
kato [Thu, 17 Dec 1998 08:42:10 +0000 (08:42 +0000)]
Sync with sys/i386/isa/wd.c revision 1.181.

25 years agoSync with sys/i386/isa/npx.c revision 1.64.
kato [Thu, 17 Dec 1998 08:41:34 +0000 (08:41 +0000)]
Sync with sys/i386/isa/npx.c revision 1.64.

25 years agoSync with sys/i386/isa/fd.c revision 1.129.
kato [Thu, 17 Dec 1998 08:40:36 +0000 (08:40 +0000)]
Sync with sys/i386/isa/fd.c revision 1.129.

25 years agoDon't return stack-based data. This may have caused
brian [Thu, 17 Dec 1998 00:28:12 +0000 (00:28 +0000)]
Don't return stack-based data.  This may have caused
server-side CHAP authentication problems in the past :-/

25 years agoFix a few problems with handling rejected transfer negotiation messages.
gibbs [Thu, 17 Dec 1998 00:06:52 +0000 (00:06 +0000)]
Fix a few problems with handling rejected transfer negotiation messages.

Use the host message loop for any unknown message types instead of performing
a reject message in the sequencer.  Pass reject messages to the host
message loop too which frees up a sequencer interrupt type slot.

Default to issuing a bus reset if initiator mode is enabled.  It seems
that the reset scsi bus bit is not defined in the same location for
all aic78xx BIOSes, so attempting to honor this setting will have to
wait until I get more information on how to detect it.

Nuke some unused variables.

25 years agoCorrectly track allocated accept target I/O ccbs so that they can be
gibbs [Thu, 17 Dec 1998 00:03:14 +0000 (00:03 +0000)]
Correctly track allocated accept target I/O ccbs so that they can be
aborted prior to disabling our lun.  This requires a second set of
links since we use the ones in the ccb_hdr during normal operations.

Nuke some unused variables.

25 years agoAt Justin's request, limit the size of buffers that can be mapped into
ken [Wed, 16 Dec 1998 21:00:06 +0000 (21:00 +0000)]
At Justin's request, limit the size of buffers that can be mapped into
and out of kernel address space (via the pass(4) and xpt(4) peripheral
drivers) to 64K (DFLTPHYS).  Some controllers, like the Adaptec 1542,
don't support more than 64K transactions.

We plan on eventually having the capability of limiting this size based
on min(MAXPHYS, controller max), but since that capability isn't here yet,
limit things to the lowest common denominator.

25 years ago<sys/types.h> isn't a prerequisite for <kvm.h>.
bde [Wed, 16 Dec 1998 19:12:14 +0000 (19:12 +0000)]
<sys/types.h> isn't a prerequisite for <kvm.h>.

25 years agoThe previous commit was bogus. <stdlib.h> was never a prerequisite
bde [Wed, 16 Dec 1998 19:04:54 +0000 (19:04 +0000)]
The previous commit was bogus.  <stdlib.h> was never a prerequisite
for <kvm.h> or kvm_getloadavg(), and <sys/types.h> was only a
prerequisite for <kvm.h> when <kvm.h> was broken.

25 years agoDeclare size_t and ssize_t if they are not already declared, so that
bde [Wed, 16 Dec 1998 18:59:47 +0000 (18:59 +0000)]
Declare size_t and ssize_t if they are not already declared, so that
<kvm.h> is self-sufficient again.

Moved typedefs and forward struct declarations out of __BEGIN_DECLS/
__END_DECLS.

Don't comment out the prototype for kvm_uread().  This was a 4 year
old kludge for previous breakage of self-sufficiency.  The prototypwe
was broken instead.

Fixed bitrot (const poisoning) in the type of kvm_uread().

Fixed order of the declaration of kvm_uread().

25 years agoStraigthen out the use of the tls and tlf callbacks.
phk [Wed, 16 Dec 1998 18:42:38 +0000 (18:42 +0000)]
Straigthen out the use of the tls and tlf callbacks.

Not tested on the if_sr, if_cx and if_ar drivers, but
expected to work just the same as it used to.

Any users of these drivers (or even better: donors
of hardware for them) please contact phk@freebsd.org
so we can test the next batch of changes to if_sppp.

25 years agoAdjust for kern.ps_strings and PS_STRINGS not being a pointer. This
bde [Wed, 16 Dec 1998 18:31:51 +0000 (18:31 +0000)]
Adjust for kern.ps_strings and PS_STRINGS not being a pointer.  This
fixes a type mismatch in the call to kvm_uread().  The bug has gone
undetected for almost 3 years because kvm_uproc()'s protoype has been
disabled for almost 4 years.

Trust sysctlbyname() to work properly if it succeeds.

Fixed style bugs in revs. 1.19 and 1.22.

25 years agoGeneralize the if_up() and if_down() functions under the names
phk [Wed, 16 Dec 1998 18:30:43 +0000 (18:30 +0000)]
Generalize the if_up() and if_down() functions under the names
if_route() and if_unroute().

This is first step towards sanitizing IFF_UP and IFF_RUNNING

25 years agoAdd missing "break"s to allow multicast routing to work.
fenner [Wed, 16 Dec 1998 18:07:11 +0000 (18:07 +0000)]
Add missing "break"s to allow multicast routing to work.

Submitted by: Amancio Hasty <hasty@rah.star-gate.com>

25 years agoProbable fix for the "cdda2wav" panics that various people have been
ken [Wed, 16 Dec 1998 18:00:39 +0000 (18:00 +0000)]
Probable fix for the "cdda2wav" panics that various people have been
reporting since this past summer.  (I think Daniel O'Conner was the first.)

The problem appears to have been something like this:

 - cdda2wav by default passes in a buffer that is close to the 128K MAXPHYS
   limit.
 - many times, the buffer is not page aligned
 - vmapbuf() truncates the address, so that it is page aligned
 - that causes the total size of the buffer to be greater than MAXPHYS,
   which of course is a bad thing.

Here's a quote from the PR (kern/9067):

==================
In particular, note bp->b_bufsize = 0x0001f950 and bp->b_data = 0xf2219960
(which does not start on a page boundary).  vunmapbuf() loops through all
the pages without any difficulty until addr reaches 0xf2239000, and then
the panic occurs.  This seems to indicate that we are exceeding MAXPHYS
since we actually started from the middle of a page (the data is being
transfered to a non page aligned location).

To complete the description, note that the system call originates from
ReadCddaMMC12() (in scsi_cmds.c of cdda2wav) with a request to read 55
audio sectors of 2352 bytes (which is calculated to fall under MAXPHYS).
This in turn ends up calling scsi_send() (in scsi-bsd.c) which calls
cam_fill_csio() and cam_send_ccb().  This results in a CAMIOCOMMAND ioctl
with a ccb function code of XPT_SCSI_IO.
==================

The fix is to change the size check in cam_periph_mapmem() so that it is
like the one in minphys().  In particular, it is something like:

if ((buffer_length + (buf_ptr & PAGE_MASK)) > MAXPHYS)
buffer is too big

My fix is based on the one in the PR, but I cleaned up a fair number of
things in cam_periph_mapmem().  The checks for each buffer to be mapped
are now in a separate loop from the actual mapping operation.  With the new
arrangement, we don't have to bother with unmapping any previously mapped
buffers if one of the checks fails.

Many thanks to James Liu for tracking this down.  I'd appreciate it if some
vm-savvy folks would look this over.  I believe this fix is correct, but I
could be wrong.

PR: kern/9067 (also, kern/8112)
Reviewed by: gibbs
Submitted by: "James T. Liu" <jtliu@phlebas.rockefeller.edu>