]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
26 years agoIf we get back too much data to fit in result, return NULL. This avoids
Warner Losh [Tue, 9 Jun 1998 05:06:27 +0000 (05:06 +0000)]
If we get back too much data to fit in result, return NULL.  This avoids
a buffer overflow, but might negatively impact those hosts who have
enough aliases to fill MAXHOSTNAMELEN * 2 characters in them.

Good candidate for merging back into -stable.  Lightly tested by me, but
it came from OpenBSD a while ago.

Obtained from: OpenBSD

26 years agoUse correct size for readlink buffer. This is the size of the
Warner Losh [Tue, 9 Jun 1998 05:02:29 +0000 (05:02 +0000)]
Use correct size for readlink buffer.  This is the size of the
buffer -1 since readlink adds its own NUL to the end.

Inspired by: Similar changes in OpenBSD

26 years agouse strncpy correctly
Warner Losh [Tue, 9 Jun 1998 05:01:38 +0000 (05:01 +0000)]
use strncpy correctly
use strncat correctly
check buffer size in graph_name
Change paths from /tmp to /var/run for server sockets

This has been building in my tree for ages, but has remained lightly
tested.
Reviewed by: jkh ages ago

26 years agoo be careful about buffer overflows
Warner Losh [Tue, 9 Jun 1998 04:58:36 +0000 (04:58 +0000)]
o be careful about buffer overflows
o Use snprintf rather than sprintf

This is only lightly tested.

26 years agoRemove unused strvisx.
Warner Losh [Tue, 9 Jun 1998 04:35:20 +0000 (04:35 +0000)]
Remove unused strvisx.

26 years agoDon't assume that hp->h_lenght == 4. Be conservative in its use.
Warner Losh [Tue, 9 Jun 1998 04:31:02 +0000 (04:31 +0000)]
Don't assume that hp->h_lenght == 4.  Be conservative in its use.
Submitted by:  J. Assange a long time ago.

26 years agoo Use snprintf rather than sprintf.
Warner Losh [Tue, 9 Jun 1998 04:28:05 +0000 (04:28 +0000)]
o Use snprintf rather than sprintf.
o Revert gid to caller before a shell so we can run setgid.
o minor style nits to make bruce happy :-)
o use strncpy correctly.

26 years agoUse mkstemp rather than mktemp for yacc's temp files. This change was made
Warner Losh [Tue, 9 Jun 1998 04:20:51 +0000 (04:20 +0000)]
Use mkstemp rather than mktemp for yacc's temp files.  This change was made
to OpenBSD a long time ago and to my tree shortly thereafter.  I think theo
made this change, or one similar to it, but I could be wrong.

26 years agoo Use snprintf rather than sprintf
Warner Losh [Tue, 9 Jun 1998 04:17:29 +0000 (04:17 +0000)]
o Use snprintf rather than sprintf
o Add more checks for buffer overflows
o Use snprintf rather than strcat/cpy and have better checks for max
  length exceeded.

Most of these changes are not exploitable buffer overruns, but it never
hurts to be safe.

Inspired by and obtained from: OpenBSD

26 years agoo Use snprintf over sprintf.
Warner Losh [Tue, 9 Jun 1998 04:13:03 +0000 (04:13 +0000)]
o Use snprintf over sprintf.
o Use strncpy correctly.
o Use enough buffer for line.

Inspired by or Obtained from: Similar changes in OpenBSD

26 years agoMake sure we pass the length - 1 to readlink, since it adds its own
Warner Losh [Tue, 9 Jun 1998 04:10:08 +0000 (04:10 +0000)]
Make sure we pass the length - 1 to readlink, since it adds its own
NUL at the end of the path.

Inspired by: OpenBSD's changes in this area by theo de raadt

26 years ago$@ is deprecated, use ${.TARGET} instead. Ditto for other single character
Warner Losh [Tue, 9 Jun 1998 04:07:23 +0000 (04:07 +0000)]
$@ is deprecated, use ${.TARGET} instead.  Ditto for other single character
$n.
PR:  conf/3273

26 years agoMake sure we don't overflow the path buffer. Exit if we do.
Warner Losh [Tue, 9 Jun 1998 03:39:38 +0000 (03:39 +0000)]
Make sure we don't overflow the path buffer.  Exit if we do.
Obtained from or inspired by: A similar change in OpenBSD by theo

26 years agoMake sure we pass the length - 1 to readlink, since it adds its own
Warner Losh [Tue, 9 Jun 1998 03:38:43 +0000 (03:38 +0000)]
Make sure we pass the length - 1 to readlink, since it adds its own
NUL at the end of the path.
Inspired by: OpenBSD's changes in this area by theo de raadt

26 years agoUpdated generated files.
Bruce Evans [Tue, 9 Jun 1998 03:36:59 +0000 (03:36 +0000)]
Updated generated files.

26 years agoDon't generate COMPAT_43 cruft if there are no COMPAT_43 syscalls.
Bruce Evans [Tue, 9 Jun 1998 03:32:05 +0000 (03:32 +0000)]
Don't generate COMPAT_43 cruft if there are no COMPAT_43 syscalls.
In particular, don't generate an include of "opt_compat.h" if it
wouldn't affect anything we create.  This will fix recent breakage
of the ibcs2 LKM.  The ibcs2 syscall files were not regenerated
properly, so the LKM didn't break immediately when we started
generating this extraneous include.

26 years agoInclude "opt_compat.h" explicitly instead of depending on
Bruce Evans [Tue, 9 Jun 1998 03:28:14 +0000 (03:28 +0000)]
Include "opt_compat.h" explicitly instead of depending on
makesyscalls.sh generating it as pollution.

Don't attempt to separate the includes by newlines.  makesyscalls.sh
just ignores newlines.

26 years agoFixed `make depend', which I broke here about a month ago by moving
Bruce Evans [Tue, 9 Jun 1998 02:55:40 +0000 (02:55 +0000)]
Fixed `make depend', which I broke here about a month ago by moving
the null default for `depend' from bsd.dep.mk to bsd.obj.mk.
bsd.kmod.mk included these files in a bad order, so it picked up
the null default.

26 years agoThe version of the softdep changes in FreeBSD broke the
Julian Elischer [Mon, 8 Jun 1998 23:55:33 +0000 (23:55 +0000)]
The version of the softdep changes in FreeBSD broke the
(doingdirectory && !newparent) case of ufs_rename().
rename("D1/X/", "D2/Y/") gives a wrong link count for D2.

Submitted by: Bruce Evans <bde@zeta.org.au>
Reviewed by: Kirk McKusick <mckusick@McKusick.COM>

26 years ago- Added entry for Xircon CreditCard Modem.
Nate Williams [Mon, 8 Jun 1998 21:10:41 +0000 (21:10 +0000)]
- Added entry for Xircon CreditCard Modem.

Submitted by:  "Mike Del" <repenting@hotmail.com>

26 years agoMy previous change could conceivably succeed at removing CHROOTDIR.
John Polstra [Mon, 8 Jun 1998 20:58:02 +0000 (20:58 +0000)]
My previous change could conceivably succeed at removing CHROOTDIR.
Ignore errors from the chflags command that follows it, just in case.

26 years agoDon't let ifunit() modify the string passed as an argument.
Julian Elischer [Mon, 8 Jun 1998 20:33:29 +0000 (20:33 +0000)]
Don't let ifunit() modify the string passed as an argument.
it may be in the text segment and write protected.

26 years agoCorrect ``set server'' usage and add mention that ``set ns'' changed
Brian Somers [Mon, 8 Jun 1998 20:23:44 +0000 (20:23 +0000)]
Correct ``set server'' usage and add mention that ``set ns'' changed
in README.changes.
Suggested by: stuart henderson <stuart@internationalschool.co.uk>

26 years agoRemove some junk left over from a previous commit.
John Dyson [Mon, 8 Jun 1998 18:18:28 +0000 (18:18 +0000)]
Remove some junk left over from a previous commit.
Submitted by: phk

26 years agoPrint the write counts if they are nonzero even if we're mounted
Bruce Evans [Mon, 8 Jun 1998 14:08:54 +0000 (14:08 +0000)]
Print the write counts if they are nonzero even if we're mounted
readonly, since they tell us about previous write activity.

Use the correct format to print the write counts.

26 years agoUpdated generated files.
Bruce Evans [Mon, 8 Jun 1998 11:08:35 +0000 (11:08 +0000)]
Updated generated files.

26 years agoFixed some style bugs in output (missing tabs and unparenthesized macros).
Bruce Evans [Mon, 8 Jun 1998 11:02:00 +0000 (11:02 +0000)]
Fixed some style bugs in output (missing tabs and unparenthesized macros).

Fixed some style bugs in source (mostly, superfluous backslashes).

26 years agoAdd LIB_SPEC (same as on i386) so that gcc knows about libc_r.
John Birrell [Mon, 8 Jun 1998 10:58:36 +0000 (10:58 +0000)]
Add LIB_SPEC (same as on i386) so that gcc knows about libc_r.

26 years agoFixed pedantic semantics errors (bitfields not of type int, signed int
Bruce Evans [Mon, 8 Jun 1998 09:47:47 +0000 (09:47 +0000)]
Fixed pedantic semantics errors (bitfields not of type int, signed int
or unsigned int (this doesn't change the struct layout, size or
alignment in any of the files changed in this commit, at least for
gcc on i386's.  Using bitfields of type u_char may affect size and
alignment but not packing)).

26 years agoFix more of my DDB breakage.
Doug Rabson [Mon, 8 Jun 1998 09:23:24 +0000 (09:23 +0000)]
Fix more of my DDB breakage.

26 years agoFix a typo which prevented i386 elf from working at all (including Linux
Doug Rabson [Mon, 8 Jun 1998 09:19:35 +0000 (09:19 +0000)]
Fix a typo which prevented i386 elf from working at all (including Linux
emulated elf binaries).

26 years agoCorrect even more ioctl breakage :(
Søren Schmidt [Mon, 8 Jun 1998 09:00:28 +0000 (09:00 +0000)]
Correct even more ioctl breakage :(
Doesn't people check their code before committing anymore, it could
at least compile  ????

26 years agoSync with sys/i386/isa/clock.c revision 1.123.
KATO Takenori [Mon, 8 Jun 1998 08:56:43 +0000 (08:56 +0000)]
Sync with sys/i386/isa/clock.c revision 1.123.

26 years agoMerge 64bit portability fixes from sys/i386 stuff.
KATO Takenori [Mon, 8 Jun 1998 08:55:47 +0000 (08:55 +0000)]
Merge 64bit portability fixes from sys/i386 stuff.

26 years agoFixed pedantic semantics errors (bitfields not of type int, signed int
Bruce Evans [Mon, 8 Jun 1998 08:50:43 +0000 (08:50 +0000)]
Fixed pedantic semantics errors (bitfields not of type int, signed int
or unsigned int (don't do this in one case where it would change the
object file)).

26 years agoMake DDB work again after I broke it :-(.
Doug Rabson [Mon, 8 Jun 1998 08:43:20 +0000 (08:43 +0000)]
Make DDB work again after I broke it :-(.

26 years agoAdd rcsid, remove unused #includes. Use err(3).
Philippe Charnier [Mon, 8 Jun 1998 06:44:05 +0000 (06:44 +0000)]
Add rcsid, remove unused #includes. Use err(3).

26 years agoFix spelling in printf().
Joseph Koshy [Mon, 8 Jun 1998 06:42:34 +0000 (06:42 +0000)]
Fix spelling in printf().

26 years agoUse warnx()+fprintf() to handle multi-line messages. Correct K&R support and
Philippe Charnier [Mon, 8 Jun 1998 06:41:47 +0000 (06:41 +0000)]
Use warnx()+fprintf() to handle multi-line messages. Correct K&R support and
KNF continuation indent rule.
Requested by: Bruce.

26 years agoUpdated yet another ioctl, and put wst in LINT to inhibit further bitrot.
Bruce Evans [Mon, 8 Jun 1998 06:18:53 +0000 (06:18 +0000)]
Updated yet another ioctl, and put wst in LINT to inhibit further bitrot.

26 years agoSpelling fixes.
Joseph Koshy [Mon, 8 Jun 1998 06:12:02 +0000 (06:12 +0000)]
Spelling fixes.

Submitted by: Josh Gilliam <josh@quick.net>
PR: 6880

26 years agoip_fil.h has 9 separate declarations of iplioctl() in a disgusting
Bruce Evans [Mon, 8 Jun 1998 06:04:12 +0000 (06:04 +0000)]
ip_fil.h has 9 separate declarations of iplioctl() in a disgusting
ifdef tangle.  The previous commit to ip_fil.h didn't change the
one that actually applies to the current FreeBSD kernel, of course.
Fixed.

Fixed style bugs in previous commit to ip_fil.h.

26 years agoReword to remove confusion between su(1)'s "-c" option and that of the shell
Joseph Koshy [Mon, 8 Jun 1998 05:29:51 +0000 (05:29 +0000)]
Reword to remove confusion between su(1)'s "-c" option and that of the shell
being invoked.  Add example that clarifies usage of "-c".

PR: 6859

26 years agoDisable attempted write combining support. This probably
John Dyson [Mon, 8 Jun 1998 04:07:51 +0000 (04:07 +0000)]
Disable attempted write combining support.  This probably
causes more trouble than help (for now.)

26 years agoFixed an ioctl which grew overnight.
Bruce Evans [Mon, 8 Jun 1998 02:43:12 +0000 (02:43 +0000)]
Fixed an ioctl which grew overnight.

26 years agoBacked out previous commit, which just consisted of a 1-byte editing
Bruce Evans [Mon, 8 Jun 1998 02:15:37 +0000 (02:15 +0000)]
Backed out previous commit, which just consisted of a 1-byte editing
error.

26 years agoFixed missing ifdefs and disordering in previous commit.
Bruce Evans [Mon, 8 Jun 1998 02:12:19 +0000 (02:12 +0000)]
Fixed missing ifdefs and disordering in previous commit.

26 years agoFixed ifdef bogotification in previous commit.
Bruce Evans [Mon, 8 Jun 1998 02:11:19 +0000 (02:11 +0000)]
Fixed ifdef bogotification in previous commit.

26 years agoFix typo: exit -> exist
Daniel O'Callaghan [Mon, 8 Jun 1998 02:00:45 +0000 (02:00 +0000)]
Fix typo: exit -> exist

26 years agoSpeed up the cleaning of CHROOTDIR by almost a factor of two, by
John Polstra [Sun, 7 Jun 1998 21:53:01 +0000 (21:53 +0000)]
Speed up the cleaning of CHROOTDIR by almost a factor of two, by
replacing two monstrous file tree walks with one monstrous tree walk
and two miniscule ones.

26 years agoAdd a member function more to the timecounters, this one is for use
Poul-Henning Kamp [Sun, 7 Jun 1998 20:36:55 +0000 (20:36 +0000)]
Add a member function more to the timecounters, this one is for use
with latch based PPS implementations.  The client that uses it will
be committed after more testing.

26 years agoFix another DEVFS/SLICE ioctl.
Doug Rabson [Sun, 7 Jun 1998 20:10:53 +0000 (20:10 +0000)]
Fix another DEVFS/SLICE ioctl.

26 years agoThis is a prototype implementation of the draft-mogul-pps-api-##.txt
Poul-Henning Kamp [Sun, 7 Jun 1998 19:44:22 +0000 (19:44 +0000)]
This is a prototype implementation of the draft-mogul-pps-api-##.txt
paper.

It will be updated along with the draft and possible subsequent
standard.

The ppbus based pps driver is updated to implement this API.

26 years agoFix some more ioctls which I missed becausese they were hidden by options
Doug Rabson [Sun, 7 Jun 1998 19:40:41 +0000 (19:40 +0000)]
Fix some more ioctls which I missed becausese they were hidden by options
which were not in LINT.

26 years agoFix the retyping of the ioctl cmd parm from int to u_long.
Søren Schmidt [Sun, 7 Jun 1998 18:44:03 +0000 (18:44 +0000)]
Fix the retyping of the ioctl cmd parm from int to u_long.

26 years agoAlways ask for homedir.
Wolfram Schneider [Sun, 7 Jun 1998 18:38:32 +0000 (18:38 +0000)]
Always ask for homedir.

PR: 6754
Submitted by: uhlar@netlab.sk

26 years agoThis commit fixes various 64bit portability problems required for
Doug Rabson [Sun, 7 Jun 1998 17:13:14 +0000 (17:13 +0000)]
This commit fixes various 64bit portability problems required for
FreeBSD/alpha.  The most significant item is to change the command
argument to ioctl functions from int to u_long.  This change brings us
inline with various other BSD versions.  Driver writers may like to
use (__FreeBSD_version == 300003) to detect this change.

The prototype FreeBSD/alpha machdep will follow in a couple of days
time.

26 years agoFix previous commit. bit 1 is the execute bit, not the read bit *blush*
Brian Somers [Sun, 7 Jun 1998 17:08:42 +0000 (17:08 +0000)]
Fix previous commit.  bit 1 is the execute bit, not the read bit *blush*
ie, BINMODE=4551 -> BINMODE=4554
Pointed out by: Bruce Evans <bde@zeta.org.au>

26 years agoPR: docs/3636
Peter Hawkins [Sun, 7 Jun 1998 15:08:50 +0000 (15:08 +0000)]
PR: docs/3636
Submitted by: Gary Palmer gpalmer@FreeBSD.ORG
Add mention of the 1024-character line length limit on the netgroup database.

26 years agoSigh, all good words are reserved words these days...
Poul-Henning Kamp [Sun, 7 Jun 1998 14:14:39 +0000 (14:14 +0000)]
Sigh, all good words are reserved words these days...
s/private/priv/

Noticed by: sos

26 years agoAdded a used include (in ifdefed code).
Bruce Evans [Sun, 7 Jun 1998 12:02:59 +0000 (12:02 +0000)]
Added a used include (in ifdefed code).

26 years agoFixed pedantic semantics errors (bitfields not of type int, signed int
Bruce Evans [Sun, 7 Jun 1998 12:00:13 +0000 (12:00 +0000)]
Fixed pedantic semantics errors (bitfields not of type int, signed int
or unsigned int).

26 years agoFixed pedantic syntax errors caused by a trailing semicolon in a macro
Bruce Evans [Sun, 7 Jun 1998 11:52:17 +0000 (11:52 +0000)]
Fixed pedantic syntax errors caused by a trailing semicolon in a macro
definition.

26 years agoNull change. Forgot to mention in previous log message that MNT_NOATIME
Bruce Evans [Sun, 7 Jun 1998 11:04:26 +0000 (11:04 +0000)]
Null change.  Forgot to mention in previous log message that MNT_NOATIME
is now ignored for special files, so that mounting root with option
noatime doesn't break reporting of idle times in programs like `w'.
The problem of execessive disk updates just to stamp atimes will be
handled for special files by only writing atimes to disk when inodes
become active.  This works well because special files are relatively
uncommon and their atimes are even more disposable at panic time than
regular files' atimes.

26 years agoCompile mtree before it is used, this is a hack, but make world has
Søren Schmidt [Sun, 7 Jun 1998 10:50:28 +0000 (10:50 +0000)]
Compile mtree before it is used, this is a hack, but make world has
failed since new functionality was introduced in mtree.

26 years agoFixed some longstanding timestamp bugs:
Bruce Evans [Sun, 7 Jun 1998 10:49:18 +0000 (10:49 +0000)]
Fixed some longstanding timestamp bugs:
1. mark atimes and mtimes of special files and fifos for update upon
   successful completion of non-null i/o, not at the beginning of the
   syscall.
2. never update file times for readonly filesystems.  They were updated
   for stats and closes but not for syncs.  The updates were of course
   only in-core and were thrown away when the inode was uncached, so
   the times sometimes appeared to go backwards.

Improved comments in code related to (1) (mostly by removing them).

Unmacroized ITIMES().  The test in (2) bloated it even more.  Don't
call getmicrotime() in the function version of it when we only need
the time in seconds.

26 years agoSync with sys/i386/isa/clock.c revision 1.122.
KATO Takenori [Sun, 7 Jun 1998 09:51:08 +0000 (09:51 +0000)]
Sync with sys/i386/isa/clock.c revision 1.122.

26 years agoAdd a "this" style argument and a "void *private" so timecounters can
Poul-Henning Kamp [Sun, 7 Jun 1998 08:40:53 +0000 (08:40 +0000)]
Add a "this" style argument and a "void *private" so timecounters can
figure out which instance to wount with.

26 years agoUse a relative libalias.so name (hardcoded at 2.5 - yeuch!).
Brian Somers [Sun, 7 Jun 1998 03:54:41 +0000 (03:54 +0000)]
Use a relative libalias.so name (hardcoded at 2.5 - yeuch!).

26 years agoSearch for libraries in dlopen() when the specified path
Brian Somers [Sun, 7 Jun 1998 03:53:08 +0000 (03:53 +0000)]
Search for libraries in dlopen() when the specified path
contains no ``/''s.
Elf already searches it seems.
Mostly submitted by: Mike Smith <mike@smith.net.au>

26 years agoMake ppp world-readable so that root can get at it over
Brian Somers [Sun, 7 Jun 1998 00:16:57 +0000 (00:16 +0000)]
Make ppp world-readable so that root can get at it over
NFS when nobody isn't in group network :-)

Pointed out by: Bruce Evans <bde@zeta.org.au>

26 years agoUpdate the write descriptor selection set for our data links
Brian Somers [Sun, 7 Jun 1998 00:16:37 +0000 (00:16 +0000)]
Update the write descriptor selection set for our data links
*after* shuffling fragments from the IP queue into the
individual link queues.

This fixes a latency problem pointed out by: Bruce Evans <bde@zeta.org.au>

26 years agoFix an alignment problem on alpha by doing a bytewise copy.
John Birrell [Sat, 6 Jun 1998 23:33:28 +0000 (23:33 +0000)]
Fix an alignment problem on alpha by doing a bytewise copy.

26 years agoMake ``set ? log'' more verbose.
Brian Somers [Sat, 6 Jun 1998 23:00:33 +0000 (23:00 +0000)]
Make ``set ? log'' more verbose.
Suggested by: Paul Dufresne <dufrp@oricom.ca>

26 years agoDon't call PunchFWHole() ifdef NO_FW_PUNCH
Brian Somers [Sat, 6 Jun 1998 21:52:37 +0000 (21:52 +0000)]
Don't call PunchFWHole() ifdef NO_FW_PUNCH
Pointed out by: "Steve Sims" <SimsS@IBM.Net>

26 years agoMake sure the default value of a dummy variable is 0
Julian Elischer [Sat, 6 Jun 1998 21:49:17 +0000 (21:49 +0000)]
Make sure the default value of a dummy variable is 0
so that it doesn't do anything.

26 years agoPut the correct pid in /var/run/tunX.pid
Brian Somers [Sat, 6 Jun 1998 20:50:57 +0000 (20:50 +0000)]
Put the correct pid in /var/run/tunX.pid
Suggested by: many

26 years agoFix wrong data type for a pointer.
Julian Elischer [Sat, 6 Jun 1998 20:45:28 +0000 (20:45 +0000)]
Fix wrong data type for a pointer.

26 years agoclean up the changes made to ipfw over the last weeks
Julian Elischer [Sat, 6 Jun 1998 19:39:10 +0000 (19:39 +0000)]
clean up the changes made to ipfw over the last weeks
(should make the ipfw lkm work again)

26 years agoFix this so that it compiles in the !__STDC__ case.
Steve Price [Sat, 6 Jun 1998 18:52:43 +0000 (18:52 +0000)]
Fix this so that it compiles in the !__STDC__ case.
Also be consistent about usage of #if ...

Pointed out by: bde

26 years agokeymap -> key_map so that the kernel will compile with
Steve Price [Sat, 6 Jun 1998 17:45:11 +0000 (17:45 +0000)]
keymap -> key_map so that the kernel will compile with
-DESKEYMAP.

PR: 6864
Submitted by: Javier Rueda <jmrueda@diatel.upm.es>

26 years agoPR: docs/5986
Peter Hawkins [Sat, 6 Jun 1998 15:37:23 +0000 (15:37 +0000)]
PR: docs/5986
Submitted by: MITSUNAGA Noriaki mitchy@er.ams.eng.osaka-u.ac.jp
Replace use of obsolete .q groff macro with .Dq

26 years agoPR: docs/5628
Peter Hawkins [Sat, 6 Jun 1998 15:16:02 +0000 (15:16 +0000)]
PR: docs/5628
Submitted by: MITSUNAGA Noriaki mitchy@er.ams.eng.osaka-u.ac.jp
Correct formatting error in display of the "-i size" option description section.

26 years agoSync PATH variable with 'dot.login'.
Jun Kuriyama [Sat, 6 Jun 1998 15:01:46 +0000 (15:01 +0000)]
Sync PATH variable with 'dot.login'.

I think here is not my area so feel free to back this out when any
troubles are happened.

Pointed out by: IIJIMA Akihiro <aki@noc.titech.ac.jp>

26 years agoAdd a warning message for a thread locking against itself. This is
John Birrell [Sat, 6 Jun 1998 07:27:06 +0000 (07:27 +0000)]
Add a warning message for a thread locking against itself. This is
not supposed to happen, but I have seen bogus g++ code that causes
it.

26 years agoSimplify the handling of thread specific data. Only track if a key
John Birrell [Sat, 6 Jun 1998 07:24:24 +0000 (07:24 +0000)]
Simplify the handling of thread specific data. Only track if a key
is allocated or not, rather than keeping a count and attempting to
know it it is in-use. POSIX says that once a key is deleted, using the
key again results in undefined behaviour.

26 years agoRe-design the thread specific key structure.
John Birrell [Sat, 6 Jun 1998 07:20:23 +0000 (07:20 +0000)]
Re-design the thread specific key structure.

26 years agonamed-xfer can't be linked static due to the duplicated symbols in
John Birrell [Sat, 6 Jun 1998 07:09:01 +0000 (07:09 +0000)]
named-xfer can't be linked static due to the duplicated symbols in
libc and libbind.

rpc.rstatd required libkvm.

Only try to build these on i386 for the time being.

26 years agoI got the last commit back to front.
John Birrell [Sat, 6 Jun 1998 07:02:27 +0000 (07:02 +0000)]
I got the last commit back to front.

26 years agoSpelling corrections.
Joseph Koshy [Sat, 6 Jun 1998 05:50:53 +0000 (05:50 +0000)]
Spelling corrections.

PR: 6868
Submitted by: Josh Gilliam <josh@quick.net>

26 years agoMake LINE30 a new style option.
KATO Takenori [Sat, 6 Jun 1998 05:25:37 +0000 (05:25 +0000)]
Make LINE30 a new style option.

26 years agoMake BS_TARG_SAFEMODE a new style option.
KATO Takenori [Sat, 6 Jun 1998 05:21:56 +0000 (05:21 +0000)]
Make BS_TARG_SAFEMODE a new style option.

26 years agoSpelling nits.
Joseph Koshy [Sat, 6 Jun 1998 04:56:13 +0000 (04:56 +0000)]
Spelling nits.

Pointed out by: Sue Blake <sue@welearn.com.au>

26 years ago-DALLLANG is now obsolete in "src/Makefile"; move to "src/release/Makefile".
Joseph Koshy [Sat, 6 Jun 1998 04:39:42 +0000 (04:39 +0000)]
-DALLLANG is now obsolete in "src/Makefile"; move to "src/release/Makefile".

PR: 6685

26 years agoDon't attempt to copy the whole slices "struct" for DIOCGSLICEINFO.
Bruce Evans [Sat, 6 Jun 1998 03:06:55 +0000 (03:06 +0000)]
Don't attempt to copy the whole slices "struct" for DIOCGSLICEINFO.
The slices "struct" isn't really a struct; we allocate only part of
it in the fully dangerously dedicated case.  Since the "struct" is
malloced, the page beyond it may not be mapped, so attempts to copy
it would crash.  This problem became larger when the full struct was
bloated from < 1K to > 3K by the addition of (mostly unused) DEVFS
tokens some time before 2.2.0 was released.

26 years agoRemoved dead code.
Bruce Evans [Sat, 6 Jun 1998 02:32:51 +0000 (02:32 +0000)]
Removed dead code.

26 years agoDon't run the target's ldconfig or build it as a tool. If target == host
Bruce Evans [Sat, 6 Jun 1998 01:13:21 +0000 (01:13 +0000)]
Don't run the target's ldconfig or build it as a tool.  If target == host
(in particular, if DESTDIR is empty or "/"), then the host's ldconfig will
be the target's ldconfig by the time it is run.

Fixed disordering of env.

Don't know too much about libcrypt.  Use exactly the same definition of
_libcrypt as lib/Makefile.

Don't build strip twice.

26 years agoFixed bogotification of the lex bootstrap in rev.1.186 and the previous
Bruce Evans [Sat, 6 Jun 1998 00:56:23 +0000 (00:56 +0000)]
Fixed bogotification of the lex bootstrap in rev.1.186 and the previous
commit - don't wander off to bootstrap mtree and include in the middle
of bootstrapping lex, and don't forget what we were doing and build
some lex obj dirs twice.

26 years agoReviewed by: Kirk Mckusick (mckusick@mckusick.com)
Julian Elischer [Fri, 5 Jun 1998 23:33:26 +0000 (23:33 +0000)]
Reviewed by: Kirk Mckusick (mckusick@mckusick.com)
Submitted by: luoqi Chen
fix a type in fsck.
(also add a comment that got picked up by mistake but is worth adding)

26 years agoFix the signal behaviour for internal states which set the thread
John Birrell [Fri, 5 Jun 1998 23:31:55 +0000 (23:31 +0000)]
Fix the signal behaviour for internal states which set the thread
state to running despite the SA_RESTART flag which is really just for
syscalls.