Eirik Nygaard [Wed, 23 Jun 2004 20:40:50 +0000 (20:40 +0000)]
This is less, correct name in upgrade file.
Noticed by: Simon 'corecode' Schubert <corecode@fs.ei.tum.de>
Matthew Dillon [Wed, 23 Jun 2004 16:45:23 +0000 (16:45 +0000)]
Fix a broken vrele() in the session tty exit code.
Submitted-by: Joerg Sonnenberger <joerg@britannica.bec.de>
Matthew Dillon [Wed, 23 Jun 2004 16:15:24 +0000 (16:15 +0000)]
Hack in the code from FreeBSD-5 to set the timings for NVIDIA/AMD chipsets.
This seems to greatly reduce READ_BIG: ILLEGAL REQUEST problems accessing
a CD on my AMD64 box (NVidia3 chipset, secondary controller, master, PIO4).
The problem occurs at a much lower frequency in overnight tests. Further
testing is ongoing.
Matthew Dillon [Wed, 23 Jun 2004 06:53:13 +0000 (06:53 +0000)]
Do a partial synch from FreeBSD-5 of the NVIDIA and NFORCE ATA setup code.
Our previous code was totally bogus.
Matthew Dillon [Wed, 23 Jun 2004 06:52:26 +0000 (06:52 +0000)]
Fix an improper DELAY in the ata tag code (but nobody should be using
the ata tag code anyway).
Matthew Dillon [Wed, 23 Jun 2004 06:51:55 +0000 (06:51 +0000)]
Check for a queued interrupt being dispatched after the ATA driver
diabled the controller interrupt and return if so. Enclose the ATA
command setup in a critical section.
Hiten Pandya [Wed, 23 Jun 2004 06:27:28 +0000 (06:27 +0000)]
Fix indentation.
Hiten Pandya [Wed, 23 Jun 2004 06:17:50 +0000 (06:17 +0000)]
Remove references to FreeBSD in the comments.
Discussed-with: Matthew Dillon <dillon@backplane.com>
Matthew Dillon [Tue, 22 Jun 2004 21:46:54 +0000 (21:46 +0000)]
Unbreak the buildworld by fixing a cc_tools dependancy on cc_prep in
Makefile.inc1 instead of trying to create a dependancy in cc_tools. The
object tree had not yet been created for cc_prep and this caused files to
be generated in /usr/src, amoung other things.
Chris Pressey [Tue, 22 Jun 2004 20:51:30 +0000 (20:51 +0000)]
Perl is no longer user during kernel build, so remove the Perl scripts
that are no longer needed for this purpose.
Confirmed-with: eirikn
Chris Pressey [Tue, 22 Jun 2004 20:37:02 +0000 (20:37 +0000)]
Make virecover more robust. Instead of aborting (and causing
subsequent RCNG scripts to fail) when the directory /var/tmp/vi.recover
doesn't exist, just re-create it.
Reviewed-by: eirikn, Devon H. O'Dell, GeekGod
Matthew Dillon [Tue, 22 Jun 2004 16:40:32 +0000 (16:40 +0000)]
Turn propolice (-fstack-protector) on by default. Propolice rearranges
the ordering of local variable declarations so pointers do not get
overwritten by buffer overflows, and checks a magic marker on return to see
if the return address has been blown away. It's a very important security
feature.
Matthew Dillon [Mon, 21 Jun 2004 21:50:02 +0000 (21:50 +0000)]
Fix -j builds for gcc-3.4. The .nx build in cc_tools was breaking
parallelism because individual source/header dependancies were being built
in parallel, and this was causing the helper programs to build in parallel
and/or generate the same header file in parallel.
Remove all direct dependancies on the helper programs. Make libgcc
dependant on cc_tools (to generate required header files), clean up
the dependancy target for cc_tools, and force make depend to build
the tools in bsd.hostprog.mk. This should result in all the tools being
available to the rest of the gcc-3.4 build.
Tested with several -j 8 builds. There could still be additional
-j issues.
Matthew Dillon [Mon, 21 Jun 2004 21:49:13 +0000 (21:49 +0000)]
Fix -j builds for gcc-3.4. The .nx build in cc_tools was breaking
parallelism because individual source/header dependancies were being built
in parallel, and this was causing the helper programs to build in parallel
and/or generate the same header file in parallel.
Remove all direct dependancies on the helper programs. Make libgcc
dependant on cc_tools (to generate required header files), clean up
the dependancy target for cc_tools, and force make depend to build
the tools. This should result in all the tools being available to the
rest of the gcc-3.4 build.
Tested with several -j 8 builds. There could still be additional
-j issues.
Chris Pressey [Mon, 21 Jun 2004 17:47:12 +0000 (17:47 +0000)]
Merge with FreeBSD (HEAD) src/usr.sbin/adduser/*:
Move towards Perl removal from the base by replacing the Perl scripts
`adduser' and `rmuser' with corresponding versions from FreeBSD 5
which are written in the Bourne shell.
Submitted-by: Tim Wickberg <me at k9mach3 dot org>
Additional-testing-by: William DeVries
Matthew Dillon [Mon, 21 Jun 2004 15:45:42 +0000 (15:45 +0000)]
Some more M_NOWAIT->M_INTWAIT. Not sure if the sym changes are correct,
but they are more correct then using M_NOWAIT.
Matthew Dillon [Mon, 21 Jun 2004 15:39:31 +0000 (15:39 +0000)]
General M_NOWAIT -> M_INTWAIT work, except in periodic timeout() routines
which can handle occassional malloc() failures and really shouldn't block.
Matthew Dillon [Mon, 21 Jun 2004 15:35:41 +0000 (15:35 +0000)]
Correct a known panic in the twe driver, M_NOWAIT -> M_INTWAIT work.
Matthew Dillon [Mon, 21 Jun 2004 06:02:21 +0000 (06:02 +0000)]
gcc-3.4 cleanups. missing break / goto labels.
Matthew Dillon [Mon, 21 Jun 2004 05:58:01 +0000 (05:58 +0000)]
gcc-3.4 cleanups. Add missing break statements, deal with goto labels,
and adjust the use of __FUNCTION__ (string concat is no longer supported).
Matthew Dillon [Mon, 21 Jun 2004 05:31:47 +0000 (05:31 +0000)]
Adding missing break after default: (gcc-3.4).
Matthew Dillon [Mon, 21 Jun 2004 05:28:53 +0000 (05:28 +0000)]
gcc-3.4 cleanup.
Matthew Dillon [Mon, 21 Jun 2004 03:48:07 +0000 (03:48 +0000)]
Hook gcc 3.4 into the buildworld. Rearrange HOST_CCVER so it becomes the
responsibility of the Makefile element that is overriding CCVER. Get rid
of the obsolete MY_CCVER magic. Fix improper use of ${.TARGET} in
cc34/cc_tools/Makefile.tools. Do not try to override _CPUCFLAGS in
bsd.hostprog.mk, which doesn't handle the generic .nx/.no case. Instead
remove -mtune, -mcpu, and -march options in NXCFLAGS and NXCXXFLAGS.
Work-by: Joerg Sonnenberger <joerg@crater.dragonflybsd.org>
Reviewed-by: Matthew Dillon <dillon@backplane.com>
Matthew Dillon [Mon, 21 Jun 2004 03:39:10 +0000 (03:39 +0000)]
Fix a bug in 2.95.x prototypes that prevents us from building the world with
gcc-3.4. Note that contrib/gcc is an 'old' contrib that came before our
new 'no-touch' policy for contrib sources so we can modify it directly.
Matthew Dillon [Mon, 21 Jun 2004 03:38:52 +0000 (03:38 +0000)]
Add a missing break after default: for gcc-3.4 support.
Matthew Dillon [Mon, 21 Jun 2004 03:31:00 +0000 (03:31 +0000)]
Remove the last vestiges of -DCCVER from 2.95.x contrib so we can remove
it from the build tree.
Matthew Dillon [Mon, 21 Jun 2004 03:25:33 +0000 (03:25 +0000)]
Fix a prototype error in gcc-3.3's com.h in order to allow gcc-3.4 to compile
it. This is a bad hack but we are going to remove 3.3 from the tree soon
anyway.
Matthew Dillon [Mon, 21 Jun 2004 02:28:34 +0000 (02:28 +0000)]
'lrint' is a gcc3.4 builtin and gcc3.4 is unhappy if we use that for anything
else.
Chris Pressey [Mon, 21 Jun 2004 01:37:42 +0000 (01:37 +0000)]
Merge with FreeBSD (HEAD) src/usr.sbin/kbdmap/*:
Move towards Perl removal from the base by replacing the Perl script
'kbdmap' with the version from FreeBSD 5, which is written in C.
Submitted-by: Tim Wickberg <me at k9mach3 dot org>
Additional-testing-by: William DeVries
Hiten Pandya [Mon, 21 Jun 2004 01:03:06 +0000 (01:03 +0000)]
Check kp (struct kinfo_proc *kp) against NULL and not 0, because it is a
pointer.
Hiten Pandya [Mon, 21 Jun 2004 00:47:57 +0000 (00:47 +0000)]
Use the kern.boottime sysctl for retrieving the system boot time as a
struct timeval. This is used to fix the STARTED column for pure kernel
threads.
Note, currently, we do not set the start time for pure threads due to
safety issues with microtime(9), although this change conditionalizes
the thread start time appropriately, so that in the future when pure
threads will account their start time, we wouldn't need to change anything
in this utility (with regard to the start time, ofcourse).
Discussed-with: Matthew Dillon <dillon@apollo.backplane.com>
Joerg Sonnenberger [Mon, 21 Jun 2004 00:20:33 +0000 (00:20 +0000)]
make ==> ${MAKE} ${.TARGET}
Joerg Sonnenberger [Mon, 21 Jun 2004 00:03:41 +0000 (00:03 +0000)]
make => ${MAKE}
Hiten Pandya [Sun, 20 Jun 2004 22:29:10 +0000 (22:29 +0000)]
Move the 'p_start' field from struct pstats (Process Statistics) into the
thread structure and call it 'td_start'. The behavior of vm_fork(9) is
retained, i.e., it still copies the start time from the parent process just
as it did before.
The 'td_start' will later be used by pure threads to indicate their start
time. It has not been committed in this round because use of the microtime()
function at such a early point in the boot process might be unsafe.
Note, there should be no problem in accessing the td_start field, unless
the process is a Zombie; due to the way Zombies are reaped, the thread will
be decoupled in kern_wait1() but the process will still be around for a
while it will not be possible to access the td_start field in such
scenarios. A little note about this has been added on top of struct proc
in <sys/proc.h> for future reference.
This work was a collaboration of Hiten Pandya <hmp@backplane.com> and
Matthew Dillon <dillon@apollo.backplane.com>
Matthew Dillon [Sun, 20 Jun 2004 22:19:51 +0000 (22:19 +0000)]
Allow CCVER to be empty, which will cause the compiled-in default compiler
to be used.
Joerg Sonnenberger [Sun, 20 Jun 2004 21:41:15 +0000 (21:41 +0000)]
Add the STABS default output hack for GCC 3.4 too.
Joerg Sonnenberger [Sun, 20 Jun 2004 20:59:21 +0000 (20:59 +0000)]
For cc_tools, recurse into cc_prep when building dependencies to get
the patched source tree.
c++filt and gcov use getopt, add the libc version when bootrapping. The
source and header file is copyied by cc_prep in that situation.
Add an explicit directory order in the top-level Makefile.
Joerg Sonnenberger [Sun, 20 Jun 2004 20:54:09 +0000 (20:54 +0000)]
Explicitly recurse into gnu/usr.bin/cc34/cc_prep for depend to ensure
that the source tree is prepared for us. This may otherwise fail for
buildworld.
Joerg Sonnenberger [Sun, 20 Jun 2004 20:49:45 +0000 (20:49 +0000)]
Export HOST_CCVER via environment to fixate it to either the specified
value or the default CCVER.
Joerg Sonnenberger [Sun, 20 Jun 2004 20:41:07 +0000 (20:41 +0000)]
Don't include _CPUCFLAGS since the host compiler (aka NXCC) might not
understand them. Add HOST_CCVER which replaces CCVER for NXCC and NXCXX.
This defaults to CCVER, but can be overriden via environment e.g. from
Makefile.inc1.
Hiten Pandya [Sun, 20 Jun 2004 17:44:11 +0000 (17:44 +0000)]
Zero-out the whole pstats structure and then copy the relevant fields,
instead of messing with caddr_t's and range-based zero'ing.
Discussed-with: Matthew Dillon <dillon@backplane.com>
Hiten Pandya [Sun, 20 Jun 2004 05:55:30 +0000 (05:55 +0000)]
Zero the interval timers on fork(2) rather than copying them to the
child process (SUSv3 compliance).
Discussed-with: Matthew Dillon <dillon@backplane.com>
Hiten Pandya [Sun, 20 Jun 2004 02:53:19 +0000 (02:53 +0000)]
Hook up the asa(1) utility to the build system.
Hiten Pandya [Sun, 20 Jun 2004 02:52:15 +0000 (02:52 +0000)]
Add the POSIXv2 asa(1) utility; it interprets FORTRAN carriage-control
characters.
Obtained-from: FreeBSD (they imported it from NetBSD)
Hiten Pandya [Sun, 20 Jun 2004 02:36:01 +0000 (02:36 +0000)]
Hook up the recently added utilities [1] to the build.
[1] = asf(1), csplit(1) and tabs(1)
Joerg Sonnenberger [Sun, 20 Jun 2004 02:22:55 +0000 (02:22 +0000)]
Define __DragonFly_cc_version for CPP
Matthew Dillon [Sun, 20 Jun 2004 02:14:58 +0000 (02:14 +0000)]
Merge from vendor branch GCC:
Bring in a trimmed down gcc-3.4-
20040618.
Submitted-by: Joerg Sonnenberger <joerg@crater.dragonflybsd.org>
Matthew Dillon [Sun, 20 Jun 2004 02:14:58 +0000 (02:14 +0000)]
Bring in a trimmed down gcc-3.4-
20040618.
Submitted-by: Joerg Sonnenberger <joerg@crater.dragonflybsd.org>
Joerg Sonnenberger [Sun, 20 Jun 2004 00:43:28 +0000 (00:43 +0000)]
Always set the _CPUCFLAGS in bsd.cpu.gcc2.mk to some known, safe
value. For the !CPUTYPE || CPUTYPE==i386 case _CPUCFLAGS wasn't
reset which breaks a CCVER=gcc34 buildworld, because the default
flags for gcc34 (-mtune=pentiumpro) doesn't work for gcc2.
Joerg Sonnenberger [Sun, 20 Jun 2004 00:20:28 +0000 (00:20 +0000)]
Use our specs, not FreeBSD's.
Joerg Sonnenberger [Sun, 20 Jun 2004 00:19:51 +0000 (00:19 +0000)]
Manually recurse into ../cc_prepend for depend to workaround
buildworld issues.
Joerg Sonnenberger [Sun, 20 Jun 2004 00:10:52 +0000 (00:10 +0000)]
Redo the ProPolice patches, there were partly broken.
Hiten Pandya [Sat, 19 Jun 2004 22:03:08 +0000 (22:03 +0000)]
IEEE Std. 1003.1-2001 (SUSv3):
Add the tabs(1) utility.
Written by Tim J. Robbins of the FreeBSD Project.
Hiten Pandya [Sat, 19 Jun 2004 21:51:54 +0000 (21:51 +0000)]
Add the csplit(1) utility, which splits files based on context, as
specified by IEEE Std. 1003.1-2001 (SUSv3).
Written by Tim J. Robbins (tjr@) of the FreeBSD Project.
Joerg Sonnenberger [Sat, 19 Jun 2004 20:38:22 +0000 (20:38 +0000)]
Fix GCC 3.4 build
Joerg Sonnenberger [Sat, 19 Jun 2004 20:36:04 +0000 (20:36 +0000)]
Fix GCC 3.4 build.
Joerg Sonnenberger [Sat, 19 Jun 2004 18:55:48 +0000 (18:55 +0000)]
Fix compilation "Label at end of compound statement" and some missing
prototyps / K&R style functions for GCC 3.4.
Chris Pressey [Sat, 19 Jun 2004 17:28:28 +0000 (17:28 +0000)]
Fix a bogus type declaration to remove a compiler warning - `group'
in this case refers to a netgroup name, and should therefore be a
char * instead of a gid_t.
Requested-by: joerg
Joerg Sonnenberger [Sat, 19 Jun 2004 17:19:50 +0000 (17:19 +0000)]
Fix GCC 3.4 build
Joerg Sonnenberger [Sat, 19 Jun 2004 16:40:36 +0000 (16:40 +0000)]
Remove the evil inline ==> rpcgen_inline CPP hack, rename the variable
properly. Use ISO C prototyps. Remove useless void casts. Do some other
style updates. Remove !defined(__DragonFly__) cases, one was checking for
__Dragonfly__, which broke buildworld with GCC 3.4. Add a break after
default: in switch statements, which is not portable and an error with
GCC 3.4 and later.
Matthew Dillon [Sat, 19 Jun 2004 16:03:01 +0000 (16:03 +0000)]
Fix an aggregious non-terminated buffer issue and also fix the retry code
(the buffer was being wiped and needed to be regenerated for the retry).
Reported-by: Vincent Labrecque <vincent@psyfreaks.ca>
Hiten Pandya [Sat, 19 Jun 2004 12:34:52 +0000 (12:34 +0000)]
Fix prototype of usage().
Joerg Sonnenberger [Sat, 19 Jun 2004 11:13:30 +0000 (11:13 +0000)]
First stage in cleaning the built-in pathes of gcc. Adapted patch from
cc3/gcc.patch and fit directly into the patching system.
Joerg Sonnenberger [Sat, 19 Jun 2004 10:34:17 +0000 (10:34 +0000)]
Include ProPolice suport for GCC 3.4.
Obtained-from: http://www.trl.ibm.com/projects/security/ssp
Additional files forgotten in first commit.
Joerg Sonnenberger [Sat, 19 Jun 2004 10:30:08 +0000 (10:30 +0000)]
Include ProPolice suport for GCC 3.4.
Obtained-from: http://www.trl.ibm.com/projects/security/ssp
Adjust Makefile.langs to use the patched common.opts from cc_prep.
David Rhodus [Fri, 18 Jun 2004 23:50:34 +0000 (23:50 +0000)]
Create twed0 entries in /dev
David Rhodus [Fri, 18 Jun 2004 21:45:06 +0000 (21:45 +0000)]
Fix installworld by making it aware of the rconfig examples.
Hiten Pandya [Fri, 18 Jun 2004 21:20:55 +0000 (21:20 +0000)]
Add a helper utility, called asf(8): Add Symbol File.
The purpose of this utility is to extract start addresses from kldstat(8)'s
output and prepare a .gdbinit style file (called .asf) with GDB commands to
load the KLD files at the right offset.
The file can then be given to the 'source' command of GDB so that it does
its magic. This is how one can invoke the asf(8) command:
# asf -k -f -s -x /modules gdb.kldsyms.asf
Note, the '-k' output invokes kldstat(8) automatically, otherwise it
accepts kldstat(8) sytle output on stdin, which can be produced by a
gdb macro so that is is equally useful in post-crash analysis.
One needs to compile the module with debugging symbols, though.
Obtained-from: FreeBSD (CVS, MAIN branch)
Matthew Dillon [Fri, 18 Jun 2004 18:36:06 +0000 (18:36 +0000)]
A sample rconfig script which completely wipes and reinstalls dragonfly
on a fresh HD.
Matthew Dillon [Fri, 18 Jun 2004 04:26:53 +0000 (04:26 +0000)]
chdir into WorkDir before running the downloaded script.
Exit with an error if both -a and a list of hosts are specified on the
command line.
Matthew Dillon [Fri, 18 Jun 2004 02:46:46 +0000 (02:46 +0000)]
Write a remote configuration utility called 'rconfig'. This initial
implementation is very basic and designed to be convienient. It allows a
CD user to scan the network for configuration servers, download a
configuration script based on a tag name, and execute that script all in
one simple command. Running the rconfig server is just as easy... you
basically just run the server -a -s and put your scripts in
/usr/local/etc/rconfig/<tagname>.sh and you are done.
Of course the scripts themselves have to be written and can certainly become
quite complex.
In order to use rconfig from a CD Boot at least one network interface must
be properly configured. The dhclient program is usually used to accomplish
this.
Joerg Sonnenberger [Thu, 17 Jun 2004 23:28:39 +0000 (23:28 +0000)]
The insn-conditions.c generated by stock GCC does some bad premature
optimizations which breaks e.g. genflags. Remove this until a better
solution is found.
David Rhodus [Thu, 17 Jun 2004 19:37:29 +0000 (19:37 +0000)]
Default the MMX/XMM kernel optimizations to the on position. Also make a
note in the loader man page that these optimizations can be turned off at
the loader by setting kern.mmxopt=1.
Joerg Sonnenberger [Thu, 17 Jun 2004 18:05:17 +0000 (18:05 +0000)]
Don't whine about malloc/realloc "poising" the code for YACC files.
Matthew Dillon [Thu, 17 Jun 2004 16:51:56 +0000 (16:51 +0000)]
Don't set prefetch mode on VIA chips, it causes problems on newer chips and
ATAPI devices.
Notified-by: David Xu <davidxu@viatech.com.cn>
Taken-From: FreeBSD-5
Hiten Pandya [Thu, 17 Jun 2004 01:30:27 +0000 (01:30 +0000)]
Spell 'written' properly.
Hiten Pandya [Wed, 16 Jun 2004 01:17:30 +0000 (01:17 +0000)]
Fix spelling in comment.
Matthew Dillon [Wed, 16 Jun 2004 01:14:10 +0000 (01:14 +0000)]
Make sure gcc_local.c and any left over patch mess is cleaned up by make clean,
and fix an improper setting of CLEANFILES for cpp and f77.
Matthew Dillon [Wed, 16 Jun 2004 01:06:45 +0000 (01:06 +0000)]
Clean up GCC3.3, rip out all the weird search paths it adds and fix a long
standing bug that was preventing it from finding the crtbegin*.o and
crtend*.o files in their new home.
In consultation with: Joerg Sonnenberger <joerg@britannica.bec.de>
Joerg Sonnenberger [Tue, 15 Jun 2004 13:43:20 +0000 (13:43 +0000)]
Fix warning
Joerg Sonnenberger [Tue, 15 Jun 2004 11:56:03 +0000 (11:56 +0000)]
Really use the host compiler in bsd.hostprog.mk
Joerg Sonnenberger [Tue, 15 Jun 2004 08:53:09 +0000 (08:53 +0000)]
Nuke lib/csu/i386 (a.out support) and copy lib/csu/i386-elf there.
Joerg Sonnenberger [Tue, 15 Jun 2004 07:53:32 +0000 (07:53 +0000)]
Add CCVER=gcc34 support to bsd.cpu.mk. Also add the magic for AMD64 support
for GCC 3.4 and backwards compatiblity for the older compilers.
Joerg Sonnenberger [Tue, 15 Jun 2004 07:35:17 +0000 (07:35 +0000)]
Add directory entries for GCC 3.4.
Matthew Dillon [Tue, 15 Jun 2004 00:30:55 +0000 (00:30 +0000)]
Fix a race with the clearing of p->p_session->s_ttyvp. NULL the pointer
out before calling vrele() rather then after.
Fix a bug with v_opencount accounting on revoke(). The underlying device
was being closed properly but v_opencount was being decremented which causes
it to go negative when close() is called on the descriptor later on. To
fix the bug we zero out v_opencount() when the underlying vnode's device
is disassociated and spec_close() now only decrements it when the device is
associated.
Reported-by: GeekGod
Testing-by: GeekGod, Hiten, David Rhodus.
Joerg Sonnenberger [Mon, 14 Jun 2004 22:27:56 +0000 (22:27 +0000)]
Build infrastructure for GCC 3.4
This is based both on the logical structure of GCC own Makefiles
and the build framework of GCC 3.3. For the moment, the generated
helper files are recreated for each directory using them to make
the dependencies simpler. libgcc_r is not included, because the
only difference is the normal version is an explicit call to
pthread_mutex_init instead of using the static init model, which
is serialized by a spinlock for all those locks.
This is not yet hooked into the normal buildworld system.
Joerg Sonnenberger [Mon, 14 Jun 2004 20:08:02 +0000 (20:08 +0000)]
Add the old CSU files to list of file to be deleted and keep the new ones.
Joerg Sonnenberger [Mon, 14 Jun 2004 19:35:36 +0000 (19:35 +0000)]
Switch from GCC-version specific crtbegin/crtend code to the version
independent code. This is binary compatible with old version. The old
object files are currently left in place.
Joerg Sonnenberger [Mon, 14 Jun 2004 18:56:12 +0000 (18:56 +0000)]
Import BSD-licensed crtbegin/crtend support.
Obtained-from: NetBSD
Note: crtend.c comes without license header. I send a mail to Jason Thorpe
and will update it according to the response. I'm assuming the normal
4-clause NetBSD license for now.
Hiten Pandya [Mon, 14 Jun 2004 15:26:56 +0000 (15:26 +0000)]
Make the procfs_validfile() function static.
Hiten Pandya [Mon, 14 Jun 2004 01:35:36 +0000 (01:35 +0000)]
If there was a cache hit, and the msf->m_flags was set to SFBA_ONFREEQ,
make sure we dequeue it properly before jumping to the 'done' label.
Reported-by: Emiel Kollof <coolvibe@gamepoint.net>
Discussed-with: Matthew Dillon <dillon@backplane.com>
Matthew Dillon [Sat, 12 Jun 2004 03:09:41 +0000 (03:09 +0000)]
start_forked_proc() should not be called when rfork() is called without
RFPROC, since no new process is being created.
Reported-by: Hiten
Testing-by: Hiten
Matthew Dillon [Sat, 12 Jun 2004 03:07:36 +0000 (03:07 +0000)]
ANSIfy the remaining K&R functions. Remove two bzero-after-malloc's in
favor of M_ZERO, replace all uses of the MALLOC() macro with malloc(), and
replace all uses of the FREE() macro with free().
This is a cleanup commit only.
Matthew Dillon [Sat, 12 Jun 2004 01:55:59 +0000 (01:55 +0000)]
Fix a bug in the reply port path related to aborts. Aborted messages are
requeued to the target port. If the target is processing the message and
replies to it, the replyport code is supposed to remove any requeued aborts
(since the target completed processing of that message).
However, this was only occuring for MSGF_ASYNC messages. It must be done for
both originator synchronous and originator asynchronous messages (abort
processing has nothing to do with how the originator sent the message).
Reported-by: David Rhodus
Testing-by: David Rhodus
Hiten Pandya [Fri, 11 Jun 2004 12:03:15 +0000 (12:03 +0000)]
Do not produce a warning if the sysctl does not exist. This happens when
the kernel is compiled without the INET6 option.
Matthew Dillon [Thu, 10 Jun 2004 22:11:39 +0000 (22:11 +0000)]
Both 'ps' and the loadav calculations got broken by thread sleeps, which
occur without knowledge by the proc and so ps/loadav thought processes
sitting in e.g. accept() were in a 'R'un state when they were actually
sleeping.
Make ps and the loadav calculator thread-aware.
Joerg Sonnenberger [Thu, 10 Jun 2004 10:31:29 +0000 (10:31 +0000)]
Add Makefile support for host programs (.nx) just like we support normal
programs with bsd.prog.mk. The difference is that bsd.hostprog.mk doesn't
support man pages, scripts and builds ${PROG}.nx.
Hiten Pandya [Wed, 9 Jun 2004 23:35:45 +0000 (23:35 +0000)]
Revert the locking of feof(3) for now; there is possibility of ill
effects due to the current implementation of funlockfile().
Inspired by rev. 1.7 of lib/libc/stdio/fclose.c.
Hiten Pandya [Wed, 9 Jun 2004 20:26:04 +0000 (20:26 +0000)]
Add some whitespace for clarity. No operational changes.
Hiten Pandya [Wed, 9 Jun 2004 19:40:59 +0000 (19:40 +0000)]
Add a 'big fat comment' about the FUNLOCKFILE(fp) implementation and why
it needs to be called before modification of the FILE flags.
Requested-by: Matthew Dillon <dillon at backplane.com>