Max Okumoto [Sat, 16 Apr 2005 10:35:02 +0000 (10:35 +0000)]
patch-7.185
Var_Subst(), Var_SubstOnly()
o remove dead code.
Max Okumoto [Sat, 16 Apr 2005 10:34:26 +0000 (10:34 +0000)]
patch-7.184
o Copy Var_Subst() into Var_SubstOnly(), the new function
is never passed NULL in the 'var' input parameter.
o Modified Var_Subst() to always expect var==NULL as input paramter.
o Update calling code by removing unused parameter.
YONETANI Tomokazu [Sat, 16 Apr 2005 08:17:02 +0000 (08:17 +0000)]
Add a debug knob debug.acpi.avoid.paths, a variant of debug.acpi.avoid
but without totally disabling execution of _INI and _STA. Maybe useful
to narrow down the portion of namespace causing problems without messing
with editing and compiling ASL. Disablement of _INI and _STA when
debug.acpi.avoid is specified has been added years before to workaround
interpreter crash on some systems, and I'm not sure whether the recent
ACPI-CA code still crashes on those systems, but I'll leave it as is
anyway for compatibility with FreeBSD.
Suggested-by: Sepherosa Ziehau
Max Okumoto [Fri, 15 Apr 2005 21:13:05 +0000 (21:13 +0000)]
Forgot to add these files to previous commit. directive_hash.c
must be regenerated if new directives are added/removed/modified.
Max Okumoto [Fri, 15 Apr 2005 21:09:15 +0000 (21:09 +0000)]
Add back .makenv compat until everyone's source tree is updated.
Max Okumoto [Fri, 15 Apr 2005 21:06:35 +0000 (21:06 +0000)]
Create VAR_ENV context, and fill it with everything in the environment.
Max Okumoto [Fri, 15 Apr 2005 21:05:52 +0000 (21:05 +0000)]
add .EXPORT: special tag and add ST_EXPORT variable.
Max Okumoto [Fri, 15 Apr 2005 21:01:27 +0000 (21:01 +0000)]
FreeBSD-Date: 2005/04/11 08:20:09
FreeBSD-Date: 2005/04/11 08:26:21
FreeBSD-Date: 2005/04/11 08:40:54
Author: harti
Rework the directive parsing code. Instead of using a lot of strcmp()s
on every line that starts with a dot use a minimal perfect hash
function and a single strcmp() on the first word after the dot
to find out whether it is really a directive call and, if yes, which
one. Then directly dispatch to a handler function for that directive
(or fall through to the dependency handling code). This makes the
directive parse a little bit more strict about the syntax: the directive
word must be followed by a character that is not alphanumerical and not
an underline (making .undefFOO illegal); .endif and .else can only be
followed by comments.
Max Okumoto [Fri, 15 Apr 2005 20:50:43 +0000 (20:50 +0000)]
Inline stripvarname() since it will be removed in the next few patches.
Max Okumoto [Fri, 15 Apr 2005 20:45:48 +0000 (20:45 +0000)]
Split VarFind() into a series of functions tailored for the different sets
of flags originally passed to VarFind(). This eliminates the code by
removing a bunch of tests.
patch-7.173
patch-7.174
patch-7.176
Matthew Dillon [Fri, 15 Apr 2005 19:08:32 +0000 (19:08 +0000)]
Implement Red-Black trees for the vnode clean/dirty buffer lists.
Implement ranged fsyncs and adjust the syncer to use the new capability.
This capability will also soon be used to replace the write_behind
heuristic. Rewrite the fsync code for all VFSs to use the new APIs
(generally simplifying them).
Get rid of B_WRITEINPROG, it is no longer useful or needed.
Get rid of B_SCANNED, it is no longer useful or needed.
Rewrite the NFS 2-phase commit protocol to take advantage of the new
Red-Black tree topology.
Add RB_SCAN() for callback-scanning of Red-Black trees. Give RB_SCAN
the ability to track the 'next' scan node and automatically fix it up
if the callback directly or indirectly or through blocking indirectly
deletes nodes in the tree while the scan is in progress.
Remove most related loop restart conditions, they are no longer necessary.
Disable filesystem background bitmap writes. This really needs to be
solved a different way and the concept does not work well with red-black
trees.
Joerg Sonnenberger [Fri, 15 Apr 2005 18:40:52 +0000 (18:40 +0000)]
Remove pccard RCNG script, the daemon config, the depency on pccard
for the other RCNG scripts, the actual daemon and the control program.
pccard_ether is used by usbd, keep it for now.
Matthew Dillon [Fri, 15 Apr 2005 18:39:13 +0000 (18:39 +0000)]
Zero the whole structure, not just the name, so we don't trip up on this
in the future.
Joerg Sonnenberger [Fri, 15 Apr 2005 18:26:58 +0000 (18:26 +0000)]
GC OLDCARD userland tools.
Joerg Sonnenberger [Fri, 15 Apr 2005 18:07:13 +0000 (18:07 +0000)]
Fix the parameter order of __stack_smash_handler. Also print the
return address on IA32 to aid debugging.
Joerg Sonnenberger [Fri, 15 Apr 2005 17:55:49 +0000 (17:55 +0000)]
Extend a local buffer to workarond a buffer overflow for now.
David Rhodus [Fri, 15 Apr 2005 17:55:29 +0000 (17:55 +0000)]
Remove some uses of register keywords.
Joerg Sonnenberger [Fri, 15 Apr 2005 07:08:27 +0000 (07:08 +0000)]
Fix a small stack disclosure in ifconf().
See FreeBSD-SA-05:04.ifconf for credits and details.
Eirik Nygaard [Thu, 14 Apr 2005 15:49:12 +0000 (15:49 +0000)]
getopt is WARNS 6 clean.
Eirik Nygaard [Thu, 14 Apr 2005 15:05:29 +0000 (15:05 +0000)]
A few more instances of leaf that should be leap.
Noticed by: Bart Thate <dunker@freebsd.nl>
Joerg Sonnenberger [Thu, 14 Apr 2005 11:22:16 +0000 (11:22 +0000)]
Ignore replies with a negative delay. Correct time of next query and
deadline for all peers in priv_settime is called.
This should fix the problems of ntpd -s with very large initial correction.
Joerg Sonnenberger [Thu, 14 Apr 2005 11:15:52 +0000 (11:15 +0000)]
leap second, not leaf second. fix totally stupid, but consistent spelling
error.
Noticed-by: corecode
Joerg Sonnenberger [Thu, 14 Apr 2005 10:41:29 +0000 (10:41 +0000)]
Merge from vendor branch NTPD:
Sync with OpenBSD. This fixes the bug of -s not working and allows
defered DNS lookups, which might be necessary in certain network
setups.
Joerg Sonnenberger [Thu, 14 Apr 2005 10:41:29 +0000 (10:41 +0000)]
Sync with OpenBSD. This fixes the bug of -s not working and allows
defered DNS lookups, which might be necessary in certain network
setups.
Joerg Sonnenberger [Thu, 14 Apr 2005 10:31:27 +0000 (10:31 +0000)]
Use corrected system time internally, adjust old offsets after
doing a relative adjust of kern.ntp.delta.
Change the default scaling interval to 0.005 and 0.010, which
means that offsets smaller than 5ms double to probe interval,
anything greater than 10ms uses the default. The value of
QSCALE_OFF_MIN can be decreases even more, once frequency correction
is implemented.
Joerg Sonnenberger [Thu, 14 Apr 2005 10:26:51 +0000 (10:26 +0000)]
Use corrected system time internally, adjust old offsets after
doing a relative adjust of kern.ntp.delta.
Change the default scaling interval to 0.005 and 0.010, which
means that offsets smaller than 5ms double to probe interval,
anything greater than 10ms uses the default. The value of
QSCALE_OFF_MIN can be decreases even more, once frequency correction
is implemented.
YONETANI Tomokazu [Thu, 14 Apr 2005 10:17:23 +0000 (10:17 +0000)]
Resurrect some of functions shared by rrestore.
Joerg Sonnenberger [Thu, 14 Apr 2005 08:14:31 +0000 (08:14 +0000)]
wether ==> whether
Noticed-by: asmodai
Joerg Sonnenberger [Thu, 14 Apr 2005 07:55:36 +0000 (07:55 +0000)]
Simplified NTP kernel interface:
- kern.ntp.delta gives the delta to apply (ns)
- kern.ntp.tick_delta is the correction applied in each tick (ns)
- kern.ntp.default_tick_delta is the default correction for each tick (ns)
- kern.ntp.big_delta is the threshold for kern.ntp.delta to use
10x kern.ntp.default_tick_delta, not the normal value
- kern.ntp.adjust can be used to change the current value of kern.ntp.delta
relatively.
- kern.ntp.next_leaf_second specifies the time_t of the next leaf second change,
kern.ntp.insert_leaf_second != 0 means a leaf second is inserted, otherwise
it is removed
All ntp_* variables are manipulated on CPU #0 with the exception of
set_timeofday. It just sets ntp_delta to 0, which is fine for the moment.
Joerg Sonnenberger [Wed, 13 Apr 2005 16:08:04 +0000 (16:08 +0000)]
Remove another argument name in a prototype.
Joerg Sonnenberger [Wed, 13 Apr 2005 16:07:15 +0000 (16:07 +0000)]
Decommon a lot of variables, makes some static, fix a small bug in rmtcall
introduced earlier.
Joerg Sonnenberger [Wed, 13 Apr 2005 15:45:37 +0000 (15:45 +0000)]
Remove a cast, correct another.
Liam J. Foy [Wed, 13 Apr 2005 15:44:03 +0000 (15:44 +0000)]
- Let receiver know if the sender accepts replies
- Use correct headers
- Use err(). In some cases an incorrect error message can be shown
- Strlcpy
- warnx() instead of errx() if sender has messages disabled (OpenBSD)
- Remove non-POSIX S_IWRITE (OpenBSD)
- General cleanup
Joerg Sonnenberger [Wed, 13 Apr 2005 15:43:36 +0000 (15:43 +0000)]
Don't write to tape, strdup it first.
Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
Joerg Sonnenberger [Wed, 13 Apr 2005 15:21:36 +0000 (15:21 +0000)]
Remove argument names from prototypes. Adjust prototypes to deal
with void * vs. char * and const vs. non-const behaviour.
Joerg Sonnenberger [Wed, 13 Apr 2005 14:32:01 +0000 (14:32 +0000)]
Move initialisation of wrote into the loop to prevent longjmp/setjmp
from clobbering it.
Joerg Sonnenberger [Wed, 13 Apr 2005 14:29:20 +0000 (14:29 +0000)]
Warning fixes.
Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
Joerg Sonnenberger [Wed, 13 Apr 2005 14:21:06 +0000 (14:21 +0000)]
Split atomic into two functions, atomic_read and atomic_write.
This makes the code more readable and helps the const correctness.
Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
Joerg Sonnenberger [Wed, 13 Apr 2005 14:12:35 +0000 (14:12 +0000)]
Compute buffer length only once per rmtcall.
Joerg Sonnenberger [Wed, 13 Apr 2005 14:10:18 +0000 (14:10 +0000)]
Remove unused functions.
Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
Joerg Sonnenberger [Wed, 13 Apr 2005 14:05:35 +0000 (14:05 +0000)]
First bunch of dump(8) cleanups.
Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
Matthew Dillon [Wed, 13 Apr 2005 04:02:08 +0000 (04:02 +0000)]
Only bump the switch_count counter when lwkt_switch() actually switches
to a different thread. Otherwise the idle loop causes this counter
to go haywire.
Matthew Dillon [Wed, 13 Apr 2005 04:00:56 +0000 (04:00 +0000)]
Optimize lwkt_send_ipiq() - the IPI based inter-cpu messaging routine.
* Add a passive version which does not initiate any actual hardware IPI.
The message will be handled the next time the target cpu polls the
queue (on each tick typically). Adjust the free() path to use this
version when freeing memory owned by another cpu.
* Add an interlock to avoid reissuing and unnecessarily stalling on
the hardware IPI if a prior hardware IPI to the target cpu has not
yet completed processing.
This feature theoretically means that two cpus can tightly couple a
large number of pipelined messages with only a single actual IPI being
sent.
* Reorganize the hystersis points in the IPIQ FIFOs.
* Change a token livelock warning into a panic if it occurs 10 times in
a row.
* Add a call to lwkt_process_ipiq() just after the AP startup code enables
a cpu, to process any messages that might have built up during startup.
There shouldn't be any, but this may avoid surprises later.
Max Okumoto [Tue, 12 Apr 2005 23:37:27 +0000 (23:37 +0000)]
Insert a '#' between the .else directive and the following comment.
Max Okumoto [Tue, 12 Apr 2005 23:35:37 +0000 (23:35 +0000)]
Put a '#' between {.else, .endif} and the comment following it.
This makes it clear that the following text is a comment. And
more important, the make parser is going to be a little more
strict about it, soon.
Chris Pressey [Tue, 12 Apr 2005 18:07:36 +0000 (18:07 +0000)]
The error message issued when a requested package cannot be found has
become garbled over time. Ungarble it.
David Xu [Tue, 12 Apr 2005 14:01:31 +0000 (14:01 +0000)]
Fix brokeness by using _pthread_xxx functions instead.
Max Okumoto [Tue, 12 Apr 2005 08:26:17 +0000 (08:26 +0000)]
Removed un-nessisary comment. We can get a zero length arg to make
if someone quotes it.
% make ''
Hiten Pandya [Mon, 11 Apr 2005 14:51:38 +0000 (14:51 +0000)]
Use the correct macro for printing TCP statistical counters when we only
have one format argument.
Hiten Pandya [Mon, 11 Apr 2005 11:00:40 +0000 (11:00 +0000)]
Ick! Seems I was under a placebo, correct a last minute typo.
Hiten Pandya [Mon, 11 Apr 2005 10:24:45 +0000 (10:24 +0000)]
Mechanical cleanup of IP per-cpu statistics code, better naming etc.
Correct a stale comment regarding initialisation of the counters.
Hiten Pandya [Mon, 11 Apr 2005 09:58:38 +0000 (09:58 +0000)]
Forced commit to note previous change, of introducing 'mycpuid' was
originally suggested by Jeffrey Hsu.
Hiten Pandya [Mon, 11 Apr 2005 09:54:22 +0000 (09:54 +0000)]
Remove the '_GD' macro hack:
I introduced it some revisions ago so that functions with locally
cached 'globaldata' can redefine it in order to minimise time
required for updating the IP/TCP per-cpu statistics counters.
Hiten Pandya [Mon, 11 Apr 2005 09:43:50 +0000 (09:43 +0000)]
Mechanical cleanup of TCP per-cpu statistics code, better naming etc.
Correct a stale comment regarding initialisation of the counters.
Hiten Pandya [Mon, 11 Apr 2005 09:33:31 +0000 (09:33 +0000)]
Add 'mycpuid', a #define for accessing mycpu->gd_cpuid.
Adjust nearby whitespace while I am here.
Matthew Dillon [Mon, 11 Apr 2005 06:05:54 +0000 (06:05 +0000)]
Clearly state when ACPI is overriding APM's device entries.
Chris Pressey [Sun, 10 Apr 2005 21:45:11 +0000 (21:45 +0000)]
Back out revision 1.31. The 'release' target alone should be
sufficient for creating a LiveCD containing no additional packages.
Discussed-with: drhodus
David Rhodus [Sun, 10 Apr 2005 20:55:38 +0000 (20:55 +0000)]
Clean up some register keyword usage.
David Rhodus [Sun, 10 Apr 2005 20:27:32 +0000 (20:27 +0000)]
Add in a 'livecd' taget for those who don't want to include the
installer program in their ISO image file.
Max Okumoto [Sun, 10 Apr 2005 10:28:21 +0000 (10:28 +0000)]
Remove previous hack, for now just use strtoull() instead of strtoumax().
Max Okumoto [Sun, 10 Apr 2005 10:18:45 +0000 (10:18 +0000)]
Added #ifdef to aid people upgrading from before March 17.
Needed for cross compile on FreeBSD 4.X. In addition, people
upgrading DFly via source, from before March 17, 2005 might need
to manually define NEED_STRTOUMAX in CFLAGS, because strtoumax()
did not exist before then.
This code should be removed if when cross compiling from FreeBSD
4.X is no longer required, and no one is upgrading from before
* March 17. Max Okumoto.
Hiten Pandya [Sun, 10 Apr 2005 00:09:32 +0000 (00:09 +0000)]
Add standard entries for libcaps, libkinfo and libkcore; they can now be
specified in manual pages using the 'Lb' mdoc macro.
Hiten Pandya [Sun, 10 Apr 2005 00:02:01 +0000 (00:02 +0000)]
Add entry for DragonFly 1.2.
Update entries for FreeBSD, i.e. 4.11, 5.4 and 6.0.
Hiten Pandya [Sat, 9 Apr 2005 23:41:26 +0000 (23:41 +0000)]
Add manual page for closefrom(2) system call and hook it to the build.
Partly obtained from OpenBSD with my modifications.
YONETANI Tomokazu [Sat, 9 Apr 2005 11:02:35 +0000 (11:02 +0000)]
Merge from FreeBSD:
revision 1.7
date: 2004-07-07 09:38:14 +0000; author: ru; state: Exp; lines: +1 -0
Arguments to options aren't necessarily separated with whitespace.
Noticed by: harti
revision 1.6
date: 2004-02-06 11:00:13 +0000; author: ru; state: Exp; lines: +3 -4
From what I can tell, install.sh supported all of the install(1)
options since revision 1.2, so removed the XXX comment now.
Fixed bogus test(1) expression that would be evaluated as a null
string test if script is run without arguments, compacted option
lists.
YONETANI Tomokazu [Sat, 9 Apr 2005 10:39:12 +0000 (10:39 +0000)]
Merge revision 1.5 from FreeBSD:
Fix typo in the BSD copyright: s/withough/without/
Max Okumoto [Sat, 9 Apr 2005 06:24:15 +0000 (06:24 +0000)]
FreeBSD-Date: 2005/04/08 10:03:40
Author: harti
Move the code for parsing .for and .if statements to the place where the
other directives are handled.
Max Okumoto [Sat, 9 Apr 2005 06:23:20 +0000 (06:23 +0000)]
FreeBSD-Date: 2005/04/07 15:39:38
Author: harti
When Parse_Error is called after the top-level Makefile is closed we
have no CURFILE anymore so we cannot print a file name or line number.
When ParseSkipLine() returns NULL (it does this when it has detected an
EOF in an .if block) try to pop the input stack and process the next line
Max Okumoto [Sat, 9 Apr 2005 05:51:42 +0000 (05:51 +0000)]
FreeBSD-Date: 2005/04/07 12:29:39
Author: harti
Handle popping of the input stack in ParseReadLine() instead of
Parse_File(). Remove a comment and a piece of code comming from
ancient times when the if-directive read like #if and not .if.
Correctly analyze the first character of a line.
Max Okumoto [Sat, 9 Apr 2005 05:50:51 +0000 (05:50 +0000)]
FreeBSD-Date: 2005/04/07 12:26:15
Author: harti
Mostly stylistic changes: remove the noPLen field from the Shell
structure - it is just the strlen() of noPrint. Inline JobCondPassSig()
in the only function JobPassSig() using it. Fix the argument types
of JobPrintCommand(). Better parsing for the .SHELL target keywords.
Joerg Sonnenberger [Fri, 8 Apr 2005 17:39:31 +0000 (17:39 +0000)]
Simply the loop based on the knowledge that fd <= fdp->fd_lastfile
(already checked) and fd >= 0 (also checked).
Discussed-with: Jeffrey Hsu
Joerg Sonnenberger [Fri, 8 Apr 2005 16:38:31 +0000 (16:38 +0000)]
Don't just undefine USE_RC_SUBR, because it does define the list
of rc scripts if it is defined and not YES. We didn't undefine USE_RCORDER,
which pulled the rc_subr port in.
Fix this by:
(a) save the old value or USE_RCORDER and USE_RC_SUBR to _DF_*.
(b) include the install-rc-script target from ports/Mk/bsd.port.mk,
but use _DF_* for the variables in the actual code.
(c) add _DF_* to the SUB_FILES list and add RC_SUBR to SUB_LIST.
(d) undefine USE_RC_SUBR and USE_RCORDER afterwards.
Problem-reported-by: Erik P. Skaalerud <erik@pentadon.com>
Joerg Sonnenberger [Fri, 8 Apr 2005 11:30:07 +0000 (11:30 +0000)]
Tell the world that we have closefrom.
Joerg Sonnenberger [Fri, 8 Apr 2005 11:26:52 +0000 (11:26 +0000)]
Use kernel's closefrom.
Joerg Sonnenberger [Fri, 8 Apr 2005 08:32:53 +0000 (08:32 +0000)]
Add userland prototype for closefrom.
Joerg Sonnenberger [Fri, 8 Apr 2005 08:31:45 +0000 (08:31 +0000)]
Regen.
Joerg Sonnenberger [Fri, 8 Apr 2005 08:31:06 +0000 (08:31 +0000)]
Add closefrom(2) syscall. It closes all file descriptors equal or greater
than the given descriptor.
This function does exit and return EINTR, when necessary. Other errors from
close() are ignored.
Matthew Dillon [Fri, 8 Apr 2005 06:40:33 +0000 (06:40 +0000)]
Bump HEAD's official release #define to 130000.
Matthew Dillon [Fri, 8 Apr 2005 06:12:11 +0000 (06:12 +0000)]
The release is 1.2, bump the HEAD of the tree to 1.3 for post-release
development.
Matthew Dillon [Fri, 8 Apr 2005 04:22:28 +0000 (04:22 +0000)]
Sync up to the latest installer packages (with netboot server capabilities!).
Matthew Dillon [Fri, 8 Apr 2005 03:25:18 +0000 (03:25 +0000)]
Add a hardwired dhcpd.conf for the installer's netboot server support.
Matthew Dillon [Fri, 8 Apr 2005 03:16:18 +0000 (03:16 +0000)]
The cdrom MNT_ROOTFS check was interfering with the NFS export handling,
preventing a system booted from CD from being able to export /. Fixed.
Max Okumoto [Thu, 7 Apr 2005 07:52:32 +0000 (07:52 +0000)]
Make should set optreset to one before it calls getopt() with a new
set of args.
o Removed unnessisary forward declaration for MainParseArgs()
o Use optreset global variable to inform getopt() that we are
parsing another arg list.
o Cleanup handling of remaining arguments after getopt()
completes parsing arguments. There is no reason to restart
getopt() for single dash case, just ignore it. And update
comment to reflect code.
o Removed missleading comments.
Max Okumoto [Thu, 7 Apr 2005 07:23:03 +0000 (07:23 +0000)]
Date: 2005/04/01 16:24:25
Date: 2005/04/05 08:32:12
Date: 2005/04/05 08:40:40
Date: 2005/04/05 08:43:02
Date: 2005/04/05 08:44:11
Date: 2005/04/05 09:09:48
Date: 2005/04/05 13:33:54
Author: harti
o Use a type-safe intrusive list to put jobs on the jobs or stoppedJobs
list instead of the generic Lst.
o The tfile field of struct Job isn't used anymore since revision
1.22 so remove it now.
o The static tfile is used only in one single function so move it
into that function.
o Remove unneccessary local prototypes.
o Remove the definition for STATIC and just use static.
o Remove lots if irritating parantheses and fix two comments.
o Put parantheses into 'if (foo && bar & flag)'. While they are not
strictly needed, they are useful for the human reader.
o There is not much sense in having an else after Punt() and _exit()
which never return. Also use pid_t for variables holding PIDs.
Max Okumoto [Thu, 7 Apr 2005 07:16:33 +0000 (07:16 +0000)]
FreeBSD-Date: 2005/04/01 13:35:35
FreeBSD-Date: 2005/04/01 16:23:48
Author: harti
Both struct Job and struct Shell are only used inside job.c so there is
no need to have them in a header file that is included by other modules.
Move them both into the c-file.
Remove a forward declaration that isn't needed anymore.
Max Okumoto [Thu, 7 Apr 2005 00:44:18 +0000 (00:44 +0000)]
FreeBSD-Date: 2005/04/01 10:53:43
FreeBSD-Date: 2005/04/01 11:12:29
FreeBSD-Date: 2005/04/01 11:20:17
FreeBSD-Date: 2005/04/01 12:31:15
FreeBSD-Date: 2005/04/01 12:42:03
FreeBSD-Date: 2005/04/01 13:02:17
FreeBSD-Date: 2005/04/01 13:06:05
FreeBSD-Date: 2005/04/01 13:25:45
Author: harti
Style: fix indentation
Max Okumoto [Thu, 7 Apr 2005 00:37:31 +0000 (00:37 +0000)]
FreeBSD-Date: 2005/03/31 11:35:56
FreeBSD-Date: 2005/03/31 11:40:53
Author: harti
Almost complete rewrite of the archive code (except for the Makefile parsing
part). Archive handling was broken at least since the move from BSD ar/ranlib
to GNU binutils because of the different archive format. This rewrite fixes
this by making make to carry around the defines for all formats (it supports)
so it can support all of them independent of the actually used one. The
supported formats are: traditional BSD (this seems to come from V7 at least,
short names only and __.SYMDEF), BSD4.4 (long names with #1/ and __.SYMDEF)
and SysV (extra name table and //). The only format not supported are broken
traditional archives where the member names are truncated to 15 characters.
Errors in the archive are not ignored anymore, but cause make to stop with
an error message. The command line option -A causes these errors to become
non-fatal. This is almost compatible with previous usage except for the
error message printed in any case.
Use a type-safe intrusive list for the archive cache.
Document the new -A flag for ignoring archive errors.
Reviewed by: Max Okumoto <okumoto@ucsd.edu> (without new error handling)
Max Okumoto [Thu, 7 Apr 2005 00:35:33 +0000 (00:35 +0000)]
Add compat code so code that uses types defined by c99
can compile on FreeBSD 4.X.
intmax_t
uintmax_t
INT64_MIN
strtoumax()
Harti has started using them in the FreeBSD code base.
Matthew Dillon [Wed, 6 Apr 2005 22:40:30 +0000 (22:40 +0000)]
add ports/polish
Matthew Dillon [Wed, 6 Apr 2005 20:52:55 +0000 (20:52 +0000)]
Update the various DragonFly supfiles, rename some of them to make their
function more obvious. Note the stable->preview label change.
Matthew Dillon [Wed, 6 Apr 2005 18:40:25 +0000 (18:40 +0000)]
Include fsid information in the mount point dump.
Matthew Dillon [Wed, 6 Apr 2005 18:39:53 +0000 (18:39 +0000)]
Correct an NFS bug related to ftruncate() operations. When the client
truncates a file it has to do a lot of magic on the buffer cache to scrap
buffers beyond the new EOF and properly handle the buffer that straddles
the truncation point. To ease the job the NFS client performs a flush.
This creates a problem, however, because the server includes attribute
information in its RPC replies (for NFSv3) which is then cached by the
client via nfs_loadattrcache(). This information is used by the client
to detect races with other entities modifying the same file but it creates
confusion when the client is in the middle of trying to do an ftruncate(),
especially if some of the buffer flushes prior to the truncate extend the
file. The server will report file sizes that don't match either the 'before'
or 'after' information that the client has.
To deal with this situation we have to check for mismatches after we flush
and loop if we find any, prior to issuing the setattrrpc(). Such mismatches
are virtually guarenteed to occur even with just a single NFS client working
the file. We also do a sanity check after issuing the stattrrpc(). This
second sanity check will only be triggered if our NFS client is racing
another client on the same file.
It's quite possible that there are still client-side bugs related to this
issue even with this commit.
Reported-by: Andreas Hauser <andy@splashground.de>
Matthew Dillon [Wed, 6 Apr 2005 03:36:27 +0000 (03:36 +0000)]
libstand is used by the boot loader, make sure the stack protector is
turned off.
YONETANI Tomokazu [Wed, 6 Apr 2005 03:25:46 +0000 (03:25 +0000)]
Bring in long-forgotten man pages for ACPI components currently available
on DragonFly.
Requested-By: Erik Wikström <erik-wikstrom [a] telia.com> and others
Acquired-from: FreeBSD
(acpi.4,v 1.50 acpi_thermal.4,v 1.6 acpi_toshiba.4,v 1.5)
David Xu [Tue, 5 Apr 2005 23:04:22 +0000 (23:04 +0000)]
Remove unused strong reference.
Matthew Dillon [Tue, 5 Apr 2005 22:37:37 +0000 (22:37 +0000)]
Add some minor in-code documentation.
Matthew Dillon [Tue, 5 Apr 2005 22:37:16 +0000 (22:37 +0000)]
Add cases for the nvidia nForce3 PRO and PRO S1. Who knows how well it
actually works.
Matthew Dillon [Tue, 5 Apr 2005 22:36:33 +0000 (22:36 +0000)]
Make sure neither the boot loader nor the kernel is compiled with the
stack protector.
Matthew Dillon [Tue, 5 Apr 2005 21:41:19 +0000 (21:41 +0000)]
Automatically calculate the kernel revision for release tags by extracting
the revision from the tag itself. Remove any DragonFly_ prefix or _Slip
suffix from the branch name. This will allow us to create a slip tag that
synchronizes the sources with the automated subversion increment. This also
reduces the work required to create a new, properly identified branch.
Automatically postpend a subversion from /usr/src/sys/conf/subvers-TAG, on
release builds only, where TAG is in the form RELEASE_X_Y.
Joerg Sonnenberger [Tue, 5 Apr 2005 17:39:22 +0000 (17:39 +0000)]
Add nv(4) man page.