Matthew Dillon [Sat, 5 Mar 2005 05:08:30 +0000 (05:08 +0000)]
Fix a number of alignment and offset bugs that were corrupting the
journaling output stream. In particular, the stream record expansion
code (journal_extend()) was not returning the correct payload pointer.
Add PUTPAGES support.
Further improve the code documentation.
Matthew Dillon [Sat, 5 Mar 2005 05:04:32 +0000 (05:04 +0000)]
Add a simple msf_buf_bytes() inline which returns the number of bytes
of data the msfbuf represents.
Liam J. Foy [Sat, 5 Mar 2005 00:08:46 +0000 (00:08 +0000)]
- Update mesg(1) to confirm to SUSv3 changes previously committed.
- Add an example to mesg(1)
Taken from: FreeBSD (rev: 1.9 and rev: 1.10)
Max Okumoto [Fri, 4 Mar 2005 23:51:28 +0000 (23:51 +0000)]
patch-7.110 var.c
constify Var_Parse() Yea!
Max Okumoto [Fri, 4 Mar 2005 23:50:28 +0000 (23:50 +0000)]
patch-7.108
VarModify(), VarSortWords()
Constify functions that used brk_string()
Max Okumoto [Fri, 4 Mar 2005 23:50:10 +0000 (23:50 +0000)]
patch-7.107
VarModify()
reindent
Max Okumoto [Fri, 4 Mar 2005 23:49:42 +0000 (23:49 +0000)]
patch-7.106
brk_string()
- constify brk_string(), input param str is never modified.
Max Okumoto [Fri, 4 Mar 2005 23:49:20 +0000 (23:49 +0000)]
patch-7.105a
modifier_S()
use better variable names
Max Okumoto [Fri, 4 Mar 2005 23:48:53 +0000 (23:48 +0000)]
patch-7.105
ParseModifier() and modifier_S()
- move 'S' modifier code into its own function.
Max Okumoto [Fri, 4 Mar 2005 23:48:35 +0000 (23:48 +0000)]
patch-7.104
ParseModifier()
- Move end of string test into loop.
- Clean up init section of 'S' modifier
- Restructure oldstring and newstring so they have the
same structure.
Max Okumoto [Fri, 4 Mar 2005 23:48:12 +0000 (23:48 +0000)]
ParseModifier()
- Renamed variable 'del' to 'delim'
- Used cp[0] instead of *cp to make it match other cp[n]
references.
Max Okumoto [Fri, 4 Mar 2005 23:44:37 +0000 (23:44 +0000)]
ParseModifier()
Reindent with 'S' modifier
Max Okumoto [Fri, 4 Mar 2005 23:43:33 +0000 (23:43 +0000)]
ParseModifier() and modifier_M()
- move 'M' and 'N' modifier code into it's own function.
- use better variable names.
patch-7.101
patch-7.101a
Matthew Dillon [Fri, 4 Mar 2005 21:37:27 +0000 (21:37 +0000)]
Fix an incorrect pointer in a journal_build_pad() that led to a panic.
Add additional assertions and code comments.
Hiten Pandya [Fri, 4 Mar 2005 17:21:14 +0000 (17:21 +0000)]
Add a typedef msf_t for struct msf_buf *.
Liam J. Foy [Fri, 4 Mar 2005 16:54:37 +0000 (16:54 +0000)]
- perror() -> warn()
Joerg Sonnenberger [Fri, 4 Mar 2005 13:49:02 +0000 (13:49 +0000)]
Don't use stdint.h on FreeBSD.
Submitted-by: Chuck Tuffli <chuck_tuffli@agilent.com>
Liam J. Foy [Fri, 4 Mar 2005 10:34:39 +0000 (10:34 +0000)]
- Write status information to stdout instead of stderr.
- Affect the first of stdin, stdout and stderr that is attached to a terminal.
This brings mesg(1) up to SUSv3 conformance.
Taken from: FreeBSD rev: 1.6
Matthew Dillon [Fri, 4 Mar 2005 06:10:09 +0000 (06:10 +0000)]
msf_buf_alloc() can be called with a NULL vm_page_t when e.g. the first
page cannot be easily located, typically due to a user buffer which is
not faulted in. Fix a NULL pointer dereference panic that occurs in that
case.
Chris Pressey [Fri, 4 Mar 2005 06:06:57 +0000 (06:06 +0000)]
Small step towards WARNS=6:
- Rename function parameters to avoid conflict with globals.
- Constify a local char *.
Matthew Dillon [Fri, 4 Mar 2005 05:58:46 +0000 (05:58 +0000)]
Remove some of the nastier debugging printfs.
Jeffrey Hsu [Fri, 4 Mar 2005 05:57:50 +0000 (05:57 +0000)]
The route in a syncache entry is cleared if the connection was successfully
completed, so there's no need to keep a separate SCF_KEEPROUTE flag.
Chris Pressey [Fri, 4 Mar 2005 05:47:03 +0000 (05:47 +0000)]
Style(9):
- Remove parameter names from userland-visible prototypes.
- No space between function name and parenthesis in prototypes.
Matthew Dillon [Fri, 4 Mar 2005 05:25:26 +0000 (05:25 +0000)]
Journaling layer work. Write the actual data associated with a VOP_WRITE
to the journal. Fix a number of bugs in the journal memory FIFO code.
This commit has a slight UIO hack. Because UIOs are currently destroyed by
I/O operations on them (the iovec's get cleared), we had to make a copy of
the uio for the journaling scan.
Note that the UIO scan currently maps the data into KVA via msf_uio_iterate().
This will be made more optimal when the I/O path is fixed at some future
date to support direct page lists.
Chris Pressey [Fri, 4 Mar 2005 05:22:57 +0000 (05:22 +0000)]
Style(9):
- ANSIfy function definitions.
Matthew Dillon [Fri, 4 Mar 2005 05:21:17 +0000 (05:21 +0000)]
msf_buf_kva() now returns a char * rather then a vm_offset_t.
Matthew Dillon [Fri, 4 Mar 2005 05:20:29 +0000 (05:20 +0000)]
Make msf_buf_kva() return the correct base address. It was not including
the base offset relative to the start of the page for MSFBUFs mapped from
user or kernel buffers.
Implement msf_uio_iterate(). This function iterates through a UIO, maps
the data into the kernel as required, and calls the specified callback
routine. It is used by kernel procedures which need to scan the data
represented by a UIO.
Matthew Dillon [Fri, 4 Mar 2005 05:17:37 +0000 (05:17 +0000)]
Pass the memfifo option to the kernel.
Jeffrey Hsu [Fri, 4 Mar 2005 04:38:47 +0000 (04:38 +0000)]
Prefer rtlookup() to rtalloc() when not saving the result of the look up.
Chris Pressey [Fri, 4 Mar 2005 04:31:11 +0000 (04:31 +0000)]
Style(9):
- Sort, format, and trim #include's.
- libutil.h and login_cap.h are located in this directory, so don't
treat them as system-wide headers. (<libutil.h> -> "libutil.h")
Tested-by: make quickworld && make installworld && reboot
Jeffrey Hsu [Fri, 4 Mar 2005 03:48:25 +0000 (03:48 +0000)]
Cosmetic changes only.
Jeffrey Hsu [Fri, 4 Mar 2005 03:37:43 +0000 (03:37 +0000)]
Change a 'char *' to a 'void *' because that field is not accessed
as characters.
Add a comment clarifying that fillrtmsg() is not an external API function.
Jeffrey Hsu [Fri, 4 Mar 2005 03:34:47 +0000 (03:34 +0000)]
Readability changes.
Jeffrey Hsu [Fri, 4 Mar 2005 03:05:59 +0000 (03:05 +0000)]
Remove (void) cast before a function call with an unused return value.
Jeffrey Hsu [Fri, 4 Mar 2005 02:54:31 +0000 (02:54 +0000)]
None of the callers of rtredirect() want to know the route that was modified,
so remove the last argument to rtredirect() in order to simplify reasoning
about route reference counts.
Chris Pressey [Fri, 4 Mar 2005 02:53:55 +0000 (02:53 +0000)]
Style(9):
- Sort options in usage statement correctly, and synch man page
SYNOPSIS.
- Comment describing what the program does, goes just before main().
- Minor adjustments to other comments.
Chris Pressey [Fri, 4 Mar 2005 02:37:44 +0000 (02:37 +0000)]
Style(9):
- Remove `register' keywords.
Chris Pressey [Fri, 4 Mar 2005 02:29:19 +0000 (02:29 +0000)]
WARNS=6 cleaning, stage 6/6:
- Add WARNS?=6 to Makefile and remove manually-passed -W flags.
Submitted-by: Sepherosa Ziehau <sepherosa_at_softhome_dot_net>
Chris Pressey [Fri, 4 Mar 2005 02:22:38 +0000 (02:22 +0000)]
Reduce warnings as part of WARNS=6 cleaning, stage 5/6:
- Use size_t instead of int, as appropriate, to avoid comparisons
between signed and unsigned values.
Jeffrey Hsu [Fri, 4 Mar 2005 02:21:49 +0000 (02:21 +0000)]
Convert the struct domain next pointer to an SLIST.
Chris Pressey [Fri, 4 Mar 2005 01:47:54 +0000 (01:47 +0000)]
Reduce warnings as part of WARNS=6 cleaning, stage 4/6 (or so):
- in6_fillscopeid() is a NOP unless __KAME__ and KAME_SCOPEID are
defined, so only define and call it if they are both defined.
Chris Pressey [Fri, 4 Mar 2005 01:22:37 +0000 (01:22 +0000)]
Reduce warnings as part of WARNS=6 cleaning, stage 3/6 (or so):
- Rename local variables to avoid conflict with library functions,
globals, and other (less deeply-nested) locals.
Submitted-by: Sepherosa Ziehau <sepherosa_at_softhome_dot_net>
Matthew Dillon [Fri, 4 Mar 2005 00:44:49 +0000 (00:44 +0000)]
This is a major revamping of our MSFBUF API. MSFBUFs are used to map
things into kernel memory so the kernel and device drivers which require
such mappings can easily get them. MSFBUFs will soon replace the buffer
cache's KVA mapping code, be used in the journaling code, get and put pages,
reverse-map local kernel buffers, as well as other things.
The eventual goal is to make KVA mappings completely optional and pass
XIOs down the I/O chain, and if a device driver needs a mapping it would
then be able to use an MSFBUF to get it. This in turn will allow the
buffer cache to manage much larger data sets and remove the absurdly low
dirty data buffer limitations we currently have.
* Replace the hash table with a simple hint in the vm_page structure.
Currently we do not avoid the overhead of populating an XIO structure or
regenerating the page list, but the use of qenter2 does avoid the invltlb
overhead. The API and internal structure is designed to allow far better
optimizations of XIO mappings in the future.
* Remove MSFBUF sharing. Sharing does not occur very often and not doing
it allows an msfbuf to inherit certain context-sensitive information
such as a supplied XIO pointer, cpu affinity (for NUMA), easier SMP
operation, etc.
* Take advantage of the recent XIO changes.
* Add API routines to generate MSFBUFs from various sources, including
page lists, XIO's, user bufs, kernel bufs, and soon UIO's.
Submitted-by: Collaboration between Hiten Pandya and Matt Dillon
Chris Pressey [Fri, 4 Mar 2005 00:11:11 +0000 (00:11 +0000)]
Reduce warnings as part of WARNS=6 cleaning, stage 2/6 (or so):
- Mark many callback function arguments as __unused.
Submitted-by: Sepherosa Ziehau <sepherosa_at_softhome_dot_net>
Matthew Dillon [Fri, 4 Mar 2005 00:01:30 +0000 (00:01 +0000)]
Also, do not use M_NOWAIT for an allocation that must succeed.
Chris Pressey [Thu, 3 Mar 2005 23:52:00 +0000 (23:52 +0000)]
Reduce warnings as part of WARNS=6 cleaning, stage 1/6:
- Fully specify all elements in static initializers.
Submitted-by: Sepherosa Ziehau <sepherosa_at_softhome_dot_net>
Max Okumoto [Thu, 3 Mar 2005 23:39:06 +0000 (23:39 +0000)]
Reduce diffs between FreeBSD
Max Okumoto [Thu, 3 Mar 2005 23:33:34 +0000 (23:33 +0000)]
Reduce diff between FreeBSD.
Chris Pressey [Thu, 3 Mar 2005 23:17:11 +0000 (23:17 +0000)]
Add the pfi_backend variable, which can be used to specify which
installer backend program to start. Also, correctly pass the name of
the rendezvous to the backend program when the CGI frontend is used.
Chris Pressey [Thu, 3 Mar 2005 22:57:39 +0000 (22:57 +0000)]
Update cdrtools package requirement to version 2.01.
Jeffrey Hsu [Thu, 3 Mar 2005 19:50:33 +0000 (19:50 +0000)]
Use malloc(M_ZERO) in preference to separate bzero() after allocation.
Max Okumoto [Thu, 3 Mar 2005 19:13:00 +0000 (19:13 +0000)]
My regression test had a bug in it. Quoting a value in a Makefile
causes make(1) to treat it as one word.
APPLE = "AAA AAA AAA" # one word
ORANGE = AAA AAA AAA # three words
Joerg Sonnenberger [Thu, 3 Mar 2005 19:07:29 +0000 (19:07 +0000)]
Remove bad line breaks.
Max Okumoto [Thu, 3 Mar 2005 18:22:58 +0000 (18:22 +0000)]
parse.c:
ParseLinkSrc()
Update the description comment of the function to reflect
the previous change.
parse.c:
Fix parsing of archive specifications on the target side
of dependency lines. It seems that nobody is actually is
using the archive-feature of make.
arch.c
Fix parsing of archives: in System V archive format the
member names is terminated with a slash. Although we are
not System V, ar has been configured to put that slash in.
This format allows filenames with trailing spaces.
Style: remove unneeded parantheses in conditionals.
Author: harti
Date: 2005/03/03 10:07:28
Date: 2005/03/03 10:09:39
Date: 2005/03/03 10:46:23
Date: 2005/03/03 11:34:04
Liam J. Foy [Thu, 3 Mar 2005 17:09:58 +0000 (17:09 +0000)]
- Remove unnecessary header
- err() magic
- Constify
- Minor Style(9)
Matthew Dillon [Thu, 3 Mar 2005 06:56:41 +0000 (06:56 +0000)]
off_t is a signed value. The last commit caused the kernel to fail with
a end < start panic.
Chris Pressey [Thu, 3 Mar 2005 03:43:05 +0000 (03:43 +0000)]
Remove 'pristine' file tree. It is no longer needed now that there is
a /etc.hdd directory on the install image.
Chris Pressey [Wed, 2 Mar 2005 23:09:00 +0000 (23:09 +0000)]
Update installer to version 1.1.4. Highlights include:
- NetBoot Installation Services support.
- Allow user to MFS-back any of their new filesystems.
- Handle cancel/error when setting root password more gracefully.
- Show "Press F1 for Help" only when there really is help to be shown.
- Display progress bar and bold chars correctly on monochrome terms.
- Coalesce multiple rc.conf writes and do them all at once at the end.
- Write keymaps, screenmaps and fonts w/o dirnames and extensions.
- Correctly install/remove prerequisite/dependent packages.
- Sources to copy during install can be specified via a conf file.
- Many and various generally-user-invisible bugfixes.
- NLS support (not currently enabled.)
- Buildable on multiple platforms.
Work-by: geekgod, cpressey, Denis Peplin, Robert Nagy, Brenden Grace,
& many others who sent in bug reports & feature suggestions
Max Okumoto [Wed, 2 Mar 2005 19:39:51 +0000 (19:39 +0000)]
Use the LST_FOREACH macro instead of the Lst_ForEach function. This
saves function calls and reduces void casting.
Date: 2005/03/02 08:30:49 and 2005/03/02 12:00:21
Author: harti
Taken-from: FreeBSD
Liam J. Foy [Wed, 2 Mar 2005 19:27:39 +0000 (19:27 +0000)]
- Back out part of previous commit. I thought free() was the prefered
method instead of leaving it up to exit().
Matthew Dillon [Wed, 2 Mar 2005 19:14:51 +0000 (19:14 +0000)]
PHOLD is a throwback to FreeBSD that we don't actually need here. This
also fixes an init lockup (due to a missing PRELE before process exit).
Reported-by: Rumko <rumcic@gmail.com>
Hiten Pandya [Wed, 2 Mar 2005 18:42:09 +0000 (18:42 +0000)]
Rename the flags for sf_buf_alloc(9) to be in line with FreeBSD:
SFBA_PCATCH -> SFB_CATCH
SFBA_QUICK -> SFB_CPUPRIVATE
Discussed-with: Matthew Dillon <dillon at apollo.backplane.com>
Liam J. Foy [Wed, 2 Mar 2005 17:32:06 +0000 (17:32 +0000)]
- Make sure we free() 'modep' after setmode()
- Add stdlib.h for both free() and exit()
Eirik Nygaard [Wed, 2 Mar 2005 16:14:03 +0000 (16:14 +0000)]
Be consistent in the use of BOOLEAN.
Eirik Nygaard [Wed, 2 Mar 2005 15:12:17 +0000 (15:12 +0000)]
Merge from vendor branch NCURSES:
Add ncurses 5.4 source code.
Eirik Nygaard [Wed, 2 Mar 2005 15:12:17 +0000 (15:12 +0000)]
Add ncurses 5.4 source code.
Joerg Sonnenberger [Wed, 2 Mar 2005 13:45:25 +0000 (13:45 +0000)]
Correct a mixup between NLSGRP and NLSOWN. Add LOCALE{DIR,GRP,OWN,MODE}.
Joerg Sonnenberger [Wed, 2 Mar 2005 11:48:15 +0000 (11:48 +0000)]
Use ULLONG_MAX, off_t is unsigned.
Max Okumoto [Wed, 2 Mar 2005 11:01:15 +0000 (11:01 +0000)]
Fixed bad result file.
Max Okumoto [Wed, 2 Mar 2005 10:58:59 +0000 (10:58 +0000)]
Add another. This one check if the behavior of ths S modifier is the same.
Note: It appears that both ${MAKE_VAR:S/^FOO/BAR} and ${MAKE_VAR:S/FOO$/BAR}
do not work.
Max Okumoto [Wed, 2 Mar 2005 10:55:37 +0000 (10:55 +0000)]
Hack update code since we now have separate src and work directories.
Joerg Sonnenberger [Wed, 2 Mar 2005 10:38:43 +0000 (10:38 +0000)]
Fix prototypes. This functions actually need a parameter.
Joerg Sonnenberger [Wed, 2 Mar 2005 10:15:46 +0000 (10:15 +0000)]
Fix warning about pre-ANSI function types.
Chris Pressey [Wed, 2 Mar 2005 07:36:32 +0000 (07:36 +0000)]
Add WARNS?=6 to Makefile.
Chris Pressey [Wed, 2 Mar 2005 07:35:32 +0000 (07:35 +0000)]
Raise WARNS to 6:
- Un-const-ify a function argument;
- Use a signed integer in a loop bounded by a signed integer.
Chris Pressey [Wed, 2 Mar 2005 07:20:44 +0000 (07:20 +0000)]
Improve data validation:
- Simplify the check for a well-formed -a argument.
- Move the check for overflowing the command table from addcmd()
(which is just a driver) into addmac() (which does the real work,
and which is called from places besides addcmd().)
- Add a check for overflowing the macro-pair table.
- In each case, use a clearer (while) loop for detecting overflow.
- In each case, error out by using errx().
Joerg Sonnenberger [Wed, 2 Mar 2005 07:16:23 +0000 (07:16 +0000)]
WARNS=6 cleanup.
Copy collate.h from libc/locale instead of using -I, because the current
ctype.h includes runetype.h and that gets picked from the same directory.
Joerg Sonnenberger [Wed, 2 Mar 2005 07:09:38 +0000 (07:09 +0000)]
Don't define isatty, it's already defined in unistd.h.
Chris Pressey [Wed, 2 Mar 2005 06:59:23 +0000 (06:59 +0000)]
Sync with FreeBSD, random.c:1.17 and random.6:1.7-1.8 (ru):
- Fix SYNOPSIS and sync program's usage output with it;
- Expand stdin/stdout;
- Order SEE ALSO;
- Trim BUGS and HISTORY;
- Sort sections.
David Xu [Wed, 2 Mar 2005 06:17:17 +0000 (06:17 +0000)]
In kern_sigtimedwait, do not mask SIGKILL because a stopped process
is always able to be killed by SIGKILL.
Joerg Sonnenberger [Wed, 2 Mar 2005 06:08:29 +0000 (06:08 +0000)]
Use isdigit instead of isnumber.
Joerg Sonnenberger [Wed, 2 Mar 2005 05:48:48 +0000 (05:48 +0000)]
Replace digittoint(c) with (c - '0'), we already know that c is a digit,
so the hex numbers can't occur. digittoint is a BSD function and not
part of any standard.
Joerg Sonnenberger [Wed, 2 Mar 2005 05:15:13 +0000 (05:15 +0000)]
Use the standard isxdigit instead of the non-standard ishexnumber.
Chris Pressey [Wed, 2 Mar 2005 05:03:20 +0000 (05:03 +0000)]
Raise WARNS to 6.
Chris Pressey [Wed, 2 Mar 2005 05:00:21 +0000 (05:00 +0000)]
Merge with FreeBSD, revision 1.16 (stefanf):
"Properly initialise 'filename' so that random -l doesn't try to open
NULL.
Const-qualify 'filename' to avoid a strdup() call due to
-Wwrite-strings silliness."
Chris Pressey [Wed, 2 Mar 2005 04:33:12 +0000 (04:33 +0000)]
Raise WARNS to 6:
- Declare tables of commands as arrays of mutable string buffers
instead of as arrays of string constants, and just copy new
data into the table, instead of malloc()ing new space for it.
Joerg Sonnenberger [Wed, 2 Mar 2005 03:13:01 +0000 (03:13 +0000)]
Use NXCC to compile a runnable executable. The libc we are building might
not be compatible.
Joerg Sonnenberger [Wed, 2 Mar 2005 00:56:21 +0000 (00:56 +0000)]
Merge from vendor branch GCC:
Add C++ ctype based on NetBSD-style ctype.h.
Joerg Sonnenberger [Wed, 2 Mar 2005 00:56:21 +0000 (00:56 +0000)]
Add C++ ctype based on NetBSD-style ctype.h.
Chris Pressey [Wed, 2 Mar 2005 00:43:42 +0000 (00:43 +0000)]
Constify filename variable.
Joerg Sonnenberger [Tue, 1 Mar 2005 23:40:57 +0000 (23:40 +0000)]
RIP Alpha libc.
Matthew Dillon [Tue, 1 Mar 2005 23:35:16 +0000 (23:35 +0000)]
Clean up the XIO API and structure. XIO no longer tries to 'track' partial
copies into or out of an XIO. It no longer adjusts xio_offset or xio_bytes
once they have been initialized. Instead, a relative offset is now passed
to API calls to handle partial copies. This makes the API a lot less confusing
and makes the XIO structure a lot more flexible, shareable, and more suitable
for use by higher level entities (buffer cache, pipe code, upcoming MSFBUF
work, etc).
Max Okumoto [Tue, 1 Mar 2005 23:32:52 +0000 (23:32 +0000)]
patch-7.100
ParseModifier()
- constify 'M' and 'N' modifiers.
Max Okumoto [Tue, 1 Mar 2005 23:29:29 +0000 (23:29 +0000)]
patch-7.99
VarParseLong()
- Move default code up into previous if-block since the
else block always returns.
Max Okumoto [Tue, 1 Mar 2005 23:28:46 +0000 (23:28 +0000)]
patch-7.98
VarParseLong()
- Merge two if-statement into a if-else-statement since the
two tests are compliments of each other.
Max Okumoto [Tue, 1 Mar 2005 23:28:18 +0000 (23:28 +0000)]
patch-7.97
VarParseLong()
- cleanup calculation of dynamic
Max Okumoto [Tue, 1 Mar 2005 23:27:53 +0000 (23:27 +0000)]
patch-7.96
VarParseLong()
- Remove dead code, and reindent
Max Okumoto [Tue, 1 Mar 2005 23:27:25 +0000 (23:27 +0000)]
patch-7.95
VarParseLong()
- Create new scope around default code at end of function.
Then copy that function above in a conditional. if (v != NULL)
Max Okumoto [Tue, 1 Mar 2005 23:26:55 +0000 (23:26 +0000)]
patch-7.94
VarParseLong()
- Create variable consumed, which contains the number of characters
that we have already consumed.
Max Okumoto [Tue, 1 Mar 2005 23:26:30 +0000 (23:26 +0000)]
patch-7.93
VarParseLong()
- renamed variable tstr to ptr in while loop.
- Create a new subscope to contain new tstr pointer which is
a const pointer, this will prove that the code in the scope
does not move the pointer.
- Move other variables from function scope into the new scope,
this scope will prob end up as a new function.