Robert Drehmel [Wed, 14 Aug 2002 23:20:48 +0000 (23:20 +0000)]
- Add the 'restrict' qualifier to the function definition of
strftime(3) for IEEE Std 1003.1-2001 compliance and remove
excessive usage of the 'const' qualifier that was neither
present in the prototype in the publice header, nor in the
local prototype just above the function definition.
- Replace the K&R function definition with a ANSI-C one.
- Update the prototype of strftime(3) in its manual page.
Robert Drehmel [Wed, 14 Aug 2002 22:59:22 +0000 (22:59 +0000)]
- Add the 'restrict' qualifier to the definitions of the string
concatenation and copy functions using the '__restrict' macro.
This is to satisfy IEEE Std 1003-1.2001.
- Use ANSI-C function definitions.
- Add the 'restrict' keyword to the manual pages, too.
Robert Drehmel [Wed, 14 Aug 2002 21:16:41 +0000 (21:16 +0000)]
- Add the 'restrict' qualifier to match the IEEE Std 1003.1-2001
prototype of the tdelete(3) function.
- Remove duplicated space.
- Use an ANSI-C function definition for tdelete(3).
- Update the manual page.
Robert Drehmel [Wed, 14 Aug 2002 21:01:04 +0000 (21:01 +0000)]
- Add the 'restrict' qualifier required by IEEE Std 1003.1-2001
to the function definition of strxfrm(3) in form of our
'__restrict' macro.
- Use an ANSI-C function definition for strxfrm(3).
- Change the manual page accordingly.
Robert Drehmel [Wed, 14 Aug 2002 20:40:35 +0000 (20:40 +0000)]
-Add the restrict required by IEEE Std 1003.1-2001 in form
of our __restrict macro to the prototypes and function
definitions of inet_pton and inet_ntop.
- Use ANSI-C function argument lists.
- Adjust the prototypes in the manual page.
Joerg Wunsch [Wed, 14 Aug 2002 09:28:19 +0000 (09:28 +0000)]
Implement the DIOCGSECTORSIZE and DIOCGMEDIASIZE ioctls, and rename
the DIOCGDINFO ioctl into DIOCGDVIRGIN. This allows the newfs(8) to
work on vinum volumes without the now removed -v option.
Robert Watson [Wed, 14 Aug 2002 01:04:43 +0000 (01:04 +0000)]
Move to a nested include of _label.h instead of mac.h in sys/sys/*.h
(Most of the places where mac.h was recursively included from another
kernel header file. net/netinet to follow.)
Ian Dowse [Tue, 13 Aug 2002 16:25:38 +0000 (16:25 +0000)]
If a timeout is specified, make sure that rcmd() completes within
the specified time. Previously, rsh could potentially hang indefinitely
at this point even when a timeout was set, for example if the server
accepts the connection and then never sends any reply.
PR: bin/20042
Submitted by: Keith White <Keith.White@site.uottawa.ca>
MFC after: 1 week
Robert Watson [Tue, 13 Aug 2002 15:54:28 +0000 (15:54 +0000)]
Define 'struct label' in _label.h rather than mac.h, which will permit
us to reduce namespace pollution by doing a nested include of _label.h
rather than mac.h. mac.h contains lots of baggage, whereas _label.h
contains much less. A follow-up sweep to change nested inclusion will
follow. The problem regarding exporting 'struct label' to userland
due to excessive exporting of kernel structures to userland still
needs to be resolved.
Expand the arguments to ffs_ext{read,write}() to their component
parts rather than use vop_{read,write}_args. Access to these
functions will ultimately not be available through the
"vop_{read,write}+IO_EXT" API but this functionality is retained
for debugging purposes for now.
if the filesystem backing /tmp did not support EAs.
The real solution is probably to have the extattrctl syscall do the
unlocking rather than depend on the filesystem to do it. Considering
that extattrctl is going to be made obsolete anyway, this has dogwash
priority.
Introduce typedefs for the member functions of struct vfsops and employ
these in the main filesystems. This does not change the resulting code
but makes the source a little bit more grepable.
Alfred Perlstein [Tue, 13 Aug 2002 08:47:17 +0000 (08:47 +0000)]
Make SYSVSEM mpsafe. Each semaphore set gets its own lock, however
there is a global lock over the undo structures because of the way
they are managed.
Switch to using SLIST instead of rolling our own linked list.
Fix several races where a permission check was done before a
copyin/copyout, if the copy happened to fault it may have been
possible to race for access to a semaphore set that one shouldn't
have access to.
Doug Barton [Tue, 13 Aug 2002 08:23:49 +0000 (08:23 +0000)]
A cooperative effort...
1. Update the code to the latest from NetBSD, which includes:
* A new command line option to suppress stat(2) errors
* Output is now done via stdio
* Fixes for bitwise OR'ing of letters
2. Andrew from NetBSD merged in our own Johan's readlink patches.
3. Andrew also merged in some brucifications.
I made some small contributions to the header cleanup, and tried
to generally improve the overall style(9)'ishness. I also documented
the new -q option and added the *LINKS to the Makefile.
All responsibility for mistakes in merging the contributions
from the above is mine.
Obtained from: Andrew Brown <atatat@NetBSD.org>, bde, johan
Alfred Perlstein [Tue, 13 Aug 2002 08:00:36 +0000 (08:00 +0000)]
Make SYSVMSG mpsafe. Right now there is a global lock over the
entire subsystem, we could move to per-message queue locks, however
the messages themselves seem to come from a global pool and to avoid
over-locking this code (locking individual queues, then the global
pool) I've opted to just do it this way.
Robert Watson [Tue, 13 Aug 2002 02:47:13 +0000 (02:47 +0000)]
Introduce support for labeling and access control of pipe objects
as part of the TrustedBSD MAC framework. Instrument the creation
and destruction of pipes, as well as relevant operations, with
necessary calls to the MAC framework. Note that the locking
here is probably not quite right yet, but fixes will be forthcoming.
Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs