peter [Sun, 18 Aug 1996 16:16:18 +0000 (16:16 +0000)]
make the declaration of the __error() function C++ safe. Otherwise,
the linker comes up with undefined variables because of the #define errno
when using _THREAD_SAFE.
peter [Sun, 18 Aug 1996 16:14:03 +0000 (16:14 +0000)]
Add __unused to the list of __dead, __dead2, __pure, __pure2 etc.
gcc-2.7 has __attribute__((unused)) as a postfix operator on function
arguments to prevent -Wunused from complaining about them. gcc < 2.7
doesn't have anything like this (and doesn't care about it either).
eg:
int foo(bar, baz)
int bar __unused; /* needed because of function pointer typing */
char *baz __unused;
{
return (1);
}
wosch [Sun, 18 Aug 1996 12:47:33 +0000 (12:47 +0000)]
new section ``Unix history''
* Saulus, Peter H. _A quarter century of UNIX_.
* Simon Garfinkel, Daniel Weise, Steven Strassmann.
_The UNIX-HATERS Hanbook_.
peter [Sat, 17 Aug 1996 10:37:28 +0000 (10:37 +0000)]
Fixes from NetBSD:
- inet_ntoa() returns a pointer to a static buffer, dont use it twice
in the same printf().
- prevent the possibility of never timing out
- Report two more ICMP error types (prohibited nets etc)
And some (commented out) enhancements that I use, but some don't like.
asami [Sat, 17 Aug 1996 10:16:02 +0000 (10:16 +0000)]
(1) Ignore patches directory without a patch with a message instead of
bombing mercilessly.
(2) If that directory has a directory called CVS, remind the user of
the existence of the "-P" option to cvs co and update.
(3) While I'm here, clean up the PATCH_DEBUG code a bit. In
particular, don't duplicate a whole bunch of code just for adding
a single "echo" statement. ;)
peter [Sat, 17 Aug 1996 07:15:38 +0000 (07:15 +0000)]
Put on my flame-proof suit and attempt to make the /etc/motd banner
editor safer. The old one was too deadly, if the blank line after the
"FreeBSD" version banner was missing, it would delete from the beginning
of the file up until the next blank line, possibly to the end of file.
This was not good.
peter [Fri, 16 Aug 1996 10:16:32 +0000 (10:16 +0000)]
restore nsccons variable from rev 1.115 that was deleted in rev 1.136
during phk's staticize/cleanup commits. pstat needs it, the MAXCONS
option is not visible anywhere else, and pstat uses it to find the bounds
of the sccons[MAXCONS] array, which varies.
jkh [Fri, 16 Aug 1996 07:53:29 +0000 (07:53 +0000)]
Go back to sigaction again now that Peter has found the problem.
Mine was just a "let's move on" kinda fix, Peter's does it right. :) Submitted-by: peter
jkh [Thu, 15 Aug 1996 21:06:52 +0000 (21:06 +0000)]
Make libncurses catch SIGWINCH and update the values for LINES and COLS.
I was perplexed when an example I'd written to show the values for these
variables changing as an xterm window was resized didn't work, and looking
into it I see that size tracking for LINES and COLS seems to be one SVR4
enhancement which didn't come across with libncurses.
jkh [Thu, 15 Aug 1996 20:15:18 +0000 (20:15 +0000)]
Add something so that you can solve the puzzle automatically by pressing
`a' (I'm using this code as a test for something else, and just happened
to add this in the process :).
smpatel [Thu, 15 Aug 1996 16:33:32 +0000 (16:33 +0000)]
Fix fdavail() so that correctly pays attention to the rlimit.
Fixes unp_externalize panic which occurs when a process is at it's
ulimit for file descriptors and tries to receive a file descriptor from
another process.
dyson [Thu, 15 Aug 1996 06:45:01 +0000 (06:45 +0000)]
Certain vnode buffer list operations were not being spl protected,
and they needed to be. Brelse for example can be called at interrupt
level, and the buffer list operations were not being protected from it.
julian [Tue, 13 Aug 1996 22:20:20 +0000 (22:20 +0000)]
Submitted by: Archie@whistle.com
if making a interface route, and it's a P2P link,
then also automatically lable it as an llinfo entry so that
gated and friends don't clobber it..
julian [Tue, 13 Aug 1996 19:48:41 +0000 (19:48 +0000)]
Submitted by: doug Rabson (dfr@render.com)
cleaning up some of the vnode usage..
(I'm sure it still needs more..)
where can one find out what each vfs call expects to be locked
on completion, and how can one find out what each layer expects
to be freed on error.?
pst [Tue, 13 Aug 1996 19:43:41 +0000 (19:43 +0000)]
Completely rewrite handling of protocol field for firewalls, things are
now completely consistent across all IP protocols and should be quite a
bit faster.
pst [Tue, 13 Aug 1996 19:43:24 +0000 (19:43 +0000)]
Completely rewrite handling of protocol field for firewalls, things are
now completely consistent across all IP protocols and should be quite a
bit faster.
Use getprotoname() extensively, performed minor cleanups of admin utility.
The admin utility could use a good kick in the pants.
Basicly, these were the minimal changes I could make to the code
to get it up to tollerable shape. There will be some future commits
to clean up the basic architecture of the firewall code, and if
I'm feeling ambitious, I may pull in changes like NAT from Linux
and make the firewall hooks comletely generic so that a user can
either load the ipfw module or the ipfilter module (cf Darren Reed).
peter [Tue, 13 Aug 1996 17:49:45 +0000 (17:49 +0000)]
Fix nasty bracketing/precedence bug. Every time something read (and
refilled) a file that was either line- or un-buffered, all files were
flushed. According to the code comment, the flush (according to ANSI)
is supposed to happen on write + line buffered output files, not _all_
files.
Obtained from: OpenBSD / Theo de Raadt, possibly from proven@cygnus.com
julian [Tue, 13 Aug 1996 07:21:45 +0000 (07:21 +0000)]
symlink support in devfs.
it only barely works so don't get too carried away..
I noticed that teh symlink is length 0..
I guess I'll fix that tomorrow..
it also sometimes panics with "cleaned vnode isn't" but it's not more
broken than it was before.. I really want to go over it with someone
who understands the lifecycle of a vnode better than I do..
joerg [Mon, 12 Aug 1996 21:31:07 +0000 (21:31 +0000)]
Fix two minor oddities introduced by my yesterday's patches:
. preserve a multi-char sequence in a small static buffer inside
pccngetc(), so it won't be clobbered later (used to happen when
breaking into DDB user Ctrl-Alt-ESC), and
. simplify the ``keystroke is present'' determination in sgetc(), thus
making pccncheck() actually working without waiting for a keystroke.
ache [Mon, 12 Aug 1996 18:38:49 +0000 (18:38 +0000)]
There is so many places where range comparation (using collate)
needed (much more than I think initially), so I forced to add
new user-visible non-standard function to libc.
ache [Mon, 12 Aug 1996 18:30:43 +0000 (18:30 +0000)]
There is so many places where range comparation (using collate)
needed (much more than I think initially), so I forced to add
new user-visible non-standard function to libc.