Tor Egge [Wed, 14 May 1997 02:44:27 +0000 (02:44 +0000)]
Bring in some improved bootp support for netboot submitted by Luigi:
- TFTP is not needed any longer.
Add some other improvements:
- swapsize is now detected if no value was specified.
- resvport is used by default
- merged code for setting/showing mount options for root/swap.
Submitted by: Luigi Rizzo <luigi@labinfo.iet.unipi.it>
Peter Wemm [Tue, 13 May 1997 10:58:14 +0000 (10:58 +0000)]
struct timespec needs to be useable with just <time.h>, but needs to not
conflict with the kernel visible one. Make the new stuff not visible if
_ANSI_SOURCE or _POSIX_SOURCE is defined (!) because the new things are
not part of the base 1003.1-1990 standard or ansi (says Bruce).
Doug Rabson [Tue, 13 May 1997 10:23:47 +0000 (10:23 +0000)]
Changes to support the kernel linker:
Add a -Bforcedynamic option which generates a dynamic object even
if no shared libraries were given in the link.
Make RRS in text section warnings conditional on "-assert pure-text"
so that I can link non-PIC kernel modules without tons of link
errors. Changes to bsd.lib.mk to follow.
Fix a couple of bugs exposed by the fact that the kernel is not
linked at zero.
Peter Wemm [Tue, 13 May 1997 09:33:24 +0000 (09:33 +0000)]
struct timespec needs to be useable with just <time.h>, but needs to not
conflict with the kernel visible one. Make the new stuff not visible if
_ANSI_SOURCE or _POSIX_SOURCE is defined (!) because the new things are
not part of the base 1003.1-1990 standard (or ansi).
Satoshi Asami [Tue, 13 May 1997 08:51:49 +0000 (08:51 +0000)]
Back out previous revision. Shlib version numbers are supposed to be
bumped only 0.1 or 1.0 between releases. (See handbook.)
Note that if you have built world in -current in the last 48 hours or
so, you should manually remove /usr/lib/libutil.so.2.3 before
rebuilding world to cleanse your system.
Tor Egge [Mon, 12 May 1997 23:00:32 +0000 (23:00 +0000)]
save/restore %ebx, %esi and %edi when calling bios routines.
Don't save/restore %ecx and %edx.
Reset segment limits when going to real mode.
Reviewed by: phk (briefly)
Tor Egge [Mon, 12 May 1997 19:02:56 +0000 (19:02 +0000)]
Use the old nfs arguments in the nfs_diskless structure, to be
compatible with boot proms made from the 2.2 source.
Convert the nfs arguments when copying to the new diskless structure.
Copy the gateway field in the diskless structure.
Andreas Klemm [Sun, 11 May 1997 19:01:24 +0000 (19:01 +0000)]
- fixed typo
- merged SMP option SMP_AUTOSTART from LINT
- enabled SMP_AUTOSTART, since it's the desired working mode and it's
reported to work now.
PR:
Reviewed by:
Submitted by:
Obtained from:
Tor Egge [Sun, 11 May 1997 18:05:39 +0000 (18:05 +0000)]
Bring in some kernel bootp support. This removes the need for netboot
to fill in the nfs_diskless structure, at the cost of some kernel
bloat. The advantage is that this code works on a wider range of
network adapters than netboot. Several new kernel options are
documented in LINT.
Obtained from: parts of the code comes from NetBSD.
Bruce Evans [Sun, 11 May 1997 12:39:43 +0000 (12:39 +0000)]
Fixed initialization of ldt[]. Unused entries were garbage. A comment
was stale.
Fixed initialization of gdt[] for the BDE_DEBUGGER case. APM entries
clobbered debugger entries if the debugger was loaded (APM is incompatible
with BDE_DEBUGGER) and unused entries were garbage if the debugger wasn't
loaded.
Murphy's Law: define LPR=true in /etc/make.conf if you don't want
tomorrow find 3 paper copies (ascii, latin1, postscript) of the
handbook in your printer (or someone else printer on the other half of
the world).
David Nugent [Sun, 11 May 1997 10:25:38 +0000 (10:25 +0000)]
Fix memory leak caused by not freeing memory returned by cgetstr()
calls. The cost is a little more up-front memory allocation, but the
effect seems minimal.
Problem noticed-by: bde
Added syslog at LOG_ERR when referencing an unknown gettytab entry
and for other cgetent() failues (circular reference et al).
Peter Wemm [Sun, 11 May 1997 05:33:08 +0000 (05:33 +0000)]
After looking around at what the other *BSD's do for their $xxxBSD$ tags,
take the easy way out and implement the beginnings of something similar.
Don't worry, the code here is *dormant* so far, some "help" from cvs is
required. This is going in now so that jdp can see what I'm working on.
This is an extension of the previous existing skeleton $FreeBSD$ code.
David Nugent [Sat, 10 May 1997 19:00:58 +0000 (19:00 +0000)]
login_getclass() -> login_getpwclass().
Updated (incorrect) entry for bash [-v].
Remove unused support for long rlim_t.
Added "bash2" to bash style ulimit handling.
David Nugent [Sat, 10 May 1997 18:55:38 +0000 (18:55 +0000)]
Summary of login.conf support changes:
o Incorporated BSDI code and enhancements, better logging for error
checking (which has been shown to be a problem, and is therefore
justified, imho); also some minor things we were missing, including
better quad_t math, which checks for under/overflows.
o setusercontext() now allows user resource limit overrides, but
does this AFTER dropping root privs, to restrict the user to
droping hard limits and set soft limits within the kernel's
allowed user limits.
o umask() only set once, and only if requested.
o add _secure_path(), and use in login.conf to guard against
symlinks etc. and non-root owned or non-user owned files being
used. Derived from BSDI contributed code.
o revamped authentication code to BSDI's latest api, which
includes deleting authenticate() and adding auth_check()
and a few other functions. This is still marked as depecated
in BSDI, but is included for completeness. No other source
in the tree uses this anyway, so it is now bracketed with
#ifdef LOGIN_CAP_AUTH which is by default not defined. Only
auth_checknologin() and auth_cat() are actually used in
module login_auth.c.
o AUTH_NONE definition removed (collided with other includes
in the tree). [bde]
o BSDI's login_getclass() now accepts a char *classname
parameter rather than struct passwd *pwd. We now do likewise,
but added login_getpwclass() for (sort of) backwards
compatiblity, namely because we handle root as a special
case for the default class. This will require quite a few
changes elsewhere in the source tree.
o We no longer pretend to support rlim_t as a long type.
o Revised code formatting to be more bsd-ish style.
Peter Wemm [Sat, 10 May 1997 17:14:53 +0000 (17:14 +0000)]
Clean up some more. Move parsing of sysctl iflist data into a single
place rather than updating the main loop's index variables from within
a subroutine and other revolting things like that. Move some more
globals into local variables.
Doug Rabson [Sat, 10 May 1997 16:59:36 +0000 (16:59 +0000)]
Implement a separate control for write gathering on NFSv3. This is turned
off for NFSv3 by default since write gathering seems to reduce performance
for NFSv3 by up to 60%.
Peter Wemm [Sat, 10 May 1997 14:47:35 +0000 (14:47 +0000)]
Grumble.. My last patchup here didn't quite work either. I hate this
program and it's use of global variables. Somehow, I managed to miss the
most obvious case.. "ifconfig ed0 10.0.0.1" failed (no "inet")