through the use of a new build directive, MACHINE_CPU, which contains a
list of the CPU generations/features for which optimizations are desired.
This feature will be extended to cover the ports tree in the future.
Currently OpenSSL provides optimizations for i386, i586 and i686-class
CPUs. Currently it has not been tested on an i386 or i486.
Teach make(1) to provide sensible defaults for MACHINE_CPU if it is not
defined (namely, the lowest common denominator CPU we support for each
architecture). Currently this is i386 for the i386 architecture and ev4
for the alpha. sys.mk also sets the variable as a last resort for
consistency with MACHINE_ARCH and bootstrapping from very old versions of
make.
Benchmarks show a significant speed increase even in the i386 case, with
additional improvements for i586 and i686 systems. For maximum performance
define MACHINE_CPU=i686 i586 i386 in /etc/make.conf.
Based on a patch submitted by: Mike Silbersack <silby@silby.com>
Reviewed by: current
revision 1.25. When evaluating the termination condition for the
iteration over all map entries, we must take care to use the kernel
versions of all pointers. The code was comparing a kernel pointer
to a pointer within a local variable, so the loop never terminated.
16 characters, only check the first 16 since that is all the kernel
records. This isn't ideal, but it is probably the best we can do.
Otherwise, "killall communicator-linux-4.76.bin" will not match
the process as the kernel only records "communicator-lin".
add missing .Xo/.Xc to the tags. This only worked due
to the off-by-one bug in the -hang lists, which I will
hopefully backport from the mdocNG shortly.
was in. This shall be MFC'd in about three days (probably not a good idea
to MFC the stylistic changes though - see below).
PR: 19978
Submitted by: Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
Patch by: roam (slightly modified by me to use NULL not NIL)
modules (via pam_putenv). The following variables will never be set in
this fashion:
SHELL, HOME, LOGNAME, MAIL, CDPATH, IFS, PATH
any variable starting with `LD_'
. remove SYSV compatibility bits
. sort #include's
. ifdef unused code
. cleanup BDECFLAGS warnings
. fixed few typos in diagnostics messages
. style(9) fixes
as part of this cleanup I have changed all K&R declarations to ANSI style.
Main reason for doing that was great mix of declarations used here - clean
ASNI , clean K&R, ifdef'ed (both).
Reviewed by: ache (mostly)
Use tolower() rather than bit or-ing
Sanity check user specified printf() format
Exit when too many input files are given
Remove register from variables
Reviewed by: markm, imp
to be the same as -ragged in the current implementation) to
-ragged. With mdocNG, -filled displays produce the correct
output, formatted and justified to both margins.
string after each successful snprintf() call. This makes apply(1) work
*correctly*, although the whole snprintf() deal really should be redone.
Bug noted by: nectar (about 3 weeks ago)
is empty. There doesn't appear to be another easy way to do this.
mobile# mkdir foo
mobile# mkdir foo/bar
mobile# mkdir bar
mobile# find . -empty
./foo/bar
./bar
is ultimately silly because no locks are held in user space while traversing
the list via kvm_reads... really, this should use the sysctl interface
which *is* protected by a lock in the kernel.
declarations & their arguments; use only one tab after types; restore the
type of argv to sync with src tree style; sort new variables under main();
fix continuation indents; remove extra blank line before free()'s. Still
to do: fix snprintf() handling as nectar & bde suggested to me.
Submitted by: bde
it to make.h so both dir.c and util.c can use it, although bde didn't
particularly like this part of the idea, IMO it's cleaner than it was.
Submitted by: bde
BDECFLAGS; ANSIfy; use EXEC instead of "exec " where appropriate; use
proper types (size_t, int); use proper variable names for certain things;
get rid of static 'cache' style stuff by moving and sanitizing the
original SHELL checking code to main(), this also makes it easier to
free() the string; rename file-scope system() to exec_shell(); use
snprintf() everywhere instead of sprintf(); actually remember to free()
other malloc()'d char pointers in main().
I left out the -s option in this revision along with getusershell()
checking because of objections made by Warner Losh <imp> and Garrett
Wollman <wollman>. I agreed with their assertions that such code was
unnecessary in a program like this.
I dare people to make this coredump now.
Some suggestions: nectar (snprintf() truncate checking)
Reviewed by: markm, eivind, jedgar
Tortured by: examples in apply(1), fuzz(1),
and a lot of random ideas I came up with
especially on strings passed from argv; rename system() to exec_shell(),
and make it static; use strlcpy() and make sure it works; use proper
type (size_t) to be passed to malloc()/realloc().
Use getusershell() to make sure the SHELL environment variable passed is
safe to use. Add new option -s to allow anal users to pass things like
perl; this option is here along with getusershell() checking since the
such checking is only intended to affect things like suidperl that might
call apply(1).
Reviewed by: markm, jhb, C. Stephen Gunn <csg@waterspout.com>
all devices are by default known by their 'cooked' name, so
my change was wrong. I thought it was a hangover from old 'block
tape device' support which hasn't worked (if it ever did) since
v6/PWB.
So, the default tape name is now the same as Linux. Far out, man....
actually NOT '.' and '..'. Apparently this isn't the case when accessing
a directory via XFS over NFS on SGI systems. Since I don't have access to
an environment like that, this will sit out in -current for at least six
weeks. However, the patch proposed by the submitter seems acceptable, so
I've decided to commit it to the tree, in the hope that it will solve some
problems without bringing up others.
PR: 23300
Submitted by: Jim Pirzyk <Jim.Pirzyk@disney.com>
the idle time instead of the atime.
This makes entries for people that have logged in but done nothing
else show up correctly.
Reviewed by: markk@knigma.org
base system, but not in BruceBSD.
o Fix up style violations of various sorts.
o Remove redundant normalization of hertz variable, as the sysctl handler
does this work (unlike when kread was used).
Submitted by: bde
no longer contains kernel specific data structures, but rather
only scalar values and structures that are already part of the
kernel/user interface, specifically rusage and rtprio. It no
longer contains proc, session, pcred, ucred, procsig, vmspace,
pstats, mtx, sigiolst, klist, callout, pasleep, or mdproc. If
any of these changed in size, ps, w, fstat, gcore, systat, and
top would all stop working. The new structure has over 200 bytes
of unassigned space for future values to be added, yet is nearly
100 bytes smaller per entry than the structure that it replaced.
and numvnodes are longs in the kernel. They should remain longs in systat,
what really needs to change is that they should be using SYSCTL_LONG rather
than SYSCTL_INT. I also changed wantfreevnodes to SYSCTL_LONG because I
happened to notice it.
I wish there was a way to find all of these automatically..
Pointed out by: bde
files. Mostly -I${.CURDIR} was needed -- especially for YACC generated
files as the new cpp does not look in the ultimate source file
(ie, the .y file)'s directory as told by the "#line" directive. Some were
misspellings of "-I${.CURDIR}" as "-I.".
o In practice: the comment indicates that all but umask and
environmental variables of the users login class are applied when su
occurs, unless -m is used to specify a class. This was incorrect;
in practice, the uid, gids, resources, and priority were set, and
then resources and priority were selectively removed. This meant
that some aspects of the user context were not set, including handling
of login events (wtmp, utmp), as well as the path specified in
login.conf.
o I changed it so that the behavior is the same, but instead,
LOGIN_SETALL is used, and appropriate flags are removed, including
the LOGIN_SETLOGIN and LOGIN_SETPATH entries that were implicitly
not present before. I also updated the comment to reflect
reality, selecting reality as the "correct" behavior.
o This has the practical benefit that as new LOGIN_SET* flags are
introduced, they are supported by su unless specifically disabled.
For example, of a LOGIN_SETLABEL flag is introduced to support
MAC labels determined by the user's login class, then su no longer
has to be modified.
o It might be desirable to have su use LOGIN_SETPATH depending on
its command line parameters, as it might or might not be
considered part of the "environment".
Obtained from: TrustedBSD Project
maxvnodes, numvnodes, freevnodes, nchstats, and numdirtybuffers.
o Make the hw.ncpu error checking code a little more rigorous by
sanity checking the returned data size.
o Didn't fix machine-dependent non-sysctl-exported variables:
intrnames, eintrnames, intrcnt, eintrcnt, as these variables are
defined and exported from machine-dependent kernel code in
assembly. This should probably be fixed somehow.
Added $FreeBSD tag (in the way the already present sccsid is
done). I've been told the rcsid stuff may be of dubious value
so I'm curious to know if folks (still) use it.
Submitted by: Brad Chisholm <blc@bsdwins.com>
a name clash with the library functions of the same name (in libncurses).
This problem was masked when building tset shared (the local symbols had
precedence), but caused tset to core dump when it was built -static.
This software is obsolete, and its successor has interested parties
looking at it. Even if nobody was looking at kgzip, this utility
still wouldn't be useful in an environment where aout kernels
are a thing of the past.
PR: 20643
Submitted by: Peter Pentchev <roam@orbitel.bg>
statistics on a per network address basis.
Teach the IPv4 and IPv6 input/output routines to log packets/bytes
against the network address connected to the flow.
Teach netstat to display the per-address stats for IP protocols
when 'netstat -i' is evoked, instead of displaying the per-interface
stats.
