NetBSD/sys/kern
jdolecek 24ba90929c Avoid using microtime(9) for atime/mtime, we don't need to have it
THAT accurate and microtime(9) is painlessly slow on i386 currently.
This speeds up small transfers much. The gain for large transfers
is less significant, but notable too.
Bottleneck was found by Andreas Persson (Re: kern/14246).

Performance improvement with PIII on 661 Mhz according to hbench (with
PIPE_MINDIRECT=8192):

buffersize     before    after
512            17        49
1024           33        110
2048           52        143
4096           77        163
8192           142       190
64K            577       662
128K           372       392
2001-10-28 20:47:15 +00:00
..
cnmagic.c only print the magic number changing if not cold 2000-12-19 04:39:19 +00:00
exec_aout.c
exec_conf.c Make the setregs hook emulation-specific, rather than executable 2001-09-18 19:36:32 +00:00
exec_ecoff.c
exec_elf32.c simplify an expression. 2001-07-29 21:24:23 +00:00
exec_elf64.c
exec_elf_common.c
exec_macho.c make a copy of the epp->ep_path, because it is too late to use it when 2001-10-28 04:46:42 +00:00
exec_script.c In check_exec(), don't bother checking P_TRACED along with 2001-06-15 17:24:19 +00:00
exec_subr.c - add exec_read_from and make exec_elf32 use it. 2001-07-14 02:08:29 +00:00
genassym.awk
genassym.sh Deal with invokers who pass us environment variable settings as part 2001-09-24 00:20:11 +00:00
genlintstub.awk Add a /* LINTSTUB: include <foo.h> */ mechanism. 2001-05-16 04:17:04 +00:00
init_main.c user maps are always pageable. 2001-08-16 01:37:50 +00:00
init_sysent.c Regen; added Linux-compatible clone(2) system call. 2001-07-01 18:07:33 +00:00
kern_acct.c
kern_allocsys.c Defopt CALLWHEEL_STATS. 2001-09-13 05:22:16 +00:00
kern_clock.c Defopt CALLWHEEL_STATS. 2001-09-13 05:22:16 +00:00
kern_descrip.c Unshare the file descriptor table and `cwdinfo' when we exec. 2001-07-18 05:34:37 +00:00
kern_exec.c Make the setregs hook emulation-specific, rather than executable 2001-09-18 19:36:32 +00:00
kern_exit.c Add `WALLSIG' which causes wait4(2) to wait for all processes, regardless 2001-07-18 19:10:26 +00:00
kern_fork.c flags was used uninitialized. 2001-07-07 23:33:54 +00:00
kern_kthread.c
kern_ktrace.c Fix a partial construction problem that can cause race conditions 2001-06-14 20:32:41 +00:00
kern_lkm.c add a DEBUG for why the lkm entry point failed. 2001-06-05 12:36:08 +00:00
kern_lock.c replace wakeup_one() with wakeup(). wakeup_one() can only be used if the 2001-09-29 21:27:49 +00:00
kern_malloc_debug.c a whole bunch of changes to improve performance and robustness under load: 2001-09-15 20:36:31 +00:00
kern_malloc.c a whole bunch of changes to improve performance and robustness under load: 2001-09-15 20:36:31 +00:00
kern_ntptime.c Fix issues areound compilation errors (ntv is a pointer in adjtime1 and needs 2001-09-16 09:55:26 +00:00
kern_physio.c a whole bunch of changes to improve performance and robustness under load: 2001-09-15 20:36:31 +00:00
kern_proc.c When freeing a session, remove the reference to it from the associated tty. 2001-02-04 22:32:24 +00:00
kern_prot.c Remove an splimp() that really has no purpose in life. 2001-04-12 03:08:42 +00:00
kern_resource.c Move maxdmap and maxsmap where they belong and make them big enough. 2001-02-06 19:54:43 +00:00
kern_sig.c Unshare signal actions on exec. 2001-07-18 05:34:58 +00:00
kern_subr.c Make "reboot" an option at the root device/filesystem prompt (we 2001-05-27 01:29:50 +00:00
kern_synch.c in ltsleep(), assert that the interlock is held (if one is given). 2001-09-25 01:38:38 +00:00
kern_sysctl.c resize the namecache hash table also when desiredvnodes changes. 2001-09-24 06:01:13 +00:00
kern_time.c Split root-only time-related system calls so that we have an upper part, that 2001-09-16 06:50:06 +00:00
kern_xxx.c
kgdb_stub.c Spell 'occurred' with two 'r's. 2001-09-16 16:34:23 +00:00
Make.tags.inc use ${.CURDIR:H:H} instead of ../.. for default setting of SYSDIR, 2001-04-18 11:21:24 +00:00
Makefile SYSDIR should be relative here, too. 2001-04-05 19:49:40 +00:00
makesyscalls.sh Fix last commit: It's ">", not "<", in this case. Okay'd by christos. 2001-07-08 19:08:48 +00:00
subr_autoconf.c In-kernel device configuration manager - allows modification 2001-07-01 02:56:20 +00:00
subr_disk.c ANSIfy. 2001-07-09 10:54:12 +00:00
subr_extent.c extent_alloc_subregion shouldn't allocate a region pasts the end of subregion. 2001-08-27 13:35:44 +00:00
subr_log.c
subr_pool.c in pool_drain(), call pool_reclaim() while we still have interrupts blocked 2001-10-21 00:06:05 +00:00
subr_prf.c Check msgbufenabled before referencing msgbufp in panic. Avoids 2001-07-10 20:40:56 +00:00
subr_prof.c Before allowing kgmon to do anything, check that we managed to allocate the 2001-04-28 12:56:00 +00:00
subr_prop.c Remove the property sysctl interface for now. 2001-10-05 19:05:05 +00:00
subr_rmap.c
subr_userconf.c bcopy -> memcpy 2001-07-18 06:51:38 +00:00
subr_xxx.c
sys_generic.c If no select collision is needed, try our best to avoid it by clearing 2001-06-28 05:45:49 +00:00
sys_pipe.c Avoid using microtime(9) for atime/mtime, we don't need to have it 2001-10-28 20:47:15 +00:00
sys_process.c seperate -> separate 2001-07-22 13:33:58 +00:00
sys_socket.c Use SB_ASYNC in struct sockbuf sb_flags field instead of SS_ASYNC in 2001-06-16 21:29:32 +00:00
syscalls.c Regen; added Linux-compatible clone(2) system call. 2001-07-01 18:07:33 +00:00
syscalls.conf Round up the sysent table size. 2000-12-09 06:15:42 +00:00
syscalls.master Linux-compatible clone(2) system call, lifted from the Linux 2001-07-01 18:06:11 +00:00
sysv_ipc.c
sysv_msg.c
sysv_sem.c make local 'nsops' same unsigned type as the value we store there in sys_semop() 2001-08-03 05:58:18 +00:00
sysv_shm.c eliminate the KERN_* error codes in favor of the traditional E* codes. 2001-03-15 06:10:32 +00:00
tty_conf.c bzero -> memset 2001-07-18 06:48:27 +00:00
tty_pty.c Add l_poll' to struct linesw' and provide an xxxpoll() entry point 2001-05-02 10:32:08 +00:00
tty_subr.c
tty_tb.c Remove unnecessary test of tp->t_linesw against NULL; they are results 2001-03-31 00:35:21 +00:00
tty_tty.c
tty.c Add l_poll' to struct linesw' and provide an xxxpoll() entry point 2001-05-02 10:32:08 +00:00
uipc_domain.c add pfctlinput2() (pfctlinput() with args). 2001-02-11 06:38:45 +00:00
uipc_mbuf2.c bcopy -> memcpy 2001-07-18 06:51:38 +00:00
uipc_mbuf.c a whole bunch of changes to improve performance and robustness under load: 2001-09-15 20:36:31 +00:00
uipc_proto.c
uipc_socket2.c Give different names for different wait channels. 2001-08-05 08:25:39 +00:00
uipc_socket.c Use lmin() instead of min(), and long for mlen & clen, to avoid integer 2001-09-29 14:16:19 +00:00
uipc_syscalls.c Spell 'occurred' with two 'r's. 2001-09-16 16:34:23 +00:00
uipc_usrreq.c Deprecate the "m_act" alias of "m_nextpkt" (m_act is a historical 2001-10-18 20:17:24 +00:00
vfs_bio.c in bwrite(), when deciding whether to convert sync writes into delayed writes, 2001-04-01 16:16:56 +00:00
vfs_cache.c in cache_lookup(), if we get a cache hit but then fail to vget() the found 2001-10-27 04:53:38 +00:00
vfs_getcwd.c Jump out of the scandir loop if VOP_READDIR returns an error, avoiding 2000-12-15 11:52:14 +00:00
vfs_init.c Use a pool cache for namei buffers -- it's faster to allocate from 2001-10-17 23:33:29 +00:00
vfs_lockf.c
vfs_lookup.c Use a pool cache for namei buffers -- it's faster to allocate from 2001-10-17 23:33:29 +00:00
vfs_subr.c when attempting to reclaim a vnode, tell the lockmgr() that it's ok 2001-10-04 05:46:45 +00:00
vfs_syscalls.c Allow userland to pass MNT_IGNORE (from enami) 2001-10-11 16:27:24 +00:00
vfs_vnops.c use shared locks instead of exclusive for VOP_READ() and VOP_READDIR(). 2001-09-21 08:02:55 +00:00
vnode_if.c regen 2001-09-15 20:37:36 +00:00
vnode_if.sh delint: vop_default_desc was declared const but not defined const. 2001-05-07 08:46:02 +00:00
vnode_if.src a whole bunch of changes to improve performance and robustness under load: 2001-09-15 20:36:31 +00:00