1531f6d32e
Any time we remove event from the queue make sure we 1. release the
...
event plist and 2. free the drvctl_event struct.
Discussed with jmcneill@.
2008-05-30 15:30:37 +00:00
13cf4bcc55
PR kern/38663 Kernel preemption can't be enabled on x86 because of amd64
...
FPU handling
Remove ifdef(i386), kernel preemption works on amd64 now.
2008-05-30 12:18:14 +00:00
1be38c90d8
do_sys_accept: release the reference to sock in few error paths.
...
Should fix PR/38790, report and test-case by Nicolas Joly.
2008-05-30 09:49:01 +00:00
a68758f8bd
sched_idle: initialise 'tci' to NULL, avoids compiler warning.
2008-05-30 08:31:42 +00:00
e98d2c1016
lwp_exit_switchaway: set l_lwpctl->lc_curcpu = EXITED, not NONE.
2008-05-29 23:29:59 +00:00
29170d3854
Simplifcation for running LWP migration. Removes double-locking in
...
mi_switch(), migration for LSONPROC is now performed via idle loop.
Handles/fixes on-CPU case in lwp_migrate(), misc.
Closes PR/38169, idea of migration via idle loop by Andrew Doran.
2008-05-29 22:33:27 +00:00
15e36ef766
sys_shmget: use the correct size variable for uobj_wirepages().
...
Adjust shm_memlock() for consistency too.
Fixes PR/38782, reported by Adam Hoka.
2008-05-29 21:38:18 +00:00
157262cae8
Explicitly compute the next interval using 64bit arithmetic, if the time
...
was either stepped backwards or the timer has overflown. This fixes
PR 26470.
2008-05-29 15:27:51 +00:00
de98844194
Mark pread/pwrite rump syscalls.
...
from Arnaud Ysmal
2008-05-29 12:01:37 +00:00
79d0501e52
Disable zero copy if MULTIPROCESSOR, until it is fixed:
...
- The TLB coherency overhead on MP systems is really expensive.
- It triggers a race in the VM system (grep kpause uvm/*).
2008-05-28 21:01:42 +00:00
801c395b47
Run shutdown hooks whether or not the kernel has panicked. Restores
...
NetBSD's shutdown behavior of more than 6 years before rev 1.176.
Ok joerg@.
It is essential that we restore some hardware to initial conditions
before rebooting, in order to avoid interfering with the BIOS
bootstrap. For example, if NetBSD gives control back to the Soekris
comBIOS while the kernel text is write-protected, the BIOS bootstrap
hangs during the power-on self-test, "POST: 0123456789bcefghip".
In principle, bus masters can also interfere with BIOS boot.
2008-05-28 15:40:58 +00:00
b5d8351e8e
PR kern/38355 lockf deadlock detection is broken after vmlocking
...
- Fix it; tested with Sun's libMicro.
- Use pool_cache.
- Use a global lock, so the deadlock detection code is safer.
2008-05-28 13:35:32 +00:00
5831c8ac63
Pull in sys/evcnt.h.
2008-05-27 22:05:50 +00:00
f79b59f700
#ifdef strikes again
2008-05-27 21:36:03 +00:00
4c634c7155
Sigh. The previous change did bad things to MySQL sysbench. Continue stealing
...
jobs from sched_nextlwp, but also do it in the idle loop. In sched_nextlwp
use trylock, in the idle LWP try harder.
2008-05-27 19:05:52 +00:00
208df81d99
Move lwp_exit_switchaway() into kern_synch.c. Instead of always switching
...
to the idle loop, pick a new LWP from the run queue.
2008-05-27 17:51:17 +00:00
aa7e99c693
Replace a couple of tsleep calls with cv_wait.
2008-05-27 17:50:03 +00:00
234470c22e
tsleep -> kpause
2008-05-27 17:49:07 +00:00
ec32985f61
Use pool_cache.
2008-05-27 17:48:27 +00:00
f90f3a01ea
Use kmem_alloc/free.
2008-05-27 17:42:14 +00:00
81fa379a0b
PR kern/38707 scheduler related deadlock during build.sh
...
- Fix performance regression inroduced by the workaround by making job
stealing a lot simpler: if the local run queue is empty, let the CPU enter
the idle loop. In the idle loop, try to steal a job from another CPU's run
queue if we are idle. If we succeed, re-enter mi_switch() immediatley to
dispatch the job.
- When stealing jobs, consider a remote CPU to have one less job in its
queue if it's currently in the idle loop. It will dispatch the job soon,
so there's no point sloshing it about.
- Introduce a few event counters to monitor what's happening with the run
queues.
- Revert the idle CPU bitmap change. It's pointless considering NUMA.
2008-05-27 14:48:52 +00:00
f985e88f2b
Start profiling clock on new process before setting it running, in case
...
there is a preemption.
2008-05-27 14:18:51 +00:00
b51f45d9e6
More fixes needed in the error paths for the chroot code to work.
2008-05-26 18:20:36 +00:00
06171502fc
Adjust and thus unify my license.
2008-05-26 17:45:51 +00:00
c9ac92b592
Use pool_cache for sockets.
2008-05-26 17:21:18 +00:00
46540aaf0e
brelse: always wakeup on b_busy, in case BC_WANTED is not set.
2008-05-26 14:56:55 +00:00
95488a6b7b
Broken assertion.
2008-05-26 12:58:24 +00:00
93e0e98369
Take the mutex pointer and waiters count out of sleepq_t: the values can
...
be or are maintained elsewhere. Now a sleepq_t is just a TAILQ_HEAD.
2008-05-26 12:08:38 +00:00
0d264cffef
PR/38745: Kouichirou Hiratsuka: chroot(8) can leak information of outside of
...
chrooted directory
2008-05-26 02:29:13 +00:00
0e41ecf58b
use PNBUF_* instead of malloc
2008-05-26 02:27:36 +00:00
c7615c48c8
PR kern/38707 scheduler related deadlock during build.sh
...
Fail sched_catchlwp() if mutex_tryenter() on the remote CPU's state fails.
Seems to work around the issue described in this PR.
XXX Stealing jobs from remote CPUs could probably be moved into the idle
loop, making the locking quite a bit simpler.
2008-05-25 23:46:55 +00:00
1cc9a3ae7e
If converting a process/thread from SCHED_OTHER to a realtime thread,
...
ignore the existing priority. If no priority is specified, give threads
the minimum user RT priority.
2008-05-25 23:34:24 +00:00
90035a10c9
sched_tick:
...
- Do timeslicing for SCHED_RR threads. At ~16Hz it's too slow but better
than nothing. XXX
- If a SCHED_OTHER thread has hogged the CPU for 1/8s without taking a
trip through mi_switch(), try to force a kernel preemption to give other
threads a chance.
2008-05-25 22:04:50 +00:00
6e0baf783e
don't forget to fill in the emulation.
2008-05-25 20:18:33 +00:00
5e4b324300
Properly fix the "hanging in tty" bug that was worked around with cv_wakeup()
...
some time again.
2008-05-25 19:22:21 +00:00
1e2888bbbd
Export device-driver and device-unit properties via drvctl
2008-05-25 15:03:01 +00:00
3a8a32076d
Add DRVGETEVENT support for /dev/drvctl, based on devmon support by
...
Jachym Holecek for Google Summer of Code. DRVGETEVENT plist is currently
limited to event type, device name, and device parent name.
2008-05-25 12:30:40 +00:00
934b677fde
Coverity CID 5015: Remove unnecessary test; if l was null we would have
...
crashed before when p = l->l_proc.
2008-05-24 18:43:02 +00:00
2847938186
Coverity CID 5019: Check before deref.
2008-05-24 16:49:30 +00:00
a2c63c0004
Coverity CID 5025: sbreserve is never called with a null socket.
2008-05-24 16:35:28 +00:00
25866fbff7
Set cpu_onproc on entry to the idle loop.
2008-05-24 12:59:06 +00:00
12da67c77e
Make msgsnd return EINVAL instead of 0, when the value of mtype is
...
less than 1.
2008-05-22 11:25:54 +00:00
697d5e2cd4
PR kern/38663 Kernel preemption can't be enabled on x86 because of amd64
...
FPU handling
Ugly hack until the amd64 fpu handling is working (which should be soon):
enable kernel preemption on i386.
2008-05-21 15:41:03 +00:00
d88444761b
Ignore return from module_load() and just try vfsop lookup again.
2008-05-20 19:30:03 +00:00
ce7cbbfb63
Back out unintentional change.
2008-05-20 19:21:23 +00:00
61270d54f1
If autoloading a module, don't consider the current working directory.
2008-05-20 19:20:38 +00:00
88435c0e48
Remove stale comment.
2008-05-20 19:16:07 +00:00
a72f5a57fb
Don't try to load a module while holding a vnode lock.
2008-05-20 17:28:59 +00:00
7bf8432671
If mount fails because the needed file system code isn't in kernel, try
...
to autoload with the needed vfsops.
2008-05-20 17:25:49 +00:00
67280de1f2
Allow module class to be passed to module_load(), as a basic sanity check
...
that we are loading the right kind of module.
2008-05-20 17:24:56 +00:00
freza