Aren/NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
168,991 Commits 606 Branches 0 Tags
Commit Graph

2560 Commits

Author SHA1 Message Date
nakayama
1ff012e906 Disable TRAPSTATS which was accidentally enabled when yamt-idlelwp branch 
was merged.

Ok by martin@.
 2008-04-16 15:31:15 +00:00
nakayama
a9ca1b36dd Remove sparc64_ipi_sync_tick. 
Since we can use counter-timer as timecounter instead of %tick on SMP kernel,
it is not necessary to sync all CPUs %tick registers.
 2008-04-14 17:54:07 +00:00
nakayama
bf95a282f1 Reduce use of curcpu() in tickintr_establish. 2008-04-14 17:43:02 +00:00
nakayama
37d841fbb8 Remove sparc64_ipi_sync_tick. 
Since we can use counter-timer as timecounter instead of %tick on SMP kernel,
it is not necessary to sync all CPUs %tick registers.
 2008-04-14 16:19:18 +00:00
nakayama
3c342df23b Add workaround for BB_ERRATA_1 on writing to TICK_CMPR register in 
next_tick().  If writing to TICK_CMPR fails, we lose hardclock interrupt
on secondary CPUs.

About BB_ERRATA_1 from comment in OpenSolaris:

/*
 * Writes to the TICK_COMPARE register sometimes fail on blackbird modules.
 * The failure occurs only when the following instruction decodes to wr or
 * wrpr.  The workaround is to immediately follow writes to TICK_COMPARE
 * with a read, thus stalling the pipe and keeping following instructions
 * from causing data corruption.  Aligning to a quadword will ensure these
 * two instructions are not split due to i$ misses.
 */
 2008-04-14 16:14:20 +00:00
ad
8d19e50577 Don't allocate or free pv_entry_t with pmap_lock held. Requested by martin@. 
Untested.
 2008-04-13 15:01:55 +00:00
nakayama
3e0d94e5d9 Remove kernel unlock in startlwp which is probably a leftover of 
vmlocking2 merge.

This may fix PR kern/38375.
 2008-04-10 14:28:28 +00:00
nakayama
c43d57923a Remove kernel locks around malloc introduced when I added SMP support. 2008-04-09 15:21:02 +00:00
nakayama
39b07dd47b Reorganize clock assignments on SMP kernel: 
- don't use separate statintr on primary CPU.
- invoke clockintr instead of statintr on secondary CPUs.

This change fixes the problem setitimer(2) didn't work on secodary CPUs.
 2008-04-09 14:58:23 +00:00
nakayama
3722b88cd3 Remove redundant #ifdef. KASSERT implies option DIAGNOSTIC. 2008-04-07 14:18:11 +00:00
cegger
09275ba76d use aprint_*_dev and device_xname 
OK martin
 2008-04-05 13:40:05 +00:00
nakayama
b1253edf64 Revise cpu_need_resched and cpu_signotify, then make them like x86's ones. 
This can avoid sending IPI to myself.
 2008-04-03 10:34:45 +00:00
tsutsui
8a19236fe5 One missed file in the previous commit: 
> Split softc and device_t for zsc(4) and its children.

Pointed out by mrg@.
 2008-03-30 01:41:36 +00:00
tsutsui
02cb47cab2 Split softc and device_t for zsc(4) and its children. 
XXX we should restructure MI APIs and make it really machine independent.
 2008-03-29 19:15:34 +00:00
tsutsui
4090172b20 Don't forget to initialize sc_dev. 2008-03-29 05:42:45 +00:00
tsutsui
50cafe93bd Use device_private(), missed in the previous. 2008-03-29 02:14:29 +00:00
nakayama
5a9b71307f Don't read fplwp content in delay slot for considering the case fplwp was 
modifiled.

Reported by Chris Ross on port-sparc64.
 2008-03-28 23:22:53 +00:00
tsutsui
03bb6cc822 Split device_t and softc for MI mk48txx(4) and intersil7170(4) clocks, 
and other related misc cosmetics.
 2008-03-28 20:26:12 +00:00
tsutsui
048fb884fc Split device_t and softc for MI mc146818 clock, 
and other related misc cosmetics.
 2008-03-28 19:05:49 +00:00
martin
65c31645cb Now that we have per-cpu-contexts, checking the context on the local cpu 
to find out if a pmap is on some mmu does not realy work any more.
 2008-03-28 18:22:59 +00:00
martin
e4eb0bc72d Use the passed cpu info for returning the frequency, not the current 
cpu
 2008-03-28 15:39:31 +00:00
martin
5405d9ca93 Restrict the "mach prom" command to the primary cpu - it hangs the machine 
hard on every other try on some machines.
 2008-03-27 15:21:35 +00:00
martin
afa2677a01 Fix a panic message and bump the number of retries untill we manage to 
send an IPI a bit up again.
 2008-03-27 15:20:47 +00:00
martin
1f9fec2e30 Get rid of delay() while waiting for IPI results; limit the waiting 
time via the cycle counter. Suggested by macallan.
 2008-03-27 14:51:02 +00:00
xtraeme
ae33dd3e78 sme_cookie is not needed when SME_DISABLE_REFRESH is set. 2008-03-26 20:21:38 +00:00
tnn
1a21682d68 Split device_t/softc for envctrl(4). 2008-03-26 19:15:49 +00:00
martin
261a6922f9 Make the debug message that triggered in PR 38297 slightly more verbose 2008-03-25 20:44:12 +00:00
martin
31c74d197f Count the number of syscalls per cpu, to avoid cache trashing around 
uvmexp.syscalls.
I wonder if we should apply the same treatment to uvmexp.{traps,intr,faults}
 2008-03-25 16:42:07 +00:00
martin
d05574d259 Remove unused symbol 2008-03-25 16:01:15 +00:00
ad
11d69830e6 +pseudo-device lockstat since there is a cycle counter. Ok martin@. 2008-03-24 13:48:18 +00:00
cube
c7dd91c86d Change previous to use aprint_error_dev. 2008-03-24 00:06:03 +00:00
macallan
3d2ef4afbb use device_xname() 2008-03-23 17:09:19 +00:00
nakayama
c853b61f50 Add more assertion. 2008-03-22 16:51:30 +00:00
nakayama
d0e8e57d7d Lock also on freeing own context, and add assertion. 2008-03-22 04:15:49 +00:00
nakayama
c7f58e76ff Insert a delay with enabling interrupts to avoid deadlocks 
when sending IPI is failed.
 2008-03-22 04:12:32 +00:00
ad
a9ca7a3734 Catch up with descriptor handling changes. See kern_descrip.c revision 
1.173 for details.
 2008-03-21 21:54:58 +00:00
nakayama
fb3ca52b49 - remove disabling interrupt in IPI. 
- unused VA bits in MMU demap is 13 bits, not 12 bits.
 2008-03-20 19:31:33 +00:00
nakayama
7249e6951e Use proper ASI macro. 2008-03-20 13:18:16 +00:00
nakayama
8aead4fcfb Fix a race when increment an interrupt level event counter. 2008-03-18 18:22:43 +00:00
nakayama
0a332c0d29 Fix a race condition in IPI handlers of save/drop fpstate, from OpenBSD. 
Reported and tested by martin@.
 2008-03-17 23:54:03 +00:00
nakayama
4eab56b382 Change statintr and schedintr similar to kern/kern_clock.c. 2008-03-17 18:31:19 +00:00
nakayama
6456819faa Make schedintr interrupt handler per-CPU. 
While there rename tickintr interrupt handler and share initialization
code with schedintr.
 2008-03-17 04:04:00 +00:00
nakayama
9db309f3e2 Use macros to specify a interrupt level. 2008-03-15 20:14:17 +00:00
nakayama
868048cf4d SMP support for sparc64 now works, so make GENERIC.MP kernel build 
by default.
 2008-03-14 16:43:16 +00:00
nakayama
9cfa53b41a #include "opt_multiprocessor.h" 2008-03-14 15:40:22 +00:00
nakayama
8d734e3bd0 Complete per-CPU TSBs and MMU contexts changes. 2008-03-14 15:40:02 +00:00
nakayama
5374009247 Improve FPU state save/clear like x86, idea from OpenBSD. 2008-03-14 15:39:18 +00:00
nakayama
74a205f37a On SMP kernel, 
- use "tick" interrupt for per-CPU statclock.
- disable "counter-timer" #1 interrupt and use it as timecounter.
 2008-03-14 15:38:36 +00:00
nakayama
ead4e6f7b9 Improve IPI handling: 
- make IPI takes two arguments.
- add IPI event counters per-CPU.
- implement IPI functions which were missing or broken.
- insert DELAY while halting primary CPU in IPI handler.
 2008-03-14 15:38:00 +00:00
cube
607ead0ef4 Split device_t and softc for all com(4) devices (well, everything that 
uses a com_softc backend).  Use proper types and ansify where appropriate.
 2008-03-14 15:09:09 +00:00