Commit Graph

160 Commits

Author SHA1 Message Date
Doug Rabson
2322ee63c2 Don't unwrap the function descriptor used as the callout argument to
fork_exit(). The MI version of fork_exit() needs a real function
descriptor, not a simple function pointer.
2001-04-19 12:35:47 +00:00
Doug Rabson
eed829bca9 Don't take the Giant mutex for clock interrupts. 2001-04-19 12:34:23 +00:00
Doug Rabson
ff37c2003c Don't panic when we try to modify the kernel pmap. 2001-04-18 15:08:37 +00:00
Doug Rabson
169915f56d Print an approximation of the function arguments in the stack trace. 2001-04-18 15:07:56 +00:00
Doug Rabson
072c3a5395 Implement a simple stack trace for DDB. This will have to be redone
if/when we change to a more modern toolchain.
2001-04-18 14:15:45 +00:00
Doug Rabson
dd85faa611 Record the right value for tf_ndirty for kernel interruptions so that
we can examine the interrupted register stack frame in DDB.
2001-04-18 14:10:43 +00:00
David E. O'Brien
c5e70d92ce Turn on kernel debugging support (DDB, INVARIANTS, INVARIANT_SUPPORT, WITNESS)
by default while SMPng is still being developed.

Submitted by:	jhb
2001-04-15 19:37:28 +00:00
John Baldwin
2fea957dc5 Rename the IPI API from smp_ipi_* to ipi_* since the smp_ prefix is just
"redundant noise" and to match the IPI constant namespace (IPI_*).

Requested by:	bde
2001-04-11 17:06:02 +00:00
David E. O'Brien
563dd74008 Reduce the emasculation of bounds_check_with_label() by one line, so we
propagate a bio error condition to the caller and above.
2001-03-29 20:26:12 +00:00
John Baldwin
1005a129e5 Convert the allproc and proctree locks from lockmgr locks to sx locks. 2001-03-28 11:52:56 +00:00
John Baldwin
192846463a Rework the witness code to work with sx locks as well as mutexes.
- Introduce lock classes and lock objects.  Each lock class specifies a
  name and set of flags (or properties) shared by all locks of a given
  type.  Currently there are three lock classes: spin mutexes, sleep
  mutexes, and sx locks.  A lock object specifies properties of an
  additional lock along with a lock name and all of the extra stuff needed
  to make witness work with a given lock.  This abstract lock stuff is
  defined in sys/lock.h.  The lockmgr constants, types, and prototypes have
  been moved to sys/lockmgr.h.  For temporary backwards compatability,
  sys/lock.h includes sys/lockmgr.h.
- Replace proc->p_spinlocks with a per-CPU list, PCPU(spinlocks), of spin
  locks held.  By making this per-cpu, we do not have to jump through
  magic hoops to deal with sched_lock changing ownership during context
  switches.
- Replace proc->p_heldmtx, formerly a list of held sleep mutexes, with
  proc->p_sleeplocks, which is a list of held sleep locks including sleep
  mutexes and sx locks.
- Add helper macros for logging lock events via the KTR_LOCK KTR logging
  level so that the log messages are consistent.
- Add some new flags that can be passed to mtx_init():
  - MTX_NOWITNESS - specifies that this lock should be ignored by witness.
    This is used for the mutex that blocks a sx lock for example.
  - MTX_QUIET - this is not new, but you can pass this to mtx_init() now
    and no events will be logged for this lock, so that one doesn't have
    to change all the individual mtx_lock/unlock() operations.
- All lock objects maintain an initialized flag.  Use this flag to export
  a mtx_initialized() macro that can be safely called from drivers.  Also,
  we on longer walk the all_mtx list if MUTEX_DEBUG is defined as witness
  performs the corresponding checks using the initialized flag.
- The lock order reversal messages have been improved to output slightly
  more accurate file and line numbers.
2001-03-28 09:03:24 +00:00
John Baldwin
0006681fe6 Switch from save/disable/restore_intr() to critical_enter/exit(). 2001-03-28 03:06:10 +00:00
John Baldwin
b944b9033a Catch up to the mtx_saveintr -> mtx_savecrit change. 2001-03-28 02:46:21 +00:00
John Baldwin
6283b7d01b - Switch from using save/disable/restore_intr to using critical_enter/exit
and change the u_int mtx_saveintr member of struct mtx to a critical_t
  mtx_savecrit.
- On the alpha we no longer need a custom _get_spin_lock() macro to avoid
  an extra PAL call, so remove it.
- Partially fix using mutexes with WITNESS in modules.  Change all the
  _mtx_{un,}lock_{spin,}_flags() macros to accept explicit file and line
  parameters and rename them to use a prefix of two underscores.  Inside
  of kern_mutex.c, generate wrapper functions for
  _mtx_{un,}lock_{spin,}_flags() (only using a prefix of one underscore)
  that are called from modules.  The macros mtx_{un,}lock_{spin,}_flags()
  are mapped to the __mtx_* macros inside of the kernel to inline the
  usual case of mutex operations and map to the internal _mtx_* functions
  in the module case so that modules will use WITNESS and KTR logging if
  the kernel is compiled with support for it.
2001-03-28 02:40:47 +00:00
John Baldwin
034dc442ad - Add the new critical_t type used to save state inside of critical
sections.
- Add implementations of the critical_enter() and critical_exit() functions
  and remove restore_intr() and save_intr().
- Remove the somewhat bogus disable_intr() and enable_intr() functions on
  the alpha as the alpha actually uses a priority level and not simple bit
  flag on the CPU.
2001-03-28 02:31:54 +00:00
Poul-Henning Kamp
f83880518b Send the remains (such as I have located) of "block major numbers" to
the bit-bucket.
2001-03-26 12:41:29 +00:00
Hajimu UMEMOTO
8b625cb701 Unbreak build on alpha.
- Move in_port_t to sys/types.h.
  - Nuke in_addr_t from each endian.h.

Reported by:	jhb
2001-03-24 15:17:27 +00:00
John Baldwin
5e115a7e04 - Define and use MAXCPU like the alpha and i386 instead of NCPUS.
- Sort the sys/mutex.h include in mp_machdep.c into a closer to correct
  location.
2001-03-24 06:22:57 +00:00
John Baldwin
197855a3ea Stick a prototype for handleclock() in machine/clock.h and include it
interrupt.c to quiet a warning.
2001-03-24 06:20:48 +00:00
Thomas Moestl
368d2edce4 Export intrnames and intrcnt as sysctls (hw.nintr, hw.intrnames and
hw.intrcnt).

Approved by:	rwatson
2001-03-23 03:45:17 +00:00
Peter Wemm
6eb39ac8fc Use a generic implementation of the Fowler/Noll/Vo hash (FNV hash).
Make the name cache hash as well as the nfsnode hash use it.

As a special tweak, create an unsigned version of register_t.  This allows
us to use a special tweak for the 64 bit versions that significantly
speeds up the i386 version (ie: int64 XOR int64 is slower than int64
XOR int32).

The code layout is a little strange for the string function, but I was
able to get between 5 to 10% improvement over the original version I
started with. The layout affects gcc code generation choices and this way
was fastest on x86 and alpha.

Note that 'CPUTYPE=p3' etc makes a fair difference to this.  It is
around 45% faster with -march=pentiumpro on a p6 cpu.
2001-03-17 09:31:06 +00:00
Doug Rabson
7ea98b2be3 Allow the config file to specify a root filesystem filename. 2001-03-09 13:45:31 +00:00
Doug Rabson
60fe99457c Adjust a comment slightly. 2001-03-09 13:44:53 +00:00
John Baldwin
5db078a9be Fix mtx_legal2block. The only time that it is bad to block on a mutex is
if we hold a spin mutex, since we can trivially get into deadlocks if we
start switching out of processes that hold spinlocks.  Checking to see if
interrupts were disabled was a sort of cheap way of doing this since most
of the time interrupts were only disabled when holding a spin lock.  At
least on the i386.  To fix this properly, use a per-process counter
p_spinlocks that counts the number of spin locks currently held, and
instead of checking to see if interrupts are disabled in the witness code,
check to see if we hold any spin locks.  Since child processes always
start up with the sched lock magically held in fork_exit(), we initialize
p_spinlocks to 1 for child processes.  Note that proc0 doesn't go through
fork_exit(), so it starts with no spin locks held.

Consulting from:	cp
2001-03-09 07:24:17 +00:00
John Baldwin
136d8f42b9 Unrevert the pmap_map() changes. They weren't broken on x86.
Sense beaten into me by:	peter
2001-03-07 05:29:21 +00:00
John Baldwin
f227364a17 - Release Giant a bit earlier on syscall exit.
- Don't try to grab Giant before postsig() in userret() as it is no longer
  needed.
- Don't grab Giant before psignal() in ast() but get the proc lock instead.
2001-03-07 03:53:39 +00:00
John Baldwin
19eb87d22a Grab the process lock while calling psignal and before calling psignal. 2001-03-07 03:37:06 +00:00
John Baldwin
ff655691d8 Use the proc lock to protect p_pptr when waking up our parent in cpu_exit()
and remove the mpfixme() message that is now fixed.
2001-03-07 03:20:15 +00:00
John Baldwin
4a01ebd482 Back out the pmap_map() change for now, it isn't completely stable on the
i386.
2001-03-07 01:04:17 +00:00
John Baldwin
f90f3aadba Don't psignal() a process from forward_hardclock() but set the appropriate
pending flag in p_sflag instead.
2001-03-06 07:40:51 +00:00
John Baldwin
968950e5d1 - Rework pmap_map() to take advantage of direct-mapped segments on
supported architectures such as the alpha.  This allows us to save
  on kernel virtual address space, TLB entries, and (on the ia64) VHPT
  entries.  pmap_map() now modifies the passed in virtual address on
  architectures that do not support direct-mapped segments to point to
  the next available virtual address.  It also returns the actual
  address that the request was mapped to.
- On the IA64 don't use a special zone of PV entries needed for early
  calls to pmap_kenter() during pmap_init().  This gets us in trouble
  because we end up trying to use the zone allocator before it is
  initialized.  Instead, with the pmap_map() change, the number of needed
  PV entries is small enough that we can get by with a static pool that is
  used until pmap_init() is complete.

Submitted by:		dfr
Debugging help:		peter
Tested by:		me
2001-03-06 06:06:42 +00:00
Doug Rabson
3af0f1028b Fix a couple of typos which became obvious when I started to actually use
this on real hardware.
2001-03-04 23:30:31 +00:00
John Baldwin
505b544d52 sched_swi -> swi_sched 2001-02-24 19:09:37 +00:00
John Baldwin
f58a8b1a54 Don't include machine/mutex.h and relocate sys/mutex.h's include to be
closer to alphabetical order and identical to that of the alpha.
2001-02-24 19:09:16 +00:00
John Baldwin
7b4a30b689 Clockframes have a trapframe stored in a cf_tf member, not ct_tf. 2001-02-24 18:57:34 +00:00
John Baldwin
7182d6578d Whitespace nits. 2001-02-24 18:41:38 +00:00
John Baldwin
ad0a541a9e Pass in process to mark ast on to aston(). 2001-02-24 18:41:17 +00:00
John Baldwin
2f01e2eaf2 Axe pcb_schednest as it is no longer used. 2001-02-22 17:09:50 +00:00
John Baldwin
938f15c7c4 Rename switch_trampoline() to fork_trampoline() on the alpha and ia64.
Suggested by:	dfr
2001-02-22 16:56:53 +00:00
John Baldwin
6a7b2863ba Don't set the sched_lock lesting level for new processes as it is no
longer used.
2001-02-22 16:53:23 +00:00
John Baldwin
6efc0dc140 Catch comments up to child_return() -> fork_return() as well. 2001-02-22 16:49:36 +00:00
John Baldwin
2217ebb4df Synch up with the other architectures:
- Remove unneeded spl()'s around mi_switch() in userret().
- Don't hold sched_lock across addupc_task().
- Remove the MD function child_return() now that the MI function
  fork_return() is used instead.
- Use TRAPF_USERMODE() instead of dinking with the trapframe directly to
  check for ast's in kernel mode.
- Check astpending(curproc) and resched_wanted() in ast() and return if
  neither is true.
- Use astoff() rather than setting the non-existent per-cpu variable
  astpending to 0 to clear an ast.
2001-02-22 16:27:03 +00:00
John Baldwin
7def4c9ab5 Use the MI fork_return() fork trampoline callout function for child
processes instead of the MD child_return().
2001-02-22 16:05:48 +00:00
John Baldwin
29182967ef - Don't dink with sched_lock in cpu_switch() since mi_switch() does this
for us.
- Change the switch_trampoline() to call fork_exit() passing in the
  required arguments instead of calling the fork trampoline callout
  function directly.
Warning: this hasn't been tested.

Looked over by:	dfr
2001-02-22 16:05:09 +00:00
John Baldwin
5f78859e9a - Axe the now unused ASS_* assertions for interrupt status.
- Use ia64_get_psr() instead of save_intr() in mtx_legal2block().
2001-02-22 15:43:42 +00:00
John Baldwin
409b7113a1 Add a inline function to read the psr. 2001-02-22 15:39:58 +00:00
John Baldwin
7a473ad644 Add a mtx_intr_enable() macro. 2001-02-22 15:37:57 +00:00
John Baldwin
66b5727c4e Axe the astpending per-cpu variable. 2001-02-22 15:37:34 +00:00
John Baldwin
3e68dabf19 Add TRAPF_PC() and TRAPF_USERMODE() macros and redefine CLKF_PC() and
CLKF_USERMODE() in terms of them.
2001-02-22 15:35:04 +00:00
John Baldwin
de0e830e6d Catch up to new MI astpending and need_resched handling. 2001-02-22 13:29:22 +00:00