- 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.
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.
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.
running in process context in order to run interrupt handlers. This
caused a big smashing of the stack on AMD K6, K5 and Intel Pentium (ie, P5)
processors because we are using npxproc as a flag to indicate whether
the state has been pushed onto the stack.
Submitted by: bde
very specific scenarios, and now that we have had net.inet.tcp.blackhole for
quite some time there is really no reason to use it any more.
(first of three commits)
of long and int64_t; and print the result as an unsigned long. This should
make the output from the bzero() test more readable, and avoid printing a
negative bandwidth. Note that this doesn't change the decision process,
since that is based on time elapsed, not on computed bandwidth.
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.
killing ipv6 and some other things.
This makes GENERIC and NEWCARD the same, with OLDCARD stuff commented
out and the NEWCARD stuff included. For the moment, pcic is commented
out (which has a old). Plus invariants. Plus ddb.
For UP, we were using $tmp_stk as a stack from the data section. If the
kernel text section grew beyond ~3MB, the data section would be pushed
beyond the temporary 4MB P==V mapping. This would cause the trampoline
up to high memory to fault. The hack workaround I did was to use all of
the page table pages that we already have while preparing the initial
P==V mapping, instead of just the first one.
For SMP, the AP bootstrap process suffered the same sort of problem and
got the same treatment.
MFC candidate - this breaks on 4.x just the same..
Thanks to: Richard Todd <rmtodd@ichotolot.servalan.com>
to call fork1() directly if we don't want out process queued right away.
This has the serendipitous side effect of saving us a call to pfind().
This makes threaded Linux apps (such as Opera) work again.
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
- 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.
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
bolted to a ne-2000 chip. This is necessary for the NetGear FA-410TX
and other cards.
This also requires you add mii to your kernel if you have an ed driver
configured.
This code will result in a couple of timeout messages for ed on the
impacted cards. Additional work will be needed, but this does work
right now, and many people need these cards.
Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
with egcs-1.1.1. bus_space_write_multi_2() had an extra operation that
should have been removed.
Remove it.
This fixes the panic when bus_space_write_multi_2() is used.
Obtained from: jake
a regular basis. Adjust our linux emulation to conform. This will
cause more dirty pages to be left for the pagedaemon to deal with,
but our new low-memory handling code can deal with it. The linux
way appears to be a trend, and we may very well make MAP_NOSYNC the
default for FreeBSD as well (once we have reasonable sequential
write-behind heuristics for random faults).
(will be MFC'd prior to 4.3 freeze)
Suggested by: Andrew Gallatin
rather than in silly places like "VFS Cluster debugging". People
should really be using COMPAT_LINUX instead of the linux module on
dynamic systems like -current.
gcc -aout -mno-underscores. The bioscall.s tweak is not an a.out
requirement really, but to work around the bugs in the antique version of
gas that used for a.out. Makefile hacks are all that is needed to
get an a.out kernel. There is no telling if it will work though.
This is little more than an academic curiosity anyway since all it is
good for is situations where the boot code is hard wired, eg: rom
bootstraps (such as the gnat box).
GENERIC:
...
size -aout kernel ; chmod 755 kernel
text data bss dec hex
3051520 368640 198688 3618848 373820
and used in C or vice versa. The elf compiler uses the same names
for both. Remove asnames.h with great prejudice; it has served its
purpose.
Note that this does not affect the ability to generate an aout kernel
due to gcc's -mno-underscores option.
moral support from: peter, jhb
to be more like Xint0x80_syscall and less like c function syscall().
- Reduce code duplication between the int0x80 and lcall handlers by
shuffling the elfags into the right place, saving the sizeof the
instruction in tf_err and jumping into the common int0x80 code.
Reviewed by: peter
depend on this. The linux ABI emulator tries to use it for some linux
binaries too. VM86 had a bigger cost than this and it was made default
a while ago.
Reviewed by: jhb, imp
the the original trapframe of the syscall, trap, or interrupt that entered
the kernel. Before SMPng, ast's were handled via a psuedo trap at the
end of doerti. With the SMPng commit, ast's were broken out into a
separate ast() function that was called from doreti to match the behavior
of other architectures. Unfortunately, when this was done, the
p_md.md_regs member of curproc was not updateda in ast(), thus when
signals are handled by userret() after an interrupt that returns to
userland, we end up using a stale trapframe that will result in the
registers from the old trapframe overwriting the real trapframe and
smashing all the registers right before we return to usermode. The saved
%cs:%eip from where we were in usermode are saved in the trapframe for
example.
- Don't use an atomic operation to update cnt.v_soft in ast(). This is
the only place the variable is written to, and sched_lock is always
held when it is written, so it is already protected and the mutex release
of sched_lock asserts a memory barrier that ensures the value will be
updated in a timely fashion.
- Don't hold sched_lock around addupc_task() as this apparently breaks
profiling badly due to sched_lock being held across copyin().
Reported by: bde (2)
scheduling an interrupt thread to run when needed. This has the side
effect of enabling support for entropy gathering from interrupts on
all architectures.
- Change the software interrupt and x86 and alpha hardware interrupt code
to use ithread_schedule() for most of their processing when scheduling
an interrupt to run.
- Remove the pesky Warning message about interrupt threads having entropy
enabled. I'm not sure why I put that in there in the first place.
- Add more error checking for parameters and change some cases that
returned EINVAL to panic on failure instead via KASSERT().
- Instead of doing a documented evil hack of setting the P_NOLOAD flag
on every interrupt thread whose pri was SWI_CLOCK, set the flag
explicity for clk_ithd's proc during start_softintr().
in mi_switch() just before calling cpu_switch() so that the first switch
after a resched request will satisfy the request.
- While I'm at it, move a few things into mi_switch() and out of
cpu_switch(), specifically set the p_oncpu and p_lastcpu members of
proc in mi_switch(), and handle the sched_lock state change across a
context switch in mi_switch().
- Since cpu_switch() no longer handles the sched_lock state change, we
have to setup an initial state for sched_lock in fork_exit() before we
release it.
always on curproc. This is needed to implement signal delivery properly
(see a future log message for kern_sig.c).
Debogotified the definition of aston(). aston() was defined in terms
of signotify() (perhaps because only the latter already operated on
a specified process), but aston() is the primitive.
Similar changes are needed in the ia64 versions of cpu.h and trap.c.
I didn't make them because the ia64 is missing the prerequisite changes
to make astpending and need_resched per-process and those changes are
too large to make without testing.
tsc_present in the right places (together with other variables of the
same linkage), and don't use messy ifdefs just to avoid exporting it in
some cases.
- All processes go into the same array of queues, with different
scheduling classes using different portions of the array. This
allows user processes to have their priorities propogated up into
interrupt thread range if need be.
- I chose 64 run queues as an arbitrary number that is greater than
32. We used to have 4 separate arrays of 32 queues each, so this
may not be optimal. The new run queue code was written with this
in mind; changing the number of run queues only requires changing
constants in runq.h and adjusting the priority levels.
- The new run queue code takes the run queue as a parameter. This
is intended to be used to create per-cpu run queues. Implement
wrappers for compatibility with the old interface which pass in
the global run queue structure.
- Group the priority level, user priority, native priority (before
propogation) and the scheduling class into a struct priority.
- Change any hard coded priority levels that I found to use
symbolic constants (TTIPRI and TTOPRI).
- Remove the curpriority global variable and use that of curproc.
This was used to detect when a process' priority had lowered and
it should yield. We now effectively yield on every interrupt.
- Activate propogate_priority(). It should now have the desired
effect without needing to also propogate the scheduling class.
- Temporarily comment out the call to vm_page_zero_idle() in the
idle loop. It interfered with propogate_priority() because
the idle process needed to do a non-blocking acquire of Giant
and then other processes would try to propogate their priority
onto it. The idle process should not do anything except idle.
vm_page_zero_idle() will return in the form of an idle priority
kernel thread which is woken up at apprioriate times by the vm
system.
- Update struct kinfo_proc to the new priority interface. Deliberately
change its size by adjusting the spare fields. It remained the same
size, but the layout has changed, so userland processes that use it
would parse the data incorrectly. The size constraint should really
be changed to an arbitrary version number. Also add a debug.sizeof
sysctl node for struct kinfo_proc.
Some things needed bits of <i386/include/lock.h> - cy.c now has its
own (only) copy of the COM_(UN)LOCK() macros, and IMASK_(UN)LOCK()
has been moved to <i386/include/apic.h> (AKA <machine/apic.h>).
Reviewed by: jhb
attributes. This is needed for AST's to be properly posted in a preemptive
kernel. They are backed by two new flags in p_sflag: PS_ASTPENDING and
PS_NEEDRESCHED. They are still accesssed by their old macros:
aston(), astoff(), etc. For completeness, an astpending() macro has been
added to check for a pending AST, and clear_resched() has been added to
clear need_resched().
- Rename syscall2() on the x86 back to syscall() to be consistent with
other architectures.
- Use swi_* function names.
- Use void * to hold cookies to handlers instead of struct intrhand *.
- In sio.c, use 'driver_name' instead of "sio" as the name of the driver
lock to minimize diffs with cy(4).
mtx_enter(lock, type) becomes:
mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks)
mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized)
similarily, for releasing a lock, we now have:
mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN.
We change the caller interface for the two different types of locks
because the semantics are entirely different for each case, and this
makes it explicitly clear and, at the same time, it rids us of the
extra `type' argument.
The enter->lock and exit->unlock change has been made with the idea
that we're "locking data" and not "entering locked code" in mind.
Further, remove all additional "flags" previously passed to the
lock acquire/release routines with the exception of two:
MTX_QUIET and MTX_NOSWITCH
The functionality of these flags is preserved and they can be passed
to the lock/unlock routines by calling the corresponding wrappers:
mtx_{lock, unlock}_flags(lock, flag(s)) and
mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN
locks, respectively.
Re-inline some lock acq/rel code; in the sleep lock case, we only
inline the _obtain_lock()s in order to ensure that the inlined code
fits into a cache line. In the spin lock case, we inline recursion and
actually only perform a function call if we need to spin. This change
has been made with the idea that we generally tend to avoid spin locks
and that also the spin locks that we do have and are heavily used
(i.e. sched_lock) do recurse, and therefore in an effort to reduce
function call overhead for some architectures (such as alpha), we
inline recursion for this case.
Create a new malloc type for the witness code and retire from using
the M_DEV type. The new type is called M_WITNESS and is only declared
if WITNESS is enabled.
Begin cleaning up some machdep/mutex.h code - specifically updated the
"optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN
and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently
need those.
Finally, caught up to the interface changes in all sys code.
Contributors: jake, jhb, jasone (in no particular order)
by myself. It solves a serious vm_map corruption problem that can occur
with the buffer cache when block sizes > 64K are used. This code has been
heavily tested in -stable but only tested somewhat on -current. An MFC
will occur in a few days. My additions include the vm_map_simplify_entry()
and minor buffer cache boundry case fix.
Make the buffer cache use a system map for buffer cache KVM rather then a
normal map.
Ensure that VM objects are not allocated for system maps. There were cases
where a buffer map could wind up with a backing VM object -- normally
harmless, but this could also result in the buffer cache blocking in places
where it assumes no blocking will occur, possibly resulting in corrupted
maps.
Fix a minor boundry case in the buffer cache size limit is reached that
could result in non-optimal code.
Add vm_map_simplify_entry() calls to prevent 'creeping proliferation'
of vm_map_entry's in the buffer cache's vm_map. Previously only a simple
linear optimization was made. (The buffer vm_map typically has only a
handful of vm_map_entry's. This stabilizes it at that level permanently).
PR: 20609
Submitted by: (Tor Egge) tegge
- If possible, context switch to the thread directly in sched_ithd(),
rather than triggering a delayed ast reschedule.
- Disable interrupts while restoring fpu state in the trap handler,
in order to ensure that we are not preempted in the middle, which
could cause migration to another cpu.
Reviewed by: peter
Tested by: peter (alpha)
to 1GB. A box of mine is running with MAXDSIZ and DFLDSIZ increased
up to 1.5GB.
Wishlist: It would be nice to warn if MAXTSIZ + MAXDSIZ + MAXSSIZ
exceeds VM_MAXUSER_ADDRESS - VM_MINUSER_ADDRESS.
incompletely converting simplelocks to mutexes (COM_LOCK() is supposed
to hide the SMP locking internals, but it now depends on mutex interfaces
being visible).
problem is that a mutex lock, prior to this change, is acquired before
the curproc is set to idleproc, so we mess ourselves up by calling
the mutex lock routine with curproc == NULL.
Moving it up after the aps_ready spin-wait has us hopefully setting it
after idleproc is setup.
Solved by: jake (the allmighty) :-)
instead of a trapframe directly. (Requested by bde.)
- Convert the alpha switch_trampoline to call fork_exit() and use the MI
fork_return() instead of child_return().
- Axe child_return().
that name as a variable. Use mtx_owned(&Giant) where appropriate
instead.
- Proc locking.
- P_FOO -> PS_FOO.
- Update comments about enable interrupts during trap and why this may be
bad if we trap while holding a spin mutex.
- Don't bother resetting p to curproc in syscall() in case we are the child
returning from fork. The child hasn't returned from fork through syscall
in a while.
- Remove fork_return() as it has been superseded by the MI version.
the alpha mp_machdep.c.
- Proc locking.
- Catch up to the P_FOO -> PS_FOO proc flags changes.
- Stick ap_init()'s prototype with the other prototypes.
- Remove the Xforwardirq IPI.
- Remove unused simplelocks.
- Don't try to psignal() from forward_statclock(), but set the appropriate
signal pending flag in p_sflag instead.
- Add in KTR_SMP tracepoints for various SMP functions. (Brought over
from the alpha port)
- Setup proc0.p_heldmtx, proc0.contested, and curproc earlier so that we
can use mutexes.
- Initialize sched_lock and Giant earlier and enter Giant during init386.
- Use suser(9) instead of checking cr_uid directly.
inline functions non-inlined. Hide parts of the mutex implementation that
should not be exposed.
Make sure that WITNESS code is not executed during boot until the mutexes
are fully initialized by SI_SUB_MUTEX (the original motivation for this
commit).
Submitted by: peter
interrupt threads to run with it always >= 1, so that malloc can
detect M_WAITOK from "interrupt" context. This is also necessary
in order to context switch from sched_ithd() directly.
Reviewed By: peter
initialization until after malloc() is safe to call, then iterate through
all mutexes and complete their initialization.
This change is necessary in order to avoid some circular bootstrapping
dependencies.
for SMP; just use the same ones as UP. These weren't used without
holding Giant anyway, and the routines that use them would have to
be protected from pre-emption to avoid migrating cpus.
pre-emptable kernel. For variables of size 4 bytes or less they compile
to a single instruction, which does not allow a process to migrate cpus
in the middle, and get the value for the "wrong" cpu.
appropriate function, rather than doing a horse-and-buggy
acquire. They now take the mutex type as an arg and can be
used with sleep as well as spin mutexes.
All calls to mtx_init() for mutexes that recurse must now include
the MTX_RECURSE bit in the flag argument variable. This change is in
preparation for an upcoming (further) mutex API cleanup.
The witness code will call panic() if a lock is found to recurse but
the MTX_RECURSE bit was not set during the lock's initialization.
The old MTX_RECURSE "state" bit (in mtx_lock) has been renamed to
MTX_RECURSED, which is more appropriate given its meaning.
The following locks have been made "recursive," thus far:
eventhandler, Giant, callout, sched_lock, possibly some others declared
in the architecture-specific code, all of the network card driver locks
in pci/, as well as some other locks in dev/ stuff that I've found to
be recursive.
Reviewed by: jhb
non-386 atomic_load_acq(). %eax is an input since its value is used in
the cmpxchg instruction, but we don't care what value it is, so setting
it to a specific value is just wasteful. Thus, it is being used without
being initialized as the warning stated, but it is ok for it to be used
because its value isn't important. Thus, we are only sort of lying when
we say it is an output only operand.
- Add "cc" to the clobber list for atomic_load_acq() since the cmpxchgl
changes ZF.
slow enough as it is, without having to constantly check that it really
is an i386 still. It was possible to compile out the conditionals for
faster cpus by leaving out 'I386_CPU', but it was not possible to
unconditionally compile for the i386. You got the runtime checking whether
you wanted it or not. This makes I386_CPU mutually exclusive with the
other cpu types, and tidies things up a little in the process.
Reviewed by: alfred, markm, phk, benno, jlemon, jhb, jake, grog, msmith,
jasone, dcs, des (and a bunch more people who encouraged it)
compiling errors where gcc would run out of registers.
- Add "cc" to the list of clobbers for micro-ops where we perform
instructions that alter %eflags.
- Use xchgl instead of cmpxchgl to release a spin lock. This could allow
for more efficient register allocation as we no longer mandate that %eax
be used.
- Reenable the optimized mutex micro-ops in the non-i386 case.
that modules can call.
- Remove the old gcc <= 2.8 versions of the atomic ops.
- Resort the order of some things in the file so that there is only
one #ifdef for KLD_MODULE, and so that all WANT_FUNCTIONS stuff is
moved to the bottom of the file.
- Remove ATOMIC_ACQ_REL() and just use explicit macros instead.
time I tinkered around here. Since INTREN is called from the interrupt
critical path now, it should not be too expensive. In this case, we
look at the bits being changed to decide which 8 bit IO port to write to
rather than unconditionally writing to both. I could probably have gone
further and only done the write if the bits actually changed, but that
seemed overkill for the usual case in interrupt threads.
[an outb is rather expensive when it has to cross the ISA bus]
