causing poor interactive performance while unnice processes were running.
The new scheme still allows nice to have an effect on priority but it is
not as dramatic as the effect of the interactivity score.
before calling it for bound thread. To avoid this problem, change
thread_schedule_upcall to not put new thread on run queue, let caller
do it, so we can tweak the new thread before setting it to run.
Reported by: pho
threads in the process have already masked the signal, so job control
is delayed. But later a thread unmasking the STOP signal should enable
job control, so in issignal(), scanning all threads in process to see
if we can direct suspend some of them, not just suspend current thread.
we can deadlock because of lock order reversals. This was not
caught because Witness ignores pool mutexes right now.
Diagnosis and help: truckman
Noticed by: pho
"maxproc limit exceeded by uid %i, please see tuning(7) and login.conf(5)."
Which will be triggered whenever a user hits his/her maxproc limit or
the systemwide maxproc limit is reached.
MFC after: 1 week
mutexes are supposed to only be used as leaf mutexes, and what appear
to be separate pool mutexes could be aliased together, it is bad idea
for a thread to attempt to hold two pool mutexes at the same time.
Slightly rearrange the code in kern_open() so that FILE_UNLOCK() is
called before calling VOP_GETVOBJECT(), which will grab the v_vnlock
mutex.
systems to fail more gracefully when a file descriptor exhaustion situation
occurs.
Original patch by: David G. Andersen <dga@lcs.mit.edu>
PR: 45353
MFC after: 1 week
because the run time exceeds the largest value a signed int can hold.
The real solution involves calculating how far we are over the limit.
To quickly solve this problem we loop removing 1/5th of the current value
until it falls below the limit. The common case requires no passes.
and run time.
- Scale the sleep and run time back via sched_interact_update() in more
places. This is to keep the statistic more accurate.
- Charge a parent one tick for forking a child.
- Add only the run time and not the sleep time to the parents kg when a
thread exits. This allows us to give a penalty for having an expensive
thread exit but does not give a bonus for having an interactive thread
exit.
- Change the SLP_RUN_THROTTLE to limit us to 4/5th and not 1/2.
- Change the SLP_RUN_MAX to two seconds. This keeps bursty interactive
applications like mozilla and openoffice in the interactive range even
through expensive tasks.
- Recalculate the slice after every sleep. This ensures that once a task
has been marked interactive it only has a slice of 1 at the risk of
giving tasks that sleep for a very brief period a longer time slice.
schedules an upcall. Signal delivering to a bound thread is same as
non-threaded process. This is intended to be used by libpthread to
implement PTHREAD_SCOPE_SYSTEM thread.
2. Simplify kse_release() a bit, remove sleep loop.
panics. Before revision 1.38, we used to just point panicstr at the
format string if panicstr was NULL, but since we now use a static
buffer for the formatted panic message, we have to be careful to
only write to it during the first panic.
Pointed out by: bde
which meant no process would run for longer than 20ms.
- Slightly redo the interactivity scorer. It follows the same algorithm but
in a slightly more correct way. Previously values above half were
incorrect.
- Lower the interactivity threshold to 20. It seems that in testing non-
interactive tasks are hardly ever near there and expensive interactive
tasks can sometimes surpass it. This area needs more testing.
- Remove an unnecessary KTR.
- Fix a case where an idle thread that had an elevated priority due to
priority prop. would be placed back on the idle queue.
- Delay setting NEEDRESCHED until userret() for threads that haad their
priority elevated while in kernel. This gives us the same context switch
optimization as SCHED_4BSD.
- Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior
more quickly.
- Inhert some of the run/slp time from the child in sched_exit_ksegrp().
- Redo some of the priority comparisons so they are more clear.
- Throttle the frequency of sched_pctcpu_update() so that rounding errors
do not make it invalid.
to the machine-independent parts of the VM. At the same time, this
introduces vm object locking for the non-i386 platforms.
Two details:
1. KSTACK_GUARD has been removed in favor of KSTACK_GUARD_PAGES. The
different machine-dependent implementations used various combinations
of KSTACK_GUARD and KSTACK_GUARD_PAGES. To disable guard page, set
KSTACK_GUARD_PAGES to 0.
2. Remove the (unnecessary) clearing of PG_ZERO in vm_thread_new. In
5.x, (but not 4.x,) PG_ZERO can only be set if VM_ALLOC_ZERO is passed
to vm_page_alloc() or vm_page_grab().
small but noticeable increase in performance for name lookup operations.
The code uses two zones, one for short names (less than 32 characters)
and one for long names (up to NAME_MAX). Since most file names are
fairly short, this saves a considerable amount of space that would
otherwise be wasted if we always allocated NAME_MAX bytes. The cutoff
value of 32 characters was picked arbitrarily and may benefit from some
tweaking; it could also be made into a tunable.
Submitted by: hmp
too small panics on PAE machines which have odd > 4GB sizes (4.5 gig
would render a 20MB of KVA for kmem_map instead of 200MB).
Submitted by: John Cagle <john.cagle@hp.com>, jeff
Reviewed by: jeff, peter, scottl, lots of USENIX folks
is currently executing when we try to remove it in exit1(). Without this,
it was possible for the callout to bogusly rearm itself and eventually
refire after the process had been free'd resulting in a panic.
PR: kern/51964
Reported by: Jilles Tjoelker <jilles@stack.nl>
Reviewed by: tegge, bde
curthread. Unlike td_flags, this field does not need any locking.
- Replace the td_inktr and td_inktrace variables with equivalent private
thread flags.
- Move TDF_OLDMASK over to the private flags field so it no longer requires
sched_lock.
second and equalizing the load between the two most imbalanced CPU. This
is intended to clear up long term load imbalances that would not be handled
by the 'pull' method in sched_choose().
- Pull out some bits of sched_choose() into a kseq_move() function that moves
an arbitrary thread from one kseq to another.
adding it to the nice tables. Therefore, in kseq_add_nice, we should
keep in mind that the load will be 1 if we are the only thread, and not
0.
- Assert that the sched lock is held in all the appropriate places.
- Increase the scope of the sched lock in sched_pctcpu_update().
- Hold the sched lock in sched_runnable(). It is not held by the caller.
"", temporarily map it to a call to extattr_list_vp() to provide
compatibility for older applications using the "" API to retrieve
EA lists.
Use VOP_LISTEXTATTR() to support extattr_list_vp() rather than
VOP_GETEXTATTR(..., "", ...).
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Asssociates Laboratories
specific attribute name. It will have the same semantics as the
older vop_getextattr() "retrieve the names" hack, returning
a buffer with ASCII nul-seperated names.
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
we were passing in a void* representing the PCB of the parent thread.
Now we pass a pointer to the parent thread itself.
The prime reason for this change is to allow cpu_set_upcall() to copy
(parts of) the trapframe instead of having it done in MI code in each
caller of cpu_set_upcall(). Copying the trapframe cannot always be
done with a simply bcopy() or may not always be optimal that way. On
ia64 specifically the trapframe contains information that is specific
to an entry into the kernel and can only be used by the corresponding
exit from the kernel. A trapframe copied verbatim from another frame
is in most cases useless without some additional normalization.
Note that this change removes the assignment to td->td_frame in some
implementations of cpu_set_upcall(). The assignment is redundant.
A previous call to cpu_thread_setup() already did the exact same
assignment. An added benefit of removing the redundant assignment is
that we can now change td_pcb without nasty side-effects.
This change officially marks the ability on ia64 for 1:1 threading.
Not tested on: amd64, powerpc
Compile & boot tested on: alpha, sparc64
Functionally tested on: i386, ia64
extattr_list_link() system calls, which return a least of extended
attributes defined for a vnode referenced by a file descriptor
or path name. Currently, we just invoke VOP_GETEXTATTR() since
it will convert a request for an empty name into a query for a
name list, which was the old (more hackish) API. At some point
in the near future, we'll push the distinction between get and
list down to the vnode operation layer, but this provides access
to the new API for applications in the short term.
Pointed out by: Dominic Giampaolo <dbg@apple.com>
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
file/directory/link, rather than using a less explicit hack on
the extattr retrieval API:
extattr_list_fd()
extattr_list_file()
extattr_list_link()
The existing API was counter-intuitive, and poorly documented.
The prototypes for these system calls are identical to
extattr_get_*(), but without a specific attribute name to
leave NULL.
Pointed out by: Dominic Giampaolo <dbg@apple.com>
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
Don't copyin() data we are about to overwrite.
Add a flag to tell userland that KSE is officially "DONE" with the
mailbox and has gone away.
Obtained from: davidxu@
we failed to put the bucket back into the general cache/container.
Also, fix a bad assumption. There was a KASSERT() that aimed to
guarantee that whenever the pcpu container's mc_starved was > 0,
that whatever the bucket we were freeing to was an empty bucket,
assuming it belonged to the pcpu container cache. However, there
is at least one case where this is not true anymore; consider:
1) All containers empty, next thread to try to alloc will touch
a pcpu container, notice it's empty, and increment the pcpu
container's mc_starved.
2) Some other thread frees an mbuf belonging to a bucket in
the general cache/container. Then it frees another mbuf
belonging to the same bucket (still in gen container).
3) Some third thread tries to allocate an mbuf from the pcpu
container and, since empty, grabs one mbuf now available
in the general cache and moves the non-empty bucket from
which it took 1 mbuf and to which the thread in (2) freed
to, and moves it to the pcpu container.
4) A final thread tries to free an mbuf belonging to the
NON-EMPTY bucket mentionned in (2) and (3) and, since
the pcpu container's mc_starved is > 0, but the bucket
is obviously non-empty, it trips on the KASSERT.
This meant that one could potentially get a panic in some
cases when out of mbufs and clusters. The problem could
be mitigated by commenting out some cv_signal() calls,
but I'm assuming that was pure coincidence and this is
the correct fix.
- Use a hash of umtx queues to queue blocked threads. We hash on pid and the
virtual address of the umtx structure. This eliminates cases where we
previously held a lock across a casuptr call.
Reviwed by: jhb (quickly)
the lameness of the kstack code. The EPC overhaul de-lame-ified the
kstack code by removing the need for contigmalloc(). We can now
allocate stacks using malloc(). We probably want to make the stacks
swappable as well so that we can make it MI. But that's another story.
why certain exceptions are made, note an inconsistency between
FreeBSD and some other implementations regarding IPC_M, and let
suser() generate our EPERM rather than forcing it ourselves.
Remove a carriage return that crept in in the last commit.
Reviewed by: gordon
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
the stack to be changed in a way incompatible with elf32_map_insert()
where we used data_buf without initializing it for when the partial
mapping resulting in a misaligned image (typical when the page size
implied by the image is not the same as the page size in use by the
kernel). Since data_buf is passed by reference to vm_map_find(), the
compiler cannot warn about it.
While here, move all local variables to the top of the function.
in the kernel, the sysctl_register() call would fail, as expected.
However, when unloading this module again, the kernel would then panic
in sysctl_unregister(). Print a message error instead.
Submitted by: Nicolai Petri <nicolai@catpipe.net>
Reviewed by: imp
Approved by: re@ (jhb)
buf_start() to avoid triggering a panic in softdep_disk_io_initiation()
if b_iocmd happened to be BIO_READ. The later initialisation of
b_iocmd in cluster_wbuild() could probably be moved to before the
buf_start() call, but this patch keeps the change as simple as
possible.
This is reported to fix occasional "softdep_disk_io_initiation: read"
panics, especially on NFS servers.
Reported by: Nick Hilliard <nick@netability.ie>
Tested by: Nick Hilliard <nick@netability.ie>
Approved by: re (rwatson)
because we could fail due to a small buffer and loop and rerun. If this
happens, then the vsnprintf() will have already taken the arguments off
the va_list. For i386 and others, this doesn't matter because the
va_list type is a passed as a copy. But on powerpc and amd64, this is
fatal because the va_list is a reference to an external structure that
keeps the vararg state due to the more complicated argument passing system.
On amd64, arguments can be passed as follows:
First 6 int/pointer type arguments go in registers, the rest go on
the memory stack.
Float and double are similar, except using SSE registers.
long double (80 bit precision) are similar except using the x87 stack.
Where the 'next argument' comes from depends on how many have been
processed so far and what type it is. For amd64, gcc keeps this state
somewhere that is referenced by the va_list.
I found a description that showed the va_copy was required here:
http://mirrors.ccs.neu.edu/cgi-bin/unixhelp/man-cgi?va_end+9
The single unix spec doesn't mention va_copy() at all.
Anyway, the problem was that the sysctl kern.geom.conf* nodes would panic
due to walking off the end of the va_arg lists in vsnprintf. A better fix
would be to have sbuf_vprintf() use a single pass and call kvprintf()
with a callback function that stored the results and grew the buffer
as needed.
Approved by: re (scottl)
size and the kernel's heap size, specifically, vm_kmem_size. This
function allows a maximum of 40% of the vm_kmem_size to be used for
vnodes and vm objects. This is a conservative bound based upon recent
problem reports. (In other words, a slight increase in this percentage
may be safe.)
Finally, machines with less than ~3GB of RAM should be unaffected
by this change, i.e., the maximum number of vnodes should remain
the same. If necessary, machines with 3GB or more of RAM can increase
the maximum number of vnodes by increasing vm_kmem_size.
Desired by: scottl
Tested by: jake
Approved by: re (rwatson,scottl)
keep the thread state variable consistent with its real state.
i.e. Don't say it's on the run queue when it isn't.
Also clarify the associated comment.
Turns a double panic back to a single panic :-/
Approved by: re@ (jhb)
prime objectives are:
o Implement a syscall path based on the epc inststruction (see
sys/ia64/ia64/syscall.s).
o Revisit the places were we need to save and restore registers
and define those contexts in terms of the register sets (see
sys/ia64/include/_regset.h).
Secundairy objectives:
o Remove the requirement to use contigmalloc for kernel stacks.
o Better handling of the high FP registers for SMP systems.
o Switch to the new cpu_switch() and cpu_throw() semantics.
o Add a good unwinder to reconstruct contexts for the rare
cases we need to (see sys/contrib/ia64/libuwx)
Many files are affected by this change. Functionally it boils
down to:
o The EPC syscall doesn't preserve registers it does not need
to preserve and places the arguments differently on the stack.
This affects libc and truss.
o The address of the kernel page directory (kptdir) had to
be unstaticized for use by the nested TLB fault handler.
The name has been changed to ia64_kptdir to avoid conflicts.
The renaming affects libkvm.
o The trapframe only contains the special registers and the
scratch registers. For syscalls using the EPC syscall path
no scratch registers are saved. This affects all places where
the trapframe is accessed. Most notably the unaligned access
handler, the signal delivery code and the debugger.
o Context switching only partly saves the special registers
and the preserved registers. This affects cpu_switch() and
triggered the move to the new semantics, which additionally
affects cpu_throw().
o The high FP registers are either in the PCB or on some
CPU. context switching for them is done lazily. This affects
trap().
o The mcontext has room for all registers, but not all of them
have to be defined in all cases. This mostly affects signal
delivery code now. The *context syscalls are as of yet still
unimplemented.
Many details went into the removal of the requirement to use
contigmalloc for kernel stacks. The details are mostly CPU
specific and limited to exception_save() and exception_restore().
The few places where we create, destroy or switch stacks were
mostly simplified by not having to construct physical addresses
and additionally saving the virtual addresses for later use.
Besides more efficient context saving and restoring, which of
course yields a noticable speedup, this also fixes the dreaded
SMP bootup problem as a side-effect. The details of which are
still not fully understood.
This change includes all the necessary backward compatibility
code to have it handle older userland binaries that use the
break instruction for syscalls. Support for break-based syscalls
has been pessimized in favor of a clean implementation. Due to
the overall better performance of the kernel, this will still
be notived as an improvement if it's noticed at all.
Approved by: re@ (jhb)
the vnode and restart the loop. Vflush() is vulnerable since it does not
hold a reference to the vnode and it holds no other locks while waiting
for the vnode lock. The vnode will no longer be on the list when the
loop is restarted.
Approved by: re (rwatson)
PT_DETACH ptrace(2) requests from functioning as advertised in the
manual page. As described in kern/35175, the PT_DETACH request will,
under certain circumstances, pass an unwanted signal on to the traced
process upan detaching from it. The PT_CONTINUE request will
sometimes fail if you make it pass a signal that has "properties" that
differ from the properties of the signal that origionally caused the
traced process to be stopped. Since PT_KILL is nothing than
PT_CONTINUE with SIGKILL, it is broken too. In the PT_KILL case, this
leads to an unkillable process.
PR: 44011
Submitted by: Mark Kettenis <kettenis@chello.nl>
Approved by: re(jhb)
netstat(1) not display it for now because its effects are not yet
completely implemented and we're about to cut 5.2-RELEASE.
This is temporary.
Approved by: re (scottl, rwatson)
- Move struct sigacts out of the u-area and malloc() it using the
M_SUBPROC malloc bucket.
- Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(),
sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared().
- Remove the p_sigignore, p_sigacts, and p_sigcatch macros.
- Add a mutex to struct sigacts that protects all the members of the struct.
- Add sigacts locking.
- Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now
that sigacts is locked.
- Several in-kernel functions such as psignal(), tdsignal(), trapsignal(),
and thread_stopped() are now MP safe.
Reviewed by: arch@
Approved by: re (rwatson)
don't add the current time to it, but leave it as clear so that when the
timer is disabled, the it_value is always clear.
Reviewed by: bde
Approved by: re (rwatson)
desired buffer is found at one of the roots more than 60% of the time.
Thus, checking both roots before performing either splay eliminates
unnecessary splays on the first tree splayed.
Approved by: re (jhb)
synchronization primitives from inside DDB is generally a bad idea,
and in this case it frequently results in panics due to DDB commands
being executed from the sio fast interrupt context on a serial
console. Replace the locking with a note that a lack of locking
means that DDB may get see inconsistent views of the mount and vnode
lists, which could also result in a panic. More frequently,
though, this avoids a panic than causes it.
Discussed with ages ago: bde
Approved by: re (scottl)
reason for the duplication was that m_freem() was meant to eventually
be optimized to hold the lock of the cache being freed to as long as
possible across frees but the difficulty of implementing said
optimization right now is too high, given that in some cases (see MAC
and non-cluster external buffers), we need to call into other subsytems,
something not permissible when the cache lock is held.
This change minimizes code duplication while keeping at least the
atomic mbuf+cluster free optimization.
Suggested by: luigi
constants in question refer to the number of label slots, not the
maximum number of policies that may be loaded. This should reduce
confusion regarding an element in the MAC sysctl MIB, as well as
make it more clear what the affect of changing the compile-time
constants is.
Approved by: re (jhb)
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
(1) Accept that we're now going to use mutexes, so don't attempt
to avoid treating them as mutexes. This cleans up locking
accessor function names some.
(2) Rename variables to _mtx, _cv, _count, simplifying the naming.
(3) Add a new form of the _busy() primitive that conditionally
makes the list busy: if there are entries on the list, bump
the busy count. If there are no entries, don't bump the busy
count. Return a boolean indicating whether or not the busy
count was bumped.
(4) Break mac_policy_list into two lists: one with the same name
holding dynamic policies, and a new list, mac_static_policy_list,
which holds policies loaded before mac_late and without the
unload flag set. The static list may be accessed without
holding the busy count, since it can't change at run-time.
(5) In general, prefer making the list busy conditionally, meaning
we pay only one mutex lock per entry point if all modules are
on the static list, rather than two (since we don't have to
lower the busy count when we're done with the framework). For
systems running just Biba or MLS, this will halve the mutex
accesses in the network stack, and may offer a substantial
performance benefits.
(6) Lay the groundwork for a dynamic-free kernel option which
eliminates all locking associated with dynamically loaded or
unloaded policies, for pre-configured systems requiring
maximum performance but less run-time flexibility.
These changes have been running for a few weeks on MAC development
branch systems.
Approved by: re (jhb)
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
does the copyin stuff and then calls the second part kern_sendit to do
the hard work. Don't bother holding Giant during the copyin phase.
The intent of this is to allow the Linux emulator to impliment send*
syscalls without using the stackgap.
argument to the functions shm{at,ctl}1 and shm_find_segment_by_shmid{x}.
The BSD semantics didn't allow the usage of shared segment after
being marked for removal through IPC_RMID.
The patch involves the following functions:
- shmat
- shmctl
- shm_find_segment_by_shmid
- shm_find_segment_by_shmidx
- linux_shmat
- linux_shmctl
Submitted by: Orlando Bassotto <orlando.bassotto@ieo-research.it>
Reviewed by: marcel
trustworthy for vnode-backed objects.
- Restore the old behavior of vm_object_page_remove() when the end
of the given range is zero. Add a comment to vm_object_page_remove()
regarding this behavior.
Reported by: iedowse
double free of a mbuf occurs and cause an immediate panic, rather
than allowing free list corruption to occur.
This code is trapped under INVARIANTS, so it should not cause any
change in default performance.
Reviewed by: a bunch of people on -net
MFC after: 1 week
fini routines instead of in fork() and wait(). This has the nice side
benefit that the proc lock of any process on the allproc list is always
valid and sched_lock doesn't have to be used to test against PRS_NEW
anymore.
uptime. Where necessary, convert it back to Unix time by adding boottime
to it. This fixes a potential problem in the accounting code, which would
compute the elapsed time incorrectly if the Unix time was stepped during
the lifetime of the process.
- Eliminate an odd, special-case feature:
if start == end == 0 then all pages are removed. Only one caller
used this feature and that caller can trivially pass the object's
size.
- Assert that the vm_object is locked on entry; don't bother testing
for a NULL vm_object.
- Style: Fix lines that are longer than 80 characters.
- Mark the ktrace() and utrace() syscalls as being MP safe.
- Validate the facs argument to ktrace() prior to doing any vnode
operations or acquiring any locks.
- Share lock the proctree lock over the entire section that calls
ktrsetchildren() and ktrops(). We already did this for process groups.
Doing it for the process case closes a small race where a process might
go away after we look it up. As a result of this, ktrstchildren() now
just asserts that the proctree lock is locked rather than acquiring the
lock itself.
- Add some missing comments to #else and #endif.
syscall return values should be cleared. The system calls
getcontext() and swapcontext() want to return 0 on success
but these contexts can be switched to at a later time so
the return values need to be cleared in the saved register
sets. Other callers of get_mcontext() would normally want
the context without clearing the return values.
Remove the i386-specific context saving from the KSE code.
get_mcontext() is not i386-specific any more.
Fix a bad pointer in the alpha get_mcontext() code. The
context was being bcopy()'d from &td->tf_frame, but tf_frame
is itself a pointer, so the thread was being copied instead.
Spotted by jake.
Glanced at by: jake
Reviewed by: bde (months ago)
Instead of applying the adjustment to processes with a start time of 1,
apply it to all processes with a start time of less than 3600.
None of this would be necessary if the start times were recorded in ticks
instead of seconds and microseconds.
don't include the kernel stacks of swapped-out threads in the page count,
but do include the alternate kernel stack. jhb provided some helpful
comments on this.
PR: 49102
- Add a parameter to vm_pageout_flush() that tells vm_pageout_flush()
whether its caller has locked the vm_object. (This is a temporary
measure to bootstrap vm_object locking.)
whose p_stats->p_start has the magic value 1, replace it with boottime.
Some users were apparently confused by the fact that ps(1) reported a
start time in early 1970 for system processes.
do all the various sigstack dances, unlock the proc lock, and finally do
the copyout. This more closely resembles the behavior of
kern_sigaltstack() and closes a small race.
- Remove Giant from osigstack as it is no longer needed.
rename them appropriately. Protect both flags with both the proc lock
and the sched_lock.
- Protect p_profthreads with the proc lock.
- Remove Giant from profil(2).
- For the 4BSD scheduler, this means that all callers of the static
function resetpriority() now always hold sched_lock, so don't lock
sched_lock explicitly in that function.
kg_nice is now protected by both. Being protected by both means that
other places in the kernel that want to read kg_nice only need one of the
two locks.
their prototypes.
- Remove sched_lock locking from kse_purge() as all callers already lock
the sched_lock before calling it.
- Hold the proc lock slightly longer to protect P_SHOULDSTOP().
o KMF_NOUPCALL
Ask kse_release to not return to userland upcall entry, but instead
direct returns to userland by using current thread's stack and return
address on stack. This flags is intended to be used by UTS in critical
region to wait another UTS thread to leave critical region, by using
kse_release with this flag to avoid spinnng and burning CPU. Also this
flags can be used by UTS to poll completed context when there is nothing
to do in userland and needn't restart from its entry like normal upcall.
o KMF_NOCOMPLETED
Ask kernel to not bring completed thread contexts back to userland when
doing upcall, this flags is intend to be used with above flag when an
upcall thread is in critical region and can not process completed contexts
at that time.
Tested by: deischen
the poll bits when there's actually something in the queue.
Otherwise, select always returned '2' when there were no items to be
read, and '3' when there were. This would preclude being able to read
in a threaded (libc_r) program, as well as checking to see if there
were pending events or not.
lock assertion to it.
- SIGPENDING() no longer needs sched_lock, so only grab sched_lock to set
the TDF_NEEDSIGCHK and TDF_ASTPENDING flags in signotify().
- Add a proc lock assertion to tdsigwakeup().
- Since we always set TDF_OLDMASK while holding the proc lock, the proc
lock is sufficient protection to check its state in postsig() and we only
need sched_lock when clearing the actual flag.
a process group.
- Call pgadjustjobc() twice in fixjobc() to avoid code duplication and
improve readability.
- Use the proc lock to protect P_SHOULDSTOP() instead of sched_lock.
- Check to see if a process is PRS_NEW with sched_lock before trying to
lock its proc lock since the lock may not be constructed yet.
since they are going on the current cpu and not their previously assigned
cpu.
- sched_runnable() should only return true in the SMP case if the other
processor has more than one thread that is runnable. We can not steal
curthread.
- Change kseq_print() to accept the cpuid instead of a kseq pointer. This
makes use of this function in ddb much easier.
the process and session. Instead, cache a true reference to the session
when we do the hold and release our reference on that session. This avoids
the need for the proc lock when dropping the reference.
protects, so don't bother locking it while we assign it to a ucred's
cr_prison.
- Fully construct the new credential for a process before assigning it to
p_ucred.
- Set p_acflag earlier while already hold the proc lock in fork1().
- Mark the realitexpire() callout MPSAFE for new processes. It was already
marked safe for proc0 a long while ago.
generation number back to a long (sizeof(u_int) != sizeof(long) on
sparc64). The alternative would have been to heavily change the libdevstat API.
Discussed with: phk, ken
don't try and convert the argument flags to malloc flags, or we risk
implicitly requesting blocking and generating witness warnings.
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
When enabled, this causes m_defrag to randomly return NULL (following
its normal failure case so that extra memory leaks are not introduced.)
Code similar to this was used to find / fix a few bugs last week.
returning some additional room in the first mbuf in a chain, and
avoiding feature-specific contents in the mbuf header. To do this:
- Modify mbuf_to_label() to extract the tag, returning NULL if not
found.
- Introduce mac_init_mbuf_tag() which does most of the work
mac_init_mbuf() used to do, except on an m_tag rather than an
mbuf.
- Scale back mac_init_mbuf() to perform m_tag allocation and invoke
mac_init_mbuf_tag().
- Replace mac_destroy_mbuf() with mac_destroy_mbuf_tag(), since
m_tag's are now GC'd deep in the m_tag/mbuf code rather than
at a higher level when mbufs are directly free()'d.
- Add mac_copy_mbuf_tag() to support m_copy_pkthdr() and related
notions.
- Generally change all references to mbuf labels so that they use
mbuf_to_label() rather than &mbuf->m_pkthdr.label. This
required no changes in the MAC policies (yay!).
- Tweak mbuf release routines to not call mac_destroy_mbuf(),
tag destruction takes care of it for us now.
- Remove MAC magic from m_copy_pkthdr() and m_move_pkthdr() --
the existing m_tag support does all this for us. Note that
we can no longer just zero the m_tag list on the target mbuf,
rather, we have to delete the chain because m_tag's will
already be hung off freshly allocated mbuf's.
- Tweak m_tag copying routines so that if we're copying a MAC
m_tag, we don't do a binary copy, rather, we initialize the
new storage and do a deep copy of the label.
- Remove use of MAC_FLAG_INITIALIZED in a few bizarre places
having to do with mbuf header copies previously.
- When an mbuf is copied in ip_input(), we no longer need to
explicitly copy the label because it will get handled by the
m_tag code now.
- No longer any weird handling of MAC labels in if_loop.c during
header copies.
- Add MPC_LOADTIME_FLAG_LABELMBUFS flag to Biba, MLS, mac_test.
In mac_test, handle the label==NULL case, since it can be
dynamically loaded.
In order to improve performance with this change, introduce the notion
of "lazy MAC label allocation" -- only allocate m_tag storage for MAC
labels if we're running with a policy that uses MAC labels on mbufs.
Policies declare this intent by setting the MPC_LOADTIME_FLAG_LABELMBUFS
flag in their load-time flags field during declaration. Note: this
opens up the possibility of post-boot policy modules getting back NULL
slot entries even though they have policy invariants of non-NULL slot
entries, as the policy might have been loaded after the mbuf was
allocated, leaving the mbuf without label storage. Policies that cannot
handle this case must be declared as NOTLATE, or must be modified.
- mac_labelmbufs holds the current cumulative status as to whether
any policies require mbuf labeling or not. This is updated whenever
the active policy set changes by the function mac_policy_updateflags().
The function iterates the list and checks whether any have the
flag set. Write access to this variable is protected by the policy
list; read access is currently not protected for performance reasons.
This might change if it causes problems.
- Add MAC_POLICY_LIST_ASSERT_EXCLUSIVE() to permit the flags update
function to assert appropriate locks.
- This makes allocation in mac_init_mbuf() conditional on the flag.
Reviewed by: sam
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
mbuf_to_label(). This permits the vast majority of entry point code
to be unaware that labels are stored in m->m_pkthdr.label, such that
we can experiment storage of labels elsewhere (such as in m_tags).
Reviewed by: sam
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
the current queue if its priority is really elevated. This needs more work
as there are cases where a next queue kse could be holding up what would
be a curr queue kse, and thus hurting interactivity. Also, when a thread
with an elevated priority has its priority lowered it should be placed
back on the next queue.
the target process exiting which causes attempts to register the kevent
to randomly fail depending on whether the target runs to completion before
the parent can call kevent(2). The bug actually effects EVFILT_PROC
events on any zombie process, but the most common manifestation is with
parents trying to monitor child processes.
MFC after: 2 weeks
Sponsored by: NTT Multimedia Communications Labs
the second kseq's run queue so that it is referenced by the kse when
it is switched out.
- Spell ksq_rslices properly.
Reported by: Ian Freislich <ianf@za.uu.net>
- Allow user adjustable min and max time slices (suggested by hiten).
- Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a
process is interactive or not much more quickly.
- Place a process on the current run queue if it is interactive or if it is
running at an interrupt thread priority due to priority prop.
- Use the 'current' timeshare queue for interrupt threads, realtime threads,
and idle threads that are running at higher priority due to priority prop.
This fixes problems where priorities would have been elevated but we would
not check the timeshare run queue until other lower priority tasks were
no longer runnable.
- Keep an array of loads indexed by the priority class as well as a global
load.
- Keep an bucket of nice values with a count of the number of kses currently
runnable with that nice value.
- Keep track of the minimum nice value of any running thread.
- Remove the unused short term sleep accounting. I was attempting to use
this for load balancing but it didn't work out.
- Define a kseq_print() for use with debugging.
- Add KTR debugging at useful places so we can easily debug slice and
priority assignment.
- Decouple the runq assignment from the kseq assignment. kseq_add now keeps
track of statistics. This is done so that the nice and load is still
tracked for the currently running process. Previously if a niced process
was added while a non nice process was running the niced process would
still get a slice since it was not aware of the unnice process.
- Make adjustments for the sched api changes.
of ksegs since they primarily operation on processes.
- KSEs take ticks so pass the kse through sched_clock().
- Add a sched_class() routine that adjusts a ksegrp pri class.
- Define a sched_fork_{kse,thread,ksegrp} and sched_exit_{kse,thread,ksegrp}
that will be used to tell the scheduler about new instances of these
structures within the same process. These will be used by THR and KSE.
- Change sched_4bsd to reflect this API update.
by allprison_mtx), a unique prison/jail identifier field, two path
fields (pr_path for reporting and pr_root vnode instance) to store
the chroot() point of each jail.
o Add jail_attach(2) to allow a process to bind to an existing jail.
o Add change_root() to perform the chroot operation on a specified
vnode.
o Generalize change_dir() to accept a vnode, and move namei() calls
to callers of change_dir().
o Add a new sysctl (security.jail.list) which is a group of
struct xprison instances that represent a snapshot of active jails.
Reviewed by: rwatson, tjr
of asserting that an mbuf has a packet header. Use it instead of hand-
rolled versions wherever applicable.
Submitted by: Hiten Pandya <hiten@unixdaemons.com>
- Treat each class specially in kseq_{choose,add,rem}. Let the rest of the
code be less aware of scheduling classes.
- Skip the interactivity calculation for non TIMESHARE ksegrps.
- Move slice and runq selection into kseq_add(). Uninline it now that it's
big.
as it could be and can do with some more cleanup. Currently its under
options LAZY_SWITCH. What this does is avoid %cr3 reloads for short
context switches that do not involve another user process. ie: we can
take an interrupt, switch to a kthread and return to the user without
explicitly flushing the tlb. However, this isn't as exciting as it could
be, the interrupt overhead is still high and too much blocks on Giant
still. There are some debug sysctls, for stats and for an on/off switch.
The main problem with doing this has been "what if the process that you're
running on exits while we're borrowing its address space?" - in this case
we use an IPI to give it a kick when we're about to reclaim the pmap.
Its not compiled in unless you add the LAZY_SWITCH option. I want to fix a
few more things and get some more feedback before turning it on by default.
This is NOT a replacement for Bosko's lazy interrupt stuff. This was more
meant for the kthread case, while his was for interrupts. Mine helps a
little for interrupts, but his helps a lot more.
The stats are enabled with options SWTCH_OPTIM_STATS - this has been a
pseudo-option for years, I just added a bunch of stuff to it.
One non-trivial change was to select a new thread before calling
cpu_switch() in the first place. This allows us to catch the silly
case of doing a cpu_switch() to the current process. This happens
uncomfortably often. This simplifies a bit of the asm code in cpu_switch
(no longer have to call choosethread() in the middle). This has been
implemented on i386 and (thanks to jake) sparc64. The others will come
soon. This is actually seperate to the lazy switch stuff.
Glanced at by: jake, jhb
to select a KSE with a slice of 0 we will update its slice and insert it
onto the next queue.
- Pass the KSE instead of the ksegrp into sched_slice(). This more
accurately reflects the behavior of the code. Slices are granted to kses.
- Add a function kseq_nice_min() which finds the smallest nice value
assigned to the kseg of any KSE on the queue.
- Rewrite the logic in sched_slice(). Add a large comment describing the
new slice selection scheme. To summarize, slices are assigned based on
the nice value. Priorities are still calculated based on the nice and
interactivity of a process. Slice sizes of 0 may be granted for KSEs
whos nice is 20 or futher away from the lowest nice on the run queue.
Other nice values are scaled across the range [min, min+20]. This fixes
ULEs bad behavior with positively niced processes.
if (p->p_numthreads > 1) and not a flag because action is only necessary
if there are other threads. The rest of the system has no need to
identify thr threaded processes.
- In kern_thread.c use thr_exit1() instead of thread_exit() if P_THREADED
is not set.
- umtx_lock() is defined as an inline in umtx.h. It tries to do an
uncontested acquire of a lock which falls back to the _umtx_lock()
system-call if that fails.
- umtx_unlock() is also an inline which falls back to _umtx_unlock() if the
uncontested unlock fails.
- Locks are keyed off of the thr_id_t of the currently running thread which
is currently just the pointer to the 'struct thread' in kernel.
- _umtx_lock() uses the proc pointer to synchronize access to blocked thread
queues which are stored in the first blocked thread.
- sys/thr.h contains the user space visible api that is intended only for
use in threading library packages.
- kern/kern_thr.c contains thr system calls and other thr specific code.
kern_sigtimedwait() which is capable of supporting all of their semantics.
- These should be POSIX compliant but more careful review is needed before
we announce this.
a follow on commit to kern_sig.c
- signotify() now operates on a thread since unmasked pending signals are
stored in the thread.
- PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
- Change all consumers to pass in a thread.
Right now this does not cause any functional changes but it will be important
later when signals can be delivered to specific threads.
of sf_buf_alloc() instead of expecting sf_buf_alloc()'s caller to map it.
The ultimate reason for this change is to enable two optimizations:
(1) that there never be more than one sf_buf mapping a vm_page at a time
and (2) 64-bit architectures can transparently use their 1-1 virtual
to physical mapping (e.g., "K0SEG") avoiding the overhead of pmap_qenter()
and pmap_qremove().
time a character is written. Use this at boot time to reject the
existing buffer contents if they are corrupt. This fixes a problem
seen on some hardware (especially laptops) where the message buffer
gets partially corrupted during a short power cycle or reset, but
the msgbuf structure is left intact so it gets reused, resulting
in random junk and control characters appearing in dmesg and
/var/log/messages.
PR: kern/28497
that the feature can be enabled during the boot process. Note the
continued limitation that FreeBSD fails so rapidly with this setting
enabled that it's hard to narrow down particular failures for
correction; we really need per-malloc type failure rates.
in a debugging feature causing M_NOWAIT allocations to fail at
a specified rate. This can be useful for detecting poor
handling of M_NOWAIT: the most frequent problems I've bumped
into are unconditional deference of the pointer even though
it's NULL, and hangs as a result of a lost event where memory
for the event couldn't be allocated. Two sysctls are added:
debug.malloc.failure_rate
How often to generate a failure: if set to 0 (default), this
feature is disabled. Otherwise, the frequency of failures --
I've been using 10 (one in ten mallocs fails), but other
popular settings might be much lower or much higher.
debug.malloc.failure_count
Number of times a coerced malloc failure has occurred as a
result of this feature. Useful for tracking what might have
happened and whether failures are being generated.
Useful possible additions: tying failure rate to malloc type,
printfs indicating the thread that experienced the coerced
failure.
Reviewed by: jeffr, jhb
additional flags argument to indicate blocking disposition, and
pass in M_NOWAIT from the IP reassembly code to indicate that
blocking is not OK when labeling a new IP fragment reassembly
queue. This should eliminate some of the WITNESS warnings that
have started popping up since fine-grained IP stack locking
started going in; if memory allocation fails, the creation of
the fragment queue will be aborted.
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
where physical addresses larger than virtual addresses, such as i386s
with PAE.
- Use this to represent physical addresses in the MI vm system and in the
i386 pmap code. This also changes the paddr parameter to d_mmap_t.
- Fix printf formats to handle physical addresses >4G in the i386 memory
detection code, and due to kvtop returning vm_paddr_t instead of u_long.
Note that this is a name change only; vm_paddr_t is still the same as
vm_offset_t on all currently supported platforms.
Sponsored by: DARPA, Network Associates Laboratories
Discussed with: re, phk (cdevsw change)
flexible process_fork, process_exec, and process_exit eventhandlers. This
reduces code duplication and also means that I don't have to go duplicate
the eventhandler locking three more times for each of at_fork, at_exec, and
at_exit.
Reviewed by: phk, jake, almost complete silence on arch@
is set to 0, it now has the same affect as setting witness_dead used to
have.
- Added a sysctl handler that allows root to change witness_watch from a
non-zero value to zero to disable witness at runtime. Note that you
can't turn witness back on once it is off. You can only turn it off as
a one-way switch.
- Added a comment describing the possible values of witness_watch.
kse_mailbox to schedule an upcall, this is useful for userland timeout
routine, for example pthread_cond_timedwait().
Also extract upcall scheduling code from kse_reassign and create
a new function called thread_switchout to include these code.
Reviewed by: julain
the devstat is for an "interior" GEOM node and register using the
name argument as a geom identity pointer. Do not put these devstat
structures on the list returned by the sysctl.
This gives us the ability to tell the two kinds of nodes apart and
leave the current "strictly physical" view of devstat intact without
modifications, yet be able to use devstat for both kinds of devices.
It also saves us bloating struct devstat with another 48 bytes of
space for the name. At least for now.
Reviewed by: ken
Add a mutex and protect the allocation and traversal of the list with it.
When we allocate a page for devstat use we drop the mutex and use
M_WAITOK this is not nice, but under the given circumstances the
best we can do.
In the sysctl handler for returning the devstat entries we do not want to
hold the mutex across copyout(9) calls, so we keep a very careful eye on
the devstat_generation count, and abandon with EBUSY if it changes under
our feet.
Specifically test for BIO_WRITE, rather than default non-read,non-deletes
as write. Make the default be DEVSTAT_NO_DATA.
Add atomic increments of the sequence[01] fields so applications using the
mmap'ed view stand a chance of detecting updates in progress.
Reviewed by: ken
but I decided that it was important for this patch to not bit-rot, and
since it is mainly moving code around, the total amount of entropy is
epsilon /phk)
This is a patch to move the common parts of linux_getcwd() back into
kern/vfs_cache.c so that the standard FreeBSD libc getcwd() can use it's
extended functionality. The linux syscall linux_getcwd() in
compat/linux/linux_getcwd.c has been rewritten to use it too. It should
be possible to simplify libc's getcwd() after this. No doubt this code
needs some cleaning up, since I've left in the sysctl variables I used
for debugging.
PR: 48169
Submitted by: James Whitwell <abacau@yahoo.com.au>
Kernel:
Change statistics to use the *uptime() timescale (ie: relative to
boottime) rather than the UTC aligned timescale. This makes the
device statistics code oblivious to clock steps.
Change timestamps to bintime format, they are cheaper.
Remove the "busy_count", and replace it with two counter fields:
"start_count" and "end_count", which are updated in the down and
up paths respectively. This removes the locking constraint on
devstat.
Add a timestamp argument to devstat_start_transaction(), this will
normally be a timestamp set by the *_bio() function in bp->bio_t0.
Use this field to calculate duration of I/O operations.
Add two timestamp arguments to devstat_end_transaction(), one is
the current time, a NULL pointer means "take timestamp yourself",
the other is the timestamp of when this transaction started (see
above).
Change calculation of busy_time to operate on "the salami principle":
Only when we are idle, which we can determine by the start+end
counts being identical, do we update the "busy_from" field in the
down path. In the up path we accumulate the timeslice in busy_time
and update busy_from.
Change the byte_* and num_* fields into two arrays: bytes[] and
operations[].
Userland:
Change the misleading "busy_time" name to be called "snap_time" and
make the time long double since that is what most users need anyway,
fill it using clock_gettime(CLOCK_MONOTONIC) to put it on the same
timescale as the kernel fields.
Change devstat_compute_etime() to operate on struct bintime.
Remove the version 2 legacy interface: the change to bintime makes
compatibility far too expensive.
Fix a bug in systat's "vm" page where boot relative busy times would
be bogus.
Bump __FreeBSD_version to 500107
Review & Collaboration by: ken
KTRFAC_DROP to track instances when ktrace events are dropped due to the
request pool being exhausted. When a thread tries to post a ktrace event
and is unable to due to no available ktrace request objects, it sets
KTRFAC_DROP in its process' p_traceflag field. The next trace event to
successfully post from that process will set the KTR_DROP flag in the
header of the request going out and clear KTRFAC_DROP.
The KTR_DROP flag is the high bit in the type field of the ktr_header
structure. Older kdump binaries will simply complain about an unknown type
when seeing an entry with KTR_DROP set. Note that KTR_DROP being set on a
record in a ktrace file does not tell you anything except that at least one
event from this process was dropped prior to this event. The user has no
way of knowing what types of events were dropped nor how many were dropped.
Requested by: phk
struct proc as p_tracecred alongside the current cache of the vnode in
p_tracep. This credential is then used for all later ktrace operations on
this file rather than using the credential of the current thread at the
time of each ktrace event.
- Now that we have multiple ktrace-related items in struct proc that are
pointers, rename p_tracep to p_tracevp to make it less ambiguous.
Requested by: rwatson (1)
- Create a new function bdone() which sets B_DONE and calls wakup(bp). This
is suitable for use as b_iodone for buf consumers who are not going
through the buf cache.
- Create a new function bwait() which waits for the buf to be done at a set
priority and with a specific wmesg.
- Replace several cases where the above functionality was implemented
without locking with the new functions.
possible for some time.
- Lock the buf before accessing fields. This should very rarely be locked.
- Assert that B_DELWRI is set after we acquire the buf. This should always
be the case now.
requiring locked bufs in vfs_bio_awrite(). Previously the buf could
have been written out by fsync before we acquired the buf lock if it
weren't for giant. The cluster_wbuild() handles this race properly but
the single write at the end of vfs_bio_awrite() would not.
- Modify flushbufqueues() so there is only one copy of the loop. Pass a
parameter in that says whether or not we should sync bufs with deps.
- Call flushbufqueues() a second time and then break if we couldn't find
any bufs without deps.
than a MAXPHYS size block ahead. Having this set too high just leaves
other processes starved for IO and screws up interactive response. Let the
users with RAID set it higher when they need it.
- If SYSCTL_OUT() fails in sysctl_kern_proc_args(), return the error
instead of ignoring it if we have new arguments for the process.
- If the new arguments for a process are too long, return ENOMEM instead of
returning success but not doing the actual copy.
Submitted by: bde
hold hold it across the check to avoid extra lock operations in the
common case.
- Copy in the new args to a temporary pargs structure before we drop the
reference to the old one. Thus, if the copyin() fails, the process
arguments are unchanged rather than being deleted. Also, p_args is no
longer NULL during the sysctl operation.
it from its pgrp to avoid leaving zombies around with p_pgrp == NULL.
This bug was apparent as a NULL-dereference in the pid selection code
in fork1().
