for nodes used in vm_radix.
On architectures supporting direct mapping, also avoid to pre-allocate
the KVA for such nodes.
In order to do so make the operations derived from vm_radix_insert()
to fail and handle all the deriving failure of those.
vm_radix-wise introduce a new function called vm_radix_replace(),
which can replace a leaf node, already present, with a new one,
and take into account the possibility, during vm_radix_insert()
allocation, that the operations on the radix trie can recurse.
This means that if operations in vm_radix_insert() recursed
vm_radix_insert() will start from scratch again.
Sponsored by: EMC / Isilon storage division
Reviewed by: alc (older version)
Reviewed by: jeff
Tested by: pho, scottl
Now the MTX_RECURSE flag can be passed to the mtx_*_flag() calls.
This helps in cases we want to narrow down to specific calls the
possibility to recurse for some locks.
Sponsored by: EMC / Isilon storage division
Reviewed by: jeff, alc
Tested by: pho
Unify the 2 concept into a real, minimal, sxlock where the shared
acquisition represent the soft busy and the exclusive acquisition
represent the hard busy.
The old VPO_WANTED mechanism becames the hard-path for this new lock
and it becomes per-page rather than per-object.
The vm_object lock becames an interlock for this functionality:
it can be held in both read or write mode.
However, if the vm_object lock is held in read mode while acquiring
or releasing the busy state, the thread owner cannot make any
assumption on the busy state unless it is also busying it.
Also:
- Add a new flag to directly shared busy pages while vm_page_alloc
and vm_page_grab are being executed. This will be very helpful
once these functions happen under a read object lock.
- Move the swapping sleep into its own per-object flag
The KPI is heavilly changed this is why the version is bumped.
It is very likely that some VM ports users will need to change
their own code.
Sponsored by: EMC / Isilon storage division
Discussed with: alc
Reviewed by: jeff, kib
Tested by: gavin, bapt (older version)
Tested by: pho, scottl
match devices where the driver class was fixed but the unit number was
wildcarded. This better matches the documented behaviour in
DEVICE_PROBE(9).
Reviewed by: imp
if NOTE_EXIT is not being monitored. The rationale is that a listener
should only get an event for exit() if they registered interest via
NOTE_EXIT. This matches the behavior on OS X.
- Don't save the exit status on process exit unless NOTE_EXIT is being
monitored.
- Add an internal EV_DROP flag that requests kqueue_scan() to free the
knote without signalling it to userland and use this when a process
exits but the fflags in the knote is zero.
Reviewed by: jmg
MFC after: 1 month
transparent layering and better fragmentation.
- Normalize functions that allocate memory to use kmem_*
- Those that allocate address space are named kva_*
- Those that operate on maps are named kmap_*
- Implement recursive allocation handling for kmem_arena in vmem.
Reviewed by: alc
Tested by: pho
Sponsored by: EMC / Isilon Storage Division
We cannot busy a page before doing pagefaults.
Infact, it can deadlock against vnode lock, as it tries to vget().
Other functions, right now, have an opposite lock ordering, like
vm_object_sync(), which acquires the vnode lock first and then
sleeps on the busy mechanism.
Before this patch is reinserted we need to break this ordering.
Sponsored by: EMC / Isilon storage division
Reported by: kib
- It does not let pages respect the LRU policy
- It bloats the active/inactive queues of few pages
Try to avoid it as much as possible with the long-term target to
completely remove it.
Use the soft-busy mechanism to protect page content accesses during
short-term operations (like uiomove_fromphys()).
After this change only vm_fault_quick_hold_pages() is still using the
hold mechanism for page content access.
There is an additional complexity there as the quick path cannot
immediately access the page object to busy the page and the slow path
cannot however busy more than one page a time (to avoid deadlocks).
Fixing such primitive can bring to complete removal of the page hold
mechanism.
Sponsored by: EMC / Isilon storage division
Discussed with: alc
Reviewed by: jeff
Tested by: pho
kern_sendfile() which is unnecessary.
The page is already wired so it will not be subjected to pagefault.
The content cannot be effectively protected as it is full of races
already.
Multiple accesses to the same indexes are serialized through vn_rdwr().
Sponsored by: EMC / Isilon storage division
Reviewed by: alc, jeff
Tested by: pho
other than the one specified by the BOOTP server. This configures NFS
using the BOOTP protocol while also respecting other root-path options such
as setting vfs.root.mountfrom in the environment or using the RB_DFLTROOT
boot option. It allows you to override the root path provided by the
server, or to supply a root path when the server provides IP configuration
but no root path info.
This maintains the historical BOOTP_NFSROOT behavior of panicking on a
failure to mount the root path provided by the server, unless you've
provided an alternative via the ROOTDEVNAME kernel option or by setting
vfs.root.mountfrom. The behavior of panicking when given no other options
is preserved because it amounts to a bit of a retry loop that could
eventually recover from a transient network or server problem.
The user can now override the root path from loader(8) even if the
kernel is compiled with BOOTP_NFSROOT. If vfs.root.mountfrom is set in
the environment it is used unconditionally -- it always overrides the
BOOTP info. If it begins with [old]nfs: then the BOOTP code uses it
instead of the server-provided info. If it specifies some other
filesystem then the bootp code will not panic like it used to and the code
in vfs_mountroot.c will invoke the right filesystem to do the mount.
If the kernel is compiled with the ROOTDEVNAME option, then that name is
used by the BOOTP code if either
* The server doesn't provide a pathname.
* The boothowto flags include RB_DFLTROOT.
The latter allows the user to compile in alternate path in ROOTDEVNAME
such as ufs:/dev/da0s1a and boot from that path by setting
boot_dftlroot=1 in loader(8) or using the '-r' option in boot(8).
The one thing not provided here is automatic failover from a
server-provided path to a compiled-in one without the user manually
requesting that. The code just isn't currently structured in a way that
makes that possible with a lot of rewrite. I think the ability to set
vfs.root.mountfrom and to use ROOTDEVNAME automatically when the server
doesn't provide a name covers the most common needs.
A set of patches submitted by Lars Eggert provided the part I couldn't
figure out by myself when I tried to do this last year; many thanks.
Reviewed by: rodrigc
must be destroyed, knlist_clear() and seldrain() calls could be
avoided, since vpollinfo was not used. More, the knlist_clear()
calling protocol requires the knlist locked, which is not true at the
call site.
Split the destruction into the helper destroy_vpollinfo_free(), and
call it when raced, instead of destroy_vpollinfo().
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
to those that are universally administered. While it is possible to
add locally administered MAC addresses, it's unclear whether those
are (expected) to be more unique than random multicast MAC addresses
or not.
With many U-Boot configurations assigning fixed and non-official MAC
addresses to ethernet ports and without setting the 'X' flag, this
change may have very little value in the embedded (development)
space. Uniqueness of the universally administered addresses is non-
existent on the (H/W) bench and questionable under the (S/W) desk.
In short: this change is aimed at production environments...
Also directly call swapper() at the end of mi_startup instead of
relying on swapper being the last thing in sysinits order.
Rationale:
- "RUN_SCHEDULER" was misleading, scheduling already takes place at that stage
- "scheduler" was misleading, the function swaps in the swapped out processes
- another SYSINIT(SI_SUB_RUN_SCHEDULER, SI_ORDER_ANY) could never be
invoked depending on its relative order with scheduler; this was not obvious
and the bug actually used to exist
Reviewed by: kib (ealier version)
MFC after: 14 days
addresses added to the UUID generator using uuid_ether_add(). The
UUID generator keeps an arbitrary number of MAC addresses, under
the assumption that they are rarely removed (= uuid_ether_del()).
This achieves the following:
1. It brings up closer to having the network stack as a loadable
module.
2. It allows the UUID generator to filter MAC addresses for best
results (= highest chance of uniqeness).
3. MAC addresses can come from anywhere, irrespactive of whether
it's used for an interface or not.
A side-effect of the change is that when no MAC addresses have been
added, a random multicast MAC address is created once and re-used if
needed. Previusly, when a random MAC address was needed, it was
created for every call. Thus, a change in behaviour is introduced
for when no MAC addresses exist.
Obtained from: Juniper Networks, Inc.
for consumption outside the vfs_aio.c.
For SIGEV_THREAD_ID and SIGEV_SIGNAL notification delivery methods,
also copy in the sigev_value, since librt event pumping loop compares
note generation number with the value passed through sigev_value.
Tested by: Petr Salinger <Petr.Salinger@seznam.cz>
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
- Add a new address space allocation method (VMFS_OPTIMAL_SPACE) for
vm_map_find() that will try to alter the alignment of a mapping to match
any existing superpage mappings of the object being mapped. If no
suitable address range is found with the necessary alignment,
vm_map_find() will fall back to using the simple first-fit strategy
(VMFS_ANY_SPACE).
- Change mmap() without MAP_FIXED, shmat(), and the GEM mapping ioctl to
use VMFS_OPTIMAL_SPACE instead of VMFS_ANY_SPACE.
Reviewed by: alc (earlier version)
MFC after: 2 weeks
Now that r253351 moved sendfile() stats to a separate struct, the
last field used in mbstat is m_mcfail, which is updated, but never
read or obtained from userland.
Submitted by: adrian, zec
Fix multiple kernel panics when VIMAGE is enabled in the kernel.
These fixes are based on patches submitted by Adrian Chadd and Marko Zec.
(1) Set curthread->td_vnet to vnet0 in device_probe_and_attach() just before calling
device_attach(). This fixes multiple VIMAGE related kernel panics
when trying to attach Bluetooth or USB Ethernet devices because
curthread->td_vnet is NULL.
(2) Set curthread->td_vnet in if_detach(). This fixes kernel panics when detaching networking
interfaces, especially USB Ethernet devices.
(3) Use VNET_DOMAIN_SET() in ng_btsocket.c
(4) In ng_unref_node() set curthread->td_vnet. This fixes kernel panics
when detaching Netgraph nodes.
error if any user wired mappings exist. Doing the invalidation
destroys the user wiring.
The change is the temporal measure to close the bug, the more proper
fix is to delegate the invalidation of the page to upper layers
always.
Reported and tested by: pho
Reviewed by: alc
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
about mount and unmount events. This is used by Juniper to implement a more
optimal implementation of NetBSD's veriexec.
This change differs from r253224 in the following way:
o The vfs_mounted handler is called before mountcheckdirs() and with
newdp locked. vp is unlocked.
o The event handlers are declared in <sys/eventhandler.h> and not in
<sys/mount.h>. The <sys/mount.h> header is used in user land code
that pretends to be kernel code and as such creates a very convoluted
environment. It's hard to untangle.
Submitted by: stevek@juniper.net
Discussed with: pjd@
Obtained from: Juniper Networks, Inc.
vfs_busy(mp);
vfs_write_suspend(mp);
which are problematic if other thread starts unmount between two
calls. The unmount starts a write, while vfs_write_suspend() drain
writers. On the other hand, unmount drains busy references, causing
the deadlock.
Add a flag argument to vfs_write_suspend and require the callers of it
to specify VS_SKIP_UNMOUNT flag, when the call is performed not in the
mount path, i.e. the covered vnode is not locked. The suspension is
not attempted if VS_SKIP_UNMOUNT is specified and unmount is in
progress.
Reported and tested by: Andreas Longwitz <longwitz@incore.de>
Sponsored by: The FreeBSD Foundation
MFC after: 3 weeks
The distance between ticks and td_swvoltick should be calculated as
an unsigned number. Previously we could end up comparing a negative
number with hogticks in which case should_yield() would give incorrect
answer.
We should probably ensure that td_swvoltick is properly initialized.
Sponsored by: HybridCluster
MFC after: 5 days
/dev/kmem and /dev/mem (in addition to traditional file permission checks).
PRIV_KMEM_READ is different from other PRIV_* checks in that it's allowed
by default.
Reviewed by: kib, mckusick
in the ithread code where we could lose ithread interrupts if
intr_event_schedule_thread() is called while the ithread is already
running. Effectively memory writes could be ordered incorrectly
such that the unatomic write of 0 to ithd->it_need (in ithread_loop)
could be delayed until after it was set to be triggered again by
intr_event_schedule_thread().
This was particularly a big problem for CAM because CAM optimizes
scheduling of ithreads such that it only signals camisr() when it
queues to an empty queue. This means that additional completion
events would not unstick CAM and quickly lead to a complete lockup
of the CAM stack.
To fix this use atomics in all places where ithd->it_need is accessed.
Submitted by: delphij, mav
Obtained from: TrueOS, iXsystems
MFC After: 1 week
If n fds were passed, it would receive the first one n times.
Reported by: Shawn Webb <lattera@gmail.com>, koobs, gleb
Tested by: koobs, gleb
Reviewed by: pjd
Issues were noted by Bruce Evans and are present on all architectures.
On i386, a counter fetch should use atomic read of 64bit value,
otherwise carry from the increment on other CPU could be lost for the
given fetch, making error of 2^32. If 64bit read (cmpxchg8b) is not
available on the machine, it cannot be SMP and it is enough to disable
preemption around read to avoid the split read.
On x86 the counter increment is not atomic on purpose, which makes it
possible for the store of the incremented result to override just
zeroed per-cpu slot. The effect would be a counter going off by
arbitrary value after zeroing. Perform the counter zeroing on the
same processor which does the increments, making the operations
mutually exclusive. On i386, same as for the fetching, if the
cmpxchg8b is not available, machine is not SMP and we disable
preemption for zeroing.
PowerPC64 is treated the same as amd64.
For other architectures, the changes made to allow the compilation to
succeed, without fixing the issues with zeroing or fetching. It
should be possible to handle them by using the 64bit loads and stores
atomic WRT preemption (assuming the architectures also converted from
using critical sections to proper asm). If architecture does not
provide the facility, using global (spin) mutex would be non-optimal
but working solution.
Noted by: bde
Sponsored by: The FreeBSD Foundation
instead of allocating new one each time
All limits are set to RLIM_INFINITY which sould be ok (even though we
care only about RLIMT_FSIZE in this case).
MFC after: 1 week
- Reconnect with some minor modifications, in particular now selsocket()
internals are adapted to use sbintime units after recent'ish calloutng
switch.
The filedesc lock may not be dropped unconditionally before exporting
fd to sbuf: fd might go away during execution. While it is ok for
DTYPE_VNODE and DTYPE_FIFO because the export is from a vrefed vnode
here, for other types it is unsafe.
Instead, drop the lock in export_fd_to_sb(), after preparing data in
memory and before writing to sbuf.
Spotted by: mjg
Suggested by: kib
Review by: kib
MFC after: 1 week
deadlkres was using a reversed test to check whether ticks had rolled over.
This meant that deadlkres could only fire after ticks had rolled over.
This test was actually unnecessary as deadlkres only ever took the
difference of ticks values which is safe even in the presence of ticks
rollover. Remove the tests entirely. Now deadlkres will properly fire
after a lock has been held after the timeout period.
MFC after: 1 month
originally inspired by the Solaris vmem detailed in the proceedings
of usenix 2001. The NetBSD version was heavily refactored for bugs
and simplicity.
- Use this resource allocator to allocate the buffer and transient maps.
Buffer cache defrags are reduced by 25% when used by filesystems with
mixed block sizes. Ultimately this may permit dynamic buffer cache
sizing on low KVA machines.
Discussed with: alc, kib, attilio
Tested by: pho
Sponsored by: EMC / Isilon Storage Division
bug where a PCI device would be powered down if it failed to probe, but
not when its driver was detached (e.g. via kldunload).
- Add a new helper method resource_list_release_active() which forcefully
releases any active resources of a specified type from a resource list.
- Add a bus_child_detached method for the PCI bus driver which forces any
active resources to be released (and whines to the console if it finds
any) and then powers the device down.
- Call pci_child_detached() if we fail to probe a device when a driver
is kldloaded. This isn't perfect but can avoid leaking resources
from a probe() routine in the kldload case.
Reviewed by: imp, brooks
MFC after: 1 month
- Call lock_init() first before setting any lock_object fields in
lock init routines. This way if the machine panics due to a duplicate
init the lock's original state is preserved.
- Somewhat similarly, don't decrement td_locks and td_slocks until after
an unlock operation has completed successfully.
provided by Isilon.
- Add an rm_assert() supporting various lock assertions similar to other
locking primitives. Because rmlocks track readers the assertions are
always fully accurate unlike rw_assert() and sx_assert().
- Flesh out the lock class methods for rmlocks to support sleeping via
condvars and rm_sleep() (but only while holding write locks), rmlock
details in 'show lock' in DDB, and the lc_owner method used by
dtrace.
- Add an internal destroyed cookie so that API functions can assert
that an rmlock is not destroyed.
- Make use of rm_assert() to add various assertions to the API (e.g.
to assert locks are held when an unlock routine is called).
- Give RM_SLEEPABLE locks their own lock class and always use the
rmlock's own lock_object with WITNESS.
- Use THREAD_NO_SLEEPING() / THREAD_SLEEPING_OK() to disallow sleeping
while holding a read lock on an rmlock.
Submitted by: andre
Obtained from: EMC/Isilon
mbuf that was fully consumed by the previous call, the mbuf ptr returned by the
current call ends up being the previous mbuf in the sb chain to the one that
contains the data we want.
This does not cause any observable issues because the mbuf copy routines happily
walk the mbuf chain to get to the data at the moff offset, which in this case
means they effectively skip over the mbuf returned by sbsndptr().
We can't adjust sb->sb_sndptr during the previous call for this case because the
next mbuf in the chain may not exist yet. We therefore need to detect the
condition and make the adjustment during the current call.
Fix by detecting the special case of moff being at the start of the next mbuf in
the chain and adjust the required accounting variables accordingly.
Reviewed by: andre
MFC after: 2 weeks
star for me. EVENTHANDLER_DEREGISTER() attempts to acquire the lock which is
held by the event handler framework while executing event handler functions,
leading to deadlock.
Move EVENTHANDLER_DEREGISTER() to alq_load_handler() and thus deregister the ALQ
shutdown_pre_sync handler at module unload time, which takes care of the
originally reported panic and fixes the deadlock introduced in r250951.
Reported by: Luiz Otavio O Souza
MFC after: 3 days
X-MFC with: 250951
In order to get some coverage of C11 atomics in kernelspace, switch at
least one piece of code in kernelspace to use C11 atomics instead of
<machine/atomic.h>.
While there, slightly improve the code by adding an assertion to prevent
the use count from going negative.
ensure all per-vnet related hhook initialisation is completed prior to any
virtualised hhook points attempting registration.
vnet_register_sysinit() requires that a stage later than SI_SUB_VNET be chosen.
There are no per-vnet initialisors in the source tree at this time which run
earlier than SI_SUB_INIT_IF. A quick audit of non-virtualised SYSINITs indicates
there are no subsystems pre SI_SUB_MBUF that would likely be interested in
registering a virtualised hhook point.
Settle on SI_SUB_MBUF as hhook's per-vnet initialisation stage as it's the first
overtly network-related initilisation stage to run after SI_SUB_VNET. If a
subsystem that initialises earlier than SI_SUB_MBUF ends up wanting to register
virtualised hhook points in future, hhook's use of SI_SUB_MBUF will need to be
revisited and would probably warrant creating a dedicated SI_SUB_HHOOK which
runs immediately after SI_SUB_VNET.
MFC after: 1 week
hook functions into hhook points which register after the modules were loaded -
potentially useful during boot or if hhook points are dynamically registered.
MFC after: 1 week
hhook_{add|remove}_hook_lookup() so that khelp (and other potential API
consumers) do not have to care when they attempt to (un)hook a particular hook
point identified by id and type.
Reviewed by: scottl
MFC after: 1 week
KASSERT during TCP hhook registration at boot. Virtualised hook points only
require extra housekeeping and sanity checking when "options VIMAGE" is present.
Reported by: bdrewery,jh,dhw
Tested by: dhw
MFC after: 1 week
X-MFC with: 251732
type and id, as compared to virtualised hook points which are now uniquely
identified by type, id and a vid (which for vimage is the pointer to the vnet
that the hhook resides in).
All hhook_head structs for both virtualised and non-virtualised hook points
coexist in hhook_head_list, and a separate list is maintained for hhook points
within each vnet to simplify some vimage-related housekeeping.
Reviewed by: scottl
MFC after: 1 week
- When looping, check for the pending suspension. Otherwise, other
usermode thread which races with the looping one, could try to
prevent the process from stopping or exiting.
- Add missed checks for the faults from casuword*(). The code is
structured in a way which makes the loops exit if the specified
address is invalid, since both fuword() and casuword() return -1 on
the fault. But if the address is mapped readonly, the typical value
read by fuword() is different from -1, while casuword() returns -1.
Absent the checks for casuword() faults, this is interpreted as the
race with other thread and causes non-interruptible spinning in the
kernel.
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
about mount and unmount events. This is used by Juniper to implement a more
optimal implementation of NetBSD's veriexec.
Submitted by: stevek@juniper.net
Obtained from: Juniper Networks, Inc
the mask of a cpuset. Also, change the cpuset's mask before updating the
masks of all children. Previously changing a cpuset's mask first required
setting the mask to a super-set of both the old and new masks and then
changing it a second time to the new mask.
- Split the bqlock into bqclean and bqdirty locks.
- Only acquire the wakeup synchronization locks when we cross a
threshold requiring them.
- Restructure the way flushbufqueues() targets work so they are more
smp friendly and sane.
Reviewed by: kib
Discussed with: mckusick, attilio
Sponsored by: EMC / Isilon Storage Division
M vfs_bio.c
of a lock within a single thread.
- Fix handling of interlocks in WITNESS by properly requiring the interlock
to be held exactly once if it is specified.
the vnode lock while iterating over the free vnode list. Instead of
yielding, pause for 1 tick. The change is reported to help in some
virtualized environments.
Submitted by: Roger Pau Monn? <roger.pau@citrix.com>
Discussed with: jilles
Tested by: pho
MFC after: 2 weeks
space, fork(2) would cause shadowing of the physical object and
copying of the shared page into private copy, effectively preventing
updates for the exported timehands structure and stopping the clock.
Specify the maximum allowed permissions for the page to be read and
execute, preventing write from the user mode.
Reported and tested by: <huanghwh@yahoo.com>
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
used as the estimation of size, to 32GB. This provides around 100K of
buffer headers and corresponding KVA for buffer map at the peak.
Sizing the cache larger is not useful, also resulting in the wasting
and exhausting of KVA for large machines.
Reported and tested by: bdrewery
Sponsored by: The FreeBSD Foundation
requires a pkthdr being present but that's not the case for either
_bus_dmamap_load_mbuf_sg() or bus_dmamap_load_mbuf_sg(9).
Reported by: sbruno
MFC after: 1 week
