host.host_ocr, examine the correct field when setting up the hardware. Also,
the offset for the capabilties register should be 0x140, not 0x240.
Submitted by: Ilya Bakulin <ilya@bakulin.de>
Pointy hat to: me
This is a fix for a regression introduced in r246293.
vm_page_clear_dirty expects the range to have DEV_BSIZE aligned boundaries,
otherwise it extends them. Thus it can happen that the whole page is
marked clean while actually having some small dirty region(s).
This commit makes the range properly aligned and ensures that only
the clean data is marked as such.
It would interesting to evaluate how much benefit clearing with DEV_BSIZE
granularity produces. Perhaps instead we should clear the whole page
when it is completely overwritten and don't bother clearing any bits
if only a portion a page is written.
Reported by: George Hartzell <hartzell@alerce.com>,
Richard Todd <rmtodd@servalan.servalan.com>
Tested by: George Hartzell <hartzell@alerce.com>,
Reviewed by: kib
MFC after: 5 days
On machines with seveal CPUs and enough RAM this can easily twice improve
ZFS performance or twice reduce CPU usage. It was disabled three years
ago due to memory and KVA exhaustion reports, but our VM subsystem got
improved a lot since that time, hopefully enough to make another try.
This is a last resort for very low memory condition in case other measures
to free memory were ineffective. Sequentially cycle through all CPUs and
extract per-CPU cache buckets into zone cache from where they can be freed.
Lock congestion is the same, whether it happens on alloc or free, so
handle it equally. Now that we have back pressure, there is no problem
to grow buckets a bit faster. Any way growth is much slower then in 9.x.
These new buckets make bucket size self-tuning more soft and precise.
Without them there are buckets for 1, 5, 13, 29, ... items. While at
bigger sizes difference about 2x is fine, at smallest ones it is 5x and
2.6x respectively. New buckets make that line look like 1, 3, 5, 9, 13,
29, reducing jumps between steps, making algorithm work softer, allocating
and freeing memory in better fitting chunks. Otherwise there is quite a
big gap between allocating 128K and 5x128K of RAM at once.
Every time system detects low memory condition decrease bucket sizes for
each zone by one item. As result, higher memory pressure will push to
smaller bucket sizes and so smaller per-CPU caches and so more efficient
memory use.
Before this change there was no force to oppose buckets growth as result
of practically inevitable zone lock conflicts, and after some run time
per-CPU caches could consume enough RAM to kill the system.
adapters. Both devices support Gigabit Ethernet and USB 2.0, and the AX88179
supports USB 3.0. The driver was written by kevlo@ and lwhsu@, with a few
bug fixes from me.
MFC after: 2 months
Fix several problems that can cause panics on kldload and kldunload.
* kproc_create(fasttrap_pid_cleanup_cb, ...) gets called before
fasttrap_provs.fth_table gets allocated. This can lead to a panic
on module load, because fasttrap_pid_cleanup_cb references
fasttrap_provs.fth_table. Move kproc_create down after the point
that fasttrap_provs.fth_table gets allocated, and modify the error
handling accordingly.
* dtrace_fasttrap_{fork,exec,exit} weren't getting NULLed until
after fasttrap_provs.fth_table got freed. That caused panics on
module unload because fasttrap_exec_exit calls
fasttrap_provider_retire, which references
fasttrap_provs.fth_table. NULL those function pointers earlier.
* There wasn't any code to destroy the
fasttrap_{tpoints,provs,procs}.fth_table mutexes on module unload,
leading to a resource leak when WITNESS is enabled. Destroy those
mutexes during fasttrap_unload().
Reviewed by: markj
Approved by: ken (mentor)
Sponsored by: Spectra Logic
MFC after: 4 weeks
It was not being correctly copied into the kernel on FreeBSD, and as a
result, probes with multiple probe sites were not being created properly.
To fix this, change the ioctl definition so that the fasttrap ioctl handler
is responsible for copying in userland data.
Submitted by: Prashanth Kumar <pra_udupi@yahoo.co.in>
MFC after: 1 month
Per POSIX, si_status should contain the value passed to exit() for
si_code==CLD_EXITED and the signal number for other si_code. This was
incorrect for CLD_EXITED and CLD_DUMPED.
This is still not fully POSIX-compliant (Austin group issue #594 says that
the full value passed to exit() shall be returned via si_status, not just
the low 8 bits) but is sufficient for a si_status-related test in libnih
(upstart, Debian/kFreeBSD).
PR: kern/184002
Reported by: Dmitrijs Ledkovs
Tested by: Dmitrijs Ledkovs
On some machines (G5 with lots of RAM), entering OF sometimes causes the
machine to hang. Once the machine is booted, currently the only entry point
into OF is through resetting the framebuffer on mode switch on these machines.
Disabling this allows the machine to stay up at the expense of less usable
consoles after X is started.
MFC after: Never, this is only a hack
some comment I wrote about these values "lying" in the negative diff, which
referes to an earlier misunderstanding about which node to read them from.
This gets at least the PPC64 kernel booting in the mac99 system model in
QEMU after bypassing the MacIO ATA driver, which apparently still has
problems.
conditionally include (but ignore failures) /boot/loader.rc.local and
/boot/menu.rc.local -- to make customizing the menu easier.
Reviewed by: alfred
Discussed on: -hackers
This fixes DPMS with KDE and radeonkms. Without this, the display would
freeze when the monitor is put into sleep state, and only resumes after
several dozens of minutes once the monitor is powered on again.
Tested by: Mathias Picker <Mathias.Picker@virtual-earth.de>
allows FPU emulation on AIM as well as providing support for the mfpvr
and lwsync instructions from userland on e500 cores. lwsync, in particular,
is required for many C++ programs to work correctly.
MFC after: 1 week
the actual FPU is enabled, while PCB_FPREGS indicates that the FPU state
structure in the PCB is valid. This separation reflects the situation on
FPU-less systems in which the FP state is used by the emulator but we don't
actually want to try to turn on the non-existant FPU.
Use this flag to save and restore FP regs properly on both AIM and Book-E.
As a side effect, this sets up hard-FP and Altivec on Book-E CPUs with such
abilities except for a trap handler to call enable_fpu()/enable_altivec().
already valid metadata found at the new location. This should allow easy
transparent recovery if first resize was done by mistake.
While there, unify metadata write code and fix minor memory leak.
MFC after: 1 month
it never did -- and fix an obvious missing line. Floating point emulation
on Book-E still needs some work but this gets it basically functional on
soft-FPU systems (hard FPU for Book-E is not yet implemented).
MFC after: 1 week
it more flexible about how the CCSR range is found. With this change, the
stock MPC85XX will boot on a Routerboard 800.
Hardware donated by: Benjamin Perrault
In "manual" mode just automatically resize provider in any direction.
In "automatic" mode allow only growth (with new metadata write); in case
of shrinking destroy the multipath device same as before since it may be
undesirable to write new metadata within old user area.
MFC after: 1 month
byte positions within the OOB area to support chips with unusual OOB
sizes such as 218 or 224 bytes.
The table for 128 byte OOB works for these but it assumes 3 bytes of ECC
per 256 byte block, and in the case of an ONFI chip the params page may
ask for something different. In other words, this is better but not
yet perfect.
to native endianness. We must also pay attention to unaligned accesses.
Copy the interesting parameters to a new struct so the rest of the code can
forget about these problems.
Submitted by: Kristof Provost <kristof@sigsegv.be> (cleanup) and me (orig).
The ONFI spec states that at least two bytes of the signature ("ONFI")
must be present, and the CRC must be correct to have a valid parameter
page. If the page is not valid there are at least two backup pages where
the data can also be found.
Submitted by: Kristof Provost <kristof@sigsegv.be> (cleanup) and me (orig).
Fixed various link related issues and 10GBaseT is now linking properly.
Modified the types for the driver tunables to be consistent with the sysctl APIs.
Approved by: davidch (mentor)
a very hard time to fully understand) with much more intuitive rights:
CAP_EVENT - when set on descriptor, the descriptor can be monitored
with syscalls like select(2), poll(2), kevent(2).
CAP_KQUEUE_EVENT - When set on a kqueue descriptor, the kevent(2)
syscall can be called on this kqueue to with the eventlist
argument set to non-NULL value; in other words the given
kqueue descriptor can be used to monitor other descriptors.
CAP_KQUEUE_CHANGE - When set on a kqueue descriptor, the kevent(2)
syscall can be called on this kqueue to with the changelist
argument set to non-NULL value; in other words it allows to
modify events monitored with the given kqueue descriptor.
Add alias CAP_KQUEUE, which allows for both CAP_KQUEUE_EVENT and
CAP_KQUEUE_CHANGE.
Add backward compatibility define CAP_POLL_EVENT which is equal to CAP_EVENT.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
sys/dev/xen/balloon/balloon.c:
Remove unused and commented out code.
Fix deadlock caused by performing a sleepable malloc
while holding the balloon mutex.
Perform proper accounting of the memory used by the domain.
Submitted by: Roger Pau Monné
Sponsored by: Citrix Systems R&D
Reviewed by: gibbs
MFC after: 2 days
sys/i386/i386/machdep.c:
sys/amd64/amd64/machdep.c:
The value reported by FreeBSD as "real memory" when booting
doesn't match what is later reported by sysctl as hw.realmem.
This is due to the fact that the value printed during the
boot process is fetched from smbios data (when possible),
and accounts for holes in physical memory. On the other
hand, the value of hw.realmem is unconditionally set to be
one larger than the highest page of the physical address
space.
Fix this by setting hw.realmem to the same value printed
during boot, this makes hw.realmem honour it's name and
account properly for physical memory present in the system.
Submitted by: Roger Pau Monné
Reviewed by: gibbs
in index 0 for the future.
- Move CAP_BINDAT and CAP_CONNECTAT rights to index 0 so we can include
CAP_LOOKUP right in them.
- Shuffle the bits around so there are no gaps. This is last chance to do
that as all moved rights are not used yet.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
CAP_SYMLINKAT and CAP_UNLINKAT capability rights make no sense without
the CAP_LOOKUP right, so include this rights.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
assigned by the IEEE. This file includes documentation on how developers
must carve up the space as well as an initial allocation for bhyve.
Sponsored by: The FreeBSD Foundation
struct mbx_common_read_write_flashrom plus 32KB and caps the actual
transfer size at 32KB. This is harmless as it is but may confuse
static code analyzer, so allocate a full 32KB instead.
Reported by: Coverity via mjacob
Submitted by: Venkata Duvvuru <VenkatKumar.Duvvuru Emulex Com>
Coverity CID: 1125820
VM subsystem twice for every written record.
Tests on 24-core system show double reduction of CPU time spent on copying
single large well-compressed file.
This patch is not really needed on illumos (while not harm either) since
their memory allocator by default uses caching for all requests up to 128K.
Reviewed by: Saso Kiselkov <skiselkov.ml@gmail.com>
Debuggers may need to change PSL_RF. Note that tf_eflags is already stored
in the signal context during signal handling and PSL_RF previously could be
modified via sigreturn, so this change should not provide any new ability
to userspace.
For background see the thread at:
http://lists.freebsd.org/pipermail/freebsd-i386/2007-September/005910.html
Reviewed by: jhb, kib
Sponsored by: DARPA, AFRL
- Do not return blindly if proto isn't ICMP.
- The dport is in network order, so fix comparisons.
- Remove ridiculous htonl(arc4random()).
- Push local variable to a narrower block.
- close cosmetic race in svc_exit();
- do not set wait timeout for idle threads if we have no use for wakeups;
- create new requested thread sooner, not only after some another thread
wakeup, that may happen later under constant load.
This field needs to be (a) set, and (b) greater than the other timeouts
(passive, active, maxquiet, etc.) It also is in microseconds, not
milliseconds.
I hope this will fix the scan hangs that people are seeing.
Obtained from: Linux iwlwifi
sources from uio. Both uio_offset and offset, and uio_resid and resid
have the same types for some time.
Add check for buflen overflow by comparing the buflen with both offset
and resid (vs. comparing with offset only, as it is currently done).
Reported and tested by: pho
Approved by: des (pseudofs maintainer)
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
The previous code simply hard-coded IWN_ANT_AB which is only correct for
some of the NICs.
Now, if the NIC is a 1-stream TX, you need to set IWN_ANT_AB and _not_
just a single antenna. The Intel 5100 firmware panics the moment the
link quality table is updated.
So!
* no secondary antenna? Set it to IWN_ANT_AB;
* two-stream device? Transmit on the full transmit antenna configuration.
Tested:
* Intel 5100, STA
* Intel 2200 (eadler)
Obtained from: Linux iwlwifi
I found a stack overflow when a coredump was taken onto a ZFS volume with
heavy network activity. 2 DSI traps, plus one DECR trap, along with several
function calls in the stack, overflowed the 4 pages. 8 page stack fixes this.
Discussed with: nwhitehorn
MFC after: 1 week
upcoming in-kernel device emulations like the HPET.
The ioctls VM_IOAPIC_ASSERT_IRQ and VM_IOAPIC_DEASSERT_IRQ are used to
manipulate the ioapic pin state.
Discussed with: grehan@
Submitted by: Tycho Nightingale (tycho.nightingale@pluribusnetworks.com)
a variant PCI bus instead of trying to shoehorn it into the PCI host bridge
adapter. Besides matching better the architecture on other platforms, this
also allows systems with multiple partitionable endpoints per PCI host
bridge to work correctly.
it can be overriden by its OFW/FDT version.
Give a chance for GPIO devices that implement the device_identify method to
attach.
Approved by: adrian (mentor)
rely only on checking the device unit to indentify the BSC unit we are
attaching to. Make use of the device base address to identify our BSC unit.
Approved by: adrian (mentor)
support.
* Extend the hardware base_params structure to include a bunch of hardware
flags indicating what is and isn't supported.
* Convert a bunch of the initial hardware configuration conditionals to
consult the base_params structure.
* Add new calibration code for temperature calibration for the Centrino 2xxx
series NICs.
* Add new bluetooth coexistence code for Centrino 2xxx series NICs.
* For NICs that support PAN (personal area networking), use a different
transmit queue and command queue setup, in preparation for said
PAN support.
* Extend the calibration array in iwn_softc to include enough space for
the new calibration types.
Tested (by myself, if not mentioned):
* Intel 4965
* Intel 5100
* Intel 6150
* Intel 2230
* Intel 2200 (eadler)
* Intel 1030
* Intel 6200
* Intel 6230
* Intel 6250
* Intel 6150
* Intel 100
What doesn't work:
* Intel 6235 - fails in calibration at startup
TODO:
* Testing on Intel 53xx series hardware
Submitted by: Cedric Gross <cg@cgross.info>
This is a terrible solution that at least behaves mostly correctly.
It walks the currently active rate table looking for rates to match.
It assumes that the code matches the setup path in the link quality
setup code (much like the previous, much simpler but even more hackish
math did.)
It's O(n), but n<15, so we're okay for the time being.
Tested:
* Intel 5100, STA - 11a, 11n, 11bg modes.
(which is a 1x2 device) panics the firmware.
But, for some 6xxx devices that require IWN_ANT_BC for the TX chainmask,
the link quality entries need to represent _that_.
So, revert this for now until I can figure out what is supposed to be
going on.
PCPU fields for curthread, by doing the same to Book-E. This closes
some potential races switching between CPUs. As a side effect, it turns out
the AIM and Book-E swtch.S implementations were the same to within a few
registers, so move that to powerpc/powerpc.
MFC after: 3 months
This fixes fallout from r256425.
Reported by: Pavel Timofeev <timp87@gmail com>
Tested by: Pavel Timofeev <timp87@gmail com>
Reviewed by: Roger Pau Monnц╘
MFC after: 3 days
add actual platform probing based on PVR. Still needs a little more work:
in particular, the CCRS setup should move here.
Also turn "bare" into a truly bare platform that doesn't pretend to know how
to do anything except get the memory map. This should also be enhanced to
process the FDT reserved memory list, but that is for another day.
assumed that the MDIO bus was a direct child of the Ethernet interface. It
may not be and indeed on many device trees is not. While here, add proper
locking for MII transactions, which may be on a bus shared by several MACs.
Hardware donated by: Benjamin Perrault
unlocked route. Use in6_rtalloc() instead of in6_rtalloc1. This helps
simplify the code and remove several now unused variables.
PR: 156283
MFC after: 2 weeks
NIC and pushed up to the driver. Unfortunately this means there's
no rate control notification done. Thus, if the rate control code
makes a decision that hits a crappy rate that can't succeed, the
rate code would never lower the rate and packet loss would continue.
So, fake some rate control notification in this case.
Without this, a far away station with low signal strength would
associate using the management rate (by default the lowest rate)
and then the EAPOL frames would go out at the current AMRR best
guess. This would result in association failing authentication.
Tested:
* Intel 5100, STA
* Intel 2230, STA
- Process ATIO queue only if interrupt status tells so;
- Do not update queue out pointers after each processed command, do it
only once at the end of the loop.
every time. The purpose of that register is unlikely output queue overflow
detection, so read it only when its last known (and probably stale now)
value signals overflow.
This reduces CPU load and lock congestion and rises bottleneck in CTL
while doing target mode via two 8Gbps ports from 100K to 120K IOPS.
- Fix LOR and possible lock recursion when handling high-power commands.
Introduce new lock to protect left power quota and list of frozen devices.
- Correct locking around xpt periph creation.
- Remove seems never used XPT_FLAG_OPEN xpt periph flag.
_KERNEL braces. Struct mount is only defined for the kernel build.
Reported and tested by: andreast
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
The hang occurred in nfsv4_setsequence() when it couldn't find an
available session slot and is fixed by checking for a forced dismount
in progress and just returning for this case.
MFC after: 1 month
shared vnode lock for VOP_PUTPAGES() as well. The only such
filesystem in the tree is ZFS, and it uses
vnode_pager_generic_putpages(), which performs the pageout with
VOP_WRITE().
Reviewed by: alc
Discussed with: avg
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
holding the vnode lock; vp->v_mount is checked first for NULL
equiality, and then dereferenced if not NULL. If vnode is reclaimed
meantime, second dereference would still give NULL. Change
VFS_PROLOGUE() to evaluate the mp once, convert MNTK_SHARED_WRITES and
MNTK_EXTENDED_SHARED tests into inline functions.
Reviewed by: alc
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
There were two bugs:
* If the initial lowest rate didn't go through the loop at least once,
the AMRR rate index would be the highest rate in the table
(eg the rix mapping to MCS15) but rate would stay at the default
value, namely 0.
This meant that the initial rate selection would be MCS15 _but_ the
node ni_txrate value would be MCS0.
* If the node is 11n, then break out of the loop correctly. Beforehand,
my initial 11n AMRR commit would immediately exit out as it would
fail the 11n check, then it would always fall through to the non-11n
rate which would then see if it was < 36mbit (ie, "72"), which would
always match. Hence, it'd always return MCS15.
Tested:
* Intel Centrino 2230 STA (local changes), STA mode
* Intel Wifi 5100, STA
This is a no-op for now!
* Add a new flag value for "there are no extra bits" for some random
field;
* Add a definition for the maximum number of calibration entries in
the calibration data cache in iwn_softc. It's not yet used.
* Add regulatory bands for the 2030 NIC.
Submitted by: Cedric Gross <cg@cgross.info>
makes FreeBSD halt but not poweroff (as expected when issuing a
shutdown from the VM manager). Fix this by using the same handler
for both "halt" and "poweroff".
NB: The "halt" signal seems to be used on XenServer only. The OSS
Xen toolstack (xl) uses "poweroff" instead.
Submitted by: Roger Pau Monné
Sponsored by: Citrix Systems R&D
Reviewed by: gibbs
MFC after: 2 days
current FreeBSD signal trampoline, but does not specifies
sv_sigcode_base, since shared page is not mapped. This results in the
zero %eip for the signal frame. Fall back to calculating %eip as
offset from the psstrings when sv_sigcode_base is not initialized.
Reported by: Rich Naill <rich@enterprisesystems.net>
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
described in the rev. 3.0 of the Kabini BKDG, document 48751.pdf.
Partially based on the patch submitted by: Dmitry Luhtionov <dmitryluhtionov@gmail.com>
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
words, every architecture is now auto-sizing the kmem arena. This revision
changes kmeminit() so that the definition of VM_KMEM_SIZE_SCALE becomes
mandatory and the definition of VM_KMEM_SIZE becomes optional.
Replace or eliminate all existing definitions of VM_KMEM_SIZE. With
auto-sizing enabled, VM_KMEM_SIZE effectively became an alternate spelling
for VM_KMEM_SIZE_MIN on most architectures. Use VM_KMEM_SIZE_MIN for
clarity.
Change kmeminit() so that the effect of defining VM_KMEM_SIZE is similar to
that of setting the tunable vm.kmem_size. Whereas the macros
VM_KMEM_SIZE_{MAX,MIN,SCALE} have had the same effect as the tunables
vm.kmem_size_{max,min,scale}, the effects of VM_KMEM_SIZE and vm.kmem_size
have been distinct. In particular, whereas VM_KMEM_SIZE was overridden by
VM_KMEM_SIZE_{MAX,MIN,SCALE} and vm.kmem_size_{max,min,scale}, vm.kmem_size
was not. Remedy this inconsistency. Now, VM_KMEM_SIZE can be used to set
the size of the kmem arena at compile-time without that value being
overridden by auto-sizing.
Update the nearby comments to reflect the kmem submap being replaced by the
kmem arena. Stop duplicating the auto-sizing formula in every machine-
dependent vmparam.h and place it in kmeminit() where auto-sizing takes
place.
Reviewed by: kib (an earlier version)
Sponsored by: EMC / Isilon Storage Division
Full MSI-X interrupt support added.
Timeout and reset handling reworked, firmware flash update test added.
Support for drives with 4KB block size added.
Changes made to avoid exposure of phys. array components by default.
Approved by: scottl (mentor), emaste (co-mentor)
menu. This is reported to save headaches on some PPC systems where unload
followed by load does not produce the desired results wherein if-given
the opportunicy to abort the initial loading sequence, you can customize
the first load.
Reviewed by: nwhitehorn, kan
Discussed on: -current
regression manages to do it)... We use a packed struct to coerce
gcc/clang into producing unaligned loads (there is not packed pointer
attribute, otherwise this would be easier)...
use _storeu_ and _loadu_ when using the structure is overkill...
be better at using types properly... Since we allocate our own key
schedule and make sure it's aligned, use the __m128i type in various
arguments to functions...
clang ignores __aligned on prototypes and gcc errors on them, leave them
in comments to document that these function arguments are require to be
aligned...
about all that changes is movdqa -> movdqu from reading the diff of the
disassembly output...
Noticed by: symbolics at gmx.com
MFC after: 3 days
fdt_next_property_offset() API. The old code would sometimes (e.g. on
the device tree supplied by the RB800 boot loader) get confused and stop
partway through listing properties.
MFC after: 1 week
