Nathan Whitehorn [Tue, 10 Feb 2015 00:55:42 +0000 (00:55 +0000)]
Set thread priorities on multithreaded CPUs so that threads holding a
spinlock are high-priority and threads waiting for a spinlock are set to
low priority.
Rui Paulo [Mon, 9 Feb 2015 23:13:50 +0000 (23:13 +0000)]
Notify devd(8) when a process crashed.
This change implements a notification (via devctl) to userland when
the kernel produces coredumps after a process has crashed.
devd can then run a specific command to produce a human readable crash
report. The command is most usually a helper that runs gdb/lldb
commands on the file/coredump pair. It's possible to use this
functionality for implementing automatic generation of crash reports.
devd(8) will be notified of the full path of the binary that crashed and
the full path of the coredump file.
Add x2APIC support. Enable it by default if CPU is capable. The
hw.x2apic_enable tunable allows disabling it from the loader prompt.
To closely repeat effects of the uncached memory ops when accessing
registers in the xAPIC mode, the x2APIC writes to MSRs are preceeded
by mfence, except for the EOI notifications. This is probably too
strict, only ICR writes to send IPI require serialization to ensure
that other CPUs see the previous actions when IPI is delivered. This
may be changed later.
In vmm justreturn IPI handler, call doreti_iret instead of doing iretd
inline, to handle corner conditions.
Note that the patch only switches LAPICs into x2APIC mode. It does not
enables FreeBSD to support > 255 CPUs, which requires parsing x2APIC
MADT entries and doing interrupts remapping, but is the required step
on the way.
Reviewed by: neel
Tested by: pho (real hardware), neel (on bhyve)
Discussed with: jhb, grehan
Sponsored by: The FreeBSD Foundation
MFC after: 2 months
Randall Stewart [Mon, 9 Feb 2015 19:28:11 +0000 (19:28 +0000)]
This fixes a bug in the way that the LLE timers for nd6
and arp were being used. They basically would pass in the
mutex to the callout_init. Because they used this method
to the callout system, it was possible to "stop" the callout.
When flushing the table and you stopped the running callout, the
callout_stop code would return 1 indicating that it was going
to stop the callout (that was about to run on the callout_wheel blocked
by the function calling the stop). Now when 1 was returned, it would
lower the reference count one extra time for the stopped timer, then
a few lines later delete the memory. Of course the callout_wheel was
stuck in the lock code and would then crash since it was accessing
freed memory. By using callout_init(c, 1) we always get a 0 back
and the reference counting bug does not rear its head. We do have
to make a few adjustments to the callouts themselves though to make
sure it does the proper thing if rescheduled as well as gets the lock.
Commented upon by hiren and sbruno
See Phabricator D1777 for more details.
Commented upon by hiren and sbruno
Reviewed by: adrian, jhb and bz
Sponsored by: Netflix Inc.
Randall Stewart [Mon, 9 Feb 2015 19:19:44 +0000 (19:19 +0000)]
This fixes two conditions that can incur when migration
is being done in the callout code and harmonizes the macro
use.:
1) The callout_active() will lie. Basically if a migration
is occuring and the callout is about to expire and the
migration has been deferred, the callout_active will no
longer return true until after the migration. This confuses
and breaks callers that are doing callout_init(&c, 1); such
as TCP.
2) The migration code had a bug in it where when migrating, if
a two calls to callout_reset came in and they both collided with
the callout on the wheel about to run, then the second call to
callout_reset would corrupt the list the callout wheel uses
putting the callout thread into a endless loop.
3) Per imp, I have fixed all the macro occurance in the code that
were for the most part being ignored.
Phabricator D1711 and looked at by lstewart and jhb and sbruno.
Reviewed by: kostikbel, imp, adrian, hselasky
MFC after: 3 days
Sponsored by: Netflix Inc.
Warner Losh [Mon, 9 Feb 2015 16:29:44 +0000 (16:29 +0000)]
Prefer install over mkdir to create the directory. Add test to ensure
the directory doesn't exist before creating it. This makes the NO_ROOT
case actually work.
Warner Losh [Mon, 9 Feb 2015 16:04:08 +0000 (16:04 +0000)]
We need to create /boot/dtb since some use cases don't create a full
root with BSD.root.mtree, so it often times will not exist. Rather
than force the latter for an installkernel, just create the directory
with a comment about why.
Mark invalid page table entries correctly for PMAP as well as for the
hypervisor. This prevents an infinite loop where processes with evicted
pages would page fault forever when PMAP decided the evicted pages on
which the process was faulting was actually present and did not need to
be restored.
Bjoern A. Zeeb [Mon, 9 Feb 2015 11:34:45 +0000 (11:34 +0000)]
Properly hide a variable under #ifdef as it is only used inside the
specific #ifdef block otherwise leaving an unused variable and breaking
other kernel builds.
Glen Barber [Mon, 9 Feb 2015 10:46:39 +0000 (10:46 +0000)]
Revert r278445.
I was going to use __FreeBSD_version to determine if
xz(1) should be multi-threaded by default, but doing
this will cause problems if/when the changes are merged
from head.
Glen Barber [Mon, 9 Feb 2015 10:42:27 +0000 (10:42 +0000)]
Enable multi-threaded xz(1) compression, after r278433.
Allow multi-threaded xz(1) to be turned off by specifying
NO_XZTHREADS, and allow number of threads to be overridden
by specifying XZ_THREADS=N.
MFC after: 1 week
X-MFC-needs: r278433
Sponsored by: The FreeBSD Foundation
Rui Paulo [Mon, 9 Feb 2015 06:20:34 +0000 (06:20 +0000)]
Merge xz 5.2.0.
This brings support for multi-threaded compression. This brings close
N times faster compression where N is the number of CPU cores.
Because of this, liblzma now depends on libthr.
Soon libarchive will be modified to use the new lzma API.
Do not mark shared structures as __packed, it leads to race condition
If structure packed as __packed clang (and probably gcc) generates
code that loads word fields (e.g. tx_pos) byte-by-byte and if it's
modified by VideoCore in the same time as ARM loads the value result
is going to be mixed combination of bytes from previous value and
new one.
Fix an extremely subtle concurrency bug triggered by running on 32-thread
POWER8 systems. During thread switch, there was a very small window when
the stack pointer was set to the stack pointer of the outgoing thread, but
after the lock on that thread had already been released.
If, during that window, the outgoing thread were rescheduled on another CPU
and begin execution and an exception were taken on the original CPU, the
trap handler and the outgoing thread would simultaneously execute on the same
stack, causing memory corruption. Fix this by making sure to release the
old thread only after cpu_switch() is done with its stack.
MFC after: 2 weeks
Sponsored by: FreeBSD Foundation
Simplify trapcode setup by placing a copy of the generic trap handler at
every possible trap address by default. This also makes sure the kernel
notices (and panics at) traps from newer CPUs that the kernel was not
expecting rather than executing gibberish memory.
Add some error checking on the supplied page size list. This makes sure
that we (a) get the correct large page size to provide to pmap and (b)
we can alert the user if running under incorrectly-configured PowerKVM
on POWER7 and POWER8 systems.
Navdeep Parhar [Sun, 8 Feb 2015 09:28:55 +0000 (09:28 +0000)]
cxgbe(4): tidy up some of the interaction between the Upper Layer
Drivers (ULDs) and the base if_cxgbe driver.
Track the per-adapter activation of ULDs in a new "active_ulds" field.
This was done pretty arbitrarily before this change -- via TOM_INIT_DONE
in adapter->flags for TOM, and the (1 << MAX_NPORTS) bit in
adapter->offload_map for iWARP.
iWARP and hw-accelerated iSCSI rely on the TOE (supported by the TOM
ULD). The rules are:
a) If the iWARP and/or iSCSI ULDs are available when TOE is enabled then
iWARP and/or iSCSI are enabled too.
b) When the iWARP and iSCSI modules are loaded they go looking for
adapters with TOE enabled and enable themselves on that adapter.
c) You cannot deactivate or unload the TOM module from underneath iWARP
or iSCSI. Any such attempt will fail with EBUSY.
Protect uninitialized scalar variable from being accessed
In a couple of cases a variable "stayopen" can be checked
unitialized. This is of no danger as the complementary
condition is false but prevent the access by switching
the checks.
Alexander Motin [Sat, 7 Feb 2015 13:19:04 +0000 (13:19 +0000)]
Teach ctld(8) to control non-iSCSI CTL ports.
This change introduces new target option "port", that assigns current target
to specified CTL port. On config application ctld(8) will apply LUN mapping
according to target configuration to specified port and bring the port up.
On shutdown cltd(8) will remove the mapping and put the port down.
This change allows to configure both iSCSI and FibreChannel targets in the
same configuration file in alike way.
Kernel side support was added earlier at r278037.
MFC after: 2 weeks
Relnotes: yes
Sponsored by: iXsystems, Inc.
Bools (that are the result of direct truncs) are lowered as whatever
the argument to the trunc was and a "and 1", causing the part of the
MBB responsible for this argument to look something like this:
but remember to (at the end of isel) replace vreg7 by vreg15. Now for
the bug. In fast isel lowering, we mistakenly mark vreg8 as the result
of the load instead of the trunc. This adds a fixup to have
vreg8 replaced by whatever the result of the load is as well, so
we end up with
Dimitry Andric [Sat, 7 Feb 2015 12:20:33 +0000 (12:20 +0000)]
Fix a number of -Wcast-qual warnings in ath's ar9300_attach.c, by making
the ia_array field of struct ar9300_ini_array const, and removing the
const-dropping casts. No functional change.
Reviewed by: adrian
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D1725
Steve Kargl [Sat, 7 Feb 2015 00:38:18 +0000 (00:38 +0000)]
Truncate the exponent 'n' of type long to a domain contained
within [INT_MIN, INT_MAX] where the magnitude of the lower
and upper bounds are sufficiently large to span the range of
scalbn[fl].
While here, remove the GNU style bug in the function declarations.
Make it possible to set (via ctl.conf(5)) and query (via ctladm islist -v)
target iSCSI offload. Add mechanism to query maximum receive data segment
size supported by chosen hardware offload module, and use it in ctld(8)
to determine the value to advertise to the other side.
MFC after: 1 month
Sponsored by: The FreeBSD Foundation
John Baldwin [Fri, 6 Feb 2015 19:41:23 +0000 (19:41 +0000)]
Change ktrdump to use the more standard -M/-N flags to specify the path
to a crash dump and kernel, respectively. The existing -m/-e flags are
still supported for backwards compatiblity but are no longer documented.
John Baldwin [Fri, 6 Feb 2015 18:19:59 +0000 (18:19 +0000)]
Revert the IPI startup sequence to match what is described in the
Intel Multiprocessor Specification v1.4. The Intel SDM claims that
the INIT IPIs here are invalid, but other systems follow the MP
spec instead.
While here, fix the IPI wait routine to accept a timeout in microseconds
instead of a raw spin count, and don't spin forever during AP startup.
Instead, panic if a STARTUP IPI is not delivered after 20 us.
Alexander Motin [Fri, 6 Feb 2015 17:43:13 +0000 (17:43 +0000)]
Add support for multiple portal groups per target.
This change allows multiple "portal-group" options to be specified per
target. Each of them may include new optional auth-group name parameter
to override per-target auth parameters for specific portal group.
John Baldwin [Fri, 6 Feb 2015 16:45:10 +0000 (16:45 +0000)]
Use direct hardware access for internal requests for KCS and SMIC. In
particular, updates to the watchdog should no longer sleep.
- Add a new IPMI_IO_LOCK for low-level I/O access. Use this for
kcs_polled_request() and smic_polled_request().
- Add a new backend callback "ipmi_driver_request" to handle a driver
request. The new callback performs the request sychronously for KCS
and SMIC. SSIF still defers the work to the worker thread since the
worker thread sleeps during request processing anyway.
- Allocate driver requests on the stack rather than using malloc().
John Baldwin [Fri, 6 Feb 2015 16:09:01 +0000 (16:09 +0000)]
Add a new device control utility for new-bus devices called devctl. This
allows the user to request administrative changes to individual devices
such as attach or detaching drivers or disabling and re-enabling devices.
- Add a new /dev/devctl2 character device which uses ioctls for device
requests. The ioctls use a common 'struct devreq' which is somewhat
similar to 'struct ifreq'.
- The ioctls identify the device to operate on via a string. This
string can either by the device's name, or it can be a bus-specific
address. (For unattached devices, a bus address is the only way to
locate a device.) Bus drivers register an eventhandler to claim
unrecognized device names that the driver recognizes as a valid address.
Two buses currently support addresses: ACPI recognizes any device
in the ACPI namespace via its full path starting with "\" and
the PCI bus driver recognizes an address specification of
'pci[<domain>:]<bus>:<slot>:<func>' (identical to the PCI selector
strings supported by pciconf).
- To make it easier to cut and paste, change the PnP location string
in the PCI bus driver to output a full PCI selector string rather
than 'slot=<slot> function=<func>'.
- Add a devctl(3) interface in libdevctl which provides a wrapper around
the ioctls and is the preferred interface for other userland code.
- Add a devctl(8) program which is a simple wrapper around the requests
supported by devctl(3).
- Add a device_is_suspended() function to check DF_SUSPENDED.
- Add a resource_unset_value() function that can be used to remove a
hint from the kernel environment. This is used to clear a
hint.<driver>.<unit>.disabled hint when re-enabling a boot-time
disabled device.
This was a local addition to the original change from NetBSD.
Being this libc there is some chance for it to interfere with
user's cget*() functions usage. The memory leak was finely
plugged by r278300.
Davide Italiano [Fri, 6 Feb 2015 03:49:31 +0000 (03:49 +0000)]
Always prefer double dashes for GNU LD long options.
I discovered this while working on llvm/lld and realized export-dynamic
only supported --. Although upstream will eventually grow to support
both - and --, switch this in our build system, because GNU ld supports
both modes, and because there's some hope lld will become the default linker
for FreeBSD in the future.