markj [Wed, 4 Dec 2013 01:40:39 +0000 (01:40 +0000)]
Enable some previously-disabled DTrace tests for umod, ufunc and usym. They
expect the installed ksh binary to be named "ksh", which is not the case
when it's installed on FreeBSD via the shells/ksh93 port. Allow for it to be
"ksh93" as well so that the tests can actually pass.
cperciva [Tue, 3 Dec 2013 21:55:57 +0000 (21:55 +0000)]
The rc system aggressively caches the contents of /etc/rc.conf in order to
improve boot performance; this produces arguably astonishing (non-)results
if /etc/rc.conf is modified during the boot process.
Since performance considerations make it infeasible to automatically detect
if the cached /etc/rc.conf parameters should be invalidated, provide a
mechanism for explicitly requesting that /etc/rc.conf be reloaded: Catch
SIGALRM and reload /etc/rc.conf if it is received.
cperciva [Tue, 3 Dec 2013 21:35:25 +0000 (21:35 +0000)]
Add a new sysctl / loader tunable kern.panic_reboot_wait_time which
defaults to PANIC_REBOOT_WAIT_TIME (a long-existing kernel config
setting). Use this now-variable value in place of the defined constant
to control how long the system waits after a panic before rebooting.
emaste [Tue, 3 Dec 2013 21:29:45 +0000 (21:29 +0000)]
lldb: Threaded inferior support for FreeBSD
This is in the process of being submitted to the upstream LLDB
repository. The thread list functionality is modelled in part on
GDBRemoteCommunicationClient.
cxgbe(4): T4_SET_SCHED_CLASS and T4_SET_SCHED_QUEUE ioctls to program
scheduling classes in the chip and to bind tx queue(s) to a scheduling
class respectively. These can be used for various kinds of tx traffic
throttling (to force selected tx queues to drain at a fixed Kbps rate,
or a % of the port's total bandwidth, or at a fixed pps rate, etc.).
emaste [Tue, 3 Dec 2013 18:12:51 +0000 (18:12 +0000)]
Workaround lldb issue with main module base address
On FreeBSD lldb sometimes reloads the the main module's (executable's)
symbols at the wrong address. Work around this for now by explicitly
reloading at base_address=0 when it happens.
A proper fix is needed but early testers have reported this issue so
this workaround should allow them to make further progress.
jhb [Tue, 3 Dec 2013 14:50:12 +0000 (14:50 +0000)]
Fix an off-by-one error in r228960. The maximum priority delta provided
by SCHED_PRI_TICKS should be SCHED_PRI_RANGE - 1 so that the resulting
priority value (before nice adjustment) is between SCHED_PRI_MIN and
SCHED_PRI_MAX, inclusive.
neel [Tue, 3 Dec 2013 00:43:31 +0000 (00:43 +0000)]
Rename 'vm_interrupt_hostcpu()' to 'vcpu_notify_event()' because the function
has outgrown its original name. Originally this function simply sent an IPI
to the host cpu that a vcpu was executing on but now it does a lot more than
just that.
attilio [Mon, 2 Dec 2013 22:34:47 +0000 (22:34 +0000)]
Right now LOCK_DEBUG is enabled if KTR is on.
This is to support LOCK_LOG_* functionality effectively in debugging
environments but it is overkill because really LOCK_DEBUG should be on
only if (KTR_COMPILE & KTR_LOCK) is true.
Fix this by applying the correct logic.
In this process, move the KTR classes to its own header to reduce
namespace pollution.
zbb [Mon, 2 Dec 2013 13:09:59 +0000 (13:09 +0000)]
Enable missing Access Flag for secondary cores on ARMv6/v7
Spotted by: Wojciech Macek <wma@semihalf.com>
Obtained from: Semihalf
> Description of fields to fill in above: 76 columns --|
> PR: If a GNATS PR is affected by the change.
> Submitted by: If someone else sent in the change.
> Reviewed by: If someone else reviewed your modification.
> Approved by: If you needed approval for this commit.
> Obtained from: If the change is from a third party.
> MFC after: N [day[s]|week[s]|month[s]]. Request a reminder email.
> Security: Vulnerability reference (one per line) or description.
> Empty fields above will be automatically removed.
kevlo [Mon, 2 Dec 2013 09:07:43 +0000 (09:07 +0000)]
The RUN_MAX_TXSZ constat is defined as Tx desc + Tx wireless info + MCLBYTES +
max padding. We were lucky in that run(4) working fine since both
rt2860_rxwi and rt2860_txwi structure sizes are the same.
pjd [Mon, 2 Dec 2013 08:21:28 +0000 (08:21 +0000)]
Please welcome casperd daemon. It (and its services) will be responsible for
giving access to functionality that is not available in capability mode
sandbox. The functionality can be precisely restricted.
Start with the following services:
- system.dns - provides API compatible to:
- gethostbyname(3),
- gethostbyname2(3),
- gethostbyaddr(3),
- getaddrinfo(3),
- getnameinfo(3),
- system.grp - provides getgrent(3)-compatible API,
- system.pwd - provides getpwent(3)-compatible API,
- system.random - allows to obtain entropy from /dev/random,
- system.sysctl - provides sysctlbyname(3-compatible API.
adrian [Mon, 2 Dec 2013 03:59:45 +0000 (03:59 +0000)]
Overhaul the iwn(4) scan infrastructure to be slightly more "correct"
for these chipsets.
* Correctly set the active/passive flag in the scan request - this is
NOT a "is the channel active|passive"; it's to do with whether we
have an SSID to actively scan for or not. The firmware takes care
of the active/passive setup of the channel.
* Calculate the active/passive dwell time based on the beacon interval
and the channel mode, rather than using a hard coded value.
* For now, hardcode the scan service_time. It's defined as:
31:22 - number of beacon intervals to come back onto the home channel
for;
0:21 - time (microseconds) to come back onto the home channel for.
When doing an active scan when the NIC is active (whether we're associated
or not - it only matters if we've setup the NIC to a destination or not)
this determines how much time to stay on the home channel for when
scanning. We can tune this based on the amount of active traffic.
For now it's 4 beacon intervals and 100 microseconds.
* Fix the "good crc threshold" setting. It differs based on the NIC
firmware. Some older firmware required a workaround; the later
firmware instead treats the field as a flag.
* Enforce that we are not sending a scan command if one is already
pending. Any time this is done is a bug and it absolutely needs
to be fixed - so be very loud.
* Add the SCAN flag to a few debug messages that are scan related but
only occuring under STATE.
Now, this does get noisy when you're scanning in an actively busy 2GHz
network as the firmware (for reason I don't quite yet understand) seems
hell bent on staying on some passive channels longer than it should.
However, it should eventually recover and complete the scan.
This is a work in progress; please let me know if things get stuck or
if things improve!
Tested:
* intel centrino 2200
* intel centrino 2230
* intel 6200
* intel 5100
* intel 4965 (gets upset, but that's a known issue)
adrian [Mon, 2 Dec 2013 03:40:51 +0000 (03:40 +0000)]
Add some code to double-check whether we're correctly populating the
TX ring according to what the firmware requires.
The firmware requires A-MPDU sub-frames to be at a very specific ring
offset - that is, the ring slot offset should be (seqno % 256.)
This holds for every NIC I've tested thus far except the 4965,
which starts erroring out here shortly before the firmware panics.
Which is good, it's doing what it's supposed to (read: capture that
we've screwed up somewhere.)
The specifics about getting this stuff right:
* the initial seqno allocation should match up with the ringid.
* .. yes, this means we can start at a ring offset that isn't zero.
* .. because we program the start seqno in the firmware message
to setup the AC.
* The initial seqno allocation may be non-zero _and_ frames may be
being transmitted during a-mpdu negotiation. I faced similar
issues on ath(4) and had to software queue frames to that node+TID
during A-MPDU negotiation.
* seqno allocation should be in lockstep with ring increments.
* If we fail to transmit some segment, no, we shouldn't reuse that
ring slot. We should just transmit a BAR (which we aren't yet
doing, sigh) and move onto the next seqno.
* In theory there shouldn't be any holes in the seqno space when
we are transmitting frames.
Tested:
* 4965 (throws problems, so yes we have to fix this);
* 5100 (seems ok);
* 6200 (seems ok);
* 2200 (seems ok);
* 2230 (seems ok).
eadler [Mon, 2 Dec 2013 03:36:44 +0000 (03:36 +0000)]
Partial revert of r258779 and r258780:
The directory sys/dev/drm2/i915 is apperently contributed code.
Revert to the broken version of this file to make future imports easier.
eadler [Mon, 2 Dec 2013 03:11:25 +0000 (03:11 +0000)]
In a situation where:
- The remote host sends a FIN
- in an ACK for a sequence number for which an ACK has already
been received
- There is still unacked data on route to the remote host
- The packet does not contain a window update
The packet may be dropped without processing the FIN flag.
PR: kern/99188
Submitted by: Staffan Ulfberg <staffan@ulfberg.se>
Discussed with: andre
MFC after: never
nwhitehorn [Mon, 2 Dec 2013 00:44:36 +0000 (00:44 +0000)]
Add new sysctl, kern.supported_abis, containing the list of FreeBSD
MACHINE_ARCH values whose binaries this kernel can run. This patch provides
a feature requested for implementing pkgng ABI identifiers in a robust
way.
The list is designed to indicate whether, say, an i386 package can be run on
the current system. If kern.supported_abis contains "i386", then the answer
is yes. Otherwise, the answer is no.
At the moment, this only supports MACHINE_ARCH and MACHINE_ARCH32. As we
gain support for more interesting combinations, this needs to become more
flexible, possibily through the sysent framework, along with the
hw.machine_arch emulation immediately preceding this code in kern_mib.c.
nwhitehorn [Sun, 1 Dec 2013 19:45:04 +0000 (19:45 +0000)]
No actual hardware supported by FreeBSD requires this SPRG save/restore
paranoia, so kill it. In particular, changes to SPRG0 are dangerous, since
that is where the PCPU pointer is kept.
nwhitehorn [Sun, 1 Dec 2013 19:43:15 +0000 (19:43 +0000)]
Rearchitect platform memory map parsing to make it less
Open Firmware-centric:
- Keep the static list of regions in platform.c instead of ofw_machdep.c
- Move various merging and sorting operations to platform.c as well
- Move apple_hacks code out of ofw_machdep.c and into platform_powermac.c,
where it belongs
- Move CHRP-specific dynamic-reconfiguration memory parsing into
platform_chrp.c instead of pretending it is shared code
nwhitehorn [Sun, 1 Dec 2013 19:03:20 +0000 (19:03 +0000)]
Open Firmware mandates that certain cross-references, in particular those
in /chosen, be ihandles. The ePAPR spec makes those cross-reference phandles,
since FDT has no concept of ihandles. Have the OF FDT CI module interpret
queries about ihandles as cross-reference phandles.
nwhitehorn [Sun, 1 Dec 2013 17:28:28 +0000 (17:28 +0000)]
The kernel stack guard pages are only below the stack pointer, not above.
Prevent erroneous detection of stack overflows on legitimate faults on the
page after this thread's stack.
nwhitehorn [Sun, 1 Dec 2013 16:02:22 +0000 (16:02 +0000)]
Make uart_cpu_powerpc work on both FDT and OFW systems. This is the last
remaining modification required to build kernels that work with both on
PowerPC.
pjd [Sun, 1 Dec 2013 09:41:06 +0000 (09:41 +0000)]
Move my simple logging API to a separate library. It is now already used
by hastctl(8), hastd(8) and auditdistd(8) and will soon be also used
by casperd(8) and its services. There is no documentation and pjdlog.h
header file is not installed in /usr/include/ to keep it private.
Unfortunately we don't have /lib/private/ at this point, only
/usr/lib/private/, so the library is installed in /lib/.
mckusick [Sun, 1 Dec 2013 07:34:21 +0000 (07:34 +0000)]
We needlessly panic when trying to flush MKDIR_PARENT dependencies.
We had previously tried to flush all MKDIR_PARENT dependencies (and
all the NEWBLOCK pagedeps) by calling ffs_update(). However this will
only resolve these dependencies in direct blocks. So very large
directories with MKDIR_PARENT dependencies in indirect blocks had
not yet gotten flushed. As the directory is in the midst of doing a
complete sync, we simply defer the checking of the MKDIR_PARENT
dependencies until the indirect blocks have been sync'ed.
adrian [Sun, 1 Dec 2013 03:53:21 +0000 (03:53 +0000)]
Migrate the sendfile_sync structure into a public(ish) API in preparation
for extending and reusing it.
The sendfile_sync wrapper is mostly just a "mbuf transaction" wrapper,
used to indicate that the backing store for a group of mbufs has completed.
It's only being used by sendfile for now and it's only implementing a
sleep/wakeup rendezvous. However, there are other potential signaling
paths (kqueue) and other potential uses (socket zero-copy write) where the
same mechanism would also be useful.
So, with that in mind:
* extract the sendfile_sync code out into sf_sync_*() methods
* teach the sf_sync_alloc method about the current config flag -
it will eventually know about kqueue.
* move the sendfile_sync code out of do_sendfile() - the only thing
it now knows about is the sfs pointer. The guts of the sync
rendezvous (setup, rendezvous/wait, free) is now done in the
syscall wrapper.
* .. and teach the 32-bit compat sendfile call the same.
This should be a no-op. It's primarily preparation work for teaching
the sendfile_sync about kqueue notification.
Tested:
* Peter Holm's sendfile stress / regression scripts
eadler [Sat, 30 Nov 2013 22:17:27 +0000 (22:17 +0000)]
Fix undefined behavior: (1 << 31) is not defined as 1 is an int and this
shifts into the sign bit. Instead use (1U << 31) which gets the
expected result.
This fix is not ideal as it assumes a 32 bit int, but does fix the issue
for most cases.
pjd [Sat, 30 Nov 2013 15:08:35 +0000 (15:08 +0000)]
Make process descriptors standard part of the kernel. rwhod(8) already
requires process descriptors to work and having PROCDESC in GENERIC
seems not enough, especially that we hope to have more and more consumers
in the base.
tuexen [Sat, 30 Nov 2013 12:51:19 +0000 (12:51 +0000)]
In
http://svnweb.freebsd.org/changeset/base/258221
I introduced a bug which initialized global locks
whenever the SCTP stack initialized. This was fixed in
http://svnweb.freebsd.org/changeset/base/258574
by rodrigc@. He just initialized the locks for
the default vnet. This fix reverts to the old
behaviour before r258221, which explicitly makes
sure it is only called once, because this works also on
other platforms.
MFC after: 3 days
X-MFC with: r258574.
nwhitehorn [Fri, 29 Nov 2013 21:59:03 +0000 (21:59 +0000)]
Use the Open Firmware-based CPU frequency determination as a generic
fallback if we can't measure CPU frequency. This is also useful on a
variety of embedded systems using FDT.