Reduce number of timehands to just two. This is useful because consumers can now be only one tc_windup() call late.
Use C99 initialization.
Tested by: pho (as part of the whole patch)
Reviewed by: jhb (same)
Discussed with: bde
Sponsored by: The FreeBSD Foundation
MFC after: 1 month
X-Differential revision: https://reviews.freebsd.org/D7302
Update #3175.
Hide the boottime and bootimebin globals, provide the getboottime(9) and getboottimebin(9) KPI. Change consumers of boottime to use the KPI. The variables were renamed to avoid shadowing issues with local variables of the same name.
Issue is that boottime* should be adjusted from tc_windup(), which
requires them to be members of the timehands structure. As a
preparation, this commit only introduces the interface.
Some uses of boottime were found doubtful, e.g. NLM uses boottime to
identify the system boot instance. Arguably the identity should not
change on the leap second adjustment, but the commit is about the
timekeeping code and the consumers were kept bug-to-bug compatible.
Tested by: pho (as part of the bigger patch)
Reviewed by: jhb (same)
Discussed with: bde
Sponsored by: The FreeBSD Foundation
MFC after: 1 month
X-Differential revision: https://reviews.freebsd.org/D7302
Update #3175.
Define `fhard` in pps_event(..) only when PPS_SYNC is defined to mute an -Wunused-but-set-variable warning
Reported by: FreeBSD_HEAD_amd64_gcc4.9 jenkins job
Sponsored by: EMC / Isilon Storage Division
Update #3175.
Use per-cpu values for base and last in tc_cpu_ticks(). The values are updated lockess, different CPUs write its own view of timecounter state. The critical section is done for safety, callers of tc_cpu_ticks() are supposed to already enter critical section, or to own a spinlock.
The change fixes sporadical reports of too high values reported for
the (W)CPU on platforms that do not provide cpu ticker and use
tc_cpu_ticks(), in particular, arm*.
Diagnosed and reviewed by: jhb
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Update #3175.
Constify the pointers to eventtimer and timecounter name strings.
The need for this appears as soon as you try to set the names to something
that isn't a "quoted literal". (I'm actually confused why quoted strings
aren't a problem as well, we must have some warning disabled.)
Update #3175.
If a specific timecounter has been chosen via sysctl, and a new timecounter with higher quality registers (presumably in a module that has just been loaded), do not undo the user's choice by switching to the new timecounter.
Document that behavior, and also the fact that there is no way to unregister
a timecounter (and thus no way to unload a module containing one).
Update #3175.
Only process the PPS event types currently enabled in pps_params.mode.
This makes the PPS API behave correctly, but isn't ideal -- we still end
up capturing PPS data for non-enabled edges, we just don't process the
data into an event that becomes visible outside of kern_tc. That's because
the event type isn't passed to pps_capture(), so it can't do the filtering.
Any solution for capture filtering is going to require touching every driver.
Update #3175.
Reimplement the ordering requirements for the timehands updates, and for timehands consumers, by using fences.
Ensure that the timehands->th_generation reset to zero is visible
before the data update is visible [*]. tc_setget() allowed data update
writes to become visible before generation (but not on TSO
architectures).
Remove tc_setgen(), tc_getgen() helpers, use atomics inline [**].
Noted by: alc [*]
Requested by: bde [**]
Reviewed by: alc, bde
Sponsored by: The FreeBSD Foundation
MFC after: 3 weeks
Update #3175.
Tweaks for r284178:
Do not include machine/atomic.h explicitely, the header is already included
by sys/systm.h.
Force inlining of tc_getgen() and tc_setgen(). The functions are used
more than once, which causes compilers with non-aggressive inlining
policies to generate calls.
Suggested by: bde
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Update #3175.
Implement a mechanism for making changes in the kernel<->driver PPS interface without breaking ABI or API compatibility with existing drivers.
The existing data structures used to communicate between the kernel and
driver portions of PPS processing contain no spare/padding fields and no
flags field or other straightforward mechanism for communicating changes
in the structures or behaviors of the code. This makes it difficult to
MFC new features added to the PPS facility. ABI compatibility is
important; out-of-tree drivers in module form are known to exist. (Note
that the existing api_version field in the pps_params structure must
contain the value mandated by RFC 2783 and any RFCs that come along after.)
These changes introduce a pair of abi-version fields which are filled in
by the driver and the kernel respectively to indicate the interface
version. The driver sets its version field before calling the new
pps_init_abi() function. That lets the kernel know how much of the
pps_state structure is understood by the driver and it can avoid using
newer fields at the end of the structure that it knows about if the driver
is a lower version. The kernel fills in its version field during the init
call, letting the driver know what features and data the kernel supports.
To implement the new version information in a way that is backwards
compatible with code from before these changes, the high bit of the
lightly-used 'kcmode' field is repurposed as a flag bit that indicates the
driver is aware of the abi versioning scheme. Basically if this bit is
clear that indicates a "version 0" driver and if it is set the driver_abi
field indicates the version.
These changes also move the recently-added 'mtx' field of pps_state from
the middle to the end of the structure, and make the kernel code that uses
this field conditional on the driver being abi version 1 or higher. It
changes the only driver currently supplying the mtx field, usb_serial, to
use pps_init_abi().
Reviewed by: hselasky@
Update #3175.
Set default thread processor affinity to all processors of the
pre-allocated set. This allows to constify the
_POSIX_Threads_Default_attributes.
Update #2514.
Return stack area via pthread_getattr_np().
Simplify
* pthread_attr_setaffinity_np(), and
* pthread_attr_getaffinity_np()
and let the scheduler do the more sophisticated error checks.
Make
* pthread_setaffinity_np(),
* pthread_getaffinity_np(),
* pthread_attr_setaffinity_np(), and
* pthread_attr_getaffinity_np()
available in all configurations.
Update #2514.
Close#3145.
Close#3168.
The timestamp are based on the uptime. There is no need for a 64-bit
seconds part. The signed 32-bit seconds part of the sbintime_t limits
the uptime to roughly 68 years.
Close#2740.
For semaphore object pointer and object validation see
POSIX_SEMAPHORE_VALIDATE_OBJECT().
Destruction or close of a busy semaphore returns an error status. The
object is not flushed.
POSIX semaphores are now available in all configurations and no longer
depend on --enable-posix.
Update #2514.
Update #3116.
