deadline_test: Made '-i' work and added help text for it

z-stream release

Add cyclicdeadline and deadline_test to rt-tests

Install cyclictest with cap_sys_rawio to access msr and cpuid registers
Fixed url in Source0

cyclictest: new CPUs with SMI counter support

Upgrade to 1.0
tests: Makefile: Assume numa_parse_cpustring_all available
tests: Add man page for rt-migrate-test
tests: Update the cyclictest man page
tests: Add missing option to hwlatdetect man page
tests: Housekeeping fix some spelling errors.
tests: hwlat.txt: smidetect renamed to hwlatdetect
tests: Remove doc/release-checklist.txt
tests: Makefile - bump version to 1.0
Add patch to install rt-migrate-test.8-man-page
The latest build also includes fixes for undocumented options in cyclictest and hwlatdetect, as well as many more documentation fixes.
Also documenting here that the new feature cyclictest: SMI count/detection via MSR/SMI counter was added in v0.97

cyclictest: stop any tracing after hitting a breaktrace threshold
Resolves: rhbz#1333762

Change spec file to compile with HAVE_PARSE_CPUSTRING_ALL=1
This makes numa_parse_cpustring_all() available which is needed for
Resolves: rhbz#1330468

Changes from v0.93 to v0.97
migrate-test: updated to latest code from rostedt
-smi description on cyclictest man page
tests: Allow for user-specified PYLIB
tests: Break out install_hwlatdetect
switch to CC and AR
number for sched_(gs)etattr() for tile
switch to Makefile
migrate-test
sched.h: do not unconditionally define syscall-numbers
migrate-test
utils: Add John Kacur to the copyright
utils: Fix some checkpatch errors in rt-utils.c
dumpmachine once in the makefile
log update
ups in timerthread before working on it
only-prerequisite
tests.spec-in
in
migrate-test for exit(1)
tests: Makefile: Bump version number to 0.97
tests: workaround poor gzip implementations
1) for failures
-tracemark option
KĂśnig (5):
migrate-test: remove space before \n
Resolves: rhbz#1283264

tests.tar file using git-archiv
archive should ignore
-watch option to watch output in realtime
checklist.tx

Maria Gleixner (2):
DHAVE_PARSE_CPUSTRING_ALL

From Boris Egorov <egorov@linux.com>
rt-migrate-test: exit early if nr_runs is non-positive
rt-migrate-test: use variables instead of macros
From Uwe Kleine-KĂśnig <u.kleine-koenig@pengutronix.de>
pi_stress: remove timestamp of compilation from version output
rt-migrate-test: make sure input parameters are converted to correct units
rt-migrate-test: sanity check --prio value

pip_stress: parameterize usleep value to work-around platform issues
From Daniel Wagner <daniel.wagner@bmw-carit.de>:
pi_stress: Store schedule attributes per thread
rt-utils: Add gettid()
rt-utils: Add helper to parse/print scheduling policies
rt-sched: Add sched_setattr/sched_getattr API
pi_stress: Use error.h for logging and debugging
error: Add debug() function
pi_stress: Remove unused TIMER_SIGNAL definition
rt-tests.h: Remove unused header file
hackbench: Don't re-assign context for each fd
From Joakim Hernberg <jbh@alchemy.lu>:
cyclictest: make affinity option only use number of online cpus
From John Kacur <jkacur@redhat.com>:
cyclictest: Add long option --laptop to preserve battery power
cyclictest: Fix help for long options only
cyclictest: Change the output from function sighand() to stderr
cyclictest: Always print an err message if write of 0 to cpu-dma_latency fails
rt_numa.h: Suppress discards 'const' qualifier warning
lib: Rework err_msg_n to output strerror after message
From Alexey Brodkin <Alexey.Brodkin@synopsys.com>:
Makefile: allow building selected tests with non-NPTL toolchain

clean up debugging comments and printfs from last release
ran hwlatdetect.py through 2to3, works for both python2 and 3

From Uwe Kleine-KĂśnig <u.kleine-koenig@pengutronix.de>:
cyclictest: Fix tracemark output when the latency threshold is hit on ARM
From Gary S. Robertson <gary.robertson@linaro.org>:
cyclictest: Restore CPU affinity function for non-NUMA builds
cyclictest: Don't offer --numa option when unavailable

From Aaron Fabbri <ajfabbri@gmail.com>:
cyclictest: CPU affinity now takes arbitrary set of cpus