* Wed Apr 29 2020 mfleming@suse.de
- Update -rt config files after merging Daniel Wagner's latest
patches.
- commit d5d34a0
* Wed Apr 29 2020 dwagner@suse.de
- Unravel "Kconfig on/off" section
Align the "Kconfig on/off" section with v5.4-rt.
- commit da56f6a
* Wed Apr 29 2020 dwagner@suse.de
- Add/update SEQCOUNT section
v5.4-rt reorganised the seqcount patches into it's own section.
- commit c14a397
* Wed Apr 29 2020 dwagner@suse.de
- Update series.conf
Move the two patches after the ARM section as it is in v5.4-rt.
- commit a7f64fb
* Tue Apr 28 2020 dwagner@suse.de
- Add link to patches.rt/signal-revert-ptrace-preempt-magic.patch
Add upstream discussion why revert is needed.
- commit 6aba1ca
* Tue Apr 28 2020 dwagner@suse.de
- Move timekeeping-split-jiffies-lock.patch to patches.suse
This patch has been added to mainline, so no need to carry it in the
- rt section.
- commit 5f7f20d
* Tue Apr 28 2020 dwagner@suse.de
- mm/compaction: Disable compact_unevictable_allowed on RT
(bsc#1162702).
- mm/compaction: really limit compact_unevictable_allowed to 0
and 1 (bsc#1162702).
- commit 366de1c
* Tue Apr 28 2020 dwagner@suse.de
- Drop workaround patches for sleeping locks under RCU-sched.
Drop following patches:
patches.rt/block-blk-mq-move-blk_queue_usage_counter_release-in.patch
patches.rt/fs-aio-simple-simple-work.patch
Both patch added a workaround to avoid taking a sleeping lock while
running under RCU-sched. Both workarounds are not necessary anymore.
The upstream fix 9e8d42a0f7eb ("percpu-refcount: Use normal instead of
RCU-sched"") reverted the offending patch and normal RCU is
used (works for RT).
The fix is already backported so we can just drop the patches.
- commit b5ea7ec
* Tue Apr 28 2020 dwagner@suse.de
- of: Rework and simplify phandle cache to use a fixed size
(bsc#1162702).
Replace patches.rt/of-allocate-free-phandle-cache-outside-of-the-devtre.patch
with rework patch from mainline.
- commit 0c9ab67
* Tue Apr 28 2020 dwagner@suse.de
- Update softirq-Add-preemptible-softirq.patch and powerpc-preempt-lazy-support.patch
The v5.4-rt version of softirq-Add-preemptible-softirq.patch
containts rcu-Acquire-RCU-lock-when-disabling-BHs.patch.
Therefore, drop the later patch.
Because the updated version of softirq-Add-preemptible-softirq.patch
also modifies the linux/interrupt.h header we update
powerpc-preempt-lazy-support.patch which merged the changes in
powerpc-Fixup-compile-and-lazy-preempt.patch. Drop this patch
too.
- commit 8f4b8c7
* Tue Apr 28 2020 dwagner@suse.de
- Replace workqueue v5.2-rt changes with v5.{4|6}-rt version
The patches for makein workqueue -rt ready are delayed. The subsystem
maintainers is not keen getting -rt patches with out -rt ready kernel.
Backport the old patch with the set of workqueue patches. This alligns
SLERT to v5.4-rt and v5.6-rt.
- commit 2cb02e2
* Tue Apr 28 2020 dwagner@suse.de
- Replace cgroup v5.2-rt fix with v5.{4|6}-rt version
The workqueue-Convert-the-locks-to-raw-type.patch was not accepted
upstream and an alternative fix was propossed but not accepted yet.
Backport the old patch with the set of cgroup patches. This alligns
SLERT to v5.4-rt and v5.6-rt.
- commit 84eeb44
* Tue Apr 28 2020 mfleming@suse.de
- Fix build errors for rt_debug by disabling CONFIG_KERNEL_HEADER_TEST
A recent commit from SLE15-SP2,
commit 0628907a58b ("- sched/nohz: Avoid disabling the tick for very short durations (bnc#754690, bsc#1158748)
causes build failures for rt_debug because that config enables
CONFIG_KERNEL_HEADER_TEST. Even though rt_debug is designed to be a
debug config we really don't want to be chasing problems caused by
non-self-contained headers. Disable it.
- commit bfd07bc
* Mon Apr 27 2020 dwagner@suse.de
- smp: Remove superfluous cond_func check in
smp_call_function_many_cond() (bsc#1162702).
- docs: kmemleak: DEBUG_KMEMLEAK_EARLY_LOG_SIZE changed names
(bsc#1162702).
- commit 5baf587
* Mon Apr 27 2020 dwagner@suse.de
- Drop -rt kmemleak patches before porting mainline version
The kmemleak patches haven been applied in mainline. Before we can
cleanly backport drop them from our queue.
- commit ec3f2cd
* Mon Apr 27 2020 dwagner@suse.de
- mm/kmemleak: turn kmemleak_lock and object->lock to
raw_spinlock_t (bsc#1162702).
- kmemleak: Do not corrupt the object_list during clean-up
(bsc#1162702).
- mm/kmemleak.c: record the current memory pool size
(bsc#1162702).
- mm: kmemleak: use the memory pool for early allocations
(bsc#1162702).
- mm: kmemleak: simple memory allocation pool for kmemleak objects
(bsc#1162702).
- mm: kmemleak: make the tool tolerant to struct scan_area
allocation failures (bsc#1162702).
- commit 56a4d88
* Mon Apr 27 2020 mkubecek@suse.cz
- fix patch metadata
- fix Patch-mainline:
patches.suse/cpuidle-menu-Allow-tick-to-be-stopped-if-PM-QoS-is-used.patch
patches.suse/mm-vmscan.c-clean-code-by-removing-unnecessary-assignment.patch
patches.suse/rcu-Allow-only-one-expedited-GP-to-run-concurrently-with-wakeups.patch
patches.suse/rcu-Fix-missed-wakeup-of-exp_wq-waiters.patch
patches.suse/sched-fair-Align-rq-avg_idle-and-rq-avg_scan_cost.patch
- commit f15a73a
* Mon Apr 27 2020 mgorman@suse.de
- sched/nohz: Avoid disabling the tick for very short durations
(bnc#754690, bsc#1158748).
- Delete patches.suse/sched-throttle-nohz.patch.
- commit 0628907
* Mon Apr 27 2020 mgorman@suse.de
- sched/cpuacct: Use __this_cpu_add() instead of this_cpu_ptr()
(bnc#1155798 (CPU scheduler functional and performance
backports)).
- commit 04cc379
* Mon Apr 27 2020 mgorman@suse.de
- sched/fair: Use __this_cpu_read() in wake_wide() (bnc#1155798
(CPU scheduler functional and performance backports)).
- sched/fair: Align rq->avg_idle and rq->avg_scan_cost
(bnc#1155798 (CPU scheduler functional and performance
backports)).
- mm/vmscan.c: clean code by removing unnecessary assignment
(git fixes (mm/vmscan)).
- rcu: Allow only one expedited GP to run concurrently with
wakeups (git fixes (rcu)).
- rcu: Fix missed wakeup of exp_wq waiters (git fixes (rcu)).
- cpuidle: menu: Allow tick to be stopped if PM QoS is used
(bnc#1158748 (network regression)).
- commit cb72208