Merge tag 'timers-core-2020-08-14' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull more timer updates from Thomas Gleixner:
 "A set of posix CPU timer changes which allows to defer the heavy work
  of posix CPU timers into task work context. The tick interrupt is
  reduced to a quick check which queues the work which is doing the
  heavy lifting before returning to user space or going back to guest
  mode. Moving this out is deferring the signal delivery slightly but
  posix CPU timers are inaccurate by nature as they depend on the tick
  so there is no real damage. The relevant test cases all passed.

  This lifts the last offender for RT out of the hard interrupt context
  tick handler, but it also has the general benefit that the actual
  heavy work is accounted to the task/process and not to the tick
  interrupt itself.

  Further optimizations are possible to break long sighand lock hold and
  interrupt disabled (on !RT kernels) times when a massive amount of
  posix CPU timers (which are unpriviledged) is armed for a
  task/process.

  This is currently only enabled for x86 because the architecture has to
  ensure that task work is handled in KVM before entering a guest, which
  was just established for x86 with the new common entry/exit code which
  got merged post 5.8 and is not the case for other KVM architectures"

* tag 'timers-core-2020-08-14' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86: Select POSIX_CPU_TIMERS_TASK_WORK
  posix-cpu-timers: Provide mechanisms to defer timer handling to task_work
  posix-cpu-timers: Split run_posix_cpu_timers()

2 files changed, 21 insertions, 0 deletions

diff --git a/include/linux/posix-timers.h b/include/linux/posix-timers.h
index e3f0f8585da4..896c16d2c5fb 100644
--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -6,6 +6,7 @@
 #include <linux/list.h>
 #include <linux/alarmtimer.h>
 #include <linux/timerqueue.h>
+#include <linux/task_work.h>
 
 struct kernel_siginfo;
 struct task_struct;
@@ -125,6 +126,16 @@ struct posix_cputimers {
 	unsigned int			expiry_active;
 };
 
+/**
+ * posix_cputimers_work - Container for task work based posix CPU timer expiry
+ * @work:	The task work to be scheduled
+ * @scheduled:  @work has been scheduled already, no further processing
+ */
+struct posix_cputimers_work {
+	struct callback_head	work;
+	unsigned int		scheduled;
+};
+
 static inline void posix_cputimers_init(struct posix_cputimers *pct)
 {
 	memset(pct, 0, sizeof(*pct));
@@ -165,6 +176,12 @@ static inline void posix_cputimers_group_init(struct posix_cputimers *pct,
 					      u64 cpu_limit) { }
 #endif
 
+#ifdef CONFIG_POSIX_CPU_TIMERS_TASK_WORK
+void posix_cputimers_init_work(void);
+#else
+static inline void posix_cputimers_init_work(void) { }
+#endif
+
 #define REQUEUE_PENDING 1
 
 /**
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 53ddc02e2e79..93ecd930efd3 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -890,6 +890,10 @@ struct task_struct {
 	/* Empty if CONFIG_POSIX_CPUTIMERS=n */
 	struct posix_cputimers		posix_cputimers;
 
+#ifdef CONFIG_POSIX_CPU_TIMERS_TASK_WORK
+	struct posix_cputimers_work	posix_cputimers_work;
+#endif
+
 	/* Process credentials: */
 
 	/* Tracer's credentials at attach: */