Merge tag 'soc-for-v3.18' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap into next/soc

SoC related changes for omaps for v3.18 merge window:

- PM changes to make the code easier to use on newer SoCs
- PM changes for newer SoCs suspend and resume and wake-up events
- Minor clean-up to remove dead Kconfig options

Note that these have a dependency to the fixes-v3.18-not-urgent
tag and is based on a commit in that series.

* tag 'soc-for-v3.18' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap: (514 commits)
  ARM: OMAP5+: Reuse OMAP4 PM code for OMAP5 and DRA7
  ARM: dts: OMAP3+: Add PRM interrupt
  ARM: omap: Remove stray ARCH_HAS_OPP references
  ARM: DRA7: Add hook in SoC initcalls to enable pm initialization
  ARM: OMAP5: Add hook in SoC initcalls to enable pm initialization
  ARM: OMAP5 / DRA7: Enable CPU RET on suspend
  ARM: OMAP5 / DRA7: PM: Provide a dummy startup function for CPU hotplug
  ARM: OMAP5 / DRA7: PM: Avoid all SAR saves
  ARM: OMAP5 / DRA7: PM: Enable Mercury retention mode on CPUx powerdomains
  ARM: OMAP5 / DRA7: PM / wakeupgen: Enables ES2 PM mode by default
  ARM: OMAP5 / DRA7: PM: Set MPUSS-EMIF clock-domain static dependency
  ARM: OMAP5 / DRA7: PM: Update CPU context register offset
  ARM: AM437x: use pdata quirks for pinctrl information
  ARM: DRA7: use pdata quirks for pinctrl information
  ARM: OMAP5: use pdata quirks for pinctrl information
  ARM: OMAP4+: PM: Use only valid low power state for CPU hotplug
  ARM: OMAP4+: PM: use only valid low power state for suspend
  ARM: OMAP4+: PM: Make logic state programmable
  ARM: OMAP2+: powerdomain: introduce logic for finding valid power domain
  ARM: OMAP2+: powerdomain: pwrdm_for_each_clkdm iterate only valid clkdms
  ...

1 files changed, 15 insertions, 1 deletions

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index afb04b9b818a..b38fb2b9e237 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -626,8 +626,22 @@ int ring_buffer_poll_wait(struct ring_buffer *buffer, int cpu,
 		work = &cpu_buffer->irq_work;
 	}
 
-	work->waiters_pending = true;
 	poll_wait(filp, &work->waiters, poll_table);
+	work->waiters_pending = true;
+	/*
+	 * There's a tight race between setting the waiters_pending and
+	 * checking if the ring buffer is empty.  Once the waiters_pending bit
+	 * is set, the next event will wake the task up, but we can get stuck
+	 * if there's only a single event in.
+	 *
+	 * FIXME: Ideally, we need a memory barrier on the writer side as well,
+	 * but adding a memory barrier to all events will cause too much of a
+	 * performance hit in the fast path.  We only need a memory barrier when
+	 * the buffer goes from empty to having content.  But as this race is
+	 * extremely small, and it's not a problem if another event comes in, we
+	 * will fix it later.
+	 */
+	smp_mb();
 
 	if ((cpu == RING_BUFFER_ALL_CPUS && !ring_buffer_empty(buffer)) ||
 	    (cpu != RING_BUFFER_ALL_CPUS && !ring_buffer_empty_cpu(buffer, cpu)))