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

Merge "omap intc changes for v3.18 merge window" from Tony Lindgren:

Interrupt code related clean-up for omap2 and 3 to make
it ready to move to drivers/irqchip. Note that this series
does not yet move the interrupt code to drivers, that will
be posted separately as a follow-up series.

Note that this branch has a dependency to patches both
in fixes-v3.18-not-urgent and soc-for-v3.18 and is based on
a merge. Without doing the merge, off-idle would not work
properly for git bisect.

* tag 'intc-for-v3.18' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap: (325 commits)
  arm: omap: intc: switch over to linear irq domain
  arm: omap: irq: get rid of ifdef hack
  arm: omap: irq: introduce omap_nr_pending
  arm: omap: irq: remove nr_irqs argument
  arm: omap: irq: remove unnecessary header
  arm: omap: irq: drop omap2_intc_handle_irq()
  arm: omap: irq: drop omap3_intc_handle_irq()
  arm: omap: irq: call set_handle_irq() from .init_irq
  arm: omap: irq: move some more code around
  arm: boot: dts: omap2/3/am33xx: drop ti,intc-size
  arm: omap: irq: drop ti,intc-size support
  arm: boot: dts: am33xx/omap3: fix intc compatible flag
  arm: omap: irq: use compatible flag to figure out number of IRQ lines
  arm: omap: irq: add specific compatibles for omap3 and am33xx devices
  arm: omap: irq: drop .handle_irq and .init_irq fields
  arm: omap: irq: use IRQCHIP_DECLARE macro
  arm: omap: irq: call set_handle_irq() from intc_of_init
  arm: omap: irq: make intc_of_init static
  arm: omap: irq: reorganize code a little bit
  arm: omap: irq: always define omap3 support
  ...

Signed-off-by: Olof Johansson <olof@lixom.net>

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)))