[Intel-gfx] [PATCH] drm/i915/execlists: Relax the locked clear_bit(IRQ_EXECLIST)

[Chris Wilson]

We only need to care about the ordering of the clearing of the bit with
the uncached CSB read in order to correctly detect a new interrupt
before the read completes. The uncached read itself acts as a full
memory barrier, so we do not to enforce another in the form of a locked
clear_bit.

Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
---
 drivers/gpu/drm/i915/intel_lrc.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 59acdd3b7a12..c9010fa881b4 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -540,7 +540,14 @@ static void intel_lrc_irq_handler(unsigned long data)
 			dev_priv->regs + i915_mmio_reg_offset(RING_CONTEXT_STATUS_BUF_LO(engine, 0));
 		unsigned int csb, head, tail;
 
-		clear_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
+		/* The write will be ordered by the uncached read (itself
+		 * a memory barrier), we do need another in the form a
+		 * locked instruction. That is the clear of IRQ_EXECLIST bit
+		 * will be visible to another cpu prior to the completion
+		 * of the CSB read. If the other cpu receives an interrupt
+		 * before then, we will redo the CSB read.
+		 */
+		__clear_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
 		csb = readl(csb_mmio);
 		head = GEN8_CSB_READ_PTR(csb);
 		tail = GEN8_CSB_WRITE_PTR(csb);
-- 
2.11.0