i965: Implement WaCsStallAtEveryFourthPipecontrol on IVB/BYT.

According to the documentation, we need to do a CS stall on every fourth
PIPE_CONTROL command to avoid GPU hangs.  The kernel does a CS stall
between batches, so we only need to count the PIPE_CONTROLs in our batches.

v2: Get the generation check right (caught by Chris Wilson),
    combine the ++ with the check (suggested by Daniel Vetter).

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index fde4177301895824b075ec9e5def20a1b5ce9c41..c1787a8cf7f2dea0f9a09b2475e77e587e4baa0a 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -868,6 +868,8 @@ struct intel_batchbuffer {
    enum brw_gpu_ring ring;
    bool needs_sol_reset;
 
+   uint8_t pipe_controls_since_last_cs_stall;
+
    struct {
       uint16_t used;
       int reloc_count;

diff --git a/src/mesa/drivers/dri/i965/intel_batchbuffer.c b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
index 2bd11d7a63e27207ea6985ea7650c9fdc136e1b4..193550bc2dd0be65dcb54c22a189b4ac78268d56 100644 (file)
--- a/src/mesa/drivers/dri/i965/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
@@ -81,6 +81,7 @@ intel_batchbuffer_reset(struct brw_context *brw)
    brw->batch.state_batch_offset = brw->batch.bo->size;
    brw->batch.used = 0;
    brw->batch.needs_sol_reset = false;
+   brw->batch.pipe_controls_since_last_cs_stall = 0;
 
    /* We don't know what ring the new batch will be sent to until we see the
     * first BEGIN_BATCH or BEGIN_BATCH_BLT.  Mark it as unknown.
@@ -433,6 +434,33 @@ gen8_add_cs_stall_workaround_bits(uint32_t *flags)
       *flags |= PIPE_CONTROL_STALL_AT_SCOREBOARD;
 }
 
+/* Implement the WaCsStallAtEveryFourthPipecontrol workaround on IVB, BYT:
+ *
+ * "Every 4th PIPE_CONTROL command, not counting the PIPE_CONTROL with
+ *  only read-cache-invalidate bit(s) set, must have a CS_STALL bit set."
+ *
+ * Note that the kernel does CS stalls between batches, so we only need
+ * to count them within a batch.
+ */
+static uint32_t
+gen7_cs_stall_every_four_pipe_controls(struct brw_context *brw, uint32_t flags)
+{
+   if (brw->gen == 7 && !brw->is_haswell) {
+      if (flags & PIPE_CONTROL_CS_STALL) {
+         /* If we're doing a CS stall, reset the counter and carry on. */
+         brw->batch.pipe_controls_since_last_cs_stall = 0;
+         return 0;
+      }
+
+      /* If this is the fourth pipe control without a CS stall, do one now. */
+      if (++brw->batch.pipe_controls_since_last_cs_stall == 4) {
+         brw->batch.pipe_controls_since_last_cs_stall = 0;
+         return PIPE_CONTROL_CS_STALL;
+      }
+   }
+   return 0;
+}
+
 /**
  * Emit a PIPE_CONTROL with various flushing flags.
  *
@@ -454,6 +482,8 @@ brw_emit_pipe_control_flush(struct brw_context *brw, uint32_t flags)
       OUT_BATCH(0);
       ADVANCE_BATCH();
    } else if (brw->gen >= 6) {
+      flags |= gen7_cs_stall_every_four_pipe_controls(brw, flags);
+
       BEGIN_BATCH(5);
       OUT_BATCH(_3DSTATE_PIPE_CONTROL | (5 - 2));
       OUT_BATCH(flags);
@@ -496,6 +526,8 @@ brw_emit_pipe_control_write(struct brw_context *brw, uint32_t flags,
       OUT_BATCH(imm_upper);
       ADVANCE_BATCH();
    } else if (brw->gen >= 6) {
+      flags |= gen7_cs_stall_every_four_pipe_controls(brw, flags);
+
       /* PPGTT/GGTT is selected by DW2 bit 2 on Sandybridge, but DW1 bit 24
        * on later platforms.  We always use PPGTT on Gen7+.
        */