[Accel-config] [PATCH v1 9/9] accel-config/test: add cache flush operation support

The Cache Flush operation, 0x20, flushes the processor caches at
the Destination Address.

Signed-off-by: Tony Zhu <tony.zhu(a)intel.com>
---
 test/dsa.c      | 56 +++++++++++++++++++++++++++++++++++++++++++++++++
 test/dsa.h      |  5 +++++
 test/dsa_test.c |  7 +++++++
 test/prep.c     | 30 ++++++++++++++++++++++++++
 4 files changed, 98 insertions(+)

diff --git a/test/dsa.c b/test/dsa.c
index ecda9a3..6cf7f80 100644
--- a/test/dsa.c
+++ b/test/dsa.c
@@ -964,6 +964,62 @@ int dsa_drain_multi_task_nodes(struct dsa_context *ctx)
 	return ret;
 }
 
+int dsa_wait_cflush(struct dsa_context *ctx, struct task *tsk)
+{
+	struct dsa_hw_desc *desc = tsk->desc;
+	struct dsa_completion_record *comp = tsk->comp;
+
+	int rc;
+
+again:
+	rc = dsa_wait_on_desc_timeout(comp, ms_timeout);
+	if (rc < 0) {
+		err("cflush desc timeout\n");
+		return DSA_STATUS_TIMEOUT;
+	}
+
+	/* re-submit if PAGE_FAULT reported by HW && BOF is off */
+	if (stat_val(comp->status) == DSA_COMP_PAGE_FAULT_NOBOF &&
+	    !(desc->flags & IDXD_OP_FLAG_BOF)) {
+		dsa_reprep_cflush(ctx, tsk);
+		goto again;
+	}
+
+	return DSA_STATUS_OK;
+}
+
+int dsa_cflush_multi_task_nodes(struct dsa_context *ctx)
+{
+	struct task_node *tsk_node = ctx->multi_task_node;
+	int ret = DSA_STATUS_OK;
+
+	while (tsk_node) {
+		tsk_node->tsk->dflags = IDXD_OP_FLAG_CRAV | IDXD_OP_FLAG_RCR;
+		if ((tsk_node->tsk->test_flags & TEST_FLAGS_BOF) && ctx->bof)
+			tsk_node->tsk->dflags |= IDXD_OP_FLAG_BOF;
+
+		dsa_prep_cflush(tsk_node->tsk);
+		tsk_node = tsk_node->next;
+	}
+
+	tsk_node = ctx->multi_task_node;
+	while (tsk_node) {
+		dsa_desc_submit(ctx, tsk_node->tsk->desc);
+		tsk_node = tsk_node->next;
+	}
+	tsk_node = ctx->multi_task_node;
+	info("Submitted all cflush jobs\n");
+
+	while (tsk_node) {
+		ret = dsa_wait_cflush(ctx, tsk_node->tsk);
+		if (ret != DSA_STATUS_OK)
+			info("Desc: %p failed with ret: %d\n",
+			     tsk_node->tsk->desc, tsk_node->tsk->comp->status);
+		tsk_node = tsk_node->next;
+	}
+	return ret;
+}
+
 int dsa_wait_batch(struct batch_task *btsk)
 {
 	int rc;
diff --git a/test/dsa.h b/test/dsa.h
index 6cceb62..e77f818 100644
--- a/test/dsa.h
+++ b/test/dsa.h
@@ -268,6 +268,9 @@ int dsa_wait_noop(struct dsa_context *ctx, struct task *tsk);
 int dsa_drain_multi_task_nodes(struct dsa_context *ctx);
 int dsa_wait_drain(struct dsa_context *ctx, struct task *tsk);
 
+int dsa_cflush_multi_task_nodes(struct dsa_context *ctx);
+int dsa_wait_cflush(struct dsa_context *ctx, struct task *tsk);
+
 int dsa_memcpy_multi_task_nodes(struct dsa_context *ctx);
 int dsa_wait_memcpy(struct dsa_context *ctx, struct task *tsk);
 
@@ -303,6 +306,8 @@ int dsa_wait_dif(struct dsa_context *ctx, struct task *tsk);
 
 void dsa_prep_noop(struct task *tsk);
 void dsa_prep_drain(struct task *tsk);
+void dsa_prep_cflush(struct task *tsk);
+void dsa_reprep_cflush(struct dsa_context *ctx, struct task *tsk);
 void dsa_prep_memcpy(struct task *tsk);
 void dsa_reprep_memcpy(struct dsa_context *ctx, struct task *tsk);
 void dsa_prep_memfill(struct task *tsk);
diff --git a/test/dsa_test.c b/test/dsa_test.c
index 9b56160..0de762c 100644
--- a/test/dsa_test.c
+++ b/test/dsa_test.c
@@ -399,6 +399,12 @@ static int test_memory(struct dsa_context *ctx, size_t buf_size,
 			if (rc != DSA_STATUS_OK)
 				return rc;
 			break;
+		case DSA_OPCODE_CFLUSH:
+			rc = dsa_cflush_multi_task_nodes(ctx);
+			if (rc != DSA_STATUS_OK)
+				return rc;
+			break;
+
 		default:
 			err("Unsupported op %#x\n", opcode);
 			return -EINVAL;
@@ -726,6 +732,7 @@ int main(int argc, char *argv[])
 	case DSA_OPCODE_COMPARE:
 	case DSA_OPCODE_COMPVAL:
 	case DSA_OPCODE_DUALCAST:
+	case DSA_OPCODE_CFLUSH:
 		rc = test_memory(dsa, buf_size, tflags, opcode, num_desc);
 		if (rc != DSA_STATUS_OK)
 			goto error;
diff --git a/test/prep.c b/test/prep.c
index 48e0a7e..c427109 100644
--- a/test/prep.c
+++ b/test/prep.c
@@ -53,6 +53,36 @@ void dsa_prep_drain(struct task *tsk)
 	tsk->comp->status = 0;
 }
 
+void dsa_prep_cflush(struct task *tsk)
+{
+	info("preparing descriptor for cflush\n");
+
+	dsa_prep_desc_common(tsk->desc, tsk->opcode, (uint64_t)(tsk->dst1),
+			     (uint64_t)(tsk->src1), tsk->xfer_size, tsk->dflags);
+	tsk->desc->completion_addr = (uint64_t)(tsk->comp);
+	tsk->comp->status = 0;
+}
+
+void dsa_reprep_cflush(struct dsa_context *ctx, struct task *tsk)
+{
+	struct dsa_completion_record *compl = tsk->comp;
+	struct dsa_hw_desc *hw = tsk->desc;
+
+	info("PF addr %#lx dir %d bc %#x\n",
+	     compl->fault_addr, compl->result,
+	     compl->bytes_completed);
+
+	hw->xfer_size -= compl->bytes_completed;
+
+	hw->src_addr += compl->bytes_completed;
+
+	resolve_page_fault(compl->fault_addr, compl->status);
+
+	compl->status = 0;
+
+	dsa_desc_submit(ctx, hw);
+}
+
 void dsa_prep_memcpy(struct task *tsk)
 {
 	info("preparing descriptor for memcpy\n");
-- 
2.27.0