bdev: Add spdk_bdev_get_current_qd to measure and return current value

The generic bdev layer has a public API spdk_bdev_get_qd() but its
value is the most recently measured value and it requires qd sampling
to be enabled. We will have bdev modules to want to wait until
all bdev_ios are aborted by a reset. Unfortunately, spdk_bdev_get_qd()
is not suitable for the custom bdev module. Furthermore,
spdk_bdev_channel::io_outstanding is not accessible from bdev modules.
Hence, add a new public API spdk_bdev_get_current_qd().
This function should be used only from the bdev module and it should
be ensured that the bdev is not unregistered during execution.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ica30a8d8fe3264e28f0772a39bdf5f9ba72933e1
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12791
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>

CHANGELOG.md

@@ -20,6 +20,9 @@ New RPCs `bdev_xnvme_create` and `bdev_xnvme_delete` were added to support the x
 A new API `spdk_bdev_for_each_bdev_io` was added to execute the function on the appropriate
 thread for each bdev_io submitted to the bdev.
 
+A new API `spdk_bdev_get_current_qd` was added to measure and return the queue depth from a
+bdev. This API is available even when queue depth sampling is disabled.
+
 ### sock
 
 Added new `ssl` based socket implementation, the code is located in module/sock/posix.

include/spdk/bdev_module.h

@@ -1313,6 +1313,27 @@ typedef void (*spdk_bdev_for_each_io_cb)(void *ctx, int rc);
 void spdk_bdev_for_each_bdev_io(struct spdk_bdev *bdev, void *ctx, spdk_bdev_io_fn fn,
 				spdk_bdev_for_each_io_cb cb);
 
+typedef void (*spdk_bdev_get_current_qd_cb)(struct spdk_bdev *bdev, uint64_t current_qd,
+		void *cb_arg, int rc);
+
+/**
+ * Measure and return the queue depth from a bdev.
+ *
+ * Note: spdk_bdev_get_qd() works only when the user enables queue depth sampling,
+ * while this new function works even when queue depth sampling is disabled.
+ * The returned queue depth may not be exact, for example, some additional I/Os may
+ * have been submitted or completed during the for_each_channel operation.
+ * This function should be used only in the bdev module and it should be ensured
+ * that the dev is not unregistered while executing the function.
+ * cb_fn is required to specify.
+ *
+ * \param bdev Block device to query.
+ * \param cb_fn Callback function to be called with queue depth measured for a bdev.
+ * \param cb_arg Argument to pass to callback function.
+ */
+void spdk_bdev_get_current_qd(struct spdk_bdev *bdev,
+			      spdk_bdev_get_current_qd_cb cb_fn, void *cb_arg);
+
 /*
  *  Macro used to register module for later initialization.
  */

lib/bdev/bdev.c

@@ -4052,6 +4052,58 @@ spdk_bdev_set_qd_sampling_period(struct spdk_bdev *bdev, uint64_t period)
 				   bdev, period);
 }
 
+struct bdev_get_current_qd_ctx {
+	uint64_t current_qd;
+	spdk_bdev_get_current_qd_cb cb_fn;
+	void *cb_arg;
+};
+
+static void
+bdev_get_current_qd_done(struct spdk_io_channel_iter *i, int status)
+{
+	struct bdev_get_current_qd_ctx *ctx = spdk_io_channel_iter_get_ctx(i);
+	void *io_dev = spdk_io_channel_iter_get_io_device(i);
+
+	ctx->cb_fn(__bdev_from_io_dev(io_dev), ctx->current_qd, ctx->cb_arg, 0);
+
+	free(ctx);
+}
+
+static void
+bdev_get_current_qd(struct spdk_io_channel_iter *i)
+{
+	struct bdev_get_current_qd_ctx *ctx = spdk_io_channel_iter_get_ctx(i);
+	struct spdk_io_channel *io_ch = spdk_io_channel_iter_get_channel(i);
+	struct spdk_bdev_channel *bdev_ch = spdk_io_channel_get_ctx(io_ch);
+
+	ctx->current_qd += bdev_ch->io_outstanding;
+
+	spdk_for_each_channel_continue(i, 0);
+}
+
+void
+spdk_bdev_get_current_qd(struct spdk_bdev *bdev, spdk_bdev_get_current_qd_cb cb_fn,
+			 void *cb_arg)
+{
+	struct bdev_get_current_qd_ctx *ctx;
+
+	assert(cb_fn != NULL);
+
+	ctx = calloc(1, sizeof(*ctx));
+	if (ctx == NULL) {
+		cb_fn(bdev, 0, cb_arg, -ENOMEM);
+		return;
+	}
+
+	ctx->cb_fn = cb_fn;
+	ctx->cb_arg = cb_arg;
+
+	spdk_for_each_channel(__bdev_to_io_dev(bdev),
+			      bdev_get_current_qd,
+			      ctx,
+			      bdev_get_current_qd_done);
+}
+
 static void
 _resize_notify(void *arg)
 {

lib/bdev/spdk_bdev.map

@@ -45,6 +45,7 @@
 	spdk_bdev_get_dif_type;
 	spdk_bdev_is_dif_head_of_md;
 	spdk_bdev_is_dif_check_enabled;
+	spdk_bdev_get_current_qd;
 	spdk_bdev_get_qd;
 	spdk_bdev_get_qd_sampling_period;
 	spdk_bdev_set_qd_sampling_period;