DPDK patches and discussions
 help / color / mirror / Atom feed
From: xuan.ding@intel.com
To: maxime.coquelin@redhat.com, chenbo.xia@intel.com
Cc: dev@dpdk.org, jiayu.hu@intel.com, xingguang.he@intel.com,
	weix.ling@intel.com, cheng1.jiang@intel.com,
	yuanx.wang@intel.com, wenwux.ma@intel.com,
	Xuan Ding <xuan.ding@intel.com>
Subject: [PATCH v8 1/2] vhost: introduce DMA vchannel unconfiguration
Date: Tue, 25 Oct 2022 08:25:39 +0000	[thread overview]
Message-ID: <20221025082540.100618-2-xuan.ding@intel.com> (raw)
In-Reply-To: <20221025082540.100618-1-xuan.ding@intel.com>

From: Xuan Ding <xuan.ding@intel.com>

Add a new API rte_vhost_async_dma_unconfigure() to unconfigure DMA
vChannels in vhost async data path. Lock protection are also added
to protect DMA vChannel configuration and unconfiguration
from concurrent calls.

Signed-off-by: Xuan Ding <xuan.ding@intel.com>
---
 doc/guides/prog_guide/vhost_lib.rst    |  5 ++
 doc/guides/rel_notes/release_22_11.rst |  5 ++
 lib/vhost/rte_vhost_async.h            | 20 +++++++
 lib/vhost/version.map                  |  3 ++
 lib/vhost/vhost.c                      | 72 ++++++++++++++++++++++++--
 5 files changed, 100 insertions(+), 5 deletions(-)

diff --git a/doc/guides/prog_guide/vhost_lib.rst b/doc/guides/prog_guide/vhost_lib.rst
index bad4d819e1..0d9eca1f7d 100644
--- a/doc/guides/prog_guide/vhost_lib.rst
+++ b/doc/guides/prog_guide/vhost_lib.rst
@@ -323,6 +323,11 @@ The following is an overview of some key Vhost API functions:
   Get device type of vDPA device, such as VDPA_DEVICE_TYPE_NET,
   VDPA_DEVICE_TYPE_BLK.
 
+* ``rte_vhost_async_dma_unconfigure(dma_id, vchan_id)``
+
+  Clean DMA vChannel finished to use. After this function is called,
+  the specified DMA vChannel should no longer be used by the Vhost library.
+
 Vhost-user Implementations
 --------------------------
 
diff --git a/doc/guides/rel_notes/release_22_11.rst b/doc/guides/rel_notes/release_22_11.rst
index 2da8bc9661..bbd1c5aa9c 100644
--- a/doc/guides/rel_notes/release_22_11.rst
+++ b/doc/guides/rel_notes/release_22_11.rst
@@ -225,6 +225,11 @@ New Features
   sysfs entries to adjust the minimum and maximum uncore frequency values,
   which works on Linux with Intel hardware only.
 
+* **Added DMA vChannel unconfiguration for async vhost.**
+
+  Added support to unconfigure DMA vChannel that is no longer used
+  by the Vhost library.
+
 * **Rewritten pmdinfo script.**
 
   The ``dpdk-pmdinfo.py`` script was rewritten to produce valid JSON only.
diff --git a/lib/vhost/rte_vhost_async.h b/lib/vhost/rte_vhost_async.h
index 1db2a10124..8f190dd44b 100644
--- a/lib/vhost/rte_vhost_async.h
+++ b/lib/vhost/rte_vhost_async.h
@@ -266,6 +266,26 @@ rte_vhost_async_try_dequeue_burst(int vid, uint16_t queue_id,
 	struct rte_mempool *mbuf_pool, struct rte_mbuf **pkts, uint16_t count,
 	int *nr_inflight, int16_t dma_id, uint16_t vchan_id);
 
+/**
+ * @warning
+ * @b EXPERIMENTAL: this API may change, or be removed, without prior notice.
+ *
+ * Unconfigure DMA vChannel in Vhost asynchronous data path.
+ * This function should be called when the specified DMA vChannel is no longer
+ * used by the Vhost library. Before this function is called, make sure there
+ * does not exist in-flight packets in DMA vChannel.
+ *
+ * @param dma_id
+ *  the identifier of DMA device
+ * @param vchan_id
+ *  the identifier of virtual DMA channel
+ * @return
+ *  0 on success, and -1 on failure
+ */
+__rte_experimental
+int
+rte_vhost_async_dma_unconfigure(int16_t dma_id, uint16_t vchan_id);
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/lib/vhost/version.map b/lib/vhost/version.map
index 7a00b65740..0b61870870 100644
--- a/lib/vhost/version.map
+++ b/lib/vhost/version.map
@@ -94,6 +94,9 @@ EXPERIMENTAL {
 	rte_vhost_async_try_dequeue_burst;
 	rte_vhost_driver_get_vdpa_dev_type;
 	rte_vhost_clear_queue;
+
+	# added in 22.11
+	rte_vhost_async_dma_unconfigure;
 };
 
 INTERNAL {
diff --git a/lib/vhost/vhost.c b/lib/vhost/vhost.c
index 8740aa2788..1bb01c2a2e 100644
--- a/lib/vhost/vhost.c
+++ b/lib/vhost/vhost.c
@@ -23,6 +23,7 @@
 
 struct virtio_net *vhost_devices[RTE_MAX_VHOST_DEVICE];
 pthread_mutex_t vhost_dev_lock = PTHREAD_MUTEX_INITIALIZER;
+pthread_mutex_t vhost_dma_lock = PTHREAD_MUTEX_INITIALIZER;
 
 struct vhost_vq_stats_name_off {
 	char name[RTE_VHOST_STATS_NAME_SIZE];
@@ -1844,19 +1845,21 @@ rte_vhost_async_dma_configure(int16_t dma_id, uint16_t vchan_id)
 	void *pkts_cmpl_flag_addr;
 	uint16_t max_desc;
 
+	pthread_mutex_lock(&vhost_dma_lock);
+
 	if (!rte_dma_is_valid(dma_id)) {
 		VHOST_LOG_CONFIG("dma", ERR, "DMA %d is not found.\n", dma_id);
-		return -1;
+		goto error;
 	}
 
 	if (rte_dma_info_get(dma_id, &info) != 0) {
 		VHOST_LOG_CONFIG("dma", ERR, "Fail to get DMA %d information.\n", dma_id);
-		return -1;
+		goto error;
 	}
 
 	if (vchan_id >= info.max_vchans) {
 		VHOST_LOG_CONFIG("dma", ERR, "Invalid DMA %d vChannel %u.\n", dma_id, vchan_id);
-		return -1;
+		goto error;
 	}
 
 	if (!dma_copy_track[dma_id].vchans) {
@@ -1868,7 +1871,7 @@ rte_vhost_async_dma_configure(int16_t dma_id, uint16_t vchan_id)
 			VHOST_LOG_CONFIG("dma", ERR,
 				"Failed to allocate vchans for DMA %d vChannel %u.\n",
 				dma_id, vchan_id);
-			return -1;
+			goto error;
 		}
 
 		dma_copy_track[dma_id].vchans = vchans;
@@ -1877,6 +1880,7 @@ rte_vhost_async_dma_configure(int16_t dma_id, uint16_t vchan_id)
 	if (dma_copy_track[dma_id].vchans[vchan_id].pkts_cmpl_flag_addr) {
 		VHOST_LOG_CONFIG("dma", INFO, "DMA %d vChannel %u already registered.\n",
 			dma_id, vchan_id);
+		pthread_mutex_unlock(&vhost_dma_lock);
 		return 0;
 	}
 
@@ -1894,7 +1898,7 @@ rte_vhost_async_dma_configure(int16_t dma_id, uint16_t vchan_id)
 			rte_free(dma_copy_track[dma_id].vchans);
 			dma_copy_track[dma_id].vchans = NULL;
 		}
-		return -1;
+		goto error;
 	}
 
 	dma_copy_track[dma_id].vchans[vchan_id].pkts_cmpl_flag_addr = pkts_cmpl_flag_addr;
@@ -1902,7 +1906,12 @@ rte_vhost_async_dma_configure(int16_t dma_id, uint16_t vchan_id)
 	dma_copy_track[dma_id].vchans[vchan_id].ring_mask = max_desc - 1;
 	dma_copy_track[dma_id].nr_vchans++;
 
+	pthread_mutex_unlock(&vhost_dma_lock);
 	return 0;
+
+error:
+	pthread_mutex_unlock(&vhost_dma_lock);
+	return -1;
 }
 
 int
@@ -2091,5 +2100,58 @@ int rte_vhost_vring_stats_reset(int vid, uint16_t queue_id)
 	return 0;
 }
 
+int
+rte_vhost_async_dma_unconfigure(int16_t dma_id, uint16_t vchan_id)
+{
+	struct rte_dma_info info;
+	struct rte_dma_stats stats = { 0 };
+
+	pthread_mutex_lock(&vhost_dma_lock);
+
+	if (!rte_dma_is_valid(dma_id)) {
+		VHOST_LOG_CONFIG("dma", ERR, "DMA %d is not found.\n", dma_id);
+		goto error;
+	}
+
+	if (rte_dma_info_get(dma_id, &info) != 0) {
+		VHOST_LOG_CONFIG("dma", ERR, "Fail to get DMA %d information.\n", dma_id);
+		goto error;
+	}
+
+	if (vchan_id >= info.max_vchans || !dma_copy_track[dma_id].vchans ||
+		!dma_copy_track[dma_id].vchans[vchan_id].pkts_cmpl_flag_addr) {
+		VHOST_LOG_CONFIG("dma", ERR, "Invalid channel %d:%u.\n", dma_id, vchan_id);
+		goto error;
+	}
+
+	if (rte_dma_stats_get(dma_id, vchan_id, &stats) != 0) {
+		VHOST_LOG_CONFIG("dma", ERR,
+				 "Failed to get stats for DMA %d vChannel %u.\n", dma_id, vchan_id);
+		goto error;
+	}
+
+	if (stats.submitted - stats.completed != 0) {
+		VHOST_LOG_CONFIG("dma", ERR,
+				 "Do not unconfigure when there are inflight packets.\n");
+		goto error;
+	}
+
+	rte_free(dma_copy_track[dma_id].vchans[vchan_id].pkts_cmpl_flag_addr);
+	dma_copy_track[dma_id].vchans[vchan_id].pkts_cmpl_flag_addr = NULL;
+	dma_copy_track[dma_id].nr_vchans--;
+
+	if (dma_copy_track[dma_id].nr_vchans == 0) {
+		rte_free(dma_copy_track[dma_id].vchans);
+		dma_copy_track[dma_id].vchans = NULL;
+	}
+
+	pthread_mutex_unlock(&vhost_dma_lock);
+	return 0;
+
+error:
+	pthread_mutex_unlock(&vhost_dma_lock);
+	return -1;
+}
+
 RTE_LOG_REGISTER_SUFFIX(vhost_config_log_level, config, INFO);
 RTE_LOG_REGISTER_SUFFIX(vhost_data_log_level, data, WARNING);
-- 
2.17.1


  reply	other threads:[~2022-10-25  8:29 UTC|newest]

Thread overview: 43+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-08-14 14:04 [PATCH v1 0/2] " xuan.ding
2022-08-14 14:04 ` [PATCH v1 1/2] " xuan.ding
2022-08-14 14:04 ` [PATCH v1 2/2] example/vhost: unconfigure DMA vchannel xuan.ding
2022-09-06  5:21 ` [PATCH v2 0/2] vhost: introduce DMA vchannel unconfiguration xuan.ding
2022-09-06  5:21   ` [PATCH v2 1/2] " xuan.ding
2022-09-26  6:06     ` Xia, Chenbo
2022-09-26  6:43       ` Ding, Xuan
2022-09-06  5:21   ` [PATCH v2 2/2] examples/vhost: unconfigure DMA vchannel xuan.ding
2022-09-29  1:32 ` [PATCH v3 0/2] vhost: introduce DMA vchannel unconfiguration xuan.ding
2022-09-29  1:32   ` [PATCH v3 1/2] " xuan.ding
2022-09-29  1:32   ` [PATCH v3 2/2] examples/vhost: unconfigure DMA vchannel xuan.ding
2022-09-29  8:27     ` Xia, Chenbo
2022-10-08  0:38       ` Ding, Xuan
2022-10-13  6:40 ` [PATCH v4 0/2] vhost: introduce DMA vchannel unconfiguration xuan.ding
2022-10-13  6:40   ` [PATCH v4 1/2] " xuan.ding
2022-10-13  8:01     ` Maxime Coquelin
2022-10-13  8:45       ` Ding, Xuan
2022-10-13  6:40   ` [PATCH v4 2/2] examples/vhost: unconfigure DMA vchannel xuan.ding
2022-10-13  8:07     ` Maxime Coquelin
2022-10-13  8:49       ` Ding, Xuan
2022-10-13  9:27 ` [PATCH v5 0/2] vhost: introduce DMA vchannel unconfiguration xuan.ding
2022-10-13  9:27   ` [PATCH v5 1/2] " xuan.ding
2022-10-13  9:27   ` [PATCH v5 2/2] examples/vhost: unconfigure DMA vchannel xuan.ding
2022-10-18 15:22 ` [PATCH v6 0/2] vhost: introduce DMA vchannel unconfiguration xuan.ding
2022-10-18 15:22   ` [PATCH v6 1/2] " xuan.ding
2022-10-19  9:28     ` Xia, Chenbo
2022-10-18 15:22   ` [PATCH v6 2/2] examples/vhost: unconfigure DMA vchannel xuan.ding
2022-10-19  2:57     ` Ling, WeiX
2022-10-20  9:11 ` [PATCH v7 0/2] vhost: introduce DMA vchannel unconfiguration xuan.ding
2022-10-20  9:11   ` [PATCH v7 1/2] " xuan.ding
2022-10-21  8:09     ` Maxime Coquelin
2022-10-21  8:22       ` Ding, Xuan
2022-10-20  9:11   ` [PATCH v7 2/2] examples/vhost: unconfigure DMA vchannel xuan.ding
2022-10-21  8:12     ` Maxime Coquelin
2022-10-25  8:25 ` [PATCH v8 0/2] vhost: introduce DMA vchannel unconfiguration xuan.ding
2022-10-25  8:25   ` xuan.ding [this message]
2022-10-26  5:13     ` [PATCH v8 1/2] " Maxime Coquelin
2022-10-26  9:02     ` Xia, Chenbo
2022-10-25  8:25   ` [PATCH v8 2/2] examples/vhost: unconfigure DMA vchannel xuan.ding
2022-10-25  9:56     ` Ling, WeiX
2022-10-26  5:14     ` Maxime Coquelin
2022-10-26  9:03     ` Xia, Chenbo
2022-10-26  9:07   ` [PATCH v8 0/2] vhost: introduce DMA vchannel unconfiguration Xia, Chenbo

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20221025082540.100618-2-xuan.ding@intel.com \
    --to=xuan.ding@intel.com \
    --cc=chenbo.xia@intel.com \
    --cc=cheng1.jiang@intel.com \
    --cc=dev@dpdk.org \
    --cc=jiayu.hu@intel.com \
    --cc=maxime.coquelin@redhat.com \
    --cc=weix.ling@intel.com \
    --cc=wenwux.ma@intel.com \
    --cc=xingguang.he@intel.com \
    --cc=yuanx.wang@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).