DPDK patches and discussions
 help / color / mirror / Atom feed
From: Shreyansh Jain <shreyansh.jain@nxp.com>
To: thomas@monjalon.net, dev@dpdk.org
Cc: hemant.agrawal@nxp.com, akhil.goyal@nxp.com,
	anatoly.burakov@intel.com,
	Shreyansh Jain <shreyansh.jain@nxp.com>
Subject: [dpdk-dev] [PATCH 2/3] bus/fslmc: optimize physical to virtual address searching
Date: Fri, 27 Apr 2018 21:55:55 +0530	[thread overview]
Message-ID: <20180427162556.22781-3-shreyansh.jain@nxp.com> (raw)
In-Reply-To: <20180427162556.22781-1-shreyansh.jain@nxp.com>

With Hotplugging memory support, the order of memseg has been changed
from physically contiguous to virtual contiguous. FSLMC bus and dpaa2
drivers depend on PA to VA address conversion when in Physical
addressing mode.

This patch creates a list of blocks requested to be pinned to the
DPAA2 mempool. For searching physical addresses, it is expected that
it would belong to this list (from hardware pool) and hence it is
less expensive than memseg walks. Though, this has marginal impact on
performance vis-a-vis legacy mode with physically contiguous memsegs.

Signed-off-by: Shreyansh Jain <shreyansh.jain@nxp.com>

--

An optimized algorithm is being worked upon based on some recent patches
in hotplugging. That would improve/recover the performance. Until that
time, this patch is to be treated a stop-gap solution.
---
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h  | 23 +++++++++++++++++
 drivers/mempool/dpaa2/dpaa2_hw_mempool.c | 43 ++++++++++++++++++++++++++++++++
 2 files changed, 66 insertions(+)

diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index c76393d45..da6e639dc 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -254,15 +254,38 @@ enum qbman_fd_format {
  */
 #define DPAA2_EQ_RESP_ALWAYS		1
 
+/* Various structures representing contiguous memory maps */
+struct dpaa2_memseg {
+	TAILQ_ENTRY(dpaa2_memseg) next;
+	char *vaddr;
+	rte_iova_t iova;
+	size_t len;
+};
+
+TAILQ_HEAD(dpaa2_memseg_list, dpaa2_memseg);
+extern struct dpaa2_memseg_list dpaa2_memsegs;
+
 #ifdef RTE_LIBRTE_DPAA2_USE_PHYS_IOVA
 extern uint8_t dpaa2_virt_mode;
 static void *dpaa2_mem_ptov(phys_addr_t paddr) __attribute__((unused));
 /* todo - this is costly, need to write a fast coversion routine */
 static void *dpaa2_mem_ptov(phys_addr_t paddr)
 {
+	struct dpaa2_memseg *ms;
+
 	if (dpaa2_virt_mode)
 		return (void *)(size_t)paddr;
 
+	/* Check if the address is already part of the memseg list internally
+	 * maintained by the dpaa2 driver.
+	 */
+	TAILQ_FOREACH(ms, &dpaa2_memsegs, next) {
+		if (paddr >= ms->iova && paddr <
+			ms->iova + ms->len)
+			return RTE_PTR_ADD(ms->vaddr, (paddr - ms->iova));
+	}
+
+	/* If not, Fallback to full memseg list searching */
 	return rte_mem_iova2virt(paddr);
 }
 
diff --git a/drivers/mempool/dpaa2/dpaa2_hw_mempool.c b/drivers/mempool/dpaa2/dpaa2_hw_mempool.c
index ce7a4c577..4c44c33cc 100644
--- a/drivers/mempool/dpaa2/dpaa2_hw_mempool.c
+++ b/drivers/mempool/dpaa2/dpaa2_hw_mempool.c
@@ -32,6 +32,13 @@
 struct dpaa2_bp_info rte_dpaa2_bpid_info[MAX_BPID];
 static struct dpaa2_bp_list *h_bp_list;
 
+/* List of all the memseg information locally maintained in dpaa2 driver. This
+ * is to optimize the PA_to_VA searches until a better mechanism (algo) is
+ * available.
+ */
+struct dpaa2_memseg_list dpaa2_memsegs
+	= TAILQ_HEAD_INITIALIZER(dpaa2_memsegs);
+
 /* Dynamic logging identified for mempool */
 int dpaa2_logtype_mempool;
 
@@ -358,6 +365,41 @@ rte_hw_mbuf_get_count(const struct rte_mempool *mp)
 	return num_of_bufs;
 }
 
+static int
+dpaa2_populate(struct rte_mempool *mp, unsigned int max_objs,
+	      void *vaddr, rte_iova_t paddr, size_t len,
+	      rte_mempool_populate_obj_cb_t *obj_cb, void *obj_cb_arg)
+{
+	struct dpaa2_memseg *ms;
+
+	/* For each memory chunk pinned to the Mempool, a linked list of the
+	 * represeted memsegs is created for searching when PA to VA
+	 * conversion is required.
+	 */
+	ms = rte_zmalloc(NULL, sizeof(struct dpaa2_memseg), 0);
+	if (!ms) {
+		DPAA2_MEMPOOL_ERR("Unable to allocate internal memory.");
+		DPAA2_MEMPOOL_WARN("Fast Physical to Virtual Addr translation would not be available.");
+		/* If the element is not added, it would only lead to failure
+		 * in searching for the element and the logic would Fallback
+		 * to traditional DPDK memseg traversal code. So, this is not
+		 * a blocking error - but, error would be printed on screen.
+		 */
+		return 0;
+	}
+
+	ms->vaddr = vaddr;
+	ms->iova = paddr;
+	ms->len = len;
+	/* Head insertions are generally faster than tail insertions as the
+	 * buffers pinned are picked from rear end.
+	 */
+	TAILQ_INSERT_HEAD(&dpaa2_memsegs, ms, next);
+
+	return rte_mempool_op_populate_default(mp, max_objs, vaddr, paddr, len,
+					       obj_cb, obj_cb_arg);
+}
+
 struct rte_mempool_ops dpaa2_mpool_ops = {
 	.name = DPAA2_MEMPOOL_OPS_NAME,
 	.alloc = rte_hw_mbuf_create_pool,
@@ -365,6 +407,7 @@ struct rte_mempool_ops dpaa2_mpool_ops = {
 	.enqueue = rte_hw_mbuf_free_bulk,
 	.dequeue = rte_dpaa2_mbuf_alloc_bulk,
 	.get_count = rte_hw_mbuf_get_count,
+	.populate = dpaa2_populate,
 };
 
 MEMPOOL_REGISTER_OPS(dpaa2_mpool_ops);
-- 
2.14.1

  parent reply	other threads:[~2018-04-27 16:10 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-04-27 16:25 [dpdk-dev] [PATCH 0/3] Optimization for DPAA/DPAA2 for PA/VA Addressing Shreyansh Jain
2018-04-27 16:25 ` [dpdk-dev] [PATCH 1/3] crypto/dpaa_sec: remove ctx based offset for PA-VA conversion Shreyansh Jain
2018-04-27 16:25 ` Shreyansh Jain [this message]
2018-04-27 16:25 ` [dpdk-dev] [PATCH 3/3] bus/dpaa: optimize physical to virtual address searching Shreyansh Jain
2018-04-27 17:20 ` [dpdk-dev] [PATCH v2 0/3] Optimization for DPAA/DPAA2 for PA/VA Addressing Shreyansh Jain
2018-04-27 17:20   ` [dpdk-dev] [PATCH v2 1/3] crypto/dpaa_sec: remove ctx based offset for PA-VA conversion Shreyansh Jain
2018-04-27 17:20   ` [dpdk-dev] [PATCH v2 2/3] bus/fslmc: optimize physical to virtual address searching Shreyansh Jain
2018-04-27 18:49     ` Thomas Monjalon
2018-04-27 19:24       ` Thomas Monjalon
2018-04-27 17:20   ` [dpdk-dev] [PATCH v2 3/3] bus/dpaa: " Shreyansh Jain
2018-04-27 19:32     ` Thomas Monjalon
2018-04-27 19:38   ` [dpdk-dev] [PATCH v2 0/3] Optimization for DPAA/DPAA2 for PA/VA Addressing Thomas Monjalon

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=20180427162556.22781-3-shreyansh.jain@nxp.com \
    --to=shreyansh.jain@nxp.com \
    --cc=akhil.goyal@nxp.com \
    --cc=anatoly.burakov@intel.com \
    --cc=dev@dpdk.org \
    --cc=hemant.agrawal@nxp.com \
    --cc=thomas@monjalon.net \
    /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).