[dpdk-dev] [PATCH v3] eal: refactor rte_eal_init into sub-functions

[dpdk-dev] [PATCH v3] eal: refactor rte_eal_init into sub-functions

[Rahul Gupta]

From: Rahul Gupta <rahulgupt@microsoft.com>

In continuation to the following email, I am sending this patch.
(https://inbox.dpdk.org/dev/20231110172523.GA17466@microsoft.com/)

Initialization requires rte_eal_init + rte_pktmbuf_pool_create which
can consume a total time of 500-600 ms:
a) For many devices FLR may take a significant chunk of time
   (200-250 ms in our use-case), this FLR is triggered during device
   probe in rte_eal_init().
b) rte_pktmbuf_pool_create() can consume up to 300-350 ms for
applications that require huge memory.

This cost is incurred on each restart (which happens in our use-case
during binary updates for servicing).
This patch provides an optimization using pthreads that applications
can use and which can save 200-230ms.

In this patch, rte_eal_init() is refactored into two parts-
a) 1st part is dependent code ie- it’s a perquisite of the FLR and
   mempool creation. So this code needs to be executed before any
   pthreads. Its named as rte_eal_init_setup()
b) 2nd part of code is independent code ie- it can execute in parallel
   to mempool creation in a pthread. Its named as rte_eal_init_async().

Existing applications requires to just call-
rte_eal_init_wait_async_complete() after rte_eal_init() unless they wish
to leverage the optimization.

If the application wants to leverage this optimization, then it should
create a thread using rte_eal_remote_launch() to schedule a task it would
like todo in parallel rte_eal_init_async(), this task can be a mbuf pool
creation using- rte_pktmbuf_pool_create()

After both the above tasks, if next operations require completion of
above thread a), then user can use rte_eal_init_wait_async_complete(),
or if user wants to just check status of that thread, then use-
rte_eal_init_async_done()
---
v2: Address Stephen Hemminger's comment
---
v3: address support for single lcore

 app/test-pmd/testpmd.c    | 22 +++++++++++++++++
 lib/eal/include/rte_eal.h | 48 ++++++++++++++++++++++++++++++++++++
 lib/eal/linux/eal.c       | 52 ++++++++++++++++++++++++++++++++++++---
 lib/eal/version.map       |  6 +++++
 4 files changed, 125 insertions(+), 3 deletions(-)

diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index 9e4e99e53b..345f3c1c04 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -4531,6 +4531,8 @@ main(int argc, char** argv)
 	portid_t port_id;
 	uint16_t count;
 	int ret;
+	int lcore_id = 0;
+	int main_lcore_id = 0;
 
 #ifdef RTE_EXEC_ENV_WINDOWS
 	signal(SIGINT, signal_handler);
@@ -4555,6 +4557,26 @@ main(int argc, char** argv)
 		rte_exit(EXIT_FAILURE, "Cannot init EAL: %s\n",
 			 rte_strerror(rte_errno));
 
+	main_lcore_id = rte_lcore_id();
+	lcore_id = rte_get_next_lcore(main_lcore_id, 0, 1);
+	/* Gives status of rte_eal_init_async() */
+	if (main_lcore_id != lcore_id)
+		while (rte_eal_init_async_done(lcore_id) == 0)
+			;
+
+	/*
+	 * Use rte_eal_init_wait_async_complete() to get return value of
+	 * rte_eal_init_async().
+	 * Or
+	 * if testpmd application dont want to know progress/status of
+	 * rte_eal_init_async() and just want to wait till it finishes
+	 * then use following function.
+	 */
+	ret = rte_eal_init_wait_async_complete();
+	if (ret < 0)
+		rte_exit(EXIT_FAILURE, "Cannot init EAL: "
+			 "rte_eal_init_async() failed: %s\n",
+			 strerror(ret));
 	/* allocate port structures, and init them */
 	init_port();
 
diff --git a/lib/eal/include/rte_eal.h b/lib/eal/include/rte_eal.h
index c2256f832e..91ed80e292 100644
--- a/lib/eal/include/rte_eal.h
+++ b/lib/eal/include/rte_eal.h
@@ -111,6 +111,54 @@ int rte_eal_iopl_init(void);
  */
 int rte_eal_init(int argc, char **argv);
 
+/**
+ * Initialize the Environment Abstraction Layer (EAL): Initial setup
+ *
+ * Its called from rte_eal_init() on MASTER lcore only and must NOT be directly
+ * called by user application.
+ * The driver dependent code is present in this function, ie before calling any other
+ * function eal library function this function must be complete successfully.
+ *
+ * return value is same as rte_eal_init().
+ */
+__rte_experimental int rte_eal_init_setup(int argc, char **argv);
+
+/**
+ * Initialize the Environment Abstraction Layer (EAL): FLR and probe device
+ *
+ * Its thread is forked by rte_eal_init() and must NOT be directly called by user application.
+ * Launched on next available worker lcore.
+ * In this function initialisation needed for memory pool creation is completed,
+ * so this code can be executed in parallel to non device related operations
+ * like mbuf pool creation.
+ *
+ * return value is same as rte_eal_init().
+ */
+__rte_experimental int rte_eal_init_async(__attribute__((unused)) void *arg);
+
+/**
+ * Initialize the Environment Abstraction Layer (EAL): Indication of rte_eal_init() completion
+ *
+ * It waits for rte_eal_init_async() to finish. It MUST be called from application,
+ * when a thread join is needed. Typically application will call this function after
+ * it performs all device independent operation (like mbuf pool creation) on initial lcore.
+ *
+ * return value is same as rte_eal_init().
+ */
+__rte_experimental int rte_eal_init_wait_async_complete(void);
+
+/**
+ * Initialize the Environment Abstraction Layer (EAL): Indication of rte_eal_init() completion
+ *
+ * It shows status of rte_eal_init_async() ie the function is executing or completed.
+ * It MUST be called from application,
+ * Typically an application will call this function when it wants to know status of
+ * rte_eal_init_async() (ie FLR and probe thread).
+ *
+ * return value is same as rte_eal_init().
+ */
+__rte_experimental int rte_eal_init_async_done(int lcore_id);
+
 /**
  * Clean up the Environment Abstraction Layer (EAL)
  *
diff --git a/lib/eal/linux/eal.c b/lib/eal/linux/eal.c
index fd422f1f62..92697f3c8c 100644
--- a/lib/eal/linux/eal.c
+++ b/lib/eal/linux/eal.c
@@ -962,9 +962,8 @@ eal_worker_thread_create(unsigned int lcore_id)
 	return ret;
 }
 
-/* Launch threads, called at application init(). */
-int
-rte_eal_init(int argc, char **argv)
+__rte_experimental
+int rte_eal_init_setup(int argc, char **argv)
 {
 	int i, fctret, ret;
 	static RTE_ATOMIC(uint32_t) run_once;
@@ -1268,7 +1267,15 @@ rte_eal_init(int argc, char **argv)
 	 */
 	rte_eal_mp_remote_launch(sync_func, NULL, SKIP_MAIN);
 	rte_eal_mp_wait_lcore();
+	return fctret;
+}
 
+__rte_experimental int
+rte_eal_init_async(__attribute__((unused)) void *arg)
+{
+	int ret = 0;
+	struct internal_config *internal_conf =
+		eal_get_internal_configuration();
 	/* initialize services so vdevs register service during bus_probe. */
 	ret = rte_service_init();
 	if (ret) {
@@ -1322,6 +1329,45 @@ rte_eal_init(int argc, char **argv)
 
 	eal_mcfg_complete();
 
+	return 0;
+}
+
+/*
+ * waits until function executing on given lcore finishes.
+ * returns value returned by the function executing on that lcore.
+ */
+__rte_experimental int
+rte_eal_init_wait_async_complete(void)
+{
+	int lcore_id = -1;
+	lcore_id = rte_lcore_id();
+	lcore_id = rte_get_next_lcore(lcore_id, 0, 1);
+	int ret = rte_eal_wait_lcore(lcore_id);
+	return ret;
+}
+
+/*
+ * returns current status of execution on a given lcore
+ */
+__rte_experimental int
+rte_eal_init_async_done(int lcore_id)
+{
+	int ret = (lcore_config[lcore_id].state);
+	return (ret == WAIT);
+}
+
+/* Launch threads, called at application init(). */
+int
+rte_eal_init(int argc, char **argv)
+{
+	int fctret = rte_eal_init_setup(argc, argv);	/* initial lcore*/
+	if (fctret < 0)
+		return fctret;
+	int lcore_id = -1;
+	lcore_id = rte_lcore_id();
+	lcore_id = rte_get_next_lcore(lcore_id, 0, 1);
+	/* running on a worker lcore */
+	rte_eal_remote_launch(rte_eal_init_async, NULL, lcore_id);
 	return fctret;
 }
 
diff --git a/lib/eal/version.map b/lib/eal/version.map
index 5e0cd47c82..8cd843109f 100644
--- a/lib/eal/version.map
+++ b/lib/eal/version.map
@@ -393,6 +393,12 @@ EXPERIMENTAL {
 	# added in 23.07
 	rte_memzone_max_get;
 	rte_memzone_max_set;
+
+	# added in 24.01
+	rte_eal_init_async;
+	rte_eal_init_async_done;
+	rte_eal_init_setup;
+	rte_eal_init_wait_async_complete;
 };
 
 INTERNAL {
-- 
2.25.1

Re: [dpdk-dev] [PATCH v3] eal: refactor rte_eal_init into sub-functions

[Bruce Richardson]

On Fri, Jan 19, 2024 at 08:58:17AM -0800, Rahul Gupta wrote:
> From: Rahul Gupta <rahulgupt@microsoft.com>
> 
> In continuation to the following email, I am sending this patch.
> (https://inbox.dpdk.org/dev/20231110172523.GA17466@microsoft.com/)
> 
> Initialization requires rte_eal_init + rte_pktmbuf_pool_create which
> can consume a total time of 500-600 ms:
> a) For many devices FLR may take a significant chunk of time
>    (200-250 ms in our use-case), this FLR is triggered during device
>    probe in rte_eal_init().
> b) rte_pktmbuf_pool_create() can consume up to 300-350 ms for
> applications that require huge memory.
> 
> This cost is incurred on each restart (which happens in our use-case
> during binary updates for servicing).
> This patch provides an optimization using pthreads that applications
> can use and which can save 200-230ms.
> 
> In this patch, rte_eal_init() is refactored into two parts-
> a) 1st part is dependent code ie- it’s a perquisite of the FLR and
>    mempool creation. So this code needs to be executed before any
>    pthreads. Its named as rte_eal_init_setup()
> b) 2nd part of code is independent code ie- it can execute in parallel
>    to mempool creation in a pthread. Its named as rte_eal_init_async().
> 
> Existing applications requires to just call-
> rte_eal_init_wait_async_complete() after rte_eal_init() unless they wish
> to leverage the optimization.
> 
This part here is a bit problematic. For something as ubiquitous as
rte_eal_init(), we should not be requiring existing application change.
Instead, can you create a new init function to replace rte_eal_init() in
the async case - and then eal_init can call that and then the
async_complete function, to avoid duplicating code.

The goal should be to have existing apps unaffected, but to allow a path to
faster startup for those that need it. In the original email you linked to
above, we had such a proposal: rte_eal_init_async() &
rte_eal_init_async_done().

Thanks,
/Bruce

Re: [dpdk-dev] [PATCH v3] eal: refactor rte_eal_init into sub-functions

[Stephen Hemminger]

On Fri, 19 Jan 2024 08:58:17 -0800
Rahul Gupta <rahulgupt@linux.microsoft.com> wrote:

> diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
> index 9e4e99e53b..345f3c1c04 100644
> --- a/app/test-pmd/testpmd.c
> +++ b/app/test-pmd/testpmd.c
> @@ -4531,6 +4531,8 @@ main(int argc, char** argv)
>  	portid_t port_id;
>  	uint16_t count;
>  	int ret;
> +	int lcore_id = 0;
> +	int main_lcore_id = 0;

These do not need to be initialized, there is assignment
before first usage.

There already exists rte_get_main_lcore() please use that