d/numam-dpdk
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

sched: cleanup comments

Browse Source 
Break block comments that exceed common practice for line length.
Shorten wording for obvious things.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
This commit is contained in:
Stephen Hemminger 2015-11-13 09:58:28 -08:00 committed by Thomas Monjalon
parent 33e8113351
commit 9bd5992515
2 changed files with 135 additions and 94 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
lib/librte_sched/rte_sched.c
View File

@ -346,7 +346,8 @@ rte_sched_port_check_params(struct rte_sched_port_params *params)
return -7;
}
/* qsize: non-zero, power of 2, no bigger than 32K (due to 16-bit read/write pointers) */
/* qsize: non-zero, power of 2,
* no bigger than 32K (due to 16-bit read/write pointers) */
for (i = 0; i < RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE; i ++) {
uint16_t qsize = params->qsize[i];
@ -1318,7 +1319,8 @@ rte_sched_port_enqueue(struct rte_sched_port *port, struct rte_mbuf **pkts, uint
#else
/* The enqueue function implements a 4-level pipeline with each stage processing
/*
* The enqueue function implements a 4-level pipeline with each stage processing
* two different packets. The purpose of using a pipeline is to hide the latency
* of prefetching the data structures. The naming convention is presented in the
* diagram below:
@ -1329,7 +1331,7 @@ rte_sched_port_enqueue(struct rte_sched_port *port, struct rte_mbuf **pkts, uint
* ----->|_______|----->|_______|----->|_______|----->|_______|----->
* p01 p11 p21 p31
*
***/
*/
int
rte_sched_port_enqueue(struct rte_sched_port *port, struct rte_mbuf **pkts, uint32_t n_pkts)
{

221
lib/librte_sched/rte_sched.h
View File

@ -42,39 +42,48 @@ extern "C" {
* @file
* RTE Hierarchical Scheduler
*
* The hierarchical scheduler prioritizes the transmission of packets from different
* users and traffic classes according to the Service Level Agreements (SLAs) defined
* for the current network node.
* The hierarchical scheduler prioritizes the transmission of packets
* from different users and traffic classes according to the Service
* Level Agreements (SLAs) defined for the current network node.
*
* The scheduler supports thousands of packet queues grouped under a 5-level hierarchy:
* The scheduler supports thousands of packet queues grouped under a
* 5-level hierarchy:
* 1. Port:
* - Typical usage: output Ethernet port;
* - Multiple ports are scheduled in round robin order with equal priority;
* - Multiple ports are scheduled in round robin order with
* equal priority;
* 2. Subport:
* - Typical usage: group of users;
* - Traffic shaping using the token bucket algorithm (one bucket per subport);
* - Traffic shaping using the token bucket algorithm
* (one bucket per subport);
* - Upper limit enforced per traffic class at subport level;
* - Lower priority traffic classes able to reuse subport bandwidth currently
* unused by higher priority traffic classes of the same subport;
* - When any subport traffic class is oversubscribed (configuration time
* event), the usage of subport member pipes with high demand for that
* traffic class pipes is truncated to a dynamically adjusted value with no
* - Lower priority traffic classes able to reuse subport
* bandwidth currently unused by higher priority traffic
* classes of the same subport;
* - When any subport traffic class is oversubscribed
* (configuration time event), the usage of subport member
* pipes with high demand for thattraffic class pipes is
* truncated to a dynamically adjusted value with no
* impact to low demand pipes;
* 3. Pipe:
* - Typical usage: individual user/subscriber;
* - Traffic shaping using the token bucket algorithm (one bucket per pipe);
* - Traffic shaping using the token bucket algorithm
* (one bucket per pipe);
* 4. Traffic class:
* - Traffic classes of the same pipe handled in strict priority order;
* - Traffic classes of the same pipe handled in strict
* priority order;
* - Upper limit enforced per traffic class at the pipe level;
* - Lower priority traffic classes able to reuse pipe bandwidth currently
* unused by higher priority traffic classes of the same pipe;
* - Lower priority traffic classes able to reuse pipe
* bandwidth currently unused by higher priority traffic
* classes of the same pipe;
* 5. Queue:
* - Typical usage: queue hosting packets from one or multiple connections
* of same traffic class belonging to the same user;
* - Weighted Round Robin (WRR) is used to service the queues within same
* pipe traffic class.
* - Typical usage: queue hosting packets from one or
* multiple connections of same traffic class belonging to
* the same user;
* - Weighted Round Robin (WRR) is used to service the
* queues within same pipe traffic class.
*
***/
*/
#include <sys/types.h>
#include <rte_mbuf.h>
@ -85,7 +94,9 @@ extern "C" {
#include "rte_red.h"
#endif
/** Number of traffic classes per pipe (as well as subport). Cannot be changed. */
/** Number of traffic classes per pipe (as well as subport).
* Cannot be changed.
*/
#define RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE 4
/** Number of queues per pipe traffic class. Cannot be changed. */
@ -96,100 +107,123 @@ extern "C" {
(RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE * \
RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS)
/** Maximum number of pipe profiles that can be defined per port. Compile-time configurable.*/
/** Maximum number of pipe profiles that can be defined per port.
* Compile-time configurable.
*/
#ifndef RTE_SCHED_PIPE_PROFILES_PER_PORT
#define RTE_SCHED_PIPE_PROFILES_PER_PORT 256
#endif
/** Ethernet framing overhead. Overhead fields per Ethernet frame:
1. Preamble: 7 bytes;
2. Start of Frame Delimiter (SFD): 1 byte;
3. Frame Check Sequence (FCS): 4 bytes;
4. Inter Frame Gap (IFG): 12 bytes.
The FCS is considered overhead only if not included in the packet length (field pkt_len
of struct rte_mbuf). */
/*
* Ethernet framing overhead. Overhead fields per Ethernet frame:
* 1. Preamble: 7 bytes;
* 2. Start of Frame Delimiter (SFD): 1 byte;
* 3. Frame Check Sequence (FCS): 4 bytes;
* 4. Inter Frame Gap (IFG): 12 bytes.
*
* The FCS is considered overhead only if not included in the packet
* length (field pkt_len of struct rte_mbuf).
*/
#ifndef RTE_SCHED_FRAME_OVERHEAD_DEFAULT
#define RTE_SCHED_FRAME_OVERHEAD_DEFAULT 24
#endif
/** Subport configuration parameters. The period and credits_per_period parameters are measured
in bytes, with one byte meaning the time duration associated with the transmission of one byte
on the physical medium of the output port, with pipe or pipe traffic class rate (measured as
percentage of output port rate) determined as credits_per_period divided by period. One credit
represents one byte. */
/*
* Subport configuration parameters. The period and credits_per_period
* parameters are measured in bytes, with one byte meaning the time
* duration associated with the transmission of one byte on the
* physical medium of the output port, with pipe or pipe traffic class
* rate (measured as percentage of output port rate) determined as
* credits_per_period divided by period. One credit represents one
* byte.
*/
struct rte_sched_subport_params {
/* Subport token bucket */
uint32_t tb_rate; /**< Subport token bucket rate (measured in bytes per second) */
uint32_t tb_size; /**< Subport token bucket size (measured in credits) */
uint32_t tb_rate; /**< Rate (measured in bytes per second) */
uint32_t tb_size; /**< Size (measured in credits) */
/* Subport traffic classes */
uint32_t tc_rate[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE]; /**< Subport traffic class rates (measured in bytes per second) */
uint32_t tc_period; /**< Enforcement period for traffic class rates (measured in milliseconds) */
uint32_t tc_rate[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
/**< Traffic class rates (measured in bytes per second) */
uint32_t tc_period;
/**< Enforcement period for rates (measured in milliseconds) */
};
/** Subport statistics */
struct rte_sched_subport_stats {
/* Packets */
uint32_t n_pkts_tc[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE]; /**< Number of packets successfully written to current
subport for each traffic class */
uint32_t n_pkts_tc_dropped[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE]; /**< Number of packets dropped by the current
subport for each traffic class due to subport queues being full or congested*/
uint32_t n_pkts_tc[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
/**< Number of packets successfully written */
uint32_t n_pkts_tc_dropped[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
/**< Number of packets dropped */
/* Bytes */
uint32_t n_bytes_tc[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE]; /**< Number of bytes successfully written to current
subport for each traffic class*/
uint32_t n_bytes_tc_dropped[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE]; /**< Number of bytes dropped by the current
subport for each traffic class due to subport queues being full or congested */
uint32_t n_bytes_tc[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
/**< Number of bytes successfully written for each traffic class */
uint32_t n_bytes_tc_dropped[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
/**< Number of bytes dropped for each traffic class */
};
/** Pipe configuration parameters. The period and credits_per_period parameters are measured
in bytes, with one byte meaning the time duration associated with the transmission of one byte
on the physical medium of the output port, with pipe or pipe traffic class rate (measured as
percentage of output port rate) determined as credits_per_period divided by period. One credit
represents one byte. */
/*
* Pipe configuration parameters. The period and credits_per_period
* parameters are measured in bytes, with one byte meaning the time
* duration associated with the transmission of one byte on the
* physical medium of the output port, with pipe or pipe traffic class
* rate (measured as percentage of output port rate) determined as
* credits_per_period divided by period. One credit represents one
* byte.
*/
struct rte_sched_pipe_params {
/* Pipe token bucket */
uint32_t tb_rate; /**< Pipe token bucket rate (measured in bytes per second) */
uint32_t tb_size; /**< Pipe token bucket size (measured in credits) */
uint32_t tb_rate; /**< Rate (measured in bytes per second) */
uint32_t tb_size; /**< Size (measured in credits) */
/* Pipe traffic classes */
uint32_t tc_rate[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE]; /**< Pipe traffic class rates (measured in bytes per second) */
uint32_t tc_period; /**< Enforcement period for pipe traffic class rates (measured in milliseconds) */
uint32_t tc_rate[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
/**< Traffic class rates (measured in bytes per second) */
uint32_t tc_period;
/**< Enforcement period (measured in milliseconds) */
#ifdef RTE_SCHED_SUBPORT_TC_OV
uint8_t tc_ov_weight; /**< Weight for the current pipe in the event of subport traffic class 3 oversubscription */
uint8_t tc_ov_weight; /**< Weight Traffic class 3 oversubscription */
#endif
/* Pipe queues */
uint8_t wrr_weights[RTE_SCHED_QUEUES_PER_PIPE]; /**< WRR weights for the queues of the current pipe */
uint8_t wrr_weights[RTE_SCHED_QUEUES_PER_PIPE]; /**< WRR weights */
};
/** Queue statistics */
struct rte_sched_queue_stats {
/* Packets */
uint32_t n_pkts; /**< Number of packets successfully written to current queue */
uint32_t n_pkts_dropped; /**< Number of packets dropped due to current queue being full or congested */
uint32_t n_pkts; /**< Packets successfully written */
uint32_t n_pkts_dropped; /**< Packets dropped */
/* Bytes */
uint32_t n_bytes; /**< Number of bytes successfully written to current queue */
uint32_t n_bytes_dropped; /**< Number of bytes dropped due to current queue being full or congested */
uint32_t n_bytes; /**< Bytes successfully written */
uint32_t n_bytes_dropped; /**< Bytes dropped */
};
/** Port configuration parameters. */
struct rte_sched_port_params {
const char *name; /**< Literal string to be associated to the current port scheduler instance */
int socket; /**< CPU socket ID where the memory for port scheduler should be allocated */
uint32_t rate; /**< Output port rate (measured in bytes per second) */
uint32_t mtu; /**< Maximum Ethernet frame size (measured in bytes). Should not include the framing overhead. */
uint32_t frame_overhead; /**< Framing overhead per packet (measured in bytes) */
uint32_t n_subports_per_port; /**< Number of subports for the current port scheduler instance*/
uint32_t n_pipes_per_subport; /**< Number of pipes for each port scheduler subport */
uint16_t qsize[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE]; /**< Packet queue size for each traffic class. All queues
within the same pipe traffic class have the same size. Queues from
different pipes serving the same traffic class have the same size. */
struct rte_sched_pipe_params *pipe_profiles; /**< Pipe profile table defined for current port scheduler instance.
Every pipe of the current port scheduler is configured using one of the
profiles from this table. */
uint32_t n_pipe_profiles; /**< Number of profiles in the pipe profile table */
const char *name; /**< String to be associated */
int socket; /**< CPU socket ID */
uint32_t rate; /**< Output port rate
* (measured in bytes per second) */
uint32_t mtu; /**< Maximum Ethernet frame size
* (measured in bytes).
* Should not include the framing overhead. */
uint32_t frame_overhead; /**< Framing overhead per packet
* (measured in bytes) */
uint32_t n_subports_per_port; /**< Number of subports */
uint32_t n_pipes_per_subport; /**< Number of pipes per subport */
uint16_t qsize[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
/**< Packet queue size for each traffic class.
* All queues within the same pipe traffic class have the same
* size. Queues from different pipes serving the same traffic
* class have the same size. */
struct rte_sched_pipe_params *pipe_profiles;
/**< Pipe profile table.
* Every pipe is configured using one of the profiles from this table. */
uint32_t n_pipe_profiles; /**< Profiles in the pipe profile table */
#ifdef RTE_SCHED_RED
struct rte_red_params red_params[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE][e_RTE_METER_COLORS]; /**< RED parameters */
#endif
@ -306,7 +340,8 @@ rte_sched_subport_read_stats(struct rte_sched_port *port,
* Pointer to pre-allocated subport statistics structure where the statistics
* counters should be stored
* @param qlen
* Pointer to pre-allocated variable where the current queue length should be stored.
* Pointer to pre-allocated variable where the current queue length
* should be stored.
* @return
* 0 upon success, error code otherwise
*/
@ -317,8 +352,8 @@ rte_sched_queue_read_stats(struct rte_sched_port *port,
uint16_t *qlen);
/**
* Scheduler hierarchy path write to packet descriptor. Typically called by the
* packet classification stage.
* Scheduler hierarchy path write to packet descriptor. Typically
* called by the packet classification stage.
*
* @param pkt
* Packet descriptor handle
@ -339,9 +374,10 @@ rte_sched_port_pkt_write(struct rte_mbuf *pkt,
uint32_t queue, enum rte_meter_color color);
/**
* Scheduler hierarchy path read from packet descriptor (struct rte_mbuf). Typically
* called as part of the hierarchical scheduler enqueue operation. The subport,
* pipe, traffic class and queue parameters need to be pre-allocated by the caller.
* Scheduler hierarchy path read from packet descriptor (struct
* rte_mbuf). Typically called as part of the hierarchical scheduler
* enqueue operation. The subport, pipe, traffic class and queue
* parameters need to be pre-allocated by the caller.
*
* @param pkt
* Packet descriptor handle
@ -364,12 +400,13 @@ enum rte_meter_color
rte_sched_port_pkt_read_color(const struct rte_mbuf *pkt);
/**
* Hierarchical scheduler port enqueue. Writes up to n_pkts to port scheduler and
* returns the number of packets actually written. For each packet, the port scheduler
* queue to write the packet to is identified by reading the hierarchy path from the
* packet descriptor; if the queue is full or congested and the packet is not written
* to the queue, then the packet is automatically dropped without any action required
* from the caller.
* Hierarchical scheduler port enqueue. Writes up to n_pkts to port
* scheduler and returns the number of packets actually written. For
* each packet, the port scheduler queue to write the packet to is
* identified by reading the hierarchy path from the packet
* descriptor; if the queue is full or congested and the packet is not
* written to the queue, then the packet is automatically dropped
* without any action required from the caller.
*
* @param port
* Handle to port scheduler instance
@ -384,14 +421,16 @@ int
rte_sched_port_enqueue(struct rte_sched_port *port, struct rte_mbuf **pkts, uint32_t n_pkts);
/**
* Hierarchical scheduler port dequeue. Reads up to n_pkts from the port scheduler
* and stores them in the pkts array and returns the number of packets actually read.
* The pkts array needs to be pre-allocated by the caller with at least n_pkts entries.
* Hierarchical scheduler port dequeue. Reads up to n_pkts from the
* port scheduler and stores them in the pkts array and returns the
* number of packets actually read. The pkts array needs to be
* pre-allocated by the caller with at least n_pkts entries.
*
* @param port
* Handle to port scheduler instance
* @param pkts
* Pre-allocated packet descriptor array where the packets dequeued from the port
* Pre-allocated packet descriptor array where the packets dequeued
* from the port
* scheduler should be stored
* @param n_pkts
* Number of packets to dequeue from the port scheduler