freebsd-dev/sys/net/netisr.h

251 lines
9.3 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
* Copyright (c) 2007-2009 Robert N. M. Watson
Rework netisr policy mechanism so that per-protocol dispatch policies can be represented: - A single policy namespace is defined, consisting of four possible policies: "default" to use the global default, "deferred" to force deferred dispatch, "direct" to employ direct dispatch where possible, and "hybrid" which makes a dynamic decision based on CPU affinity, ordering, etc. Routines are implemented to convert between strings and an integer namespace. - A new global variable, netisr_dispatch_policy, subsumes existing global variables for direct dispatch, forced direct dispatch, etc, and is used for explicit policy interpretation and composition. Old variables remain so that they can be exported by legacy sysctls for use by old netstat(1) binaries. A new sysctl and tunable, netisr.dispatch.policy, accepts the above strings for specifying a global policy default. - The protocol registration structure, netisr_handler, grows an nh_dispatch field, which accepts a per-policy policy override. The default value is '0', which corresponds to "default", meaning that protocols will accept the global default policy unless otherwise specified. - Policies are now interpreted and composed explicitly at various points in packet dispatch; protocol policies override global policies. - Protocols grow the ability to express a non-opinion about affinity even when implenting m2cpuid by returning NETISR_CPUID_NONE. In that case, the framework falls back on source ordering, rather than simply using the current CPU. These changes are in support of allowing link layer re-dispatch based on RSS or similar hashes provided by NICs, especially in the case where the number of hardware receive queues matches hardware core count, rather than hardware thread count, requiring further software redistributeon. (i.e., on RMI XLR). MFC after: 3 weeks Reviewed by: bz Sponsored by: Juniper Networks, Inc.
2011-05-24 12:34:19 +00:00
* Copyright (c) 2010-2011 Juniper Networks, Inc.
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
* All rights reserved.
1994-05-24 10:09:53 +00:00
*
* This software was developed by Robert N. M. Watson under contract
* to Juniper Networks, Inc.
*
1994-05-24 10:09:53 +00:00
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
1994-05-24 10:09:53 +00:00
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
1994-05-24 10:09:53 +00:00
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
1999-08-28 01:08:13 +00:00
* $FreeBSD$
1994-05-24 10:09:53 +00:00
*/
1994-08-21 05:11:48 +00:00
#ifndef _NET_NETISR_H_
#define _NET_NETISR_H_
1994-05-24 10:09:53 +00:00
/*
* The netisr (network interrupt service routine) provides a deferred
* execution evironment in which (generally inbound) network processing can
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
* take place. Protocols register handlers which will be executed directly,
* or via deferred dispatch, depending on the circumstances.
1994-05-24 10:09:53 +00:00
*
* Historically, this was implemented by the BSD software ISR facility; it is
* now implemented via a software ithread (SWI).
1994-05-24 10:09:53 +00:00
*/
/*
* Protocol numbers, which are encoded in monitoring applications and kernel
* modules. Internally, these are used in bit shift operations so must have
* a value 0 < proto < 32; we currently further limit at compile-time to 16
* for array-sizing purposes.
*/
#define NETISR_IP 1
#define NETISR_IGMP 2 /* IGMPv3 output queue */
#define NETISR_ROUTE 3 /* routing socket */
#define NETISR_ARP 4 /* same as AF_LINK */
#define NETISR_ETHER 5 /* ethernet input */
#define NETISR_IPV6 6
#define NETISR_EPAIR 8 /* if_epair(4) */
#define NETISR_IP_DIRECT 9 /* direct-dispatch IPv4 */
#define NETISR_IPV6_DIRECT 10 /* direct-dispatch IPv6 */
1994-05-24 10:09:53 +00:00
/*
* Protocol ordering and affinity policy constants. See the detailed
* discussion of policies later in the file.
*/
#define NETISR_POLICY_SOURCE 1 /* Maintain source ordering. */
#define NETISR_POLICY_FLOW 2 /* Maintain flow ordering. */
#define NETISR_POLICY_CPU 3 /* Protocol determines CPU placement. */
Rework netisr policy mechanism so that per-protocol dispatch policies can be represented: - A single policy namespace is defined, consisting of four possible policies: "default" to use the global default, "deferred" to force deferred dispatch, "direct" to employ direct dispatch where possible, and "hybrid" which makes a dynamic decision based on CPU affinity, ordering, etc. Routines are implemented to convert between strings and an integer namespace. - A new global variable, netisr_dispatch_policy, subsumes existing global variables for direct dispatch, forced direct dispatch, etc, and is used for explicit policy interpretation and composition. Old variables remain so that they can be exported by legacy sysctls for use by old netstat(1) binaries. A new sysctl and tunable, netisr.dispatch.policy, accepts the above strings for specifying a global policy default. - The protocol registration structure, netisr_handler, grows an nh_dispatch field, which accepts a per-policy policy override. The default value is '0', which corresponds to "default", meaning that protocols will accept the global default policy unless otherwise specified. - Policies are now interpreted and composed explicitly at various points in packet dispatch; protocol policies override global policies. - Protocols grow the ability to express a non-opinion about affinity even when implenting m2cpuid by returning NETISR_CPUID_NONE. In that case, the framework falls back on source ordering, rather than simply using the current CPU. These changes are in support of allowing link layer re-dispatch based on RSS or similar hashes provided by NICs, especially in the case where the number of hardware receive queues matches hardware core count, rather than hardware thread count, requiring further software redistributeon. (i.e., on RMI XLR). MFC after: 3 weeks Reviewed by: bz Sponsored by: Juniper Networks, Inc.
2011-05-24 12:34:19 +00:00
/*
* Protocol dispatch policy constants; selects whether and when direct
* dispatch is permitted.
*/
#define NETISR_DISPATCH_DEFAULT 0 /* Use global default. */
#define NETISR_DISPATCH_DEFERRED 1 /* Always defer dispatch. */
#define NETISR_DISPATCH_HYBRID 2 /* Allow hybrid dispatch. */
#define NETISR_DISPATCH_DIRECT 3 /* Always direct dispatch. */
/*
* Monitoring data structures, exported by sysctl(2).
*
* Three sysctls are defined. First, a per-protocol structure exported by
* net.isr.proto.
*/
#define NETISR_NAMEMAXLEN 32
struct sysctl_netisr_proto {
u_int snp_version; /* Length of struct. */
char snp_name[NETISR_NAMEMAXLEN]; /* nh_name */
u_int snp_proto; /* nh_proto */
u_int snp_qlimit; /* nh_qlimit */
u_int snp_policy; /* nh_policy */
u_int snp_flags; /* Various flags. */
Rework netisr policy mechanism so that per-protocol dispatch policies can be represented: - A single policy namespace is defined, consisting of four possible policies: "default" to use the global default, "deferred" to force deferred dispatch, "direct" to employ direct dispatch where possible, and "hybrid" which makes a dynamic decision based on CPU affinity, ordering, etc. Routines are implemented to convert between strings and an integer namespace. - A new global variable, netisr_dispatch_policy, subsumes existing global variables for direct dispatch, forced direct dispatch, etc, and is used for explicit policy interpretation and composition. Old variables remain so that they can be exported by legacy sysctls for use by old netstat(1) binaries. A new sysctl and tunable, netisr.dispatch.policy, accepts the above strings for specifying a global policy default. - The protocol registration structure, netisr_handler, grows an nh_dispatch field, which accepts a per-policy policy override. The default value is '0', which corresponds to "default", meaning that protocols will accept the global default policy unless otherwise specified. - Policies are now interpreted and composed explicitly at various points in packet dispatch; protocol policies override global policies. - Protocols grow the ability to express a non-opinion about affinity even when implenting m2cpuid by returning NETISR_CPUID_NONE. In that case, the framework falls back on source ordering, rather than simply using the current CPU. These changes are in support of allowing link layer re-dispatch based on RSS or similar hashes provided by NICs, especially in the case where the number of hardware receive queues matches hardware core count, rather than hardware thread count, requiring further software redistributeon. (i.e., on RMI XLR). MFC after: 3 weeks Reviewed by: bz Sponsored by: Juniper Networks, Inc.
2011-05-24 12:34:19 +00:00
u_int snp_dispatch; /* Dispatch policy. */
u_int _snp_ispare[6];
};
/*
* Flags for sysctl_netisr_proto.snp_flags.
*/
#define NETISR_SNP_FLAGS_M2FLOW 0x00000001 /* nh_m2flow */
#define NETISR_SNP_FLAGS_M2CPUID 0x00000002 /* nh_m2cpuid */
#define NETISR_SNP_FLAGS_DRAINEDCPU 0x00000004 /* nh_drainedcpu */
/*
* Next, a structure per-workstream, with per-protocol data, exported as
* net.isr.workstream.
*/
struct sysctl_netisr_workstream {
u_int snws_version; /* Length of struct. */
u_int snws_flags; /* Various flags. */
u_int snws_wsid; /* Workstream ID. */
u_int snws_cpu; /* nws_cpu */
u_int _snws_ispare[12];
};
/*
* Flags for sysctl_netisr_workstream.snws_flags
*/
#define NETISR_SNWS_FLAGS_INTR 0x00000001 /* nws_intr_event */
/*
* Finally, a per-workstream-per-protocol structure, exported as
* net.isr.work.
*/
struct sysctl_netisr_work {
u_int snw_version; /* Length of struct. */
u_int snw_wsid; /* Workstream ID. */
u_int snw_proto; /* Protocol number. */
u_int snw_len; /* nw_len */
u_int snw_watermark; /* nw_watermark */
u_int _snw_ispare[3];
uint64_t snw_dispatched; /* nw_dispatched */
uint64_t snw_hybrid_dispatched; /* nw_hybrid_dispatched */
uint64_t snw_qdrops; /* nw_qdrops */
uint64_t snw_queued; /* nw_queued */
uint64_t snw_handled; /* nw_handled */
uint64_t _snw_llspare[7];
};
#ifdef _KERNEL
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
/*-
* Protocols express ordering constraints and affinity preferences by
* implementing one or neither of nh_m2flow and nh_m2cpuid, which are used by
* netisr to determine which per-CPU workstream to assign mbufs to.
*
* The following policies may be used by protocols:
*
* NETISR_POLICY_SOURCE - netisr should maintain source ordering without
* advice from the protocol. netisr will ignore any
* flow IDs present on the mbuf for the purposes of
* work placement.
*
* NETISR_POLICY_FLOW - netisr should maintain flow ordering as defined by
* the mbuf header flow ID field. If the protocol
* implements nh_m2flow, then netisr will query the
* protocol in the event that the mbuf doesn't have a
* flow ID, falling back on source ordering.
*
* NETISR_POLICY_CPU - netisr will delegate all work placement decisions to
* the protocol, querying nh_m2cpuid for each packet.
*
* Protocols might make decisions about work placement based on an existing
* calculated flow ID on the mbuf, such as one provided in hardware, the
* receive interface pointed to by the mbuf (if any), the optional source
* identifier passed at some dispatch points, or even parse packet headers to
* calculate a flow. Both protocol handlers may return a new mbuf pointer
* for the chain, or NULL if the packet proves invalid or m_pullup() fails.
*
* XXXRW: If we eventually support dynamic reconfiguration, there should be
* protocol handlers to notify them of CPU configuration changes so that they
* can rebalance work.
*/
struct mbuf;
2010-03-01 00:43:05 +00:00
typedef void netisr_handler_t(struct mbuf *m);
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
typedef struct mbuf *netisr_m2cpuid_t(struct mbuf *m, uintptr_t source,
u_int *cpuid);
typedef struct mbuf *netisr_m2flow_t(struct mbuf *m, uintptr_t source);
typedef void netisr_drainedcpu_t(u_int cpuid);
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
Rework netisr policy mechanism so that per-protocol dispatch policies can be represented: - A single policy namespace is defined, consisting of four possible policies: "default" to use the global default, "deferred" to force deferred dispatch, "direct" to employ direct dispatch where possible, and "hybrid" which makes a dynamic decision based on CPU affinity, ordering, etc. Routines are implemented to convert between strings and an integer namespace. - A new global variable, netisr_dispatch_policy, subsumes existing global variables for direct dispatch, forced direct dispatch, etc, and is used for explicit policy interpretation and composition. Old variables remain so that they can be exported by legacy sysctls for use by old netstat(1) binaries. A new sysctl and tunable, netisr.dispatch.policy, accepts the above strings for specifying a global policy default. - The protocol registration structure, netisr_handler, grows an nh_dispatch field, which accepts a per-policy policy override. The default value is '0', which corresponds to "default", meaning that protocols will accept the global default policy unless otherwise specified. - Policies are now interpreted and composed explicitly at various points in packet dispatch; protocol policies override global policies. - Protocols grow the ability to express a non-opinion about affinity even when implenting m2cpuid by returning NETISR_CPUID_NONE. In that case, the framework falls back on source ordering, rather than simply using the current CPU. These changes are in support of allowing link layer re-dispatch based on RSS or similar hashes provided by NICs, especially in the case where the number of hardware receive queues matches hardware core count, rather than hardware thread count, requiring further software redistributeon. (i.e., on RMI XLR). MFC after: 3 weeks Reviewed by: bz Sponsored by: Juniper Networks, Inc.
2011-05-24 12:34:19 +00:00
#define NETISR_CPUID_NONE ((u_int)-1) /* No affinity returned. */
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
/*
* Data structure describing a protocol handler.
*/
struct netisr_handler {
const char *nh_name; /* Character string protocol name. */
netisr_handler_t *nh_handler; /* Protocol handler. */
netisr_m2flow_t *nh_m2flow; /* Query flow for untagged packet. */
netisr_m2cpuid_t *nh_m2cpuid; /* Query CPU to process mbuf on. */
netisr_drainedcpu_t *nh_drainedcpu; /* Callback when drained a queue. */
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
u_int nh_proto; /* Integer protocol ID. */
u_int nh_qlimit; /* Maximum per-CPU queue depth. */
u_int nh_policy; /* Work placement policy. */
Rework netisr policy mechanism so that per-protocol dispatch policies can be represented: - A single policy namespace is defined, consisting of four possible policies: "default" to use the global default, "deferred" to force deferred dispatch, "direct" to employ direct dispatch where possible, and "hybrid" which makes a dynamic decision based on CPU affinity, ordering, etc. Routines are implemented to convert between strings and an integer namespace. - A new global variable, netisr_dispatch_policy, subsumes existing global variables for direct dispatch, forced direct dispatch, etc, and is used for explicit policy interpretation and composition. Old variables remain so that they can be exported by legacy sysctls for use by old netstat(1) binaries. A new sysctl and tunable, netisr.dispatch.policy, accepts the above strings for specifying a global policy default. - The protocol registration structure, netisr_handler, grows an nh_dispatch field, which accepts a per-policy policy override. The default value is '0', which corresponds to "default", meaning that protocols will accept the global default policy unless otherwise specified. - Policies are now interpreted and composed explicitly at various points in packet dispatch; protocol policies override global policies. - Protocols grow the ability to express a non-opinion about affinity even when implenting m2cpuid by returning NETISR_CPUID_NONE. In that case, the framework falls back on source ordering, rather than simply using the current CPU. These changes are in support of allowing link layer re-dispatch based on RSS or similar hashes provided by NICs, especially in the case where the number of hardware receive queues matches hardware core count, rather than hardware thread count, requiring further software redistributeon. (i.e., on RMI XLR). MFC after: 3 weeks Reviewed by: bz Sponsored by: Juniper Networks, Inc.
2011-05-24 12:34:19 +00:00
u_int nh_dispatch; /* Dispatch policy. */
u_int nh_ispare[4]; /* For future use. */
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
void *nh_pspare[4]; /* For future use. */
};
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
/*
* Register, unregister, and other netisr handler management functions.
*/
void netisr_clearqdrops(const struct netisr_handler *nhp);
void netisr_getqdrops(const struct netisr_handler *nhp,
u_int64_t *qdropsp);
void netisr_getqlimit(const struct netisr_handler *nhp, u_int *qlimitp);
void netisr_register(const struct netisr_handler *nhp);
int netisr_setqlimit(const struct netisr_handler *nhp, u_int qlimit);
void netisr_unregister(const struct netisr_handler *nhp);
#ifdef VIMAGE
void netisr_register_vnet(const struct netisr_handler *nhp);
void netisr_unregister_vnet(const struct netisr_handler *nhp);
#endif
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
/*
* Process a packet destined for a protocol, and attempt direct dispatch.
* Supplemental source ordering information can be passed using the _src
* variant.
*/
int netisr_dispatch(u_int proto, struct mbuf *m);
int netisr_dispatch_src(u_int proto, uintptr_t source, struct mbuf *m);
int netisr_queue(u_int proto, struct mbuf *m);
int netisr_queue_src(u_int proto, uintptr_t source, struct mbuf *m);
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
/*
* Provide a default implementation of "map an ID to a CPU ID".
*/
u_int netisr_default_flow2cpu(u_int flowid);
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
/*
* Utility routines to return the number of CPUs participting in netisr, and
* to return a mapping from a number to a CPU ID that can be used with the
* scheduler.
*/
u_int netisr_get_cpucount(void);
u_int netisr_get_cpuid(u_int cpunumber);
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
/*
* Interfaces between DEVICE_POLLING and netisr.
*/
void netisr_sched_poll(void);
void netisr_poll(void);
void netisr_pollmore(void);
1994-08-21 05:11:48 +00:00
Reimplement the netisr framework in order to support parallel netisr threads: - Support up to one netisr thread per CPU, each processings its own workstream, or set of per-protocol queues. Threads may be bound to specific CPUs, or allowed to migrate, based on a global policy. In the future it would be desirable to support topology-centric policies, such as "one netisr per package". - Allow each protocol to advertise an ordering policy, which can currently be one of: NETISR_POLICY_SOURCE: packets must maintain ordering with respect to an implicit or explicit source (such as an interface or socket). NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work, as well as allowing protocols to provide a flow generation function for mbufs without flow identifers (m2flow). Falls back on NETISR_POLICY_SOURCE if now flow ID is available. NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for each packet handled by netisr (m2cpuid). - Provide utility functions for querying the number of workstreams being used, as well as a mapping function from workstream to CPU ID, which protocols may use in work placement decisions. - Add explicit interfaces to get and set per-protocol queue limits, and get and clear drop counters, which query data or apply changes across all workstreams. - Add a more extensible netisr registration interface, in which protocols declare 'struct netisr_handler' structures for each registered NETISR_ type. These include name, handler function, optional mbuf to flow ID function, optional mbuf to CPU ID function, queue limit, and ordering policy. Padding is present to allow these to be expanded in the future. If no queue limit is declared, then a default is used. - Queue limits are now per-workstream, and raised from the previous IFQ_MAXLEN default of 50 to 256. - All protocols are updated to use the new registration interface, and with the exception of netnatm, default queue limits. Most protocols register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use NETISR_POLICY_FLOW, and will therefore take advantage of driver- generated flow IDs if present. - Formalize a non-packet based interface between interface polling and the netisr, rather than having polling pretend to be two protocols. Provide two explicit hooks in the netisr worker for start and end events for runs: netisr_poll() and netisr_pollmore(), as well as a function, netisr_sched_poll(), to allow the polling code to schedule netisr execution. DEVICE_POLLING still embeds single-netisr assumptions in its implementation, so for now if it is compiled into the kernel, a single and un-bound netisr thread is enforced regardless of tunable configuration. In the default configuration, the new netisr implementation maintains the same basic assumptions as the previous implementation: a single, un-bound worker thread processes all deferred work, and direct dispatch is enabled by default wherever possible. Performance measurement shows a marginal performance improvement over the old implementation due to the use of batched dequeue. An rmlock is used to synchronize use and registration/unregistration using the framework; currently, synchronized use is disabled (replicating current netisr policy) due to a measurable 3%-6% hit in ping-pong micro-benchmarking. It will be enabled once further rmlock optimization has taken place. However, in practice, netisrs are rarely registered or unregistered at runtime. A new man page for netisr will follow, but since one doesn't currently exist, it hasn't been updated. This change is not appropriate for MFC, although the polling shutdown handler should be merged to 7-STABLE. Bump __FreeBSD_version. Reviewed by: bz
2009-06-01 10:41:38 +00:00
#endif /* !_KERNEL */
#endif /* !_NET_NETISR_H_ */