freebsd-dev/sys/dev/netmap/netmap_mem2.h

166 lines
6.8 KiB
C
Raw Normal View History

/*
* Copyright (C) 2012-2014 Matteo Landi, Luigi Rizzo, Giuseppe Lettieri. All rights reserved.
*
* 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* 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.
*/
/*
* $FreeBSD$
*
* (New) memory allocator for netmap
*/
/*
* This allocator creates three memory pools:
* nm_if_pool for the struct netmap_if
* nm_ring_pool for the struct netmap_ring
* nm_buf_pool for the packet buffers.
*
* that contain netmap objects. Each pool is made of a number of clusters,
* multiple of a page size, each containing an integer number of objects.
* The clusters are contiguous in user space but not in the kernel.
* Only nm_buf_pool needs to be dma-able,
* but for convenience use the same type of allocator for all.
*
* Once mapped, the three pools are exported to userspace
* as a contiguous block, starting from nm_if_pool. Each
* cluster (and pool) is an integral number of pages.
* [ . . . ][ . . . . . .][ . . . . . . . . . .]
* nm_if nm_ring nm_buf
*
* The userspace areas contain offsets of the objects in userspace.
* When (at init time) we write these offsets, we find out the index
* of the object, and from there locate the offset from the beginning
* of the region.
*
* The invididual allocators manage a pool of memory for objects of
* the same size.
* The pool is split into smaller clusters, whose size is a
* multiple of the page size. The cluster size is chosen
* to minimize the waste for a given max cluster size
* (we do it by brute force, as we have relatively few objects
* per cluster).
*
* Objects are aligned to the cache line (64 bytes) rounding up object
* sizes when needed. A bitmap contains the state of each object.
* Allocation scans the bitmap; this is done only on attach, so we are not
* too worried about performance
*
* For each allocator we can define (thorugh sysctl) the size and
* number of each object. Memory is allocated at the first use of a
* netmap file descriptor, and can be freed when all such descriptors
* have been released (including unmapping the memory).
* If memory is scarce, the system tries to get as much as possible
* and the sysctl values reflect the actual allocation.
* Together with desired values, the sysctl export also absolute
* min and maximum values that cannot be overridden.
*
* struct netmap_if:
* variable size, max 16 bytes per ring pair plus some fixed amount.
* 1024 bytes should be large enough in practice.
*
* In the worst case we have one netmap_if per ring in the system.
*
* struct netmap_ring
* variable size, 8 byte per slot plus some fixed amount.
* Rings can be large (e.g. 4k slots, or >32Kbytes).
* We default to 36 KB (9 pages), and a few hundred rings.
*
* struct netmap_buffer
* The more the better, both because fast interfaces tend to have
* many slots, and because we may want to use buffers to store
* packets in userspace avoiding copies.
* Must contain a full frame (eg 1518, or more for vlans, jumbo
* frames etc.) plus be nicely aligned, plus some NICs restrict
* the size to multiple of 1K or so. Default to 2K
*/
#ifndef _NET_NETMAP_MEM2_H_
#define _NET_NETMAP_MEM2_H_
/* We implement two kinds of netmap_mem_d structures:
*
* - global: used by hardware NICS;
*
* - private: used by VALE ports.
*
* In both cases, the netmap_mem_d structure has the same lifetime as the
* netmap_adapter of the corresponding NIC or port. It is the responsibility of
* the client code to delete the private allocator when the associated
* netmap_adapter is freed (this is implemented by the NAF_MEM_OWNER flag in
* netmap.c). The 'refcount' field counts the number of active users of the
* structure. The global allocator uses this information to prevent/allow
* reconfiguration. The private allocators release all their memory when there
* are no active users. By 'active user' we mean an existing netmap_priv
* structure holding a reference to the allocator.
*/
extern struct netmap_mem_d nm_mem;
2015-07-10 05:51:36 +00:00
void netmap_mem_get_lut(struct netmap_mem_d *, struct netmap_lut *);
vm_paddr_t netmap_mem_ofstophys(struct netmap_mem_d *, vm_ooffset_t);
Update to the current version of netmap. Mostly bugfixes or features developed in the past 6 months, so this is a 10.1 candidate. Basically no user API changes (some bugfixes in sys/net/netmap_user.h). In detail: 1. netmap support for virtio-net, including in netmap mode. Under bhyve and with a netmap backend [2] we reach over 1Mpps with standard APIs (e.g. libpcap), and 5-8 Mpps in netmap mode. 2. (kernel) add support for multiple memory allocators, so we can better partition physical and virtual interfaces giving access to separate users. The most visible effect is one additional argument to the various kernel functions to compute buffer addresses. All netmap-supported drivers are affected, but changes are mechanical and trivial 3. (kernel) simplify the prototype for *txsync() and *rxsync() driver methods. All netmap drivers affected, changes mostly mechanical. 4. add support for netmap-monitor ports. Think of it as a mirroring port on a physical switch: a netmap monitor port replicates traffic present on the main port. Restrictions apply. Drive carefully. 5. if_lem.c: support for various paravirtualization features, experimental and disabled by default. Most of these are described in our ANCS'13 paper [1]. Paravirtualized support in netmap mode is new, and beats the numbers in the paper by a large factor (under qemu-kvm, we measured gues-host throughput up to 10-12 Mpps). A lot of refactoring and additional documentation in the files in sys/dev/netmap, but apart from #2 and #3 above, almost nothing of this stuff is visible to other kernel parts. Example programs in tools/tools/netmap have been updated with bugfixes and to support more of the existing features. This is meant to go into 10.1 so we plan an MFC before the Aug.22 deadline. A lot of this code has been contributed by my colleagues at UNIPI, including Giuseppe Lettieri, Vincenzo Maffione, Stefano Garzarella. MFC after: 3 days.
2014-08-16 15:00:01 +00:00
int netmap_mem_finalize(struct netmap_mem_d *, struct netmap_adapter *);
int netmap_mem_init(void);
void netmap_mem_fini(void);
Update to the current version of netmap. Mostly bugfixes or features developed in the past 6 months, so this is a 10.1 candidate. Basically no user API changes (some bugfixes in sys/net/netmap_user.h). In detail: 1. netmap support for virtio-net, including in netmap mode. Under bhyve and with a netmap backend [2] we reach over 1Mpps with standard APIs (e.g. libpcap), and 5-8 Mpps in netmap mode. 2. (kernel) add support for multiple memory allocators, so we can better partition physical and virtual interfaces giving access to separate users. The most visible effect is one additional argument to the various kernel functions to compute buffer addresses. All netmap-supported drivers are affected, but changes are mechanical and trivial 3. (kernel) simplify the prototype for *txsync() and *rxsync() driver methods. All netmap drivers affected, changes mostly mechanical. 4. add support for netmap-monitor ports. Think of it as a mirroring port on a physical switch: a netmap monitor port replicates traffic present on the main port. Restrictions apply. Drive carefully. 5. if_lem.c: support for various paravirtualization features, experimental and disabled by default. Most of these are described in our ANCS'13 paper [1]. Paravirtualized support in netmap mode is new, and beats the numbers in the paper by a large factor (under qemu-kvm, we measured gues-host throughput up to 10-12 Mpps). A lot of refactoring and additional documentation in the files in sys/dev/netmap, but apart from #2 and #3 above, almost nothing of this stuff is visible to other kernel parts. Example programs in tools/tools/netmap have been updated with bugfixes and to support more of the existing features. This is meant to go into 10.1 so we plan an MFC before the Aug.22 deadline. A lot of this code has been contributed by my colleagues at UNIPI, including Giuseppe Lettieri, Vincenzo Maffione, Stefano Garzarella. MFC after: 3 days.
2014-08-16 15:00:01 +00:00
struct netmap_if * netmap_mem_if_new(struct netmap_adapter *);
void netmap_mem_if_delete(struct netmap_adapter *, struct netmap_if *);
int netmap_mem_rings_create(struct netmap_adapter *);
void netmap_mem_rings_delete(struct netmap_adapter *);
Update to the current version of netmap. Mostly bugfixes or features developed in the past 6 months, so this is a 10.1 candidate. Basically no user API changes (some bugfixes in sys/net/netmap_user.h). In detail: 1. netmap support for virtio-net, including in netmap mode. Under bhyve and with a netmap backend [2] we reach over 1Mpps with standard APIs (e.g. libpcap), and 5-8 Mpps in netmap mode. 2. (kernel) add support for multiple memory allocators, so we can better partition physical and virtual interfaces giving access to separate users. The most visible effect is one additional argument to the various kernel functions to compute buffer addresses. All netmap-supported drivers are affected, but changes are mechanical and trivial 3. (kernel) simplify the prototype for *txsync() and *rxsync() driver methods. All netmap drivers affected, changes mostly mechanical. 4. add support for netmap-monitor ports. Think of it as a mirroring port on a physical switch: a netmap monitor port replicates traffic present on the main port. Restrictions apply. Drive carefully. 5. if_lem.c: support for various paravirtualization features, experimental and disabled by default. Most of these are described in our ANCS'13 paper [1]. Paravirtualized support in netmap mode is new, and beats the numbers in the paper by a large factor (under qemu-kvm, we measured gues-host throughput up to 10-12 Mpps). A lot of refactoring and additional documentation in the files in sys/dev/netmap, but apart from #2 and #3 above, almost nothing of this stuff is visible to other kernel parts. Example programs in tools/tools/netmap have been updated with bugfixes and to support more of the existing features. This is meant to go into 10.1 so we plan an MFC before the Aug.22 deadline. A lot of this code has been contributed by my colleagues at UNIPI, including Giuseppe Lettieri, Vincenzo Maffione, Stefano Garzarella. MFC after: 3 days.
2014-08-16 15:00:01 +00:00
void netmap_mem_deref(struct netmap_mem_d *, struct netmap_adapter *);
This new version of netmap brings you the following: - netmap pipes, providing bidirectional blocking I/O while moving 100+ Mpps between processes using shared memory channels (no mistake: over one hundred million. But mind you, i said *moving* not *processing*); - kqueue support (BHyVe needs it); - improved user library. Just the interface name lets you select a NIC, host port, VALE switch port, netmap pipe, and individual queues. The upcoming netmap-enabled libpcap will use this feature. - optional extra buffers associated to netmap ports, for applications that need to buffer data yet don't want to make copies. - segmentation offloading for the VALE switch, useful between VMs. and a number of bug fixes and performance improvements. My colleagues Giuseppe Lettieri and Vincenzo Maffione did a substantial amount of work on these features so we owe them a big thanks. There are some external repositories that can be of interest: https://code.google.com/p/netmap our public repository for netmap/VALE code, including linux versions and other stuff that does not belong here, such as python bindings. https://code.google.com/p/netmap-libpcap a clone of the libpcap repository with netmap support. With this any libpcap client has access to most netmap feature with no recompilation. E.g. tcpdump can filter packets at 10-15 Mpps. https://code.google.com/p/netmap-ipfw a userspace version of ipfw+dummynet which uses netmap to send/receive packets. Speed is up in the 7-10 Mpps range per core for simple rulesets. Both netmap-libpcap and netmap-ipfw will be merged upstream at some point, but while this happens it is useful to have access to them. And yes, this code will be merged soon. It is infinitely better than the version currently in 10 and 9. MFC after: 3 days
2014-02-15 04:53:04 +00:00
int netmap_mem_get_info(struct netmap_mem_d *, u_int *size, u_int *memflags, uint16_t *id);
ssize_t netmap_mem_if_offset(struct netmap_mem_d *, const void *vaddr);
This new version of netmap brings you the following: - netmap pipes, providing bidirectional blocking I/O while moving 100+ Mpps between processes using shared memory channels (no mistake: over one hundred million. But mind you, i said *moving* not *processing*); - kqueue support (BHyVe needs it); - improved user library. Just the interface name lets you select a NIC, host port, VALE switch port, netmap pipe, and individual queues. The upcoming netmap-enabled libpcap will use this feature. - optional extra buffers associated to netmap ports, for applications that need to buffer data yet don't want to make copies. - segmentation offloading for the VALE switch, useful between VMs. and a number of bug fixes and performance improvements. My colleagues Giuseppe Lettieri and Vincenzo Maffione did a substantial amount of work on these features so we owe them a big thanks. There are some external repositories that can be of interest: https://code.google.com/p/netmap our public repository for netmap/VALE code, including linux versions and other stuff that does not belong here, such as python bindings. https://code.google.com/p/netmap-libpcap a clone of the libpcap repository with netmap support. With this any libpcap client has access to most netmap feature with no recompilation. E.g. tcpdump can filter packets at 10-15 Mpps. https://code.google.com/p/netmap-ipfw a userspace version of ipfw+dummynet which uses netmap to send/receive packets. Speed is up in the 7-10 Mpps range per core for simple rulesets. Both netmap-libpcap and netmap-ipfw will be merged upstream at some point, but while this happens it is useful to have access to them. And yes, this code will be merged soon. It is infinitely better than the version currently in 10 and 9. MFC after: 3 days
2014-02-15 04:53:04 +00:00
struct netmap_mem_d* netmap_mem_private_new(const char *name,
u_int txr, u_int txd, u_int rxr, u_int rxd, u_int extra_bufs, u_int npipes,
int* error);
2015-07-10 05:51:36 +00:00
void netmap_mem_delete(struct netmap_mem_d *);
//#define NM_DEBUG_MEM_PUTGET 1
#ifdef NM_DEBUG_MEM_PUTGET
#define netmap_mem_get(nmd) \
do { \
__netmap_mem_get(nmd, __FUNCTION__, __LINE__); \
} while (0)
#define netmap_mem_put(nmd) \
do { \
__netmap_mem_put(nmd, __FUNCTION__, __LINE__); \
} while (0)
void __netmap_mem_get(struct netmap_mem_d *, const char *, int);
void __netmap_mem_put(struct netmap_mem_d *, const char *, int);
#else /* !NM_DEBUG_MEM_PUTGET */
void netmap_mem_get(struct netmap_mem_d *);
void netmap_mem_put(struct netmap_mem_d *);
#endif /* !NM_DEBUG_PUTGET */
Update to the current version of netmap. Mostly bugfixes or features developed in the past 6 months, so this is a 10.1 candidate. Basically no user API changes (some bugfixes in sys/net/netmap_user.h). In detail: 1. netmap support for virtio-net, including in netmap mode. Under bhyve and with a netmap backend [2] we reach over 1Mpps with standard APIs (e.g. libpcap), and 5-8 Mpps in netmap mode. 2. (kernel) add support for multiple memory allocators, so we can better partition physical and virtual interfaces giving access to separate users. The most visible effect is one additional argument to the various kernel functions to compute buffer addresses. All netmap-supported drivers are affected, but changes are mechanical and trivial 3. (kernel) simplify the prototype for *txsync() and *rxsync() driver methods. All netmap drivers affected, changes mostly mechanical. 4. add support for netmap-monitor ports. Think of it as a mirroring port on a physical switch: a netmap monitor port replicates traffic present on the main port. Restrictions apply. Drive carefully. 5. if_lem.c: support for various paravirtualization features, experimental and disabled by default. Most of these are described in our ANCS'13 paper [1]. Paravirtualized support in netmap mode is new, and beats the numbers in the paper by a large factor (under qemu-kvm, we measured gues-host throughput up to 10-12 Mpps). A lot of refactoring and additional documentation in the files in sys/dev/netmap, but apart from #2 and #3 above, almost nothing of this stuff is visible to other kernel parts. Example programs in tools/tools/netmap have been updated with bugfixes and to support more of the existing features. This is meant to go into 10.1 so we plan an MFC before the Aug.22 deadline. A lot of this code has been contributed by my colleagues at UNIPI, including Giuseppe Lettieri, Vincenzo Maffione, Stefano Garzarella. MFC after: 3 days.
2014-08-16 15:00:01 +00:00
#define NETMAP_MEM_PRIVATE 0x2 /* allocator uses private address space */
#define NETMAP_MEM_IO 0x4 /* the underlying memory is mmapped I/O */
This new version of netmap brings you the following: - netmap pipes, providing bidirectional blocking I/O while moving 100+ Mpps between processes using shared memory channels (no mistake: over one hundred million. But mind you, i said *moving* not *processing*); - kqueue support (BHyVe needs it); - improved user library. Just the interface name lets you select a NIC, host port, VALE switch port, netmap pipe, and individual queues. The upcoming netmap-enabled libpcap will use this feature. - optional extra buffers associated to netmap ports, for applications that need to buffer data yet don't want to make copies. - segmentation offloading for the VALE switch, useful between VMs. and a number of bug fixes and performance improvements. My colleagues Giuseppe Lettieri and Vincenzo Maffione did a substantial amount of work on these features so we owe them a big thanks. There are some external repositories that can be of interest: https://code.google.com/p/netmap our public repository for netmap/VALE code, including linux versions and other stuff that does not belong here, such as python bindings. https://code.google.com/p/netmap-libpcap a clone of the libpcap repository with netmap support. With this any libpcap client has access to most netmap feature with no recompilation. E.g. tcpdump can filter packets at 10-15 Mpps. https://code.google.com/p/netmap-ipfw a userspace version of ipfw+dummynet which uses netmap to send/receive packets. Speed is up in the 7-10 Mpps range per core for simple rulesets. Both netmap-libpcap and netmap-ipfw will be merged upstream at some point, but while this happens it is useful to have access to them. And yes, this code will be merged soon. It is infinitely better than the version currently in 10 and 9. MFC after: 3 days
2014-02-15 04:53:04 +00:00
uint32_t netmap_extra_alloc(struct netmap_adapter *, uint32_t *, uint32_t n);
#endif