freebsd-skq/sys/net/debugnet.h

276 lines
8.5 KiB
C
Raw Normal View History

Split out a more generic debugnet(4) from netdump(4) Debugnet is a simplistic and specialized panic- or debug-time reliable datagram transport. It can drive a single connection at a time and is currently unidirectional (debug/panic machine transmit to remote server only). It is mostly a verbatim code lift from netdump(4). Netdump(4) remains the only consumer (until the rest of this patch series lands). The INET-specific logic has been extracted somewhat more thoroughly than previously in netdump(4), into debugnet_inet.c. UDP-layer logic and up, as much as possible as is protocol-independent, remains in debugnet.c. The separation is not perfect and future improvement is welcome. Supporting INET6 is a long-term goal. Much of the diff is "gratuitous" renaming from 'netdump_' or 'nd_' to 'debugnet_' or 'dn_' -- sorry. I thought keeping the netdump name on the generic module would be more confusing than the refactoring. The only functional change here is the mbuf allocation / tracking. Instead of initiating solely on netdump-configured interface(s) at dumpon(8) configuration time, we watch for any debugnet-enabled NIC for link activation and query it for mbuf parameters at that time. If they exceed the existing high-water mark allocation, we re-allocate and track the new high-water mark. Otherwise, we leave the pre-panic mbuf allocation alone. In a future patch in this series, this will allow initiating netdump from panic ddb(4) without pre-panic configuration. No other functional change intended. Reviewed by: markj (earlier version) Some discussion with: emaste, jhb Objection from: marius Differential Revision: https://reviews.freebsd.org/D21421
2019-10-17 16:23:03 +00:00
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2019 Isilon Systems, LLC.
* Copyright (c) 2005-2014 Sandvine Incorporated
* Copyright (c) 2000 Darrell Anderson <anderson@cs.duke.edu>
* 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.
*
* $FreeBSD$
*/
/*
* Debugnet provides a reliable, bidirectional, UDP-encapsulated datagram
* transport while a machine is in a debug state. (N-1 CPUs stopped,
* interrupts disabled, may or may not be in a panic(9) state.) Only one
* stream may be active at a time. A dedicated server must be running to
* accept connections.
*/
#pragma once
#include <sys/types.h>
#include <netinet/in.h>
/*
* Debugnet protocol details.
*/
#define DEBUGNET_HERALD 1 /* Connection handshake. */
#define DEBUGNET_FINISHED 2 /* Close the connection. */
#define DEBUGNET_DATA 3 /* Contains data. */
struct debugnet_msg_hdr {
uint32_t mh_type; /* Debugnet message type. */
uint32_t mh_seqno; /* Match acks with msgs. */
uint64_t mh_offset; /* Offset in fragment. */
uint32_t mh_len; /* Attached data (bytes). */
uint32_t mh_aux2; /* Consumer-specific. */
} __packed;
struct debugnet_ack {
uint32_t da_seqno; /* Match acks with msgs. */
} __packed;
#define DEBUGNET_MAX_IN_FLIGHT 64
#ifdef _KERNEL
/*
* Hook API for network drivers.
*/
enum debugnet_ev {
DEBUGNET_START,
DEBUGNET_END,
};
struct ifnet;
struct mbuf;
typedef void debugnet_init_t(struct ifnet *, int *nrxr, int *ncl, int *clsize);
typedef void debugnet_event_t(struct ifnet *, enum debugnet_ev);
typedef int debugnet_transmit_t(struct ifnet *, struct mbuf *);
typedef int debugnet_poll_t(struct ifnet *, int);
struct debugnet_methods {
debugnet_init_t *dn_init;
debugnet_event_t *dn_event;
debugnet_transmit_t *dn_transmit;
debugnet_poll_t *dn_poll;
};
#define DEBUGNET_SUPPORTED_NIC(ifp) \
((ifp)->if_debugnet_methods != NULL && (ifp)->if_type == IFT_ETHER)
struct debugnet_pcb; /* opaque */
Split out a more generic debugnet(4) from netdump(4) Debugnet is a simplistic and specialized panic- or debug-time reliable datagram transport. It can drive a single connection at a time and is currently unidirectional (debug/panic machine transmit to remote server only). It is mostly a verbatim code lift from netdump(4). Netdump(4) remains the only consumer (until the rest of this patch series lands). The INET-specific logic has been extracted somewhat more thoroughly than previously in netdump(4), into debugnet_inet.c. UDP-layer logic and up, as much as possible as is protocol-independent, remains in debugnet.c. The separation is not perfect and future improvement is welcome. Supporting INET6 is a long-term goal. Much of the diff is "gratuitous" renaming from 'netdump_' or 'nd_' to 'debugnet_' or 'dn_' -- sorry. I thought keeping the netdump name on the generic module would be more confusing than the refactoring. The only functional change here is the mbuf allocation / tracking. Instead of initiating solely on netdump-configured interface(s) at dumpon(8) configuration time, we watch for any debugnet-enabled NIC for link activation and query it for mbuf parameters at that time. If they exceed the existing high-water mark allocation, we re-allocate and track the new high-water mark. Otherwise, we leave the pre-panic mbuf allocation alone. In a future patch in this series, this will allow initiating netdump from panic ddb(4) without pre-panic configuration. No other functional change intended. Reviewed by: markj (earlier version) Some discussion with: emaste, jhb Objection from: marius Differential Revision: https://reviews.freebsd.org/D21421
2019-10-17 16:23:03 +00:00
/*
* Debugnet consumer API.
*/
struct debugnet_conn_params {
struct ifnet *dc_ifp;
in_addr_t dc_client;
in_addr_t dc_server;
in_addr_t dc_gateway;
uint16_t dc_herald_port;
uint16_t dc_client_port;
Split out a more generic debugnet(4) from netdump(4) Debugnet is a simplistic and specialized panic- or debug-time reliable datagram transport. It can drive a single connection at a time and is currently unidirectional (debug/panic machine transmit to remote server only). It is mostly a verbatim code lift from netdump(4). Netdump(4) remains the only consumer (until the rest of this patch series lands). The INET-specific logic has been extracted somewhat more thoroughly than previously in netdump(4), into debugnet_inet.c. UDP-layer logic and up, as much as possible as is protocol-independent, remains in debugnet.c. The separation is not perfect and future improvement is welcome. Supporting INET6 is a long-term goal. Much of the diff is "gratuitous" renaming from 'netdump_' or 'nd_' to 'debugnet_' or 'dn_' -- sorry. I thought keeping the netdump name on the generic module would be more confusing than the refactoring. The only functional change here is the mbuf allocation / tracking. Instead of initiating solely on netdump-configured interface(s) at dumpon(8) configuration time, we watch for any debugnet-enabled NIC for link activation and query it for mbuf parameters at that time. If they exceed the existing high-water mark allocation, we re-allocate and track the new high-water mark. Otherwise, we leave the pre-panic mbuf allocation alone. In a future patch in this series, this will allow initiating netdump from panic ddb(4) without pre-panic configuration. No other functional change intended. Reviewed by: markj (earlier version) Some discussion with: emaste, jhb Objection from: marius Differential Revision: https://reviews.freebsd.org/D21421
2019-10-17 16:23:03 +00:00
const void *dc_herald_data;
uint32_t dc_herald_datalen;
/*
* Consistent with debugnet_send(), aux paramaters to debugnet
* functions are provided host-endian (but converted to
* network endian on the wire).
*/
uint32_t dc_herald_aux2;
uint64_t dc_herald_offset;
/*
* If NULL, debugnet is a unidirectional channel from panic machine to
* remote server (like netdump).
*
* If handler is non-NULL, packets received on the client port that are
* not just tx acks are forwarded to the provided handler.
*
* The mbuf chain will have all non-debugnet framing headers removed
* (ethernet, inet, udp). It will start with a debugnet_msg_hdr, of
* which the header is guaranteed to be contiguous. If m_pullup is
* used, the supplied in-out mbuf pointer should be updated
* appropriately.
*
* If the handler frees the mbuf chain, it should set the mbuf pointer
* to NULL. Otherwise, the debugnet input framework will free the
* chain.
*
* The handler should ACK receieved packets with debugnet_ack_output.
*/
void (*dc_rx_handler)(struct debugnet_pcb *, struct mbuf **);
};
Split out a more generic debugnet(4) from netdump(4) Debugnet is a simplistic and specialized panic- or debug-time reliable datagram transport. It can drive a single connection at a time and is currently unidirectional (debug/panic machine transmit to remote server only). It is mostly a verbatim code lift from netdump(4). Netdump(4) remains the only consumer (until the rest of this patch series lands). The INET-specific logic has been extracted somewhat more thoroughly than previously in netdump(4), into debugnet_inet.c. UDP-layer logic and up, as much as possible as is protocol-independent, remains in debugnet.c. The separation is not perfect and future improvement is welcome. Supporting INET6 is a long-term goal. Much of the diff is "gratuitous" renaming from 'netdump_' or 'nd_' to 'debugnet_' or 'dn_' -- sorry. I thought keeping the netdump name on the generic module would be more confusing than the refactoring. The only functional change here is the mbuf allocation / tracking. Instead of initiating solely on netdump-configured interface(s) at dumpon(8) configuration time, we watch for any debugnet-enabled NIC for link activation and query it for mbuf parameters at that time. If they exceed the existing high-water mark allocation, we re-allocate and track the new high-water mark. Otherwise, we leave the pre-panic mbuf allocation alone. In a future patch in this series, this will allow initiating netdump from panic ddb(4) without pre-panic configuration. No other functional change intended. Reviewed by: markj (earlier version) Some discussion with: emaste, jhb Objection from: marius Differential Revision: https://reviews.freebsd.org/D21421
2019-10-17 16:23:03 +00:00
/*
* Open a stream to the specified server's herald port.
Split out a more generic debugnet(4) from netdump(4) Debugnet is a simplistic and specialized panic- or debug-time reliable datagram transport. It can drive a single connection at a time and is currently unidirectional (debug/panic machine transmit to remote server only). It is mostly a verbatim code lift from netdump(4). Netdump(4) remains the only consumer (until the rest of this patch series lands). The INET-specific logic has been extracted somewhat more thoroughly than previously in netdump(4), into debugnet_inet.c. UDP-layer logic and up, as much as possible as is protocol-independent, remains in debugnet.c. The separation is not perfect and future improvement is welcome. Supporting INET6 is a long-term goal. Much of the diff is "gratuitous" renaming from 'netdump_' or 'nd_' to 'debugnet_' or 'dn_' -- sorry. I thought keeping the netdump name on the generic module would be more confusing than the refactoring. The only functional change here is the mbuf allocation / tracking. Instead of initiating solely on netdump-configured interface(s) at dumpon(8) configuration time, we watch for any debugnet-enabled NIC for link activation and query it for mbuf parameters at that time. If they exceed the existing high-water mark allocation, we re-allocate and track the new high-water mark. Otherwise, we leave the pre-panic mbuf allocation alone. In a future patch in this series, this will allow initiating netdump from panic ddb(4) without pre-panic configuration. No other functional change intended. Reviewed by: markj (earlier version) Some discussion with: emaste, jhb Objection from: marius Differential Revision: https://reviews.freebsd.org/D21421
2019-10-17 16:23:03 +00:00
*
* If all goes well, the server will send ACK from a different port to our ack
* port. This allows servers to somewhat gracefully handle multiple debugnet
* clients. (Clients are limited to single connections.)
*
* Returns zero on success, or errno.
*/
int debugnet_connect(const struct debugnet_conn_params *,
struct debugnet_pcb **pcb_out);
/*
* Free a debugnet stream that was previously successfully opened.
*
* No attempt is made to cleanly terminate communication with the remote
* server. Consumers should first send an empty DEBUGNET_FINISHED message, or
* otherwise let the remote know they are signing off.
*/
void debugnet_free(struct debugnet_pcb *);
/*
* Send a message, with common debugnet_msg_hdr header, to the connected remote
* server.
*
* - mhtype translates directly to mh_type (e.g., DEBUGNET_DATA, or some other
* protocol-specific type).
* - Data and datalen describe the attached data; datalen may be zero.
* - If auxdata is NULL, mh_offset's initial value and mh_aux2 will be zero.
* Otherwise, mh_offset's initial value will be auxdata->dp_offset_start and
* mh_aux2 will have the value of auxdata->dp_aux2.
*
* Returns zero on success, or an errno on failure.
*/
struct debugnet_proto_aux {
uint64_t dp_offset_start;
uint32_t dp_aux2;
};
int debugnet_send(struct debugnet_pcb *, uint32_t mhtype, const void *data,
uint32_t datalen, const struct debugnet_proto_aux *auxdata);
/*
* A simple wrapper around the above when no data or auxdata is needed.
*/
static inline int
debugnet_sendempty(struct debugnet_pcb *pcb, uint32_t mhtype)
{
return (debugnet_send(pcb, mhtype, NULL, 0, NULL));
}
/*
* Full-duplex RX should ACK received messages.
*/
int debugnet_ack_output(struct debugnet_pcb *, uint32_t seqno /*net endian*/);
/*
* Check and/or wait for further packets.
*/
void debugnet_network_poll(struct debugnet_pcb *);
Split out a more generic debugnet(4) from netdump(4) Debugnet is a simplistic and specialized panic- or debug-time reliable datagram transport. It can drive a single connection at a time and is currently unidirectional (debug/panic machine transmit to remote server only). It is mostly a verbatim code lift from netdump(4). Netdump(4) remains the only consumer (until the rest of this patch series lands). The INET-specific logic has been extracted somewhat more thoroughly than previously in netdump(4), into debugnet_inet.c. UDP-layer logic and up, as much as possible as is protocol-independent, remains in debugnet.c. The separation is not perfect and future improvement is welcome. Supporting INET6 is a long-term goal. Much of the diff is "gratuitous" renaming from 'netdump_' or 'nd_' to 'debugnet_' or 'dn_' -- sorry. I thought keeping the netdump name on the generic module would be more confusing than the refactoring. The only functional change here is the mbuf allocation / tracking. Instead of initiating solely on netdump-configured interface(s) at dumpon(8) configuration time, we watch for any debugnet-enabled NIC for link activation and query it for mbuf parameters at that time. If they exceed the existing high-water mark allocation, we re-allocate and track the new high-water mark. Otherwise, we leave the pre-panic mbuf allocation alone. In a future patch in this series, this will allow initiating netdump from panic ddb(4) without pre-panic configuration. No other functional change intended. Reviewed by: markj (earlier version) Some discussion with: emaste, jhb Objection from: marius Differential Revision: https://reviews.freebsd.org/D21421
2019-10-17 16:23:03 +00:00
/*
* PCB accessors.
*/
/*
* Get the 48-bit MAC address of the discovered next hop (gateway, or
* destination server if it is on the same segment.
*/
const unsigned char *debugnet_get_gw_mac(const struct debugnet_pcb *);
/*
* Callbacks from core mbuf code.
*/
void debugnet_any_ifnet_update(struct ifnet *);
/*
* DDB parsing helper for common debugnet options.
*
* -s <server> [-g <gateway -c <localip> -i <interface>]
*
* Order is not significant. Interface is an online interface that supports
* debugnet and can route to the debugnet server. The other parameters are all
* IP addresses. Only the server parameter is required. The others are
* inferred automatically from the routing table, if not explicitly provided.
*
* Provides basic '-h' using provided 'cmd' string.
*
* Returns zero on success, or errno.
*/
struct debugnet_ddb_config {
struct ifnet *dd_ifp; /* not ref'd */
in_addr_t dd_client;
in_addr_t dd_server;
in_addr_t dd_gateway;
bool dd_has_client : 1;
bool dd_has_gateway : 1;
};
int debugnet_parse_ddb_cmd(const char *cmd,
struct debugnet_ddb_config *result);
Split out a more generic debugnet(4) from netdump(4) Debugnet is a simplistic and specialized panic- or debug-time reliable datagram transport. It can drive a single connection at a time and is currently unidirectional (debug/panic machine transmit to remote server only). It is mostly a verbatim code lift from netdump(4). Netdump(4) remains the only consumer (until the rest of this patch series lands). The INET-specific logic has been extracted somewhat more thoroughly than previously in netdump(4), into debugnet_inet.c. UDP-layer logic and up, as much as possible as is protocol-independent, remains in debugnet.c. The separation is not perfect and future improvement is welcome. Supporting INET6 is a long-term goal. Much of the diff is "gratuitous" renaming from 'netdump_' or 'nd_' to 'debugnet_' or 'dn_' -- sorry. I thought keeping the netdump name on the generic module would be more confusing than the refactoring. The only functional change here is the mbuf allocation / tracking. Instead of initiating solely on netdump-configured interface(s) at dumpon(8) configuration time, we watch for any debugnet-enabled NIC for link activation and query it for mbuf parameters at that time. If they exceed the existing high-water mark allocation, we re-allocate and track the new high-water mark. Otherwise, we leave the pre-panic mbuf allocation alone. In a future patch in this series, this will allow initiating netdump from panic ddb(4) without pre-panic configuration. No other functional change intended. Reviewed by: markj (earlier version) Some discussion with: emaste, jhb Objection from: marius Differential Revision: https://reviews.freebsd.org/D21421
2019-10-17 16:23:03 +00:00
/* Expose sysctl variables for netdump(4) to alias. */
extern int debugnet_npolls;
extern int debugnet_nretries;
extern int debugnet_arp_nretries;
/*
* Conditionally-defined macros for device drivers so we can avoid ifdef
* wrappers in every single implementation.
*/
#ifdef DEBUGNET
#define DEBUGNET_DEFINE(driver) \
static debugnet_init_t driver##_debugnet_init; \
static debugnet_event_t driver##_debugnet_event; \
static debugnet_transmit_t driver##_debugnet_transmit; \
static debugnet_poll_t driver##_debugnet_poll; \
\
static struct debugnet_methods driver##_debugnet_methods = { \
.dn_init = driver##_debugnet_init, \
.dn_event = driver##_debugnet_event, \
.dn_transmit = driver##_debugnet_transmit, \
.dn_poll = driver##_debugnet_poll, \
}
#define DEBUGNET_NOTIFY_MTU(ifp) debugnet_any_ifnet_update(ifp)
#define DEBUGNET_SET(ifp, driver) \
(ifp)->if_debugnet_methods = &driver##_debugnet_methods
#else /* !DEBUGNET || !INET */
#define DEBUGNET_DEFINE(driver)
#define DEBUGNET_NOTIFY_MTU(ifp)
#define DEBUGNET_SET(ifp, driver)
#endif /* DEBUGNET && INET */
#endif /* _KERNEL */