* [PATCH 7/9] use connect instead of bind
From: Jens Osterkamp @ 2010-10-22 11:26 UTC (permalink / raw)
To: e1000-eedc, virtualization, evb; +Cc: chrisw, Gerhard Stenzel, Jens Osterkamp
In-Reply-To: <1287746820-6036-1-git-send-email-jens@linux.vnet.ibm.com>
modifies the setup of the netlink socket in drv_cfg.c to use pid 0 instead
of the processes pid.
Also replaces bind with connect to allow the reception of netlink messages
from libvirt.
Preparation patch for communication between libvirt and lldpad.
Signed-off-by: Gerhard Stenzel <gstenzel@linux.vnet.ibm.com>
---
drv_cfg.c | 5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/drv_cfg.c b/drv_cfg.c
index 6c02555..23c11f3 100644
--- a/drv_cfg.c
+++ b/drv_cfg.c
@@ -71,9 +71,9 @@ static int init_socket(void)
memset((void *)&snl, 0, sizeof(struct sockaddr_nl));
snl.nl_family = AF_NETLINK;
- snl.nl_pid = getpid();
+ snl.nl_pid = 0;
- if (bind(sd, (struct sockaddr *)&snl, sizeof(struct sockaddr_nl)) < 0) {
+ if (connect(sd, (struct sockaddr *)&snl, sizeof(struct sockaddr_nl)) < 0) {
close(sd);
return -EIO;
}
@@ -195,6 +195,7 @@ static struct nlmsghdr *get_msg(unsigned int seq)
nlh = NULL;
break;
}
+ fprintf(stderr, "%s:%s: nlmsg_type: %d\n", __FILE__, __FUNCTION__, nlh->nlmsg_type);
if ((nlh->nlmsg_type == RTM_GETDCB ||
nlh->nlmsg_type == RTM_SETDCB) &&
nlh->nlmsg_seq == seq) {
--
1.7.1
^ permalink raw reply related
* [PATCH 6/9] add libnl dependency to configure.ac
From: Jens Osterkamp @ 2010-10-22 11:26 UTC (permalink / raw)
To: e1000-eedc, virtualization, evb; +Cc: chrisw, Jens Osterkamp
In-Reply-To: <1287746820-6036-1-git-send-email-jens@linux.vnet.ibm.com>
Signed-off-by: Jens Osterkamp <jens@linux.vnet.ibm.com>
---
configure.ac | 5 +++++
1 files changed, 5 insertions(+), 0 deletions(-)
diff --git a/configure.ac b/configure.ac
index 619c6c4..8b9d08a 100644
--- a/configure.ac
+++ b/configure.ac
@@ -12,6 +12,11 @@ PKG_CHECK_MODULES(LIBCONFIG, libconfig,
AC_SUBST(LIBCONFIG_CFLAGS)
AC_SUBST(LIBCONFIG_LIBS)
+PKG_CHECK_MODULES(LIBNL, libnl-1 >= 1.1)
+AC_SUBST(LIBNL_CFLAGS)
+AC_SUBST(LIBNL_LIBS)
+
+AC_CHECK_LIB(nl, rtnl_link_get_by_name)
AC_CONFIG_SUBDIRS([libconfig-1.3.2])
AC_CONFIG_FILES([Makefile include/version.h lldpad.spec lldpad.pc dcbd.pc])
AC_OUTPUT
--
1.7.1
^ permalink raw reply related
* [PATCH 5/9] VDP commandline interface
From: Jens Osterkamp @ 2010-10-22 11:26 UTC (permalink / raw)
To: e1000-eedc, virtualization, evb; +Cc: chrisw, Jens Osterkamp
In-Reply-To: <1287746820-6036-1-git-send-email-jens@linux.vnet.ibm.com>
This patch implements the command line interface to control the VDP module.
In station role, it allows to register a new VSI (guest interface) profile
and query its state. In bridge role, it allows to query the state of all
registered profiles.
With lldptool it is possible to set the role from default "station" to
"bridge".
The configuration of the role for a port is saved to lldpads config file.
Signed-off-by: Jens Osterkamp <jens@linux.vnet.ibm.com>
---
Makefile.am | 12 +-
include/lldp_vdp_clif.h | 39 +++++
include/lldp_vdp_cmds.h | 44 +++++
lldp_vdp_clif.c | 136 ++++++++++++++++
lldp_vdp_cmds.c | 408 +++++++++++++++++++++++++++++++++++++++++++++++
lldptool.c | 2 +
6 files changed, 636 insertions(+), 5 deletions(-)
create mode 100644 include/lldp_vdp_clif.h
create mode 100644 include/lldp_vdp_cmds.h
create mode 100644 lldp_vdp_clif.c
create mode 100644 lldp_vdp_cmds.c
diff --git a/Makefile.am b/Makefile.am
index 4b69389..9a3baf4 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -37,7 +37,8 @@ lldpad_include_HEADERS = include/dcb_types.h include/dcbtool.h \
include/dcb_osdep.h include/clif.h include/lldp_dcbx_cmds.h include/common.h \
include/lldpad.h include/os.h include/includes.h include/lldp_mand_cmds.h \
include/clif_msgs.h include/lldp_basman_cmds.h include/lldp_8023_cmds.h \
-include/lldp_med_cmds.h include/lldp_dcbx_cfg.h include/lldp_evb_cmds.h
+include/lldp_med_cmds.h include/lldp_dcbx_cfg.h include/lldp_evb_cmds.h \
+include/lldp_vdp_cmds.h
noinst_HEADERS = include/config.h include/ctrl_iface.h \
include/dcb_driver_if_types.h include/dcb_driver_interface.h \
@@ -47,7 +48,7 @@ include/event_iface.h include/messages.h include/parse_cli.h include/version.h \
include/lldptool_cli.h include/list.h \
include/lldp_mand_clif.h include/lldp_basman_clif.h include/lldp_med_clif.h \
include/lldp_8023_clif.h include/lldp_dcbx_clif.h include/lldptool.h \
-include/lldp_rtnl.h include/lldp_evb_clif.h
+include/lldp_rtnl.h include/lldp_evb_clif.h include/lldp_vdp_clif.h
lldpad_SOURCES = lldpad.c config.c drv_cfg.c ctrl_iface.c event_iface.c eloop.c \
common.c os_unix.c lldp_dcbx_cmds.c log.c lldpad_shm.c \
@@ -64,20 +65,21 @@ lldp_dcbx_cfg.c include/lldp_dcbx_cfg.h \
lldp_util.c include/lldp_util.h \
lldp_mand.c include/lldp_mand.h \
lldp_mand_cmds.c lldp_basman_cmds.c lldp_8023_cmds.c lldp_med_cmds.c \
-lldp_evb_cmds.c \
+lldp_evb_cmds.c lldp_vdp_cmds.c \
lldp_tlv.c include/lldp_tlv.h \
lldp_basman.c include/lldp_basman.h \
lldp_med.c include/lldp_med.h \
lldp_8023.c include/lldp_8023.h \
lldp_evb.c include/lldp_evb.h \
-lldp_vdp.c include/lldp_vdp.h
+lldp_vdp.c include/lldp_vdp.h \
+lldp_vdp_cmds.h
dcbtool_SOURCES = dcbtool.c clif.c dcbtool_cmds.c parse_cli.l \
$(lldpad_include_HEADERS) $(noinst_HEADERS)
lldptool_SOURCES = lldptool.c clif.c lldptool_cmds.c common.c os_unix.c \
lldp_mand_clif.c lldp_basman_clif.c lldp_med_clif.c lldp_8023_clif.c \
-lldp_dcbx_clif.c lldp_evb_clif.c $(lldpad_include_HEADERS) \
+lldp_dcbx_clif.c lldp_evb_clif.c lldp_vdp_clif.c $(lldpad_include_HEADERS) \
$(noinst_HEADERS)
nltest_SOURCES = nltest.c nltest.h
diff --git a/include/lldp_vdp_clif.h b/include/lldp_vdp_clif.h
new file mode 100644
index 0000000..adb6333
--- /dev/null
+++ b/include/lldp_vdp_clif.h
@@ -0,0 +1,39 @@
+/*******************************************************************************
+
+ implementation of VDP according to IEEE 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#ifndef _LLDP_VDP_CLIF_H
+#define _LLDP_VDP_CLIF_H
+
+struct lldp_module *vdp_cli_register(void);
+void vdp_cli_unregister(struct lldp_module *);
+int vdp_print_tlv(u32, u16, char *);
+
+#define VDP_BUF_SIZE 256
+
+#define VDP_PREFIX "vdp"
+#define ARG_VDP_MODE "mode"
+#define ARG_VDP_ROLE "role"
+
+#endif
diff --git a/include/lldp_vdp_cmds.h b/include/lldp_vdp_cmds.h
new file mode 100644
index 0000000..5e64c27
--- /dev/null
+++ b/include/lldp_vdp_cmds.h
@@ -0,0 +1,44 @@
+/*******************************************************************************
+
+ implementation of VDP according to IEEE 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#ifndef _LLDP_VDP_CMDS_H
+#define _LLDP_VDP_CMDS_H
+
+struct arg_handlers *vdp_get_arg_handlers();
+
+enum {
+ MODE = 0,
+ MGRID,
+ TYPEID,
+ TYPEIDVERSION,
+ INSTANCEID,
+ MAC,
+ VLAN,
+};
+
+#define VAL_STATION "station"
+#define VAL_BRIDGE "bridge"
+
+#endif
diff --git a/lldp_vdp_clif.c b/lldp_vdp_clif.c
new file mode 100644
index 0000000..aa357e9
--- /dev/null
+++ b/lldp_vdp_clif.c
@@ -0,0 +1,136 @@
+/*******************************************************************************
+
+ implementation of VDP according to IEEE 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include "includes.h"
+#include "common.h"
+#include <stdio.h>
+#include <syslog.h>
+#include <sys/un.h>
+#include <sys/stat.h>
+#include "lldp_mod.h"
+#include "lldptool.h"
+#include "lldp.h"
+#include "lldp_vdp.h"
+#include "lldp_vdp_clif.h"
+
+void vdp_print_cfg_tlv(u16, char *info);
+int vdp_print_help();
+
+u32 vdp_lookup_tlv_name(char *tlvid_str);
+
+static const struct lldp_mod_ops vdp_ops_clif = {
+ .lldp_mod_register = vdp_cli_register,
+ .lldp_mod_unregister = vdp_cli_unregister,
+ .print_tlv = vdp_print_tlv,
+ .lookup_tlv_name = vdp_lookup_tlv_name,
+ .print_help = vdp_print_help,
+};
+
+struct type_name_info vdp_tlv_names[] = {
+ { ((LLDP_MOD_VDP) << 8) | LLDP_VDP_SUBTYPE,
+ "VDP protocol configuration",
+ "vdp", vdp_print_cfg_tlv },
+ { INVALID_TLVID, NULL, NULL }
+};
+
+int vdp_print_help()
+{
+ struct type_name_info *tn = &vdp_tlv_names[0];
+
+ while (tn->type != INVALID_TLVID) {
+ if (tn->key && strlen(tn->key) && tn->name) {
+ printf(" %s", tn->key);
+ if (strlen(tn->key)+3 <= 8)
+ printf("\t");
+ printf("\t: %s\n", tn->name);
+ }
+ tn++;
+ }
+
+ return 0;
+}
+
+struct lldp_module *vdp_cli_register(void)
+{
+ struct lldp_module *mod;
+
+ mod = malloc(sizeof(*mod));
+ if (!mod) {
+ fprintf(stderr, "failed to malloc module data\n");
+ return NULL;
+ }
+ mod->id = LLDP_MOD_VDP;
+ mod->ops = &vdp_ops_clif;
+
+ return mod;
+}
+
+void vdp_cli_unregister(struct lldp_module *mod)
+{
+ free(mod);
+}
+
+void vdp_print_cfg_tlv(u16 len, char *info)
+{
+ /* TODO: this should print out all associated VSI mac/vlan pairs */
+ printf("This should print out all associated VSI mac/vlan pairs !\n");
+
+ return;
+}
+
+/* return 1: if it printed the TLV
+ * 0: if it did not
+ */
+int vdp_print_tlv(u32 tlvid, u16 len, char *info)
+{
+ struct type_name_info *tn = &vdp_tlv_names[0];
+
+ while (tn->type != INVALID_TLVID) {
+ if (tlvid == tn->type) {
+ printf("%s\n", tn->name);
+ if (tn->print_info) {
+ printf("\t");
+ tn->print_info(len-4, info);
+ }
+ return 1;
+ }
+ tn++;
+ }
+
+ return 0;
+}
+
+u32 vdp_lookup_tlv_name(char *tlvid_str)
+{
+ struct type_name_info *tn = &vdp_tlv_names[0];
+
+ while (tn->type != INVALID_TLVID) {
+ if (!strcasecmp(tn->key, tlvid_str))
+ return tn->type;
+ tn++;
+ }
+ return INVALID_TLVID;
+}
+
diff --git a/lldp_vdp_cmds.c b/lldp_vdp_cmds.c
new file mode 100644
index 0000000..e2c578c
--- /dev/null
+++ b/lldp_vdp_cmds.c
@@ -0,0 +1,408 @@
+/******************************************************************************
+
+ implementation of VDP according to IEEE 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include "includes.h"
+#include "common.h"
+#include <stdio.h>
+#include <syslog.h>
+#include <sys/un.h>
+#include <sys/stat.h>
+#include <arpa/inet.h>
+#include "lldpad.h"
+#include "ctrl_iface.h"
+#include "lldp.h"
+#include "lldp_vdp.h"
+#include "lldp_mand_clif.h"
+#include "lldp_vdp_clif.h"
+#include "lldp_vdp_cmds.h"
+#include "lldp/ports.h"
+#include "messages.h"
+#include "libconfig.h"
+#include "config.h"
+#include "clif_msgs.h"
+#include "lldp/states.h"
+
+static int get_arg_tlvtxenable(struct cmd *, char *, char *, char *);
+static int set_arg_tlvtxenable(struct cmd *, char *, char *, char *);
+
+static int get_arg_mode(struct cmd *, char *, char *, char *);
+static int set_arg_mode(struct cmd *, char *, char *, char *);
+
+static int get_arg_role(struct cmd *, char *, char *, char *);
+static int set_arg_role(struct cmd *, char *, char *, char *);
+
+static struct arg_handlers arg_handlers[] = {
+ { ARG_VDP_MODE, get_arg_mode, set_arg_mode },
+ { ARG_VDP_ROLE, get_arg_role, set_arg_role },
+ { ARG_TLVTXENABLE, get_arg_tlvtxenable, set_arg_tlvtxenable },
+ { NULL }
+};
+
+static int get_arg_tlvtxenable(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ int value;
+ char *s;
+ char arg_path[VDP_BUF_SIZE];
+
+ if (cmd->cmd != cmd_gettlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case ((LLDP_MOD_VDP) << 8) | LLDP_VDP_SUBTYPE:
+ snprintf(arg_path, sizeof(arg_path), "%s.%s",
+ VDP_PREFIX, arg);
+
+ if (get_cfg(cmd->ifname, arg_path, (void *)&value,
+ CONFIG_TYPE_BOOL))
+ value = false;
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ if (value)
+ s = VAL_YES;
+ else
+ s = VAL_NO;
+
+ sprintf(obuf, "%02x%s%04x%s", (unsigned int) strlen(arg), arg,
+ (unsigned int) strlen(s), s);
+
+ return cmd_success;
+}
+
+static int set_arg_tlvtxenable(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ int value;
+ char arg_path[VDP_BUF_SIZE];
+
+ if (cmd->cmd != cmd_settlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case ((LLDP_MOD_VDP) << 8) | LLDP_VDP_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ if (!strcasecmp(argvalue, VAL_YES))
+ value = 1;
+ else if (!strcasecmp(argvalue, VAL_NO))
+ value = 0;
+ else
+ return cmd_invalid;
+
+ snprintf(arg_path, sizeof(arg_path), "%s.%s", VDP_PREFIX, arg);
+
+ if (set_cfg(cmd->ifname, arg_path, (void *)&value, CONFIG_TYPE_BOOL))
+ return cmd_failed;
+
+ return cmd_success;
+}
+
+static int get_arg_mode(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ char *s, *t;
+ struct vsi_profile *np;
+ struct vdp_data *vd;
+ int count=0;
+
+ if (cmd->cmd != cmd_gettlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case ((LLDP_MOD_VDP) << 8) | LLDP_VDP_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ vd = vdp_data(cmd->ifname);
+ if (!vd) {
+ LLDPAD_ERR("%s(%i): vdp_data for %s not found !\n", __func__, __LINE__,
+ cmd->ifname);
+ free(t);
+ return cmd_invalid;
+ }
+
+ LIST_FOREACH(np, &vd->profile_head, profile) {
+ count++;
+ }
+
+ s = t = malloc(count*VDP_BUF_SIZE);
+ if (!s)
+ return cmd_invalid;
+ memset(s, 0, count*VDP_BUF_SIZE);
+
+ LIST_FOREACH(np, &vd->profile_head, profile) {
+ PRINT_PROFILE(t, np);
+ }
+
+ sprintf(obuf, "%02x%s%04x%s", (unsigned int) strlen(arg), arg,
+ (unsigned int) strlen(s), s);
+
+ free(s);
+
+ return cmd_success;
+}
+
+static void str2instance(struct vsi_profile *profile, char *buffer)
+{
+ int i, j = 0;
+
+ for(i=0; i <= strlen(buffer); i++) {
+ if (buffer[i] == '-') {
+ continue;
+ }
+
+ if ((sscanf(&buffer[i], "%02x", &profile->instance[j]) == 1) ||
+ (sscanf(&buffer[i], "%02X", &profile->instance[j]) == 1)) {
+ i++;
+ j++;
+ continue;
+ }
+ }
+
+}
+
+/* INSTANCE_STRLEN = strlen("fa9b7fff-b0a0-4893e-beef4ff18f8f") */
+#define INSTANCE_STRLEN 32
+
+int instance2str(const u8 *p, char *dst, size_t size)
+{
+ if (dst && size > INSTANCE_STRLEN) {
+ snprintf(dst, size, "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x",
+ p[0], p[1], p[2], p[3],
+ p[4], p[5], p[6], p[7],
+ p[8], p[9], p[10], p[11],
+ p[12], p[13], p[14], p[15]);
+ return 0;
+ }
+ return -1;
+}
+
+static void vdp_fill_profile(struct vsi_profile *profile, char *buffer, int field)
+{
+ switch(field) {
+ case MODE:
+ profile->mode = atoi(buffer);
+ break;
+ case MGRID:
+ profile->mgrid = atoi(buffer);
+ break;
+ case TYPEID:
+ profile->id = atoi(buffer);
+ break;
+ case TYPEIDVERSION:
+ profile->version = atoi(buffer);
+ break;
+ case INSTANCEID:
+ str2instance(profile, buffer);
+ break;
+ case MAC:
+ str2mac(buffer, &profile->mac[0], MAC_ADDR_LEN);
+ break;
+ case VLAN:
+ profile->vlan = atoi(buffer);
+ break;
+ default:
+ LLDPAD_ERR("Unknown field in buffer !\n");
+ break;
+ }
+}
+
+static struct vsi_profile *vdp_parse_mode_line(char * argvalue)
+{
+ int i, arglen, field;
+ char *cmdstring, *buf;
+ char *buffer;
+ struct vsi_profile *profile;
+
+ profile = malloc(sizeof(struct vsi_profile));
+ if (!profile)
+ return NULL;
+ memset(profile, 0, sizeof(struct vsi_profile));
+
+ arglen = strlen(argvalue);
+ cmdstring = argvalue;
+ buffer = malloc(arglen);
+ if (!buffer)
+ goto out_free;
+ buf = buffer;
+ field = 0;
+
+ for (i=0; i <= arglen; i++) {
+ *buffer = *cmdstring;
+
+ if ((*cmdstring == ',') || (*cmdstring == '\0')) {
+ *buffer++ = '\0';
+ vdp_fill_profile(profile, buf, field);
+ field++;
+ buffer = buf;
+ memset(buffer, 0, arglen);
+ cmdstring++;
+ continue;
+ }
+
+ buffer++;
+ cmdstring++;
+ }
+
+ return profile;
+
+out_free:
+ free(profile);
+ return NULL;
+}
+
+static int set_arg_mode(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ int arglen;
+ struct vsi_profile *profile, *p;
+
+ arglen = strlen(argvalue);
+
+ if (cmd->cmd != cmd_settlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case ((LLDP_MOD_VDP) << 8) | LLDP_VDP_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ profile = vdp_parse_mode_line(argvalue);
+ profile->port = port_find_by_name(cmd->ifname);
+
+ if (!profile->port) {
+ free(profile);
+ return cmd_invalid;
+ }
+
+ p = vdp_add_profile(profile);
+
+ if (!p) {
+ free(profile);
+ return cmd_invalid;
+ }
+
+ vdp_somethingChangedLocal(profile, VDP_PROFILE_REQ);
+ vdp_vsi_sm_station(p);
+
+ return cmd_success;
+}
+
+static int get_arg_role(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ char *p;
+ char arg_path[VDP_BUF_SIZE];
+
+ if (cmd->cmd != cmd_gettlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case ((LLDP_MOD_VDP) << 8) | LLDP_VDP_SUBTYPE:
+ snprintf(arg_path, sizeof(arg_path), "%s.%s",
+ VDP_PREFIX, arg);
+
+ if (get_cfg(cmd->ifname, arg_path, (void *)&p,
+ CONFIG_TYPE_STRING))
+ return cmd_failed;
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ sprintf(obuf, "%02x%s%04x%s", (unsigned int) strlen(arg), arg,
+ (unsigned int) strlen(p), p);
+
+ return cmd_success;
+}
+
+static int set_arg_role(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ struct vdp_data *vd;
+ char arg_path[VDP_BUF_SIZE];
+
+ if (cmd->cmd != cmd_settlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case ((LLDP_MOD_VDP) << 8) | LLDP_VDP_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ vd = vdp_data(cmd->ifname);
+
+ if (!vd) {
+ LLDPAD_ERR("%s(%i): could not find vdp_data for %s !\n",
+ __FILE__, __LINE__, cmd->ifname);
+ return cmd_invalid;
+ }
+
+ if (!strcasecmp(argvalue, VAL_BRIDGE)) {
+ vd->role = VDP_ROLE_BRIDGE;
+ } else if (!strcasecmp(argvalue, VAL_STATION)) {
+ vd->role = VDP_ROLE_STATION;
+ } else {
+ return cmd_invalid;
+ }
+
+ snprintf(arg_path, sizeof(arg_path), "%s.%s", VDP_PREFIX, arg);
+
+ char *p = &argvalue[0];
+ if (set_cfg(cmd->ifname, arg_path, (void *)&p, CONFIG_TYPE_STRING))
+ return cmd_failed;
+
+ return cmd_success;
+}
+
+
+struct arg_handlers *vdp_get_arg_handlers()
+{
+ return &arg_handlers[0];
+}
diff --git a/lldptool.c b/lldptool.c
index 7e166fe..bfed101 100644
--- a/lldptool.c
+++ b/lldptool.c
@@ -40,6 +40,7 @@
#include "lldp_8023_clif.h"
#include "lldp_dcbx_clif.h"
#include "lldp_evb_clif.h"
+#include "lldp_vdp_clif.h"
#include "lldptool.h"
#include "lldptool_cli.h"
#include "lldp_mod.h"
@@ -158,6 +159,7 @@ struct lldp_module *(*register_tlv_table[])(void) = {
med_cli_register,
dcbx_cli_register,
evb_cli_register,
+ vdp_cli_register,
NULL,
};
--
1.7.1
^ permalink raw reply related
* [PATCH 4/9] implementation of IEEE 802.1Qbg in lldpad, part 2
From: Jens Osterkamp @ 2010-10-22 11:26 UTC (permalink / raw)
To: e1000-eedc, virtualization, evb; +Cc: chrisw, Jens Osterkamp
In-Reply-To: <1287746820-6036-1-git-send-email-jens@linux.vnet.ibm.com>
This is the implementation of the edge control protocol (ECP) and VSI
discovery protocol (VDP) currently being specified in IEEE 802.1Qbg.
This implementation extends the infrastructure defined lldpad to send and
receive ECP frames with a new (yet to be defined) ethertype.
Received frames are validated and analyzed before the content is handed to the
upper layer protocol (ULP, VDP in this case) for further processing. Frames
to be transmitted are compiled from VSI (guest interface) profiles registered
on a interface.
Reception and transmission of ECP frames is controlled by RX and TX state
machines, timeouts are handled timeout functions.
The patch still contains a lot of debug code to allow low-level protocol
analysis.
VDP serves as the upper layer protocol (ULP) for TLVs communicated via the
ECP protocol.
For this it registers as a new module in lldpad. The VDP module supports a
station and a bridge role. As a station, new VSI (virtual station interface)
profiles can be registered to the VDP module using lldptool or libvirt.
These profiles are then announced to an adjacent switch. Transmitted profiles
are processed to the desired state by the VDP station state machine.
As a bridge, the VDP module waits for new profiles received in TLVs by ECP.
The received profiles are processed to the desired state by a VDP bridge
state machine.
VDP module parameters are stored in the "vdp" section under the appropriate
interface.
The patch still contains a lot of debug code to allow analysis of VDP
protocol behavior.
Signed-off-by: Jens Osterkamp <jens@linux.vnet.ibm.com>
---
Makefile.am | 10 +-
ecp/ecp.c | 102 +++++
ecp/ecp.h | 101 +++++
ecp/ecp_rx.c | 610 ++++++++++++++++++++++++++
ecp/ecp_tx.c | 506 ++++++++++++++++++++++
include/lldp.h | 1 +
include/lldp_evb.h | 6 +
include/lldp_vdp.h | 159 +++++++
lldp/l2_packet.h | 2 +
lldp/ports.h | 11 +-
lldp_vdp.c | 1196 ++++++++++++++++++++++++++++++++++++++++++++++++++++
lldpad.c | 2 +
12 files changed, 2699 insertions(+), 7 deletions(-)
create mode 100644 ecp/ecp.c
create mode 100644 ecp/ecp.h
create mode 100644 ecp/ecp_rx.c
create mode 100644 ecp/ecp_tx.c
create mode 100644 include/lldp_vdp.h
create mode 100644 lldp_vdp.c
diff --git a/Makefile.am b/Makefile.am
index d59a6fa..4b69389 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -56,6 +56,8 @@ $(lldpad_include_HEADERS) $(noinst_HEADERS) \
lldp/ports.c lldp/agent.c lldp/l2_packet_linux.c lldp/tx.c \
lldp/rx.c lldp/agent.h lldp/l2_packet.h lldp/mibdata.h lldp/ports.h \
lldp/states.h \
+ecp/ecp.c ecp/ecp_tx.c \
+ecp/ecp_rx.c \
include/lldp.h include/lldp_mod.h \
lldp_dcbx.c include/lldp_dcbx.h tlv_dcbx.c include/tlv_dcbx.h \
lldp_dcbx_cfg.c include/lldp_dcbx_cfg.h \
@@ -67,16 +69,16 @@ lldp_tlv.c include/lldp_tlv.h \
lldp_basman.c include/lldp_basman.h \
lldp_med.c include/lldp_med.h \
lldp_8023.c include/lldp_8023.h \
-lldp_evb.c include/lldp_evb.h
-
-
+lldp_evb.c include/lldp_evb.h \
+lldp_vdp.c include/lldp_vdp.h
dcbtool_SOURCES = dcbtool.c clif.c dcbtool_cmds.c parse_cli.l \
$(lldpad_include_HEADERS) $(noinst_HEADERS)
lldptool_SOURCES = lldptool.c clif.c lldptool_cmds.c common.c os_unix.c \
lldp_mand_clif.c lldp_basman_clif.c lldp_med_clif.c lldp_8023_clif.c \
-lldp_dcbx_clif.c lldp_evb_clif.c $(lldpad_include_HEADERS) $(noinst_HEADERS)
+lldp_dcbx_clif.c lldp_evb_clif.c $(lldpad_include_HEADERS) \
+$(noinst_HEADERS)
nltest_SOURCES = nltest.c nltest.h
diff --git a/ecp/ecp.c b/ecp/ecp.c
new file mode 100644
index 0000000..44c3ba0
--- /dev/null
+++ b/ecp/ecp.c
@@ -0,0 +1,102 @@
+/*******************************************************************************
+
+ implementation of ECP according to 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include <net/if.h>
+#include <sys/queue.h>
+#include <sys/socket.h>
+#include <sys/ioctl.h>
+#include <sys/utsname.h>
+#include <linux/if_bridge.h>
+#include "lldp.h"
+#include "lldp_evb.h"
+#include "lldp_vdp.h"
+#include "messages.h"
+#include "config.h"
+#include "common.h"
+#include "lldp/l2_packet.h"
+#include "ecp/ecp.h"
+
+/* ecp_init - initialize ecp module
+ * @ifname: interface for which the module is initialized
+ *
+ * returns 0 on success, -1 on error
+ *
+ * finds the port to the interface name, sets up the receive handle for
+ * incoming ecp frames and initializes the ecp rx and tx state machines.
+ * should usually be called when a successful exchange of EVB TLVs has been
+ * made and ECP and VDP protocols are supported by both sides.
+ */
+int ecp_init(char *ifname)
+{
+ struct vdp_data *vd;
+
+ LLDPAD_DBG("%s(%i): starting ECP for if %s !\n", __func__, __LINE__, ifname);
+
+ vd = vdp_data(ifname);
+
+ if (!vd) {
+ LLDPAD_ERR("%s(%i): unable to find vd %s ! \n", __func__, __LINE__, ifname);
+ goto fail;
+ }
+
+ vd->ecp.l2 = l2_packet_init(vd->ifname, NULL, ETH_P_ECP,
+ ecp_rx_ReceiveFrame, vd, 1);
+ if (!vd->ecp.l2) {
+ LLDPAD_ERR("ERROR: Failed to open register layer 2 access to "
+ "ETH_P_ECP\n");
+ goto fail;
+ }
+
+ ecp_tx_run_sm(vd);
+ ecp_rx_run_sm(vd);
+
+ return 0;
+
+fail:
+ return -1;
+}
+
+int ecp_deinit(char *ifname)
+{
+ struct vdp_data *vd;
+
+ LLDPAD_DBG("%s(%i): stopping ECP for if %s !\n", __func__, __LINE__, ifname);
+
+ vd = vdp_data(ifname);
+
+ if (!vd) {
+ LLDPAD_ERR("%s(%i): unable to find vd %s ! \n", __func__, __LINE__, ifname);
+ goto fail;
+ }
+
+ ecp_tx_stop_ackTimer(vd);
+
+ l2_packet_deinit(vd->ecp.l2);
+
+ return 0;
+
+fail:
+ return -1;
+}
diff --git a/ecp/ecp.h b/ecp/ecp.h
new file mode 100644
index 0000000..cc9ca2b
--- /dev/null
+++ b/ecp/ecp.h
@@ -0,0 +1,101 @@
+/*******************************************************************************
+
+ implementation of ECP according to 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#ifndef _ECP_H
+#define _ECP_H
+
+#include "lldp_mod.h"
+#include "include/lldp_vdp.h"
+
+#define ECP_SUBTYPE 0x0
+
+#define ECP_MAX_RETRIES 3
+#define ECP_SEQUENCE_NR_START 0x0
+
+#define ECP_TRANSMISSION_TIMER EVB_RTM(RTE)*EVB_RTG
+#define ECP_TRANSMISSION_DIVIDER 10000
+
+typedef enum {
+ ECP_REQUEST = 0,
+ ECP_ACK
+} ecp_mode;
+
+struct ecp {
+ struct l2_packet_data *l2;
+ int sequence;
+ int retries;
+ int ackReceived;
+ int ackTimerExpired;
+ u16 lastSequence;
+ u16 seqECPDU;
+ struct portrx rx;
+ struct porttx tx;
+ struct portstats stats;
+};
+
+struct ecp_hdr {
+ u8 oui[3];
+ u8 pad1;
+ u16 subtype;
+ u8 mode;
+ u16 seqnr;
+} __attribute__ ((__packed__));
+
+enum {
+ ECP_TX_INIT_TRANSMIT,
+ ECP_TX_TRANSMIT_ECPDU,
+ ECP_TX_WAIT_FOR_ACK,
+ ECP_TX_REQUEST_PDU
+};
+
+static const char *ecp_tx_states[] = {
+ "ECP_TX_IDLE",
+ "ECP_TX_INIT_TRANSMIT",
+ "ECP_TX_TRANSMIT_ECPDU",
+ "ECP_TX_WAIT_FOR_ACK",
+ "ECP_TX_REQUEST_PDU"
+};
+
+enum {
+ ECP_RX_IDLE,
+ ECP_RX_INIT_RECEIVE,
+ ECP_RX_RECEIVE_WAIT,
+ ECP_RX_RECEIVE_ECPDU,
+ ECP_RX_SEND_ACK,
+ ECP_RX_RESEND_ACK,
+};
+
+static const char *ecp_rx_states[] = {
+ "ECP_RX_IDLE",
+ "ECP_RX_INIT_RECEIVE",
+ "ECP_RX_RECEIVE_WAIT",
+ "ECP_RX_RECEIVE_ECPDU",
+ "ECP_RX_SEND_ACK",
+ "ECP_RX_RESEND_ACK",
+};
+
+void ecp_rx_ReceiveFrame(void *, unsigned int, const u8 *, size_t );
+
+#endif /* _ECP_H */
diff --git a/ecp/ecp_rx.c b/ecp/ecp_rx.c
new file mode 100644
index 0000000..299377f
--- /dev/null
+++ b/ecp/ecp_rx.c
@@ -0,0 +1,610 @@
+/*******************************************************************************
+
+ implementation of ECP according to 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include "dcb_osdep.h"
+#include "lldp/ports.h"
+#include "lldp/l2_packet.h"
+#include "messages.h"
+#include "lldp.h"
+#include "lldpad.h"
+#include "lldp_mod.h"
+#include "clif_msgs.h"
+#include "lldp_mand.h"
+#include "lldp_vdp.h"
+
+/* ecp_rx_Initialize - initializes the ecp rx state machine
+ * @vd: vd for the state machine
+ *
+ * no return value
+ *
+ * initialize some variables, get rid of old frame if necessary
+ */
+void ecp_rx_Initialize(struct vdp_data *vd)
+{
+ vd->ecp.rx.rcvFrame = false;
+ vd->ecp.rx.badFrame = false;
+
+ vd->ecp.ackReceived = 0;
+
+ if (vd->ecp.rx.framein) {
+ free(vd->ecp.rx.framein);
+ vd->ecp.rx.framein = NULL;
+ }
+ vd->ecp.rx.sizein = 0;
+
+ return;
+}
+
+/* ecp_rx_freeFrame - free up received frame
+ * @vd: vd for the state machine
+ *
+ * no return value
+ *
+ * frees up an old received frame, set pointer to NULL and size to 0.
+ */
+void ecp_rx_freeFrame(struct vdp_data *vd)
+{
+ free(vd->ecp.rx.framein);
+ vd->ecp.rx.framein = NULL;
+ vd->ecp.rx.sizein = 0;
+}
+
+/* ecp_print_framein - print raw received frame
+ * @vd: vd for the state machine
+ *
+ * no return value
+ *
+ * prints out a raw version of a received frame. useful for low-level protocol
+ * debugging.
+ */
+void ecp_print_framein(struct vdp_data *vd)
+{
+ int i;
+ char *s, *t;
+
+ s = t = malloc(256);
+
+ if (!s) {
+ LLDPAD_ERR("%s(%i): unable to allocate string !\n", __func__, __LINE__);
+ }
+
+ for (i=0; i < vd->ecp.rx.sizein; i++) {
+ int c;
+ c = sprintf(s, "%02x ", vd->ecp.rx.framein[i]);
+ s += c;
+ if (!((i+1) % 16)) {
+ LLDPAD_DBG("%s\n", t);
+ s = t;
+ }
+ }
+
+ LLDPAD_DBG("%s\n", t);
+
+ free(t);
+}
+
+/* ecp_rx_SendAckFrame - send ack frame
+ * @vd: port used by ecp
+ *
+ * currently always returns 0
+ *
+ * copies current received frame over to frame out, fills in address of this
+ * port and set mode field to ACK. used by ecp_rx_send_ack_frame.
+ */
+int ecp_rx_SendAckFrame(struct vdp_data *vd)
+{
+ u16 tlv_offset = 0;
+ struct ecp_hdr *ecp_hdr;
+ struct l2_ethhdr *hdr;
+ u8 own_addr[ETH_ALEN];
+
+ LLDPAD_DBG("%s(%i)-%s: acking frame \n", __func__, __LINE__, vd->ifname);
+
+ assert(vd->ecp.rx.framein && vd->ecp.rx.sizein);
+
+ /* copy over to frameout */
+ vd->ecp.tx.frameout = (u8 *)malloc(ETH_FRAME_LEN);
+ memcpy(vd->ecp.tx.frameout, vd->ecp.rx.framein, vd->ecp.rx.sizein);
+ vd->ecp.tx.sizeout = vd->ecp.rx.sizein;
+
+ /* use my own addr to send ACK */
+ hdr = (struct l2_ethhdr *)vd->ecp.tx.frameout;
+ l2_packet_get_own_src_addr(vd->ecp.l2,(u8 *)&own_addr);
+ memcpy(hdr->h_source, &own_addr, ETH_ALEN);
+
+ tlv_offset = sizeof(struct l2_ethhdr);
+ ecp_hdr = (struct ecp_hdr *)&vd->ecp.tx.frameout[tlv_offset];
+ ecp_hdr->mode = ECP_ACK;
+
+ return 0;
+}
+
+/* ecp_rx_send_ack_frame - send out ack frame for received frame
+ * @vd: vd for the state machine
+ *
+ * no return value
+ *
+ * creates an ack frame for a just received frame, prints the about to be
+ * sent frame and finally transmits it.
+ */
+void ecp_rx_send_ack_frame(struct vdp_data *vd)
+{
+ ecp_rx_SendAckFrame(vd);
+
+ ecp_print_frameout(vd);
+
+ ecp_txFrame(vd);
+
+ return;
+}
+
+/* ecp_rx_ReceiveFrame - receive ecp frame
+ * @ctx: rx callback context, struct vd * in this case
+ * @ifindex: index of interface
+ * @buf: buffer which contains the frame just received
+ * @len: size of buffer (frame)
+ *
+ * no return value
+ *
+ * creates a local copy of the buffer and checks the header. keeps some
+ * statistics about ecp frames. Checks if it is a request or an ack frame and branches
+ * to ecp rx or ecp tx state machine.
+ */
+void ecp_rx_ReceiveFrame(void *ctx, unsigned int ifindex, const u8 *buf, size_t len)
+{
+
+ struct vdp_data *vd = NULL;
+ struct port *port;
+ u8 frame_error = 0;
+ u16 tlv_offset;
+ struct l2_ethhdr *hdr;
+ struct l2_ethhdr example_hdr,*ex;
+ struct ecp_hdr *ecp_hdr;
+
+ if (ctx)
+ vd = (struct vdp_data *)ctx;
+
+ port = port_find_by_name(vd->ifname);
+
+ LLDPAD_DBG("%s(%i)-%s: received packet with size %i\n", __func__, __LINE__,
+ vd->ifname, (int) len);
+
+ if (port->adminStatus == disabled || port->adminStatus == enabledTxOnly)
+ return;
+
+ if (vd->ecp.rx.framein &&
+ vd->ecp.rx.sizein == len &&
+ (memcmp(buf, vd->ecp.rx.framein, len) == 0)) {
+ vd->ecp.stats.statsFramesInTotal++;
+ return;
+ }
+
+ if (vd->ecp.rx.framein)
+ free(vd->ecp.rx.framein);
+
+ vd->ecp.rx.framein = (u8 *)malloc(len);
+ if (vd->ecp.rx.framein == NULL) {
+ LLDPAD_ERR("ERROR - could not allocate memory for rx'ed frame\n");
+ return;
+ }
+ memcpy(vd->ecp.rx.framein, buf, len);
+
+ vd->ecp.rx.sizein = (u16)len;
+ ex = &example_hdr;
+ memcpy(ex->h_dest, multi_cast_source, ETH_ALEN);
+ ex->h_proto = htons(ETH_P_ECP);
+ hdr = (struct l2_ethhdr *)vd->ecp.rx.framein;
+
+ if ((memcmp(hdr->h_dest,ex->h_dest, ETH_ALEN) != 0)) {
+ LLDPAD_ERR("ERROR multicast address error in incoming frame. "
+ "Dropping frame.\n");
+ frame_error++;
+ free(vd->ecp.rx.framein);
+ vd->ecp.rx.framein = NULL;
+ vd->ecp.rx.sizein = 0;
+ return;
+ }
+
+ if (hdr->h_proto != example_hdr.h_proto) {
+ LLDPAD_ERR("ERROR Ethertype not ECP ethertype but ethertype "
+ "'%x' in incoming frame.\n", htons(hdr->h_proto));
+ frame_error++;
+ free(vd->ecp.rx.framein);
+ vd->ecp.rx.framein = NULL;
+ vd->ecp.rx.sizein = 0;
+ return;
+ }
+
+ if (!frame_error) {
+ vd->ecp.stats.statsFramesInTotal++;
+ vd->ecp.rx.rcvFrame = 1;
+ }
+
+ tlv_offset = sizeof(struct l2_ethhdr);
+
+ ecp_hdr = (struct ecp_hdr *)&vd->ecp.rx.framein[tlv_offset];
+
+ vd->ecp.seqECPDU = ntohs(ecp_hdr->seqnr);
+
+ switch(ecp_hdr->mode) {
+ case ECP_REQUEST:
+ vd->ecp.ackReceived = false;
+ ecp_print_framein(vd);
+ ecp_rx_run_sm(vd);
+ break;
+ case ECP_ACK:
+ vd->ecp.ackReceived = true;
+ LLDPAD_DBG("%s(%i)-%s: received ack frame \n", __func__, __LINE__, vd->ifname);
+ ecp_print_framein(vd);
+ ecp_tx_run_sm(vd);
+ vd->ecp.ackReceived = false;
+ break;
+ default:
+ LLDPAD_ERR("ERROR: unknown mode %i!\n", ecp_hdr->mode);
+ return;
+ }
+
+ ecp_rx_freeFrame(vd);
+}
+
+/* ecp_rx_validateFrame - validates received frame
+ * @vd: vdp_data used by ecp
+ *
+ * no return value
+ *
+ * checks wether received frame has correct subtype and mode
+ */
+
+void ecp_rx_validateFrame(struct vdp_data *vd)
+{
+ u16 tlv_offset = 0;
+ struct ecp_hdr *ecp_hdr;
+
+ LLDPAD_DBG("%s(%i)-%s: validating frame \n", __func__, __LINE__, vd->ifname);
+
+ assert(vd->ecp.rx.framein && vd->ecp.rx.sizein);
+
+ tlv_offset = sizeof(struct l2_ethhdr);
+
+ ecp_hdr = (struct ecp_hdr *)&vd->ecp.rx.framein[tlv_offset];
+
+ LLDPAD_DBG("%s(%i)-%s: ecp packet with subtype %04x, mode %02x, sequence nr %04x\n",
+ __func__, __LINE__, vd->ifname, ecp_hdr->subtype, ecp_hdr->mode, ntohs(ecp_hdr->seqnr));
+
+ if (ecp_hdr->subtype != ECP_SUBTYPE) {
+ LLDPAD_ERR("ERROR: unknown subtype !\n");
+ return;
+ }
+
+ if ((ecp_hdr->oui[0] != 0x0) || (ecp_hdr->oui[1] != 0x1b) ||
+ (ecp_hdr->oui[2] != 0x3f)) {
+ LLDPAD_ERR("ERROR: incorrect OUI 0x%02x%02x%02x !\n",
+ ecp_hdr->oui[0], ecp_hdr->oui[1], ecp_hdr->oui[2]);
+ return;
+ }
+
+ switch(ecp_hdr->mode) {
+ case ECP_REQUEST:
+ break;
+ case ECP_ACK:
+ break;
+ default:
+ LLDPAD_ERR("ERROR: unknown mode %i!\n", ecp_hdr->mode);
+ return;
+ }
+
+ /* FIXME: also done in ecp_rx_ReceiveFrame,
+ * are both necessary ? */
+ vd->ecp.seqECPDU = ntohs(ecp_hdr->seqnr);
+}
+
+/* ecp_rx_validate_frame - wrapper around ecp_rx_validateFrame
+ * @vd: currently used port
+ *
+ * no return value
+ *
+ * sets rcvFrame to false and validates frame. used in ECP_RX_RECEIVE_ECPDU
+ * state of ecp_rx_run_sm
+ */
+void ecp_rx_validate_frame(struct vdp_data *vd)
+{
+ vd->ecp.rx.rcvFrame = false;
+ ecp_rx_validateFrame(vd);
+ return;
+}
+
+/* ecp_rx_ProcessFrame - process received ecp frames
+ * @vd: currently used port
+ *
+ * no return value
+ *
+ * walks through the packed vsi tlvs in an ecp frame, extracts them
+ * and passes them to the VDP ULP with vdp_indicate.
+ */
+void ecp_rx_ProcessFrame(struct vdp_data *vd)
+{
+ u16 tlv_cnt = 0;
+ u8 tlv_type = 0;
+ u16 tlv_length = 0;
+ u16 tlv_offset = 0;
+ u16 *tlv_head_ptr = NULL;
+ u8 frame_error = 0;
+ bool tlv_stored = false;
+ struct ecp_hdr *ecp_hdr;
+
+ LLDPAD_DBG("%s(%i)-%s: processing frame \n", __func__, __LINE__, vd->ifname);
+
+ assert(vd->ecp.rx.framein && vd->ecp.rx.sizein);
+
+ tlv_offset = sizeof(struct l2_ethhdr);
+
+ ecp_hdr = (struct ecp_hdr *)&vd->ecp.rx.framein[tlv_offset];
+
+ LLDPAD_DBG("%s(%i)-%s: ecp packet with subtype %04x, mode %02x, sequence nr %04x\n",
+ __func__, __LINE__, vd->ifname, ecp_hdr->subtype, ecp_hdr->mode, ntohs(ecp_hdr->seqnr));
+
+ if (ecp_hdr->mode == ECP_ACK)
+ return;
+
+ /* processing of VSI_TLVs starts here */
+
+ tlv_offset += sizeof(struct ecp_hdr);
+
+ do {
+ tlv_cnt++;
+ if (tlv_offset > vd->ecp.rx.sizein) {
+ LLDPAD_ERR("%s(%i)-%s: ERROR: Frame overrun! tlv_offset %i, sizein %i cnt %i\n",
+ __func__, __LINE__, vd->ifname, tlv_offset, vd->ecp.rx.sizein, tlv_cnt);
+ frame_error++;
+ goto out;
+ }
+
+ tlv_head_ptr = (u16 *)&vd->ecp.rx.framein[tlv_offset];
+ tlv_length = htons(*tlv_head_ptr) & 0x01FF;
+ tlv_type = (u8)(htons(*tlv_head_ptr) >> 9);
+
+ u16 tmp_offset = tlv_offset + tlv_length;
+ if (tmp_offset > vd->ecp.rx.sizein) {
+ LLDPAD_ERR("ERROR: Frame overflow error: offset=%d, "
+ "rx.size=%d \n", tmp_offset, vd->ecp.rx.sizein);
+ frame_error++;
+ goto out;
+ }
+
+ u8 *info = (u8 *)&vd->ecp.rx.framein[tlv_offset +
+ sizeof(*tlv_head_ptr)];
+
+ struct unpacked_tlv *tlv = create_tlv();
+
+ if (!tlv) {
+ LLDPAD_ERR("ERROR: Failed to malloc space for "
+ "incoming TLV. \n");
+ goto out;
+ }
+
+ if ((tlv_length == 0) && (tlv->type != TYPE_0)) {
+ LLDPAD_ERR("ERROR: tlv_length == 0\n");
+ free_unpkd_tlv(tlv);
+ goto out;
+ }
+
+ tlv->type = tlv_type;
+ tlv->length = tlv_length;
+ tlv->info = (u8 *)malloc(tlv_length);
+ if (tlv->info) {
+ memset(tlv->info,0, tlv_length);
+ memcpy(tlv->info, info, tlv_length);
+ } else {
+ LLDPAD_ERR("ERROR: Failed to malloc space for incoming "
+ "TLV info \n");
+ free_unpkd_tlv(tlv);
+ goto out;
+ }
+
+ /* Validate the TLV */
+ tlv_offset += sizeof(*tlv_head_ptr) + tlv_length;
+
+ if (tlv->type == TYPE_127) { /* private TLV */
+ /* give VSI TLV to VDP */
+ if (!vdp_indicate(vd, tlv, ecp_hdr->mode))
+ tlv_stored = true;
+ else {
+ /* TODO: put it in a list and try again later until
+ * timer and retries have expired */
+ tlv_stored = false;
+ }
+ }
+
+ if ((tlv->type != TYPE_0) && !tlv_stored) {
+ LLDPAD_DBG("\n%s: allocated TLV (%u) "
+ " was not stored! (%p)\n", __func__, tlv->type,
+ tlv);
+ tlv = free_unpkd_tlv(tlv);
+ vd->ecp.stats.statsTLVsUnrecognizedTotal++;
+ }
+
+ tlv = NULL;
+ tlv_stored = false;
+ } while(tlv_type != 0);
+
+out:
+ if (frame_error) {
+ vd->ecp.stats.statsFramesDiscardedTotal++;
+ vd->ecp.stats.statsFramesInErrorsTotal++;
+ vd->ecp.rx.badFrame = true;
+ }
+
+ return;
+}
+
+/* ecp_rx_change_state - changes the ecp rx sm state
+ * @vd: currently used port
+ * @newstate: new state for the sm
+ *
+ * no return value
+ *
+ * checks state transistion for consistency and finally changes the state of
+ * the profile.
+ */
+void ecp_rx_change_state(struct vdp_data *vd, u8 newstate)
+{
+ switch(newstate) {
+ case ECP_RX_IDLE:
+ break;
+ case ECP_RX_INIT_RECEIVE:
+ break;
+ case ECP_RX_RECEIVE_WAIT:
+ assert((vd->ecp.rx.state == ECP_RX_INIT_RECEIVE) ||
+ (vd->ecp.rx.state == ECP_RX_IDLE) ||
+ (vd->ecp.rx.state == ECP_RX_SEND_ACK) ||
+ (vd->ecp.rx.state == ECP_RX_RESEND_ACK));
+ break;
+ case ECP_RX_RECEIVE_ECPDU:
+ assert(vd->ecp.rx.state == ECP_RX_RECEIVE_WAIT);
+ break;
+ case ECP_RX_SEND_ACK:
+ assert(vd->ecp.rx.state == ECP_RX_RECEIVE_ECPDU);
+ break;
+ case ECP_RX_RESEND_ACK:
+ assert(vd->ecp.rx.state == ECP_RX_RECEIVE_ECPDU);
+ break;
+ default:
+ LLDPAD_ERR("ERROR: The ECP_RX State Machine is broken!\n");
+ log_message(MSG_ERR_RX_SM_INVALID, "%s", vd->ifname);
+ }
+
+ LLDPAD_DBG("%s(%i)-%s: state change %s -> %s\n", __func__, __LINE__,
+ vd->ifname, ecp_rx_states[vd->ecp.rx.state], ecp_rx_states[newstate]);
+
+ vd->ecp.rx.state = newstate;
+}
+
+/* ecp_set_rx_state - sets the ecp rx sm state
+ * @vd: currently used port
+ *
+ * returns true or false
+ *
+ * switches the state machine to the next state depending on the input
+ * variables. returns true or false depending on wether the state machine
+ * can be run again with the new state or can stop at the current state.
+ */
+bool ecp_set_rx_state(struct vdp_data *vd)
+{
+ struct port *port = port_find_by_name(vd->ifname);
+
+ if (port->portEnabled == false) {
+ ecp_rx_change_state(vd, ECP_RX_IDLE);
+ }
+
+ switch(vd->ecp.rx.state) {
+ case ECP_RX_IDLE:
+ if (port->portEnabled == true) {
+ ecp_rx_change_state(vd, ECP_RX_INIT_RECEIVE);
+ return true;
+ }
+ return false;
+ case ECP_RX_INIT_RECEIVE:
+ if ((port->adminStatus == enabledRxTx) ||
+ (port->adminStatus == enabledRxOnly)) {
+ ecp_rx_change_state(vd, ECP_RX_RECEIVE_WAIT);
+ return true;
+ }
+ return false;
+ case ECP_RX_RECEIVE_WAIT:
+ if ((port->adminStatus == disabled) ||
+ (port->adminStatus == enabledTxOnly)) {
+ ecp_rx_change_state(vd, ECP_RX_IDLE);
+ return true;
+ }
+ if (vd->ecp.rx.rcvFrame == true) {
+ ecp_rx_change_state(vd, ECP_RX_RECEIVE_ECPDU);
+ return true;
+ }
+ return false;
+ case ECP_RX_RECEIVE_ECPDU:
+ if (vd->ecp.seqECPDU == vd->ecp.lastSequence) {
+ LLDPAD_DBG("%s(%i):-(%s) seqECPDU %x, lastSequence %x\n", __func__, __LINE__,
+ vd->ifname, vd->ecp.seqECPDU, vd->ecp.lastSequence);
+ ecp_rx_change_state(vd, ECP_RX_RESEND_ACK);
+ return true;
+ }
+ if (vd->ecp.seqECPDU != vd->ecp.lastSequence) {
+ ecp_rx_change_state(vd, ECP_RX_RESEND_ACK);
+ return true;
+ }
+ return false;
+ case ECP_RX_SEND_ACK:
+ case ECP_RX_RESEND_ACK:
+ ecp_rx_change_state(vd, ECP_RX_RECEIVE_WAIT);
+ return false;
+ default:
+ LLDPAD_ERR("ERROR: The ECP_RX State Machine is broken!\n");
+ log_message(MSG_ERR_RX_SM_INVALID, "%s", vd->ifname);
+ return false;
+ }
+}
+
+/* ecp_rx_run_sm - state machine for ecp rx
+ * @vd: currently used port
+ *
+ * no return value
+ *
+ * runs the state machine for ecp rx.
+ */
+void ecp_rx_run_sm(struct vdp_data *vd)
+{
+ ecp_set_rx_state(vd);
+ do {
+ LLDPAD_DBG("%s(%i)-%s: ecp_rx - %s\n", __func__, __LINE__,
+ vd->ifname, ecp_rx_states[vd->ecp.tx.state]);
+
+ switch(vd->ecp.rx.state) {
+ case ECP_RX_IDLE:
+ break;
+ case ECP_RX_INIT_RECEIVE:
+ ecp_rx_Initialize(vd);
+ break;
+ case ECP_RX_RECEIVE_WAIT:
+ break;
+ case ECP_RX_RECEIVE_ECPDU:
+ ecp_rx_validate_frame(vd);
+ break;
+ case ECP_RX_SEND_ACK:
+ ecp_rx_ProcessFrame(vd);
+ break;
+ case ECP_RX_RESEND_ACK:
+ ecp_rx_ProcessFrame(vd);
+ if (!vd->ecp.ackReceived) {
+ ecp_rx_send_ack_frame(vd);
+ ecp_rx_freeFrame(vd);
+ }
+ break;
+ default:
+ LLDPAD_ERR("ERROR: The ECP_RX State Machine is broken!\n");
+ log_message(MSG_ERR_TX_SM_INVALID, "%s", vd->ifname);
+ }
+ } while (ecp_set_rx_state(vd) == true);
+
+}
diff --git a/ecp/ecp_tx.c b/ecp/ecp_tx.c
new file mode 100644
index 0000000..a87f3ac
--- /dev/null
+++ b/ecp/ecp_tx.c
@@ -0,0 +1,506 @@
+/*******************************************************************************
+
+ implementation of ECP according to 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include "dcb_osdep.h"
+#include "lldp/ports.h"
+#include "lldp/l2_packet.h"
+#include "eloop.h"
+#include "messages.h"
+#include "lldpad.h"
+#include "lldp_tlv.h"
+#include "lldp_mod.h"
+#include "lldp_mand.h"
+#include "lldp_evb.h"
+#include "include/lldp_vdp.h"
+
+void ecp_tx_run_sm(struct vdp_data *);
+
+/* ecp_somethingChangedLocal - set flag if port has changed
+ * @vd: port to set the flag for
+ * @mode: mode to set the flag to
+ *
+ * no return value
+ *
+ * set the localChange flag with a mode to indicate a port has changed.
+ * used to signal an ecpdu needs to be sent out.
+ */
+
+void ecp_somethingChangedLocal(struct vdp_data *vd, int mode)
+{
+ if (!vd)
+ return;
+
+ vd->ecp.tx.localChange |= mode;
+
+ return;
+}
+
+/* ecp_print_frameout - print outbound frame
+ * @vd: currently used port
+ *
+ * no return value
+ *
+ * prints a raw dump of an outbound ecp frame. useful for low-level protocol
+ * debugging.
+ */
+void ecp_print_frameout(struct vdp_data *vd)
+{
+ int i;
+ char *s, *t;
+
+ s = t = malloc(256);
+
+ if (!s) {
+ LLDPAD_ERR("%s(%i): unable to allocate string !\n", __func__, __LINE__);
+ }
+
+ for (i=0; i < vd->ecp.tx.sizeout; i++) {
+ int c;
+ c = sprintf(s, "%02x ", vd->ecp.tx.frameout[i]);
+ s += c;
+ if (!((i+1) % 16)) {
+ LLDPAD_DBG("%s", t);
+ s = t;
+ }
+ }
+
+ LLDPAD_DBG("%s\n", t);
+
+ free(t);
+}
+
+/* ecp_build_ECPDU - create an ecp protocol data unit
+ * @vd: currently used port
+ * @mode: mode to create pdu with (REQ or ACK)
+ *
+ * returns true on success, false on failure
+ *
+ * creates the frame header with the ports mac address, the ecp header with REQ
+ * or ACK mode, plus a list of packed TLVs created from the profiles on this
+ * port.
+ */
+bool ecp_build_ECPDU(struct vdp_data *vd, int mode)
+{
+ struct l2_ethhdr eth;
+ struct ecp_hdr ecp_hdr;
+ u8 own_addr[ETH_ALEN];
+ u32 fb_offset = 0;
+ u32 datasize = 0;
+ struct packed_tlv *ptlv = NULL;
+ struct vsi_profile *p;
+
+ /* TODO: different multicast address for sending ECP over S-channel (multi_cast_source_s)
+ * S-channels to implement later */
+ memcpy(eth.h_dest, multi_cast_source, ETH_ALEN);
+ l2_packet_get_own_src_addr(vd->ecp.l2,(u8 *)&own_addr);
+ memcpy(eth.h_source, &own_addr, ETH_ALEN);
+ eth.h_proto = htons(ETH_P_ECP);
+ vd->ecp.tx.frameout = (u8 *)malloc(ETH_FRAME_LEN);
+ if (vd->ecp.tx.frameout == NULL) {
+ LLDPAD_ERR("InfoECPDU: Failed to malloc frame buffer \n");
+ return false;
+ }
+ memset(vd->ecp.tx.frameout,0,ETH_FRAME_LEN);
+ memcpy(vd->ecp.tx.frameout, (void *)ð, sizeof(struct l2_ethhdr));
+ fb_offset += sizeof(struct l2_ethhdr);
+
+ ecp_hdr.oui[0] = 0x0;
+ ecp_hdr.oui[1] = 0x1b;
+ ecp_hdr.oui[2] = 0x3f;
+
+ ecp_hdr.pad1 = 0x0;
+
+ ecp_hdr.subtype = ECP_SUBTYPE;
+ switch(mode) {
+ case VDP_PROFILE_REQ:
+ ecp_hdr.mode = ECP_REQUEST;
+ break;
+ case VDP_PROFILE_ACK:
+ ecp_hdr.mode = ECP_ACK;
+ break;
+ default:
+ LLDPAD_ERR("%s(%i): unknown mode for %s !\n", __func__, __LINE__,
+ vd->ifname);
+ }
+
+ vd->ecp.lastSequence++;
+ ecp_hdr.seqnr = htons(vd->ecp.lastSequence);
+
+ if ((sizeof(struct ecp_hdr)+fb_offset) > ETH_MAX_DATA_LEN)
+ goto error;
+ memcpy(vd->ecp.tx.frameout+fb_offset, (void *)&ecp_hdr, sizeof(struct ecp_hdr));
+ datasize += sizeof(struct ecp_hdr);
+ fb_offset += sizeof(struct ecp_hdr);
+
+ /* create packed_tlvs for all profiles on this interface */
+ LIST_FOREACH(p, &vd->profile_head, profile) {
+ if(!p) {
+ LLDPAD_ERR("%s(%i): list vd->profile_head empty !\n", __func__, __LINE__);
+ continue;
+ }
+
+ if (p->localChange != mode) {
+ LLDPAD_ERR("%s(%i): not sending out profile !\n", __func__, __LINE__);
+ continue;
+ }
+
+ ptlv = vdp_gettlv(vd, p);
+
+ if (!ptlv) {
+ LLDPAD_ERR("%s(%i): ptlv not created !\n", __func__, __LINE__);
+ continue;
+ }
+
+ if (ptlv) {
+ if ((ptlv->size+fb_offset) > ETH_MAX_DATA_LEN)
+ goto error;
+ memcpy(vd->ecp.tx.frameout+fb_offset,
+ ptlv->tlv, ptlv->size);
+ datasize += ptlv->size;
+ fb_offset += ptlv->size;
+ }
+
+ ptlv = free_pkd_tlv(ptlv);
+ }
+
+ /* The End TLV marks the end of the LLDP PDU */
+ ptlv = pack_end_tlv();
+ if (!ptlv || ((ptlv->size + fb_offset) > ETH_MAX_DATA_LEN))
+ goto error;
+ memcpy(vd->ecp.tx.frameout + fb_offset, ptlv->tlv, ptlv->size);
+ datasize += ptlv->size;
+ fb_offset += ptlv->size;
+ ptlv = free_pkd_tlv(ptlv);
+
+ if (datasize > ETH_MAX_DATA_LEN)
+ goto error;
+
+ if (datasize < ETH_MIN_DATA_LEN)
+ vd->ecp.tx.sizeout = ETH_MIN_PKT_LEN;
+ else
+ vd->ecp.tx.sizeout = fb_offset;
+
+ return true;
+
+error:
+ ptlv = free_pkd_tlv(ptlv);
+ if (vd->ecp.tx.frameout)
+ free(vd->ecp.tx.frameout);
+ vd->ecp.tx.frameout = NULL;
+ LLDPAD_ERR("InfoECPDU: packed TLV too large for tx frame\n");
+ return false;
+}
+
+/* ecp_tx_Initialize - initializes the ecp tx state machine
+ * @vd: currently used port
+ *
+ * no return value
+ *
+ * initializes some variables for the ecp tx state machine.
+ */
+void ecp_tx_Initialize(struct vdp_data *vd)
+{
+ if (vd->ecp.tx.frameout) {
+ free(vd->ecp.tx.frameout);
+ vd->ecp.tx.frameout = NULL;
+ }
+ vd->ecp.tx.localChange = VDP_PROFILE_REQ;
+ vd->ecp.lastSequence = ECP_SEQUENCE_NR_START;
+ vd->ecp.stats.statsFramesOutTotal = 0;
+ vd->ecp.ackTimerExpired = false;
+ vd->ecp.retries = 0;
+
+ struct port *port = port_find_by_name(vd->ifname);
+ l2_packet_get_port_state(vd->ecp.l2, (u8 *)&(port->portEnabled));
+
+ return;
+}
+
+/* ecp_txFrame - transmit ecp frame
+ * @vd: currently used port
+ *
+ * returns the number of characters sent on success, -1 on failure
+ *
+ * sends out the frame stored in the frameout structure using l2_packet_send.
+ */
+u8 ecp_txFrame(struct vdp_data *vd)
+{
+ int status = 0;
+
+ status = l2_packet_send(vd->ecp.l2, (u8 *)&multi_cast_source,
+ htons(ETH_P_ECP),vd->ecp.tx.frameout,vd->ecp.tx.sizeout);
+ vd->ecp.stats.statsFramesOutTotal++;
+
+ free(vd->ecp.tx.frameout);
+
+ return status;
+}
+
+/* ecp_tx_create_frame - create ecp frame
+ * @vd: currently used port
+ *
+ * no return value
+ *
+ *
+ */
+void ecp_tx_create_frame(struct vdp_data *vd)
+{
+ /* send REQs */
+ if (vd->ecp.tx.localChange & VDP_PROFILE_REQ) {
+ LLDPAD_DBG("%s(%i)-%s: sending REQs\n", __func__, __LINE__, vd->ifname);
+ ecp_build_ECPDU(vd, VDP_PROFILE_REQ);
+ ecp_print_frameout(vd);
+ ecp_txFrame(vd);
+ }
+
+ /* send ACKs */
+ if (vd->ecp.tx.localChange & VDP_PROFILE_ACK) {
+ LLDPAD_DBG("%s(%i)-%s: sending ACKs\n", __func__, __LINE__, vd->ifname);
+ ecp_build_ECPDU(vd, VDP_PROFILE_ACK);
+ ecp_print_frameout(vd);
+ ecp_txFrame(vd);
+ }
+
+ vd->ecp.tx.localChange = 0;
+ return;
+}
+
+/* ecp_timeout_handler - handles the ack timer expiry
+ * @eloop_data: data structure of event loop
+ * @user_ctx: user context, port here
+ *
+ * no return value
+ *
+ * called when the ECP ack timer has expired. sets a flag and calls the ECP
+ * state machine.
+ */
+static void ecp_tx_timeout_handler(void *eloop_data, void *user_ctx)
+{
+ struct vdp_data *vd;
+
+ vd = (struct vdp_data *) user_ctx;
+
+ vd->ecp.ackTimerExpired = true;
+
+ LLDPAD_DBG("%s(%i)-%s: timer expired\n", __func__, __LINE__,
+ vd->ifname);
+
+ ecp_tx_run_sm(vd);
+}
+
+/* ecp_tx_stop_ackTimer - stop the ECP ack timer
+ * @vd: currently used port
+ *
+ * returns the number of removed handlers
+ *
+ * stops the ECP ack timer. used when a ack frame for the port has been
+ * received.
+ */
+int ecp_tx_stop_ackTimer(struct vdp_data *vd)
+{
+ LLDPAD_DBG("%s(%i)-%s: stopping timer\n", __func__, __LINE__,
+ vd->ifname);
+
+ return eloop_cancel_timeout(ecp_tx_timeout_handler, NULL, (void *) vd);
+}
+
+/* ecp_tx_start_ackTimer - starts the ECP ack timer
+ * @profile: profile to process
+ *
+ * returns 0 on success, -1 on error
+ *
+ * starts the ack timer when a frame has been sent out.
+ */
+static void ecp_tx_start_ackTimer(struct vdp_data *vd)
+{
+ unsigned int secs, usecs;
+
+ vd->ecp.ackTimerExpired = false;
+
+ secs = ECP_TRANSMISSION_TIMER / ECP_TRANSMISSION_DIVIDER;
+ usecs = ECP_TRANSMISSION_TIMER % ECP_TRANSMISSION_DIVIDER;
+
+ LLDPAD_DBG("%s(%i)-%s: starting timer\n", __func__, __LINE__,
+ vd->ifname);
+
+ eloop_register_timeout(secs, usecs, ecp_tx_timeout_handler, NULL, (void *) vd);
+}
+
+/* ecp_tx_change_state - changes the ecp tx sm state
+ * @vd: currently used port
+ * @newstate: new state for the sm
+ *
+ * no return value
+ *
+ * checks state transistion for consistency and finally changes the state of
+ * the profile.
+ */
+static void ecp_tx_change_state(struct vdp_data *vd, u8 newstate)
+{
+ switch(newstate) {
+ case ECP_TX_INIT_TRANSMIT:
+ break;
+ case ECP_TX_TRANSMIT_ECPDU:
+ assert((vd->ecp.tx.state == ECP_TX_INIT_TRANSMIT) ||
+ (vd->ecp.tx.state == ECP_TX_WAIT_FOR_ACK) ||
+ (vd->ecp.tx.state == ECP_TX_REQUEST_PDU));
+ break;
+ case ECP_TX_WAIT_FOR_ACK:
+ assert(vd->ecp.tx.state == ECP_TX_TRANSMIT_ECPDU);
+ break;
+ case ECP_TX_REQUEST_PDU:
+ assert(vd->ecp.tx.state == ECP_TX_WAIT_FOR_ACK);
+ break;
+ default:
+ LLDPAD_ERR("ERROR: The ECP_TX State Machine is broken!\n");
+ log_message(MSG_ERR_TX_SM_INVALID, "%s", vd->ifname);
+ }
+
+ LLDPAD_DBG("%s(%i)-%s: state change %s -> %s\n", __func__, __LINE__,
+ vd->ifname, ecp_tx_states[vd->ecp.tx.state], ecp_tx_states[newstate]);
+
+ vd->ecp.tx.state = newstate;
+ return;
+}
+
+/* ecp_set_tx_state - sets the ecp tx sm state
+ * @vd: currently used port
+ *
+ * returns true or false
+ *
+ * switches the state machine to the next state depending on the input
+ * variables. returns true or false depending on wether the state machine
+ * can be run again with the new state or can stop at the current state.
+ */
+static bool ecp_set_tx_state(struct vdp_data *vd)
+{
+ struct port *port = port_find_by_name(vd->ifname);
+
+ if (!port) {
+ LLDPAD_ERR("%s(%i): port not found !\n", __func__, __LINE__);
+ return 0;
+ }
+
+ if ((port->portEnabled == false) && (port->prevPortEnabled == true)) {
+ LLDPAD_ERR("set_tx_state: port was disabled\n");
+ ecp_tx_change_state(vd, ECP_TX_INIT_TRANSMIT);
+ }
+ port->prevPortEnabled = port->portEnabled;
+
+ switch (vd->ecp.tx.state) {
+ case ECP_TX_INIT_TRANSMIT:
+ if (port->portEnabled && ((port->adminStatus == enabledRxTx) ||
+ (port->adminStatus == enabledTxOnly)) && vd->ecp.tx.localChange) {
+ ecp_tx_change_state(vd, ECP_TX_TRANSMIT_ECPDU);
+ return true;
+ }
+ return false;
+ case ECP_TX_TRANSMIT_ECPDU:
+ if ((port->adminStatus == disabled) ||
+ (port->adminStatus == enabledRxOnly)) {
+ ecp_tx_change_state(vd, ECP_TX_INIT_TRANSMIT);
+ return true;
+ }
+ ecp_tx_change_state(vd, ECP_TX_WAIT_FOR_ACK);
+ return true;
+ case ECP_TX_WAIT_FOR_ACK:
+ if (vd->ecp.ackTimerExpired) {
+ vd->ecp.retries++;
+ if (vd->ecp.retries < ECP_MAX_RETRIES) {
+ ecp_somethingChangedLocal(vd, VDP_PROFILE_REQ);
+ ecp_tx_change_state(vd, ECP_TX_TRANSMIT_ECPDU);
+ return true;
+ }
+ if (vd->ecp.retries == ECP_MAX_RETRIES) {
+ LLDPAD_DBG("%s(%i)-%s: 1 \n", __func__, __LINE__,
+ vd->ifname);
+ ecp_tx_change_state(vd, ECP_TX_REQUEST_PDU);
+ return true;
+ }
+ }
+ if (vd->ecp.ackReceived && vd->ecp.seqECPDU == vd->ecp.lastSequence) {
+ vd->ecp.ackReceived = false;
+ ecp_tx_change_state(vd, ECP_TX_REQUEST_PDU);
+ return true;
+ }
+ return false;
+ case ECP_TX_REQUEST_PDU:
+ if (vd->ecp.tx.localChange & VDP_PROFILE_REQ) {
+ ecp_tx_change_state(vd, ECP_TX_TRANSMIT_ECPDU);
+ return true;
+ }
+ return false;
+ default:
+ LLDPAD_ERR("ERROR: The TX State Machine is broken!\n");
+ log_message(MSG_ERR_TX_SM_INVALID, "%s", vd->ifname);
+ return false;
+ }
+}
+
+/* ecp_tx_run_sm - state machine for ecp tx
+ * @vd: currently used vdp_data
+ *
+ * no return value
+ *
+ * runs the state machine for ecp tx.
+ */
+void ecp_tx_run_sm(struct vdp_data *vd)
+{
+ do {
+ LLDPAD_DBG("%s(%i)-%s: ecp_tx - %s\n", __func__, __LINE__,
+ vd->ifname, ecp_tx_states[vd->ecp.tx.state]);
+
+ switch(vd->ecp.tx.state) {
+ case ECP_TX_INIT_TRANSMIT:
+ ecp_tx_Initialize(vd);
+ break;
+ case ECP_TX_TRANSMIT_ECPDU:
+ ecp_tx_create_frame(vd);
+ ecp_tx_start_ackTimer(vd);
+ break;
+ case ECP_TX_WAIT_FOR_ACK:
+ if (vd->ecp.ackReceived) {
+ LLDPAD_DBG("%s(%i)-%s: ECP_TX_WAIT_FOR_ACK ackReceived\n", __func__, __LINE__,
+ vd->ifname);
+ LLDPAD_DBG("%s(%i)-%s: seqECPDU %x lastSequence %x \n", __func__, __LINE__,
+ vd->ifname, vd->ecp.seqECPDU, vd->ecp.lastSequence);
+ vd->ecp.tx.localChange = 0;
+ ecp_tx_stop_ackTimer(vd);
+ }
+ break;
+ case ECP_TX_REQUEST_PDU:
+ vd->ecp.retries = 0;
+ LLDPAD_DBG("%s(%i)-%s: ECP_TX_REQUEST_PDU lastSequence %x\n", __func__, __LINE__,
+ vd->ifname, vd->ecp.lastSequence);
+ break;
+ default:
+ LLDPAD_ERR("%s(%i): ERROR The TX State Machine is broken!\n", __func__,
+ __LINE__);
+ log_message(MSG_ERR_TX_SM_INVALID, "%s", vd->ifname);
+ }
+ } while (ecp_set_tx_state(vd) == true);
+
+ return;
+}
diff --git a/include/lldp.h b/include/lldp.h
index 0371844..476773c 100644
--- a/include/lldp.h
+++ b/include/lldp.h
@@ -190,6 +190,7 @@ enum {
/* IEEE 802.1Qbg subtype */
#define LLDP_EVB_SUBTYPE 0
+#define LLDP_VDP_SUBTYPE 0x2
/* forwarding mode */
#define LLDP_EVB_CAPABILITY_FORWARD_STANDARD (1 << 7)
diff --git a/include/lldp_evb.h b/include/lldp_evb.h
index 667f9ad..d028c21 100644
--- a/include/lldp_evb.h
+++ b/include/lldp_evb.h
@@ -37,6 +37,12 @@ typedef enum {
EVB_CONFIRMATION
} evb_state;
+#define RTE 13
+/* retransmission granularity (RTG) in microseconds */
+#define EVB_RTG 10
+/* retransmission multiplier (RTM) */
+#define EVB_RTM(rte) (2<<(RTE-1))
+
struct tlv_info_evb {
u8 oui[3];
u8 sub;
diff --git a/include/lldp_vdp.h b/include/lldp_vdp.h
new file mode 100644
index 0000000..b97d8c0
--- /dev/null
+++ b/include/lldp_vdp.h
@@ -0,0 +1,159 @@
+/*******************************************************************************
+
+ implementation of according to IEEE 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#ifndef _LLDP_VDP_H
+#define _LLDP_VDP_H
+
+#include "lldp_mod.h"
+#include "ecp/ecp.h"
+
+#define LLDP_MOD_VDP OUI_IEEE_8021Qbg+1
+
+#define VDP_MODE_PREASSOCIATE 0x0
+#define VDP_MODE_PREASSOCIATE_WITH_RR 0x1
+#define VDP_MODE_ASSOCIATE 0x2
+#define VDP_MODE_DEASSOCIATE 0x3
+
+#define VDP_RESPONSE_SUCCESS 0x0
+#define VDP_RESPONSE_INVALID_FORMAT 0x1
+#define VDP_RESPONSE_INSUFF_RESOURCES 0x2
+#define VDP_RESPONSE_UNUSED_VTID 0x3
+#define VDP_RESPONSE_VTID_VIOLATION 0x4
+#define VDP_RESPONSE_VTID_VER_VIOLATION 0x5
+#define VDP_RESPONSE_OUT_OF_SYNC 0x6
+
+enum {
+ VDP_PROFILE_NOCHANGE = 0,
+ VDP_PROFILE_REQ,
+ VDP_PROFILE_ACK,
+ VDP_PROFILE_NACK,
+};
+
+#define VDP_MACVLAN_FORMAT_1 1
+
+#define VDP_TRANSMISSION_TIMER 3*EVB_RTM(RTE)*EVB_RTG
+#define VDP_TRANSMISSION_DIVIDER 10000
+
+#define VDP_ROLE_STATION 0
+#define VDP_ROLE_BRIDGE 1
+
+enum {
+ VSI_UNASSOCIATED = 0,
+ VSI_ASSOC_PROCESSING,
+ VSI_ASSOCIATED,
+ VSI_PREASSOC_PROCESSING,
+ VSI_PREASSOCIATED,
+ VSI_DEASSOC_PROCESSING,
+ VSI_EXIT,
+};
+
+static char *vsi_states[] = {
+ "VSI_UNASSOCIATED",
+ "VSI_ASSOC_PROCESSING",
+ "VSI_ASSOCIATED",
+ "VSI_PREASSOC_PROCESSING",
+ "VSI_PREASSOCIATED",
+ "VSI_DEASSOC_PROCESSING",
+ "VSI_EXIT"
+};
+
+struct mac_vlan {
+ u8 mac[6];
+ u16 vlan;
+} __attribute__ ((__packed__));
+
+struct tlv_info_vdp {
+ u8 oui[3];
+ u8 sub;
+ u8 mode;
+ u8 response;
+ u8 mgrid;
+ u8 id[3];
+ u8 version;
+ u8 instance[16];
+ u8 format;
+ u16 entries;
+ struct mac_vlan mac_vlan;
+} __attribute__ ((__packed__));
+
+struct vsi_profile {
+ int mode;
+ int response;
+ u8 mgrid;
+ int id;
+ u8 version;
+ u8 instance[16];
+ u8 mac[6]; /* TODO: currently only one MAC/VLAN pair supported, more later */
+ u16 vlan;
+ struct port *port;
+ int ackTimerExpired;
+ int ackReceived;
+ int state;
+ int localChange;
+ LIST_ENTRY(vsi_profile) profile;
+};
+
+struct vdp_data {
+ char ifname[IFNAMSIZ];
+ struct ecp ecp;
+ struct unpacked_tlv *vdp;
+ int role;
+ LIST_HEAD(profile_head, vsi_profile) profile_head;
+ LIST_ENTRY(vdp_data) entry;
+};
+
+struct vdp_user_data {
+ LIST_HEAD(vdp_head, vdp_data) head;
+};
+
+struct lldp_module *vdp_register(void);
+void vdp_unregister(struct lldp_module *mod);
+struct vdp_data *vdp_data(char *ifname);
+struct packed_tlv *vdp_gettlv(struct vdp_data *vd, struct vsi_profile *profile);
+void vdp_vsi_sm_station(struct vsi_profile *profile);
+struct vsi_profile *vdp_add_profile(struct vsi_profile *profile);
+
+#define MAC_ADDR_STRLEN 18
+#define INSTANCE_STRLEN 32
+
+#define PRINT_PROFILE(s, p) \
+{ int c; \
+ c = sprintf(s, "\nmode: %i\n", p->mode); s += c; \
+ c = sprintf(s, "response: %i\n", p->response); s += c; \
+ c = sprintf(s, "state: %i\n", p->state); s += c; \
+ c = sprintf(s, "mgrid: %i\n", p->mgrid); s += c; \
+ c = sprintf(s, "id: %x\n", p->id); \
+ s += c; \
+ c = sprintf(s, "version: %i\n", p->version); s += c; \
+ char instance[INSTANCE_STRLEN+2]; \
+ instance2str(p->instance, instance, sizeof(instance)); \
+ c = sprintf(s, "instance: %s\n", &instance[0]); s += c; \
+ char macbuf[MAC_ADDR_STRLEN+1]; \
+ mac2str(p->mac, macbuf, MAC_ADDR_STRLEN); \
+ c = sprintf(s, "mac: %s\n", macbuf); s += c; \
+ c = sprintf(s, "vlan: %i\n\n", p->vlan); s += c; \
+}
+
+#endif /* _LLDP_VDP_H */
diff --git a/lldp/l2_packet.h b/lldp/l2_packet.h
index 16f3683..0962429 100644
--- a/lldp/l2_packet.h
+++ b/lldp/l2_packet.h
@@ -36,6 +36,8 @@
#define ETH_P_LLDP 0x88cc
+/* TODO: use extended ethertype until final ethertype is available */
+#define ETH_P_ECP 0x88b7
#define ETH_FRAME_LEN 1514
diff --git a/lldp/ports.h b/lldp/ports.h
index 0138efe..44dc5f1 100644
--- a/lldp/ports.h
+++ b/lldp/ports.h
@@ -142,15 +142,20 @@ struct port {
u8 portEnabled;
u8 prevPortEnabled;
u8 adminStatus;
- u8 rxChanges;
- u16 lldpdu;
+
+ /* protocol specific */
struct l2_packet_data *l2;
struct portrx rx;
struct porttx tx;
- struct porttlvs tlvs;
struct portstats stats;
struct porttimers timers;
+ u8 rxChanges;
+ u16 lldpdu;
struct msap msap;
+
+ /* not sure */
+ struct porttlvs tlvs;
+
struct port *next;
};
diff --git a/lldp_vdp.c b/lldp_vdp.c
new file mode 100644
index 0000000..d66be6e
--- /dev/null
+++ b/lldp_vdp.c
@@ -0,0 +1,1196 @@
+/*******************************************************************************
+
+ implementation of VDP according to IEEE 802.1Qbg
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include <net/if.h>
+#include <sys/queue.h>
+#include <sys/socket.h>
+#include <sys/ioctl.h>
+#include <sys/utsname.h>
+#include <linux/if_bridge.h>
+#include "lldp.h"
+#include "lldp_vdp.h"
+#include "ecp/ecp.h"
+#include "eloop.h"
+#include "lldp_evb.h"
+#include "messages.h"
+#include "config.h"
+#include "common.h"
+#include "lldp_vdp_clif.h"
+#include "lldp_vdp_cmds.h"
+
+/* vdp_data - searches vdp_data in the list of modules for this port
+ * @ifname: interface name to search for
+ *
+ * returns vdp_data on success, NULL on error
+ *
+ * searches the list of user_data for the VDP module user_data.
+ */
+struct vdp_data *vdp_data(char *ifname)
+{
+ struct vdp_user_data *ud;
+ struct vdp_data *vd = NULL;
+
+ ud = find_module_user_data_by_if(ifname, &lldp_head, LLDP_MOD_VDP);
+ if (ud) {
+ LIST_FOREACH(vd, &ud->head, entry) {
+ if (!strncmp(ifname, vd->ifname, IFNAMSIZ))
+ return vd;
+ }
+ }
+ return NULL;
+}
+
+/* vdp_free_tlv - free tlv in vdp_data
+ * @vd: vdp_data
+ *
+ * no return value
+ *
+ * frees up tlv in vdp_data. used in vdp_free_data.
+ */
+static void vdp_free_tlv(struct vdp_data *vd)
+{
+ if (vd) {
+ FREE_UNPKD_TLV(vd, vdp);
+ }
+}
+
+/* vdp_free_data - frees up vdp data
+ * @ud: user data structure
+ *
+ * no return value
+ *
+ * removes vd_structure from the user_data list. frees up tlv in vdp_data.
+ * used in vdp_unregister.
+ */
+static void vdp_free_data(struct vdp_user_data *ud)
+{
+ struct vdp_data *vd;
+ if (ud) {
+ while (!LIST_EMPTY(&ud->head)) {
+ vd = LIST_FIRST(&ud->head);
+ LIST_REMOVE(vd, entry);
+ vdp_free_tlv(vd);
+ free(vd);
+ }
+ }
+}
+
+/* vdp_print_profile - print a vsi profile
+ * @profile: profile to print
+ *
+ * no return value
+ *
+ * prints the contents of a profile first to a string using the PRINT_PROFILE
+ * macro, and then to the screen. Used for debug purposes.
+ */
+static inline void vdp_print_profile(struct vsi_profile *profile)
+{
+ LLDPAD_DBG("profile:\n");
+
+ LLDPAD_DBG("mode: %i\n", profile->mode);
+ LLDPAD_DBG("response: %i\n", profile->response);
+ LLDPAD_DBG("state: %i\n", profile->state);
+ LLDPAD_DBG("mgrid: %i\n", profile->mgrid);
+ LLDPAD_DBG("id: %x\n", profile->id);
+ LLDPAD_DBG("version: %i\n", profile->version);
+
+ char instance[INSTANCE_STRLEN+2];
+ instance2str(profile->instance, instance, sizeof(instance));
+ LLDPAD_DBG("instance: %s\n", &instance[0]);
+
+ char macbuf[MAC_ADDR_STRLEN+1];
+ mac2str(profile->mac, macbuf, MAC_ADDR_STRLEN);
+ LLDPAD_DBG("mac: %s\n", macbuf);
+
+ LLDPAD_DBG("vlan: %i\n", profile->vlan);
+}
+
+/* vdp_somethingChangedLocal - set flag if profile has changed
+ * @profile: profile to set the flag for
+ * @mode: mode to set the flag to
+ *
+ * no return value
+ *
+ * set the localChange flag with a mode to indicate a profile has changed.
+ * used next time when a ecpdu with profiles is sent out.
+ */
+void vdp_somethingChangedLocal(struct vsi_profile *profile, int mode)
+{
+ profile->localChange = mode;
+}
+
+/* vdp_ackTimer_expired - checks for expired ack timer
+ * @profile: profile to be checked
+ *
+ * returns true or false
+ *
+ * returns value of profile->ackTimerExpired, true if ack timer has expired,
+ * false otherwise.
+ */
+static bool vdp_ackTimer_expired(struct vsi_profile *profile)
+{
+ return profile->ackTimerExpired;
+}
+
+/* vdp_timeout_handler - handles the ack timer expiry
+ * @eloop_data: data structure of event loop
+ * @user_ctx: user context, profile here
+ *
+ * no return value
+ *
+ * called when the VDP ack timer has expired. sets a flag and calls the VDP
+ * state machine.
+ */
+void vdp_timeout_handler(void *eloop_data, void *user_ctx)
+{
+ struct vsi_profile *profile;
+
+ profile = (struct vsi_profile *) user_ctx;
+
+ profile->ackTimerExpired = true;
+
+ LLDPAD_DBG("%s(%i)-%s: timer expired\n", __func__, __LINE__,
+ profile->port->ifname);
+
+ vdp_vsi_sm_station(profile);
+}
+
+/* vdp_stop_ackTimer - stop the VDP ack timer
+ * @profile: profile to process
+ *
+ * returns the number of removed handlers
+ *
+ * stops the VDP ack timer. used when a ack frame for the profile has been
+ * received.
+ */
+static int vdp_stop_ackTimer(struct vsi_profile *profile)
+{
+ LLDPAD_DBG("%s(%i)-%s: stopping timer\n", __func__, __LINE__,
+ profile->port->ifname);
+
+ return eloop_cancel_timeout(vdp_timeout_handler, NULL, (void *) profile);
+}
+
+/* vdp_start_ackTimer - starts the VDP ack timer
+ * @profile: profile to process
+ *
+ * returns 0 on success, -1 on error
+ *
+ * starts the ack timer when a frame has been sent out.
+ */
+static int vdp_start_ackTimer(struct vsi_profile *profile)
+{
+ unsigned int secs, usecs;
+
+ profile->ackTimerExpired = false;
+
+ secs = VDP_TRANSMISSION_TIMER / VDP_TRANSMISSION_DIVIDER;
+ usecs = VDP_TRANSMISSION_TIMER % VDP_TRANSMISSION_DIVIDER;
+
+ LLDPAD_DBG("%s(%i)-%s: starting timer\n", __func__, __LINE__,
+ profile->port->ifname);
+
+ return eloop_register_timeout(secs, usecs, vdp_timeout_handler, NULL, (void *) profile);
+}
+
+/* vdp_vsi_change_station_state - changes the VDP station sm state
+ * @profile: profile to process
+ * @newstate: new state for the sm
+ *
+ * no return value
+ *
+ * actually changes the state of the profile
+ */
+void vdp_vsi_change_station_state(struct vsi_profile *profile, u8 newstate)
+{
+ switch(newstate) {
+ case VSI_UNASSOCIATED:
+ break;
+ case VSI_ASSOC_PROCESSING:
+ assert((profile->state == VSI_PREASSOCIATED) ||
+ (profile->state == VSI_UNASSOCIATED));
+ break;
+ case VSI_ASSOCIATED:
+ assert(profile->state == VSI_ASSOC_PROCESSING);
+ break;
+ case VSI_PREASSOC_PROCESSING:
+ assert(profile->state == VSI_UNASSOCIATED);
+ break;
+ case VSI_PREASSOCIATED:
+ assert(profile->state == VSI_PREASSOC_PROCESSING);
+ break;
+ case VSI_DEASSOC_PROCESSING:
+ assert((profile->state == VSI_PREASSOCIATED) ||
+ (profile->state == VSI_ASSOCIATED));
+ break;
+ case VSI_EXIT:
+ assert((profile->state == VSI_ASSOC_PROCESSING) ||
+ (profile->state == VSI_PREASSOC_PROCESSING) ||
+ (profile->state == VSI_DEASSOC_PROCESSING) ||
+ (profile->state == VSI_PREASSOCIATED) ||
+ (profile->state == VSI_ASSOCIATED));
+ break;
+ default:
+ LLDPAD_ERR("ERROR: The VDP station State Machine is broken!\n");
+ break;
+ }
+
+ LLDPAD_DBG("%s(%i)-%s: state change %s -> %s\n", __func__, __LINE__,
+ profile->port->ifname, vsi_states[profile->state], vsi_states[newstate]);
+
+ profile->state = newstate;
+}
+
+/* vdp_vsi_set_station_state - sets the vdp sm station state
+ * @profile: profile to process
+ *
+ * returns true or false
+ *
+ * switches the state machine to the next state depending on the input
+ * variables. returns true or false depending on wether the state machine
+ * can be run again with the new state or can stop at the current state.
+ */
+static bool vdp_vsi_set_station_state(struct vsi_profile *profile)
+{
+ switch(profile->state) {
+ case VSI_UNASSOCIATED:
+ if ((profile->mode == VDP_MODE_PREASSOCIATE) ||
+ (profile->mode == VDP_MODE_PREASSOCIATE_WITH_RR)) {
+ vdp_vsi_change_station_state(profile, VSI_PREASSOC_PROCESSING);
+ return true;
+ } else if (profile->mode == VDP_MODE_ASSOCIATE) {
+ vdp_vsi_change_station_state(profile, VSI_ASSOC_PROCESSING);
+ return true;
+ }
+ return false;
+ case VSI_ASSOC_PROCESSING:
+ if (profile->ackReceived) {
+ vdp_vsi_change_station_state(profile, VSI_ASSOCIATED);
+ return true;
+ } else if (!profile->ackReceived && vdp_ackTimer_expired(profile)) {
+ vdp_vsi_change_station_state(profile, VSI_EXIT);
+ return true;
+ }
+ return false;
+ case VSI_ASSOCIATED:
+ if (profile->mode == VDP_MODE_PREASSOCIATE) {
+ vdp_vsi_change_station_state(profile, VSI_PREASSOC_PROCESSING);
+ return true;
+ } else if (profile->mode == VDP_MODE_DEASSOCIATE) {
+ vdp_vsi_change_station_state(profile, VSI_DEASSOC_PROCESSING);
+ return true;
+ }
+ return false;
+ case VSI_PREASSOC_PROCESSING:
+ if (profile->ackReceived) {
+ vdp_vsi_change_station_state(profile, VSI_PREASSOCIATED);
+ return true;
+ } else if (vdp_ackTimer_expired(profile)) {
+ vdp_vsi_change_station_state(profile, VSI_EXIT);
+ return true;
+ }
+ case VSI_PREASSOCIATED:
+ if (profile->mode == VDP_MODE_DEASSOCIATE) {
+ vdp_vsi_change_station_state(profile, VSI_DEASSOC_PROCESSING);
+ return true;
+ }
+ if (profile->mode == VDP_MODE_ASSOCIATE) {
+ vdp_vsi_change_station_state(profile, VSI_ASSOC_PROCESSING);
+ return true;
+ }
+ return false;
+ case VSI_DEASSOC_PROCESSING:
+ if ((profile->ackReceived) || vdp_ackTimer_expired(profile)) {
+ vdp_vsi_change_station_state(profile, VSI_EXIT);
+ return true;
+ }
+ return false;
+ case VSI_EXIT:
+ return false;
+ default:
+ LLDPAD_ERR("ERROR: The VSI RX State Machine is broken!\n");
+ log_message(MSG_ERR_RX_SM_INVALID, "");
+ return false;
+ }
+}
+
+/* vdp_vsi_sm_station - state machine for vdp station role
+ * @profile: profile for which the state is processed
+ *
+ * no return value
+ *
+ * runs the state machine for the station role of VDP.
+ */
+void vdp_vsi_sm_station(struct vsi_profile *profile)
+{
+ struct vdp_data *vd = vdp_data(profile->port->ifname);
+
+ vdp_vsi_set_station_state(profile);
+ do {
+ LLDPAD_DBG("%s(%i)-%s: station - %s\n", __func__, __LINE__,
+ profile->port->ifname, vsi_states[profile->state]);
+
+ switch(profile->state) {
+ case VSI_UNASSOCIATED:
+ break;
+ case VSI_ASSOC_PROCESSING:
+ vdp_somethingChangedLocal(profile, VDP_PROFILE_REQ);
+ ecp_somethingChangedLocal(vd, VDP_PROFILE_REQ);
+ ecp_tx_run_sm(vd);
+ vdp_start_ackTimer(profile);
+ break;
+ case VSI_ASSOCIATED:
+ vdp_stop_ackTimer(profile);
+ /* TODO:
+ * vsiError = ProcRxandSetCfg(remoteTLV, localtlv, vsistate);
+ * if (!vsiError) vsistate=ASSOCIATED */
+ break;
+ case VSI_PREASSOC_PROCESSING:
+ /* send out profile */
+ vdp_somethingChangedLocal(profile, VDP_PROFILE_REQ);
+ ecp_somethingChangedLocal(vd, VDP_PROFILE_REQ);
+ ecp_tx_run_sm(vd);
+ vdp_start_ackTimer(profile);
+ break;
+ case VSI_PREASSOCIATED:
+ profile->ackReceived = false;
+ vdp_somethingChangedLocal(profile, VDP_PROFILE_NOCHANGE);
+ vdp_stop_ackTimer(profile);
+ /* TODO vsiError = ProcRxandSetCfg(remoteTLV, localtlv, vsistate);
+ * if (!vsiError) vsistate=PREASSOCIATED */
+ break;
+ case VSI_DEASSOC_PROCESSING:
+ vdp_somethingChangedLocal(profile, VDP_PROFILE_REQ);
+ ecp_somethingChangedLocal(vd, VDP_PROFILE_REQ);
+ ecp_tx_run_sm(vd);
+ vdp_start_ackTimer(profile);
+ break;
+ case VSI_EXIT:
+ /* TODO: send DEASSOC here ? */
+ vdp_stop_ackTimer(profile);
+ vdp_remove_profile(profile);
+ break;
+ default:
+ LLDPAD_ERR("ERROR: The VSI RX station State Machine is broken!\n");
+ log_message(MSG_ERR_TX_SM_INVALID, "");
+ }
+ } while (vdp_vsi_set_station_state(profile) == true);
+
+}
+
+/* vdp_vsi_change_bridge_state - changes the VDP bridge sm state
+ * @profile: profile to process
+ * @newstate: new state for the sm
+ *
+ * no return value
+ *
+ * actually changes the state of the profile
+ */
+static void vdp_vsi_change_bridge_state(struct vsi_profile *profile, u8 newstate)
+{
+ switch(newstate) {
+ case VSI_UNASSOCIATED:
+ break;
+ case VSI_ASSOC_PROCESSING:
+ assert((profile->state == VSI_UNASSOCIATED) ||
+ (profile->state == VSI_PREASSOCIATED) ||
+ (profile->state == VSI_ASSOCIATED));
+ break;
+ case VSI_ASSOCIATED:
+ assert(profile->state == VSI_ASSOC_PROCESSING);
+ break;
+ case VSI_PREASSOC_PROCESSING:
+ assert((profile->state == VSI_UNASSOCIATED) ||
+ (profile->state == VSI_PREASSOCIATED) ||
+ (profile->state == VSI_ASSOCIATED));
+ break;
+ case VSI_PREASSOCIATED:
+ assert(profile->state == VSI_PREASSOC_PROCESSING);
+ break;
+ case VSI_DEASSOC_PROCESSING:
+ assert((profile->state == VSI_UNASSOCIATED) ||
+ (profile->state == VSI_PREASSOCIATED) ||
+ (profile->state == VSI_ASSOCIATED));
+ break;
+ case VSI_EXIT:
+ assert((profile->state == VSI_DEASSOC_PROCESSING) ||
+ (profile->state == VSI_PREASSOC_PROCESSING) ||
+ (profile->state == VSI_ASSOC_PROCESSING));
+ break;
+ default:
+ LLDPAD_ERR("ERROR: The VDP bridge State Machine is broken!\n");
+ break;
+ }
+ profile->state = newstate;
+}
+
+/* vdp_vsi_set_bridge_state - sets the vdp sm bridge state
+ * @profile: profile to process
+ *
+ * returns true or false
+ *
+ * switches the state machine to the next state depending on the input
+ * variables. returns true or false depending on wether the state machine
+ * can be run again with the new state or can stop at the current state.
+ */
+static bool vdp_vsi_set_bridge_state(struct vsi_profile *profile)
+{
+ switch(profile->state) {
+ case VSI_UNASSOCIATED:
+ if ((profile->mode == VDP_MODE_DEASSOCIATE)) /* || (INACTIVE)) */ {
+ vdp_vsi_change_bridge_state(profile, VSI_DEASSOC_PROCESSING);
+ return true;
+ } else if (profile->mode == VDP_MODE_ASSOCIATE) {
+ vdp_vsi_change_bridge_state(profile, VSI_ASSOC_PROCESSING);
+ return true;
+ } else if (profile->mode == VDP_MODE_PREASSOCIATE) {
+ vdp_vsi_change_bridge_state(profile, VSI_PREASSOC_PROCESSING);
+ return true;
+ }
+ return false;
+ case VSI_ASSOC_PROCESSING:
+ /* TODO: handle error case
+ if (!vsiError) ||
+ (vsiError && vsiState == Assoc) {
+ */
+ if (profile->mode == VDP_MODE_ASSOCIATE) {
+ vdp_vsi_change_bridge_state(profile, VSI_ASSOCIATED);
+ return true;
+ }
+ return false;
+ case VSI_ASSOCIATED:
+ if (profile->mode == VDP_MODE_ASSOCIATE) /* || ( INACTIVE )*/ {
+ vdp_vsi_change_bridge_state(profile, VSI_DEASSOC_PROCESSING);
+ return true;
+ } else if (profile->mode == VDP_MODE_PREASSOCIATE) {
+ vdp_vsi_change_bridge_state(profile, VSI_PREASSOC_PROCESSING);
+ return true;
+ } else if (profile->mode == VDP_MODE_ASSOCIATE) {
+ vdp_vsi_change_bridge_state(profile, VSI_ASSOC_PROCESSING);
+ return true;
+ }
+ return false;
+ case VSI_PREASSOC_PROCESSING:
+ if (profile->response != VDP_RESPONSE_SUCCESS) {
+ vdp_vsi_change_bridge_state(profile, VSI_EXIT);
+ return true;
+ }
+ vdp_vsi_change_bridge_state(profile, VSI_PREASSOCIATED);
+ return false;
+ case VSI_PREASSOCIATED:
+ if (profile->mode == VDP_MODE_ASSOCIATE) {
+ vdp_vsi_change_bridge_state(profile, VSI_ASSOC_PROCESSING);
+ return true;
+ } else if (profile->mode == VDP_MODE_DEASSOCIATE ) {
+ vdp_vsi_change_bridge_state(profile, VSI_DEASSOC_PROCESSING);
+ return true;
+ } else if (profile->mode == VDP_MODE_PREASSOCIATE ) {
+ vdp_vsi_change_bridge_state(profile, VSI_PREASSOC_PROCESSING);
+ return true;
+ }
+ return false;
+ case VSI_DEASSOC_PROCESSING:
+ vdp_vsi_change_bridge_state(profile, VSI_EXIT);
+ return false;
+ case VSI_EXIT:
+ return false;
+ default:
+ LLDPAD_ERR("ERROR: The VSI RX State Machine (bridge) is broken!\n");
+ log_message(MSG_ERR_RX_SM_INVALID, "");
+ return false;
+ }
+}
+
+/* vdp_vsi_sm_bridge - state machine for vdp bridge role
+ * @profile: profile for which the state is processed
+ *
+ * no return value
+ *
+ * runs the state machine for the bridge role of VDP.
+ */
+static void vdp_vsi_sm_bridge(struct vsi_profile *profile)
+{
+ struct vdp_data *vd = vdp_data(profile->port->ifname);
+
+ vdp_vsi_set_bridge_state(profile);
+ do {
+ LLDPAD_DBG("%s(%i)-%s: bridge - %s\n", __func__, __LINE__,
+ profile->port->ifname, vsi_states[profile->state]);
+ switch(profile->state) {
+ case VSI_UNASSOCIATED:
+ break;
+ case VSI_ASSOC_PROCESSING:
+ /* TODO: vsiError = ProcRxandSetCfg(remoteTLV, localtlv, vsistate);
+ * if (vsiError)
+ * txTLV(Assoc NACK)
+ * else
+ * txTLV(Assoc ACK)
+ */
+ break;
+ case VSI_ASSOCIATED:
+ break;
+ case VSI_PREASSOC_PROCESSING:
+ /* TODO: vsiError = ProcRxandSetCfg(remoteTLV, localtlv, vsistate);
+ * if (vsiError)
+ * txTLV(PreAssoc NACK)
+ * else
+ * txTLV(PreAssoc ACK)
+ */
+ /* for now, we always succeed */
+ profile->response = VDP_RESPONSE_SUCCESS;
+ LLDPAD_DBG("%s(%i)-%s: framein %p, sizein %i\n", __func__, __LINE__,
+ profile->port->ifname, vd->ecp.rx.framein,
+ vd->ecp.rx.sizein);
+ ecp_rx_send_ack_frame(profile->port);
+ break;
+ case VSI_PREASSOCIATED:
+ LLDPAD_DBG("%s(%i)-%s: \n", __func__, __LINE__, profile->port->ifname);
+ break;
+ case VSI_DEASSOC_PROCESSING:
+ /* TODO: txTLV(DeAssoc ACK) */
+ break;
+ case VSI_EXIT:
+ vdp_remove_profile(profile);
+ break;
+ default:
+ LLDPAD_ERR("ERROR: The VSI RX bridge State Machine is broken!\n");
+ log_message(MSG_ERR_TX_SM_INVALID, "");
+ }
+ } while (vdp_vsi_set_bridge_state(profile) == true);
+
+}
+
+/*
+ * vdp_print_vsi_tlv - print the raw contents of a VSI TLV
+ * @tlv: the unpacked tlv which gets printed
+ *
+ * No return value
+ *
+ * used for protocol debug purposes
+ */
+static void vdp_print_vsi_tlv(struct unpacked_tlv *tlv)
+{
+ int i;
+
+ LLDPAD_DBG("%s:type %i, length %i, info:\n", __func__, tlv->type, tlv->length);
+
+ for (i=0; i < tlv->length; i++) {
+ LLDPAD_DBG("%02x ", tlv->info[i]);
+ if (!((i+1) % 16))
+ LLDPAD_DBG("\n");
+ }
+
+ LLDPAD_DBG("\n");
+}
+
+/*
+ * vdp_validate_tlv - validates vsi tlvs
+ * @vdp: decoded vsi tlv
+ *
+ * Returns 0 on success, 1 on error
+ *
+ * checks the contents of an already decoded vsi tlv for inconsistencies
+ */
+static int vdp_validate_tlv(struct tlv_info_vdp *vdp)
+{
+ if (ntoh24(vdp->oui) != OUI_IEEE_8021Qbg) {
+ LLDPAD_DBG("vdp->oui %06x \n", ntoh24(vdp->oui));
+ goto out_err;
+ }
+
+ if (vdp->sub != LLDP_VDP_SUBTYPE) {
+ LLDPAD_DBG("vdp->sub %02x \n", vdp->sub);
+ goto out_err;
+ }
+
+ if ((vdp->mode < VDP_MODE_PREASSOCIATE) ||
+ (vdp->mode > VDP_MODE_DEASSOCIATE)) {
+ LLDPAD_DBG("Unknown mode %02x in vsi tlv !\n", vdp->mode);
+ goto out_err;
+ }
+
+ if ((vdp->response < VDP_RESPONSE_SUCCESS) ||
+ (vdp->response > VDP_RESPONSE_OUT_OF_SYNC)) {
+ LLDPAD_DBG("Unknown response %02x \n", vdp->response);
+ goto out_err;
+ }
+
+ if (vdp->format != VDP_MACVLAN_FORMAT_1) {
+ LLDPAD_DBG("Unknown format %02x in vsi tlv !\n", vdp->format);
+ goto out_err;
+ }
+
+ if (ntohs(vdp->entries) != 1) {
+ LLDPAD_DBG("Multiple entries %02x in vsi tlv !\n", vdp->entries);
+ goto out_err;
+ }
+
+ return 0;
+
+out_err:
+ return 1;
+}
+
+/*
+ * vdp_indicate - receive VSI TLVs from ECP
+ * @port: the port on which the tlv was received
+ * @tlv: the unpacked tlv to receive
+ * @ecp_mode: the mode under which the tlv was received (ACK or REQ)
+ *
+ * Returns 0 on success
+ *
+ * receives a vsi tlv and creates a profile. Take appropriate action
+ * depending on the role of the (receive) port
+ */
+int vdp_indicate(struct vdp_data *vd, struct unpacked_tlv *tlv, int ecp_mode)
+{
+ struct tlv_info_vdp *vdp;
+ struct vsi_profile *p, *profile;
+ struct port *port = port_find_by_name(vd->ifname);
+
+ LLDPAD_DBG("%s(%i): indicating vdp for for %s !\n", __func__, __LINE__, vd->ifname);
+
+ if (!port) {
+ LLDPAD_ERR("%s(%i): port not found for %s !\n", __func__, __LINE__, vd->ifname);
+ goto out_err;
+ }
+
+ vdp = malloc(sizeof(struct tlv_info_vdp));
+
+ if (!vdp) {
+ LLDPAD_ERR("%s(%i): unable to allocate vdp !\n", __func__, __LINE__);
+ goto out_err;
+ }
+
+ memset(vdp, 0, sizeof(struct tlv_info_vdp));
+ memcpy(vdp, tlv->info, tlv->length);
+
+ if (vdp_validate_tlv(vdp)) {
+ LLDPAD_ERR("%s(%i): Invalid TLV received !\n", __func__, __LINE__);
+ goto out_vdp;
+ }
+
+ profile = malloc(sizeof(struct vsi_profile));
+
+ if (!profile) {
+ LLDPAD_ERR("%s(%i): unable to allocate profile !\n", __func__, __LINE__);
+ goto out_vdp;
+ }
+
+ memset(profile, 0, sizeof(struct vsi_profile));
+
+ profile->mode = vdp->mode;
+ profile->response = vdp->response;
+
+ profile->mgrid = vdp->mgrid;
+ profile->id = ntoh24(vdp->id);
+ profile->version = vdp->version;
+ memcpy(&profile->instance, &vdp->instance, 16);
+ memcpy(&profile->mac, &vdp->mac_vlan.mac, MAC_ADDR_LEN);
+ profile->vlan = ntohs(vdp->mac_vlan.vlan);
+
+ profile->port = port;
+
+ if (vd->role == VDP_ROLE_STATION) {
+ /* do we have the profile already ? */
+ LIST_FOREACH(p, &vd->profile_head, profile) {
+ if (vdp_profile_equal(p, profile)) {
+ LLDPAD_DBG("%s(%i): station: profile found, localChange %i ackReceived %i!\n",
+ __func__, __LINE__, p->localChange, p->ackReceived);
+
+ p->ackReceived = true;
+
+ vdp_vsi_sm_station(p);
+ } else {
+ LLDPAD_DBG("%s(%i): station: profile not found !\n", __func__, __LINE__);
+ /* ignore profile */
+ }
+ }
+ }
+
+ if (vd->role == VDP_ROLE_BRIDGE) {
+ /* do we have the profile already ? */
+ LIST_FOREACH(p, &vd->profile_head, profile) {
+ if (vdp_profile_equal(p, profile)) {
+ break;
+ }
+ }
+
+ if (p) {
+ LLDPAD_DBG("%s(%i): bridge: profile found !\n", __func__, __LINE__);
+ } else {
+ LLDPAD_DBG("%s(%i): bridge: profile not found !\n", __func__, __LINE__);
+ /* put it in the list */
+ profile->state = VSI_UNASSOCIATED;
+ LIST_INSERT_HEAD(&vd->profile_head, profile, profile );
+ }
+
+ vdp_vsi_sm_bridge(profile);
+ }
+
+ return 0;
+
+out_vdp:
+ free(vdp);
+out_err:
+ LLDPAD_ERR("%s(%i): error !\n", __func__, __LINE__);
+ return 1;
+
+}
+
+/*
+ * vdp_bld_vsi_tlv - build the VDP VSI TLV
+ * @vd: vdp_data structure for this port
+ * @profile: profile the vsi tlv is created from
+ *
+ * Returns 0 on success, ENOMEM otherwise
+ *
+ * creates a vdp structure from an existing profile
+ */
+static int vdp_bld_vsi_tlv(struct vdp_data *vd, struct vsi_profile *profile)
+{
+ int rc = 0;
+ struct unpacked_tlv *tlv = NULL;
+ struct tlv_info_vdp vdp;
+
+ FREE_UNPKD_TLV(vd, vdp);
+
+ memset(&vdp, 0, sizeof(vdp));
+
+ hton24(vdp.oui, OUI_IEEE_8021Qbg);
+ vdp.sub = LLDP_VDP_SUBTYPE;
+ vdp.mode = profile->mode;
+ vdp.response = 0;
+ vdp.mgrid = profile->mgrid;
+ hton24(vdp.id, profile->id);
+ vdp.version = profile->version;
+ memcpy(&vdp.instance,&profile->instance, 16);
+ vdp.format = VDP_MACVLAN_FORMAT_1;
+ vdp.entries = htons(1);
+ memcpy(&vdp.mac_vlan.mac,&profile->mac, MAC_ADDR_LEN);
+ vdp.mac_vlan.vlan = htons(profile->vlan);
+
+ tlv = create_tlv();
+ if (!tlv)
+ goto out_err;
+
+ tlv->type = ORG_SPECIFIC_TLV;
+ tlv->length = sizeof(vdp);
+ tlv->info = (u8 *)malloc(tlv->length);
+ if(!tlv->info) {
+ free(tlv);
+ tlv = NULL;
+ rc = ENOMEM;
+ goto out_err;
+ }
+ memcpy(tlv->info, &vdp, tlv->length);
+
+ vd->vdp = tlv;
+
+out_err:
+ return rc;
+}
+
+/* vdp_bld_tlv - builds a tlv from a profile
+ * @vd: vdp_data structure for this port
+ * @profile: profile the vsi tlv is created from
+ *
+ * returns 0 on success, != 0 on error
+ *
+ * wrapper function around vdp_bld_vsi_tlv. adds some checks and calls
+ * vdp_bld_vsi_tlv.
+*/
+
+static int vdp_bld_tlv(struct vdp_data *vd, struct vsi_profile *profile)
+{
+ int rc = 0;
+
+ if (!port_find_by_name(vd->ifname)) {
+ rc = EEXIST;
+ goto out_err;
+ }
+
+ if (!init_cfg()) {
+ rc = ENOENT;
+ goto out_err;
+ }
+
+ if (vdp_bld_vsi_tlv(vd, profile)) {
+ LLDPAD_ERR("%s:%s:vdp_bld_vsi_tlv() failed\n",
+ __func__, vd->ifname);
+ rc = EINVAL;
+ goto out_err_destroy;
+ }
+
+out_err_destroy:
+ destroy_cfg();
+
+out_err:
+ return rc;
+}
+
+/* vdp_gettlv - get the tlv for a profile
+ * @port: the port on which the tlv was received
+ * @profile: profile the vsi tlv is created from
+ *
+ * returns 0 on success
+ *
+ * this is the interface function called from ecp_build_ECPDU. It returns the
+ * packed tlv for a profile.
+ */
+struct packed_tlv *vdp_gettlv(struct vdp_data *vd, struct vsi_profile *profile)
+{
+ int size;
+ struct packed_tlv *ptlv = NULL;
+
+ /* frees the unpacked_tlv in vdp_data
+ * also done in vdp_bld_vsi_tlv */
+ vdp_free_tlv(vd);
+
+ if (vdp_bld_tlv(vd, profile)) {
+ LLDPAD_ERR("%s:%s vdp_bld_tlv failed\n",
+ __func__, vd->ifname);
+ goto out_err;
+ }
+
+ size = TLVSIZE(vd->vdp);
+
+ if (!size) {
+ LLDPAD_ERR("%s(%i): size %i of unpacked_tlv not correct !\n", __func__, __LINE__,
+ size);
+ goto out_err;
+ }
+
+ ptlv = create_ptlv();
+ if (!ptlv)
+ goto out_err;
+
+ ptlv->tlv = malloc(size);
+ if (!ptlv->tlv)
+ goto out_free;
+
+ ptlv->size = 0;
+ PACK_TLV_AFTER(vd->vdp, ptlv, size, out_free);
+
+ return ptlv;
+
+out_free:
+ ptlv = free_pkd_tlv(ptlv);
+out_err:
+ LLDPAD_ERR("%s:%s: failed\n", __func__, vd->ifname);
+ return NULL;
+}
+
+/* vdp_profile_equal - checks for equality of 2 profiles
+ * @p1: profile 1
+ * @p2: profile 2
+ *
+ * returns 1 on success, 0 on error
+ *
+ * compares mgrid, id, version, instance, mac and vlan of 2 profiles to find
+ * out if they are equal.
+ */
+int vdp_profile_equal(struct vsi_profile *p1, struct vsi_profile *p2)
+{
+ if (p1->mgrid != p2->mgrid)
+ return 0;
+
+ if (p1->id != p2->id)
+ return 0;
+
+ if (p1->version != p2->version)
+ return 0;
+
+ if (memcmp(p1->instance, p2->instance, 16))
+ return 0;
+
+ if (memcmp(p1->mac, p2->mac, MAC_ADDR_LEN))
+ return 0;
+
+ if (p1->vlan != p2->vlan)
+ return 0;
+
+ return 1;
+}
+
+/* vdp_add_profile - adds a profile to a per port list
+ * @profile: profile to add
+ *
+ * returns the profile that has been found or added, NULL otherwise.
+ *
+ * main interface function which adds a profile to a list kept on a per-port
+ * basis. Checks if the profile is already in the list, adds it if necessary.
+ */
+struct vsi_profile *vdp_add_profile(struct vsi_profile *profile)
+{
+ struct vsi_profile *p;
+ struct vdp_data *vd;
+
+ LLDPAD_DBG("%s(%i): adding vdp profile for %s !\n", __func__, __LINE__,
+ profile->port->ifname);
+
+ vd = vdp_data(profile->port->ifname);
+ if (!vd) {
+ LLDPAD_ERR("%s(%i): Could not find vdp_data for %s !\n", __func__, __LINE__,
+ profile->port->ifname);
+ return NULL;
+ }
+
+ vdp_print_profile(profile);
+
+ /* loop over all existing profiles and check wether
+ * one for this combination already exists. If yes, check,
+ * if the MAC/VLAN pair already exists. If not, add it.
+ * Note: currently only one MAC/VLAN pair supported ! */
+ LIST_FOREACH(p, &vd->profile_head, profile) {
+ if (p) {
+ if (vdp_profile_equal(p, profile)) {
+ if (p->mode == profile->mode) {
+ LLDPAD_DBG("%s(%i): profile already exists, ignoring !\n",
+ __func__, __LINE__);
+ return NULL;
+ } else {
+ LLDPAD_DBG("%s(%i): taking new mode !\n", __func__,
+ __LINE__);
+ p->mode = profile->mode;
+ return p;
+ }
+ }
+ }
+ }
+
+ LIST_INSERT_HEAD(&vd->profile_head, profile, profile );
+
+ return profile;
+}
+
+/* vdp_remove_profile - remove a profile from a per port list
+ * @profile: profile to remove
+ *
+ * returns 0 if removal was successful, -1 if removal failed
+ *
+ * function used in the state machines to remove a profile from a list kept on
+ * a per-port basis. Checks if the profile is in the list, removes it if there.
+ */
+int vdp_remove_profile(struct vsi_profile *profile)
+{
+ struct vsi_profile *p;
+ struct vdp_data *vd;
+
+ LLDPAD_DBG("%s(%i): removing vdp profile on %s !\n", __func__, __LINE__,
+ profile->port->ifname);
+
+ vd = vdp_data(profile->port->ifname);
+ if (!vd) {
+ LLDPAD_ERR("%s(%i): Could not find vdp_data for %s !\n", __func__, __LINE__,
+ profile->port->ifname);
+ return -1;
+ }
+
+ /* loop over all existing profiles and check wether
+ * it exists. If yes, remove it. */
+ LIST_FOREACH(p, &vd->profile_head, profile) {
+ if (p) {
+ vdp_print_profile(p);
+ if (vdp_profile_equal(p, profile)) {
+ LIST_REMOVE(p, profile);
+ free(p);
+ }
+ } else {
+ return -1;
+ }
+ }
+
+ return 0;
+}
+
+/* vdp_ifdown - tear down vdp structures for a interface
+ * @ifname: name of the interface
+ *
+ * no return value
+ *
+ * interface function to lldpad. tears down vdp specific structures if
+ * interface "ifname" goes down.
+ */
+void vdp_ifdown(char *ifname)
+{
+ struct vdp_data *vd;
+ struct vsi_profile *p;
+
+ LLDPAD_DBG("%s called on interface %s !\n", __func__, ifname);
+
+ vd = vdp_data(ifname);
+ if (!vd)
+ goto out_err;
+
+ if (ecp_deinit(ifname))
+ goto out_err;
+
+ LIST_REMOVE(vd, entry);
+
+ LIST_FOREACH(p, &vd->profile_head, profile) {
+ vdp_stop_ackTimer(p);
+ LIST_REMOVE(p, profile);
+ free(p);
+ }
+ vdp_free_tlv(vd);
+ free(vd);
+ LLDPAD_INFO("%s:%s vdp data removed\n", __func__, ifname);
+ return;
+out_err:
+ LLDPAD_ERR("%s:%s vdp data remove failed\n", __func__, ifname);
+
+ return;
+}
+
+/* vdp_ifup - build up vdp structures for a interface
+ * @ifname: name of the interface
+ *
+ * no return value
+ *
+ * interface function to lldpad. builds up vdp specific structures if
+ * interface "ifname" goes down.
+ */
+void vdp_ifup(char *ifname)
+{
+ char *p;
+ struct vdp_data *vd;
+ struct vdp_user_data *ud;
+
+ LLDPAD_DBG("%s(%i): starting VDP for if %s !\n", __func__, __LINE__, ifname);
+
+ vd = vdp_data(ifname);
+ if (vd) {
+ LLDPAD_WARN("%s:%s vdp data already exists !\n", __func__, ifname);
+ return;
+ }
+
+ /* not found, alloc/init per-port module data */
+ vd = (struct vdp_data *) calloc(1, sizeof(struct vdp_data));
+ if (!vd) {
+ LLDPAD_ERR("%s:%s malloc %ld failed\n",
+ __func__, ifname, sizeof(*vd));
+ goto out_err;
+ }
+ strncpy(vd->ifname, ifname, IFNAMSIZ);
+
+ if (!init_cfg()) {
+ free(vd);
+ goto out_err;
+ }
+
+ vd->role = VDP_ROLE_STATION;
+
+ if (!get_cfg(ifname, "vdp.role", (void *)&p,
+ CONFIG_TYPE_STRING)) {
+ if (!strcasecmp(p, VAL_BRIDGE)) {
+ vd->role = VDP_ROLE_BRIDGE;
+ }
+ }
+
+ LLDPAD_DBG("%s: configured for %s mode !\n", ifname,
+ (vd->role ==VDP_ROLE_BRIDGE) ? "bridge" : "station");
+
+ LIST_INIT(&vd->profile_head);
+
+ ud = find_module_user_data_by_if(ifname, &lldp_head, LLDP_MOD_VDP);
+ LIST_INSERT_HEAD(&ud->head, vd, entry);
+
+ ecp_init(ifname);
+
+ LLDPAD_DBG("%s:%s vdp added\n", __func__, ifname);
+ return;
+
+out_err:
+ LLDPAD_ERR("%s:%s vdp adding failed\n", __func__, ifname);
+ return;
+}
+
+static const struct lldp_mod_ops vdp_ops = {
+ .lldp_mod_register = vdp_register,
+ .lldp_mod_unregister = vdp_unregister,
+ .lldp_mod_ifup = vdp_ifup,
+ .lldp_mod_ifdown = vdp_ifdown,
+ .get_arg_handler = vdp_get_arg_handlers,
+};
+
+/* vdp_register - register vdp module to lldpad
+ * @none
+ *
+ * returns lldp_module struct on success, NULL on error
+ *
+ * allocates a module structure with vdp module information and returns it
+ * to lldpad.
+ */
+struct lldp_module *vdp_register(void)
+{
+ struct lldp_module *mod;
+ struct vdp_user_data *ud;
+
+ mod = malloc(sizeof(*mod));
+ if (!mod) {
+ LLDPAD_ERR("failed to malloc module data\n");
+ log_message(MSG_ERR_SERVICE_START_FAILURE,
+ "%s", "failed to malloc module data");
+ goto out_err;
+ }
+ ud = malloc(sizeof(struct vdp_user_data));
+ if (!ud) {
+ free(mod);
+ LLDPAD_ERR("failed to malloc module user data\n");
+ log_message(MSG_ERR_SERVICE_START_FAILURE,
+ "%s", "failed to malloc module user data");
+ goto out_err;
+ }
+ LIST_INIT(&ud->head);
+ mod->id = LLDP_MOD_VDP;
+ mod->ops = &vdp_ops;
+ mod->data = ud;
+ LLDPAD_DBG("%s:done\n", __func__);
+ return mod;
+
+out_err:
+ LLDPAD_ERR("%s:failed\n", __func__);
+ return NULL;
+}
+
+/* vdp_unregister - unregister vdp module from lldpad
+ * @none
+ *
+ * no return value
+ *
+ * frees vdp module structure.
+ */
+void vdp_unregister(struct lldp_module *mod)
+{
+ if (mod->data) {
+ vdp_free_data((struct vdp_user_data *) mod->data);
+ free(mod->data);
+ }
+ free(mod);
+ LLDPAD_DBG("%s:done\n", __func__);
+}
+
+
diff --git a/lldpad.c b/lldpad.c
index 571da31..c0938af 100644
--- a/lldpad.c
+++ b/lldpad.c
@@ -50,6 +50,7 @@
#include "lldp_med.h"
#include "lldp_8023.h"
#include "lldp_evb.h"
+#include "lldp_vdp.h"
#include "config.h"
#include "lldpad_shm.h"
#include "clif.h"
@@ -65,6 +66,7 @@ struct lldp_module *(*register_tlv_table[])(void) = {
med_register,
ieee8023_register,
evb_register,
+ vdp_register,
NULL,
};
--
1.7.1
^ permalink raw reply related
* [PATCH 3/9] BUGFIX: check for existence of ifup
From: Jens Osterkamp @ 2010-10-22 11:26 UTC (permalink / raw)
To: e1000-eedc, virtualization, evb; +Cc: chrisw, Jens Osterkamp
In-Reply-To: <1287746820-6036-1-git-send-email-jens@linux.vnet.ibm.com>
added an additional check for existence of the ifup operation.
If it does not exist, just skip it.
Signed-off-by: Jens Osterkamp <jens@linux.vnet.ibm.com>
---
config.c | 6 ++++--
1 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/config.c b/config.c
index 5cfac83..26afe3b 100644
--- a/config.c
+++ b/config.c
@@ -370,8 +370,10 @@ void init_ports(void)
} else if (check_link_status(p->if_name)) {
add_adapter(p->if_name);
- LIST_FOREACH(np, &lldp_head, lldp)
- np->ops->lldp_mod_ifup(p->if_name);
+ LIST_FOREACH(np, &lldp_head, lldp) {
+ if (np->ops->lldp_mod_ifup)
+ np->ops->lldp_mod_ifup(p->if_name);
+ }
set_lldp_port_enable_state(p->if_name, 1);
}
p++;
--
1.7.1
^ permalink raw reply related
* [PATCH 2/9] implementation of IEEE 802.1Qbg in lldpad, part 1
From: Jens Osterkamp @ 2010-10-22 11:26 UTC (permalink / raw)
To: e1000-eedc, virtualization, evb; +Cc: chrisw, Jens Osterkamp
In-Reply-To: <1287746820-6036-1-git-send-email-jens@linux.vnet.ibm.com>
This patch contains the first part of an initial implementation of the
IEEE 802.1Qbg standard: it implements code for the exchange of EVB
capabilities between a host with virtual machines and an adjacent switch.
For this it adds a new EVB TLV to LLDP.
Exchange of EVB TLV may be enabled or disabled on a per port basis.
Information about the information negotiated by the protocol can be
queried on the commandline with lldptool.
This patch adds support for querying and setting parameters used in
the exchange of EVB TLV messages.
The parameters that can be set are:
- forwarding mode
- host protocol capabilities (RTE, ECP, VDP)
- no. of supported VSIs
- retransmission timer exponent (RTE)
The parameters are implemented as a local policy: all frames received by
an adjacent switch are validated against this policy and taken over where
appropriate. Negotiated parameters are stored in lldpads config, picked up
again and used at the next start.
The patch applies to lldpad 0.9.38 and still contains code to log protocol
activity more verbosely than it would be necessary in the final version.
Signed-off-by: Jens Osterkamp <jens@linux.vnet.ibm.com>
---
Makefile.am | 10 +-
include/lldp.h | 19 ++
include/lldp_evb.h | 80 ++++++
include/lldp_evb_clif.h | 51 ++++
include/lldp_evb_cmds.h | 31 +++
include/lldp_tlv.h | 1 +
lldp_evb.c | 672 +++++++++++++++++++++++++++++++++++++++++++++++
lldp_evb_clif.c | 226 ++++++++++++++++
lldp_evb_cmds.c | 522 ++++++++++++++++++++++++++++++++++++
lldpad.c | 2 +
lldptool.c | 2 +
11 files changed, 1612 insertions(+), 4 deletions(-)
create mode 100644 include/lldp_evb.h
create mode 100644 include/lldp_evb_clif.h
create mode 100644 include/lldp_evb_cmds.h
create mode 100644 lldp_evb.c
create mode 100644 lldp_evb_clif.c
create mode 100644 lldp_evb_cmds.c
diff --git a/Makefile.am b/Makefile.am
index 743e16f..d59a6fa 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -37,7 +37,7 @@ lldpad_include_HEADERS = include/dcb_types.h include/dcbtool.h \
include/dcb_osdep.h include/clif.h include/lldp_dcbx_cmds.h include/common.h \
include/lldpad.h include/os.h include/includes.h include/lldp_mand_cmds.h \
include/clif_msgs.h include/lldp_basman_cmds.h include/lldp_8023_cmds.h \
-include/lldp_med_cmds.h include/lldp_dcbx_cfg.h
+include/lldp_med_cmds.h include/lldp_dcbx_cfg.h include/lldp_evb_cmds.h
noinst_HEADERS = include/config.h include/ctrl_iface.h \
include/dcb_driver_if_types.h include/dcb_driver_interface.h \
@@ -47,7 +47,7 @@ include/event_iface.h include/messages.h include/parse_cli.h include/version.h \
include/lldptool_cli.h include/list.h \
include/lldp_mand_clif.h include/lldp_basman_clif.h include/lldp_med_clif.h \
include/lldp_8023_clif.h include/lldp_dcbx_clif.h include/lldptool.h \
-include/lldp_rtnl.h
+include/lldp_rtnl.h include/lldp_evb_clif.h
lldpad_SOURCES = lldpad.c config.c drv_cfg.c ctrl_iface.c event_iface.c eloop.c \
common.c os_unix.c lldp_dcbx_cmds.c log.c lldpad_shm.c \
@@ -62,10 +62,12 @@ lldp_dcbx_cfg.c include/lldp_dcbx_cfg.h \
lldp_util.c include/lldp_util.h \
lldp_mand.c include/lldp_mand.h \
lldp_mand_cmds.c lldp_basman_cmds.c lldp_8023_cmds.c lldp_med_cmds.c \
+lldp_evb_cmds.c \
lldp_tlv.c include/lldp_tlv.h \
lldp_basman.c include/lldp_basman.h \
lldp_med.c include/lldp_med.h \
-lldp_8023.c include/lldp_8023.h
+lldp_8023.c include/lldp_8023.h \
+lldp_evb.c include/lldp_evb.h
@@ -74,7 +76,7 @@ $(lldpad_include_HEADERS) $(noinst_HEADERS)
lldptool_SOURCES = lldptool.c clif.c lldptool_cmds.c common.c os_unix.c \
lldp_mand_clif.c lldp_basman_clif.c lldp_med_clif.c lldp_8023_clif.c \
-lldp_dcbx_clif.c $(lldpad_include_HEADERS) $(noinst_HEADERS)
+lldp_dcbx_clif.c lldp_evb_clif.c $(lldpad_include_HEADERS) $(noinst_HEADERS)
nltest_SOURCES = nltest.c nltest.h
diff --git a/include/lldp.h b/include/lldp.h
index 66532bd..0371844 100644
--- a/include/lldp.h
+++ b/include/lldp.h
@@ -45,6 +45,8 @@
/* Telecommunications Industry Association TR-41 Committee */
#define OUI_TIA_TR41 0x0012bb
+#define OUI_IEEE_8021Qbg 0x001b3f
+
/* IEEE 802.3AB Clause 9: TLV Types */
#define CHASSIS_ID_TLV 1
#define PORT_ID_TLV 2
@@ -186,5 +188,22 @@ enum {
#define LLDP_8023_LINKAGG_CAPABLE (1 << 0)
#define LLDP_8023_LINKAGG_ENABLED (1 << 1)
+/* IEEE 802.1Qbg subtype */
+#define LLDP_EVB_SUBTYPE 0
+
+/* forwarding mode */
+#define LLDP_EVB_CAPABILITY_FORWARD_STANDARD (1 << 7)
+#define LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY (1 << 6)
+
+/* EVB supported protocols */
+#define LLDP_EVB_CAPABILITY_PROTOCOL_RTE (1 << 2)
+#define LLDP_EVB_CAPABILITY_PROTOCOL_ECP (1 << 1)
+#define LLDP_EVB_CAPABILITY_PROTOCOL_VDP (1 << 0)
+
+/* EVB specific values */
+#define LLDP_EVB_DEFAULT_MAX_VSI 65535
+#define LLDP_EVB_DEFAULT_SVSI 3295
+#define LLDP_EVB_DEFAULT_RTE 15
+
void somethingChangedLocal(char *ifname);
#endif /* _LLDP_H */
diff --git a/include/lldp_evb.h b/include/lldp_evb.h
new file mode 100644
index 0000000..667f9ad
--- /dev/null
+++ b/include/lldp_evb.h
@@ -0,0 +1,80 @@
+/*******************************************************************************
+
+ implementation of EVB TLVs for LLDP
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#ifndef _LLDP_EVB_H
+#define _LLDP_EVB_H
+
+#include "lldp_mod.h"
+
+#define LLDP_MOD_EVB OUI_IEEE_8021Qbg
+#define LLDP_OUI_SUBTYPE { 0x00, 0x1b, 0x3f, 0x00 }
+
+typedef enum {
+ EVB_OFFER_CAPABILITIES = 0,
+ EVB_CONFIGURE,
+ EVB_CONFIRMATION
+} evb_state;
+
+struct tlv_info_evb {
+ u8 oui[3];
+ u8 sub;
+ /* supported forwarding mode */
+ u8 smode;
+ /* supported capabilities */
+ u8 scap;
+ /* currently configured forwarding mode */
+ u8 cmode;
+ /* currently configured capabilities */
+ u8 ccap;
+ /* supported no. of vsi */
+ u16 svsi;
+ /* currently configured no. of vsi */
+ u16 cvsi;
+ /* retransmission exponent */
+ u8 rte;
+} __attribute__ ((__packed__));
+
+struct evb_data {
+ char ifname[IFNAMSIZ];
+ struct unpacked_tlv *evb;
+ struct tlv_info_evb *tie;
+ /* local policy */
+ struct tlv_info_evb *policy;
+ int state;
+ LIST_ENTRY(evb_data) entry;
+};
+
+struct evb_user_data {
+ LIST_HEAD(evb_head, evb_data) head;
+};
+
+struct lldp_module *evb_register(void);
+void evb_unregister(struct lldp_module *mod);
+struct packed_tlv *evb_gettlv(struct port *port);
+void evb_ifdown(char *);
+void evb_ifup(char *);
+struct evb_data *evb_data(char *ifname);
+
+#endif /* _LLDP_EVB_H */
diff --git a/include/lldp_evb_clif.h b/include/lldp_evb_clif.h
new file mode 100644
index 0000000..acaee5e
--- /dev/null
+++ b/include/lldp_evb_clif.h
@@ -0,0 +1,51 @@
+/*******************************************************************************
+
+ implementation of EVB TLVs for LLDP
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#ifndef _LLDP_EVB_CLIF_H
+#define _LLDP_EVB_CLIF_H
+
+struct lldp_module *evb_cli_register(void);
+void evb_cli_unregister(struct lldp_module *);
+int evb_print_tlv(u32, u16, char *);
+
+#define EVB_BUF_SIZE 256
+
+#define ARG_EVB_FORWARDING_MODE "fmode"
+
+#define VAL_EVB_FMODE_BRIDGE "bridge"
+#define VAL_EVB_FMODE_REFLECTIVE_RELAY "reflectiverelay"
+
+#define ARG_EVB_CAPABILITIES "capabilities"
+
+#define VAL_EVB_CAPA_RTE "rte"
+#define VAL_EVB_CAPA_ECP "ecp"
+#define VAL_EVB_CAPA_VDP "vdp"
+#define VAL_EVB_CAPA_NONE "none"
+
+#define ARG_EVB_VSIS "vsis"
+
+#define ARG_EVB_RTE "rte"
+
+#endif
diff --git a/include/lldp_evb_cmds.h b/include/lldp_evb_cmds.h
new file mode 100644
index 0000000..1367e5d
--- /dev/null
+++ b/include/lldp_evb_cmds.h
@@ -0,0 +1,31 @@
+/*******************************************************************************
+
+ implementation of EVB TLVs for LLDP
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#ifndef _LLDP_EVB_CMDS_H
+#define _LLDP_EVB_CMDS_H
+
+struct arg_handlers *evb_get_arg_handlers();
+
+#endif
diff --git a/include/lldp_tlv.h b/include/lldp_tlv.h
index a32cc71..fe3a75a 100644
--- a/include/lldp_tlv.h
+++ b/include/lldp_tlv.h
@@ -144,6 +144,7 @@ int tlv_ok(struct unpacked_tlv *tlv);
#define TLVID_8021(sub) TLVID(OUI_IEEE_8021, (sub))
#define TLVID_8023(sub) TLVID(OUI_IEEE_8023, (sub))
#define TLVID_MED(sub) TLVID(OUI_TIA_TR41, (sub))
+#define TLVID_8021Qbg(sub) TLVID(OUI_IEEE_8021Qbg, (sub))
/* the size in bytes needed for a packed tlv from unpacked tlv */
#define TLVSIZE(t) ((t) ? (2 + (t)->length) : 0)
diff --git a/lldp_evb.c b/lldp_evb.c
new file mode 100644
index 0000000..50ee17d
--- /dev/null
+++ b/lldp_evb.c
@@ -0,0 +1,672 @@
+/*******************************************************************************
+
+ implementation of EVB TLVs for LLDP
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include <net/if.h>
+#include <sys/queue.h>
+#include <sys/socket.h>
+#include <sys/ioctl.h>
+#include <sys/utsname.h>
+#include <linux/if_bridge.h>
+#include <string.h>
+#include "lldp.h"
+#include "lldp_evb.h"
+#include "messages.h"
+#include "config.h"
+#include "common.h"
+#include "lldp_mand_clif.h"
+#include "lldp_evb_clif.h"
+#include "lldp_evb_cmds.h"
+
+extern struct lldp_head lldp_head;
+
+struct evb_data *evb_data(char *ifname)
+{
+ struct evb_user_data *ud;
+ struct evb_data *ed = NULL;
+
+ ud = find_module_user_data_by_if(ifname, &lldp_head, LLDP_MOD_EVB);
+ if (ud) {
+ LIST_FOREACH(ed, &ud->head, entry) {
+ if (!strncmp(ifname, ed->ifname, IFNAMSIZ))
+ return ed;
+ }
+ }
+ return NULL;
+}
+
+static void evb_print_tlvinfo(struct tlv_info_evb *tie)
+{
+ LLDPAD_INFO("%s(%i): supported forwarding mode: %02x\n", __FILE__, __LINE__, tie->smode);
+ LLDPAD_INFO("%s(%i): configured forwarding mode: %02x\n", __FILE__, __LINE__, tie->cmode);
+ LLDPAD_INFO("%s(%i): supported capabilities: %02x\n", __FILE__, __LINE__, tie->scap);
+ LLDPAD_INFO("%s(%i): configured capabilities: %02x\n", __FILE__, __LINE__, tie->ccap);
+ LLDPAD_INFO("%s(%i): supported no. of vsis: %04i\n", __FILE__, __LINE__, tie->svsi);
+ LLDPAD_INFO("%s(%i): configured no. of vsis: %04i\n", __FILE__, __LINE__, tie->cvsi);
+ LLDPAD_INFO("%s(%i): rte: %02i\n", __FILE__, __LINE__, tie->rte);
+}
+
+static void evb_dump_tlv(struct unpacked_tlv *tlv)
+{
+ int i;
+ char *s, *t;
+
+ s = t = malloc(256);
+
+ if (!s) {
+ LLDPAD_ERR("%s(%i): unable to allocate string !\n", __func__, __LINE__);
+ }
+
+ LLDPAD_DBG("%s:type %i, length %i, info ", __func__, tlv->type, tlv->length);
+
+ for (i=0; i < tlv->length; i++) {
+ int c;
+ c = sprintf(s, "%02x ", tlv->info[i]);
+ s += c;
+ }
+
+ LLDPAD_DBG("%s\n", t);
+
+ free(t);
+}
+
+/*
+ * evb_bld_cfg_tlv - build the EVB TLV
+ * @ed: the evb data struct
+ *
+ * Returns 0 on success
+ */
+static int evb_bld_cfg_tlv(struct evb_data *ed)
+{
+ int rc = 0;
+ int i;
+ struct unpacked_tlv *tlv = NULL;
+
+ /* free ed->evb if it exists */
+ FREE_UNPKD_TLV(ed, evb);
+
+ if (!is_tlv_txenabled(ed->ifname, TLVID_8021Qbg(LLDP_EVB_SUBTYPE))) {
+ LLDPAD_ERR("%s:%s:EVB tx is currently disabled !\n",
+ __func__, ed->ifname);
+ rc = EINVAL;
+ goto out_err;
+ }
+
+ if (ed->tie->smode != ed->policy->smode) {
+ ed->tie->smode = ed->policy->smode;
+ ed->state = EVB_OFFER_CAPABILITIES;
+ }
+
+ tlv = create_tlv();
+ if (!tlv)
+ goto out_err;
+
+ tlv->type = ORG_SPECIFIC_TLV;
+ tlv->length = sizeof(struct tlv_info_evb);
+ tlv->info = (u8 *)malloc(tlv->length);
+ if(!tlv->info) {
+ free(tlv);
+ tlv = NULL;
+ rc = ENOMEM;
+ goto out_err;
+ }
+ memcpy(tlv->info, ed->tie, tlv->length);
+
+ LLDPAD_DBG("%s(%i): TLV about to be sent out:", __func__, __LINE__);
+ evb_dump_tlv(tlv);
+
+ ed->evb = tlv;
+out_err:
+ return rc;
+}
+
+static void evb_free_tlv(struct evb_data *ed)
+{
+ if (ed)
+ FREE_UNPKD_TLV(ed, evb);
+}
+
+/* evb_init_cfg_tlv:
+ *
+ * fill up tlv_info_evb structure with reasonable info
+ */
+static int evb_init_cfg_tlv(struct evb_data *ed)
+{
+ char arg_path[EVB_BUF_SIZE];
+ char *param;
+
+ /* load policy from config */
+ ed->policy = (struct tlv_info_evb *) calloc(1, sizeof(struct tlv_info_evb));
+
+ if (!ed->policy)
+ return ENOMEM;
+
+ /* set defaults */
+ hton24(ed->policy->oui, LLDP_MOD_EVB);
+ ed->policy->smode = LLDP_EVB_CAPABILITY_FORWARD_STANDARD;
+ ed->policy->scap = LLDP_EVB_CAPABILITY_PROTOCOL_RTE | LLDP_EVB_CAPABILITY_PROTOCOL_ECP |
+ LLDP_EVB_CAPABILITY_PROTOCOL_VDP;
+ ed->policy->cmode = 0;
+ ed->policy->ccap = 0;
+ ed->policy->svsi = LLDP_EVB_DEFAULT_SVSI;
+ ed->policy->rte = LLDP_EVB_DEFAULT_RTE;
+
+ /* pull forwarding mode into policy */
+ snprintf(arg_path, sizeof(arg_path), "%s%08x.fmode",
+ TLVID_PREFIX, TLVID_8021Qbg(LLDP_EVB_SUBTYPE));
+
+ if (get_cfg(ed->ifname, arg_path, (void *) ¶m, CONFIG_TYPE_STRING)) {
+ LLDPAD_WARN("%s:%s: loading EVB policy for forwarding mode failed, using default.\n",
+ __func__, ed->ifname);
+ } else {
+ if (strcasestr(param, VAL_EVB_FMODE_BRIDGE)) {
+ ed->policy->smode = LLDP_EVB_CAPABILITY_FORWARD_STANDARD;
+ }
+
+ if (strcasestr(param, VAL_EVB_FMODE_REFLECTIVE_RELAY)) {
+ ed->policy->smode = LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY;
+ }
+
+ LLDPAD_DBG("%s:%s: policy param fmode = %s.\n", __func__, ed->ifname, param);
+ LLDPAD_DBG("%s:%s: policy param smode = %x.\n", __func__, ed->ifname, ed->policy->smode);
+ }
+
+ /* pull capabilities into policy */
+ snprintf(arg_path, sizeof(arg_path), "%s%08x.capabilities",
+ TLVID_PREFIX, TLVID_8021Qbg(LLDP_EVB_SUBTYPE));
+
+ if (get_cfg(ed->ifname, arg_path, (void *) ¶m, CONFIG_TYPE_STRING)) {
+ LLDPAD_WARN("%s:%s: loading EVB policy for capabilities failed, using default.\n",
+ __func__, ed->ifname);
+ } else {
+ if (strcasestr(param, VAL_EVB_CAPA_RTE)) {
+ ed->policy->scap |= LLDP_EVB_CAPABILITY_PROTOCOL_RTE;
+ }
+
+ if (strcasestr(param, VAL_EVB_CAPA_ECP)) {
+ ed->policy->scap |= LLDP_EVB_CAPABILITY_PROTOCOL_ECP;
+ }
+
+ if (strcasestr(param, VAL_EVB_CAPA_VDP)) {
+ ed->policy->scap |= LLDP_EVB_CAPABILITY_PROTOCOL_VDP;
+ }
+
+ LLDPAD_DBG("%s:%s: policy param capabilities = %s.\n", __func__, ed->ifname, param);
+ LLDPAD_DBG("%s:%s: policy param scap = %x.\n", __func__, ed->ifname, ed->policy->scap);
+ }
+
+ /* pull rte into policy */
+ snprintf(arg_path, sizeof(arg_path), "%s%08x.rte",
+ TLVID_PREFIX, TLVID_8021Qbg(LLDP_EVB_SUBTYPE));
+
+ if (get_cfg(ed->ifname, arg_path, (void *) ¶m, CONFIG_TYPE_STRING)) {
+ LLDPAD_WARN("%s:%s: loading EVB policy for rte failed, using default.\n",
+ __func__, ed->ifname);
+ } else {
+ ed->policy->rte = atoi(param);
+
+ LLDPAD_DBG("%s:%s: policy param rte = %s.\n", __func__, ed->ifname, param);
+ LLDPAD_DBG("%s:%s: policy param rte = %i.\n", __func__, ed->ifname, ed->policy->rte);
+ }
+
+ /* pull vsis into policy */
+ snprintf(arg_path, sizeof(arg_path), "%s%08x.vsis",
+ TLVID_PREFIX, TLVID_8021Qbg(LLDP_EVB_SUBTYPE));
+
+ if (get_cfg(ed->ifname, arg_path, (void *) ¶m, CONFIG_TYPE_STRING)) {
+ LLDPAD_WARN("%s:%s: loading EVB policy for vsis failed, using default.\n",
+ __func__, ed->ifname);
+ } else {
+ ed->policy->svsi = atoi(param);
+
+ LLDPAD_DBG("%s:%s: policy param vsis = %s.\n", __func__, ed->ifname, param);
+ LLDPAD_DBG("%s:%s: policy param vsis = %i.\n", __func__, ed->ifname, ed->policy->svsi);
+ }
+
+ /* load last used EVB TLV ... */
+ ed->tie = (struct tlv_info_evb *) calloc(1, sizeof(struct tlv_info_evb));
+
+ if (!ed->tie) {
+ free(ed->policy);
+ return ENOMEM;
+ }
+
+ if (get_config_tlvinfo_bin(ed->ifname, TLVID_8021Qbg(LLDP_EVB_SUBTYPE),
+ (void *)ed->tie, sizeof(struct tlv_info_evb))) {
+ LLDPAD_INFO("%s:%s: loading last used EVB TLV failed, using default.\n",
+ __func__, ed->ifname);
+ hton24(ed->tie->oui, LLDP_MOD_EVB);
+ ed->tie->smode = ed->policy->smode;
+ ed->tie->cmode = 0x0;
+ ed->tie->scap = ed->policy->scap;
+ ed->tie->ccap = 0x0;
+ ed->tie->svsi = LLDP_EVB_DEFAULT_SVSI;
+ ed->tie->cvsi = 0x0;
+ ed->tie->rte = LLDP_EVB_DEFAULT_RTE;
+ } else {
+ LLDPAD_INFO("%s(%i): loaded last used EVB TLV from file.\n", __FILE__, __LINE__);
+ }
+
+ return 0;
+}
+
+static int evb_bld_tlv(struct evb_data *ed)
+{
+ int rc = 0;
+
+ if (!port_find_by_name(ed->ifname)) {
+ rc = EEXIST;
+ goto out_err;
+ }
+
+ if (!init_cfg()) {
+ rc = ENOENT;
+ goto out_err;
+ }
+
+ if (evb_bld_cfg_tlv(ed)) {
+ LLDPAD_ERR("%s:%s:evb_bld_cfg_tlv() failed\n",
+ __func__, ed->ifname);
+ rc = EINVAL;
+ goto out_err_destroy;
+ }
+
+ return rc;
+
+out_err_destroy:
+ destroy_cfg();
+out_err:
+ return rc;
+}
+
+static void evb_free_data(struct evb_user_data *ud)
+{
+ struct evb_data *ed;
+ if (ud) {
+ while (!LIST_EMPTY(&ud->head)) {
+ ed = LIST_FIRST(&ud->head);
+ LIST_REMOVE(ed, entry);
+ evb_free_tlv(ed);
+ free(ed);
+ }
+ }
+}
+
+struct packed_tlv *evb_gettlv(struct port *port)
+{
+ int size;
+ struct evb_data *ed;
+ struct packed_tlv *ptlv = NULL;
+
+ ed = evb_data(port->ifname);
+ if (!ed)
+ goto out_err;
+
+ evb_free_tlv(ed);
+
+ if (evb_bld_tlv(ed)) {
+ LLDPAD_ERR("%s:%s evb_bld_tlv failed\n",
+ __func__, port->ifname);
+ goto out_err;
+ }
+
+ size = TLVSIZE(ed->evb);
+
+ if (!size)
+ goto out_err;
+
+ ptlv = create_ptlv();
+ if (!ptlv)
+ goto out_err;
+
+ ptlv->tlv = malloc(size);
+ if (!ptlv->tlv)
+ goto out_free;
+
+ ptlv->size = 0;
+ PACK_TLV_AFTER(ed->evb, ptlv, size, out_free);
+ return ptlv;
+
+out_free:
+ /* FIXME: free function returns pointer ? */
+ ptlv = free_pkd_tlv(ptlv);
+out_err:
+ LLDPAD_ERR("%s:%s: failed\n", __func__, port->ifname);
+ return NULL;
+}
+
+/* evb_check_and_fill
+ *
+ * checks values received in TLV and takes over some values
+ */
+int evb_check_and_fill(struct evb_data *ed, struct tlv_info_evb *tie)
+{
+ /* sanity check of received data in tie */
+ if ((tie->smode & (LLDP_EVB_CAPABILITY_FORWARD_STANDARD |
+ LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY)) == 0) {
+ LLDPAD_ERR("Neither standard nor rr set as forwarding mode !");
+ return TLV_ERR;
+ }
+
+ if ((tie->svsi < 0) || (tie->svsi > LLDP_EVB_DEFAULT_MAX_VSI)) {
+ LLDPAD_ERR("nr of supported vsis (%i) exceeds allow value range !", tie->svsi);
+ return TLV_ERR;
+ }
+
+ if ((tie->cvsi < 0) || (tie->cvsi > LLDP_EVB_DEFAULT_MAX_VSI)) {
+ LLDPAD_ERR("nr of configured vsis (%i) exceeds allow value range !", tie->cvsi);
+ return TLV_ERR;
+ }
+
+ /* check bridge capabilities against local policy*/
+ /* if bridge supports RR and we support it as well, request it
+ * by setting smode in tlv to be sent out (ed->tie->smode) */
+ if ( (tie->smode & ed->policy->smode) ==
+ LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY ) {
+ ed->tie->smode = LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY;
+ } else {
+ ed->tie->smode = LLDP_EVB_CAPABILITY_FORWARD_STANDARD;
+ }
+
+ /* If both sides support RTE, set it */
+ if ((tie->scap & ed->policy->scap) & LLDP_EVB_CAPABILITY_PROTOCOL_RTE)
+ ed->tie->scap |= LLDP_EVB_CAPABILITY_PROTOCOL_RTE;
+
+ /* If both sides support ECP, set it */
+ if ((tie->scap & ed->policy->scap) & LLDP_EVB_CAPABILITY_PROTOCOL_ECP)
+ ed->tie->scap |= LLDP_EVB_CAPABILITY_PROTOCOL_ECP;
+
+ /* If both sides support VDP, set it */
+ if ((tie->scap & ed->policy->scap) & LLDP_EVB_CAPABILITY_PROTOCOL_VDP)
+ ed->tie->scap |= LLDP_EVB_CAPABILITY_PROTOCOL_VDP;
+
+ /* If supported caps include VDP take over min value of both */
+ if (ed->tie->scap & LLDP_EVB_CAPABILITY_PROTOCOL_VDP)
+ ed->tie->cvsi = MIN(ed->policy->svsi,tie->svsi);
+
+ /* If both sides support RTE and value offer is > 0, set it */
+ if ((ed->tie->scap & LLDP_EVB_CAPABILITY_PROTOCOL_RTE) &&
+ (tie->rte > 0) && (ed->policy->rte > 0))
+ ed->tie->rte = MAX(ed->policy->rte,tie->rte);
+
+ if (set_config_tlvinfo_bin(ed->ifname, TLVID_8021Qbg(LLDP_EVB_SUBTYPE),
+ (void *)ed->tie, sizeof(struct tlv_info_evb))) {
+ LLDPAD_ERR("%s(%i): error saving tlv_info_evb !\n", __FILE__, __LINE__);
+ } else {
+ LLDPAD_INFO("%s(%i): saved tlv_info_evb to config !\n", __FILE__, __LINE__);
+ }
+
+ /* maybe switch has already set the mode based on the saved info sent
+ * out on ifup */
+
+ if (tie->cmode == ed->tie->smode)
+ ed->tie->cmode = tie->cmode;
+
+ return TLV_OK;
+}
+
+/* evb_compare
+ *
+ * compare our own and received tlv_info_evb
+ */
+static int evb_compare(struct evb_data *ed, struct tlv_info_evb *tie)
+{
+ LLDPAD_DBG("%s(%i): \n", __func__, __LINE__);
+
+ if (ed->tie->cmode == tie->cmode)
+ return 0;
+ else
+ return 1;
+}
+
+/* evb_statemachine:
+ *
+ * handle possible states during EVB capabilities exchange
+ *
+ * possible states: EVB_OFFER_CAPABILITIES
+ * EVB_CONFIGURE
+ * EVB_CONFIRMATION
+ */
+static void evb_statemachine(struct evb_data *ed, struct tlv_info_evb *tie)
+{
+ switch(ed->state) {
+ case EVB_OFFER_CAPABILITIES:
+ /* waiting for valid packets to pour in
+ * if valid packet was received,
+ * - check parameters with what we have offered for this if,
+ * - fill structure with data,
+ * - enable local tx
+ * - switch to EVB_CONFIGURE
+ */
+ LLDPAD_DBG("%s: state -> EVB_OFFER_CAPABILITIES\n", __func__);
+ if (evb_check_and_fill(ed, tie) != TLV_OK) {
+ LLDPAD_ERR("Invalid contents of EVB Cfg TLV !\n");
+ return;
+ }
+ somethingChangedLocal(ed->ifname); /* trigger tx with new values */
+ ed->state = EVB_CONFIGURE;
+ break;
+ case EVB_CONFIGURE:
+ /* we received a valid packet, if contents is same with our local settings
+ * we can switch state to EVB_CONFIRMATION.*/
+ LLDPAD_DBG("%s: state -> EVB_CONFIGURE\n", __func__);
+ if (evb_compare(ed, tie)) {
+ ed->state= EVB_OFFER_CAPABILITIES;
+ } else {
+ LLDPAD_ERR("tlv_info_evb now equal !\n");
+ ed->state = EVB_CONFIRMATION;
+ }
+ somethingChangedLocal(ed->ifname);
+ break;
+ case EVB_CONFIRMATION:
+ /* we are already in confirmation and received a new packet with
+ * different parameters ? Check parameters. switch state back to
+ * EVB_CONFIGURE ? */
+ LLDPAD_DBG("%s: state -> EVB_CONFIRMATION\n", __func__);
+ break;
+ default:
+ LLDPAD_ERR("EVB statemachine reached invalid state !\n");
+ break;
+ }
+}
+
+/*
+ * evb_rchange: process RX TLV LLDPDU
+ *
+ * TLV not consumed on error
+ */
+static int evb_rchange(struct port *port, struct unpacked_tlv *tlv)
+{
+ int i;
+ struct evb_data *ed;
+ struct tlv_info_evb *tie = (struct tlv_info_evb *) tlv->info;
+ u8 oui_subtype[OUI_SUB_SIZE] = LLDP_OUI_SUBTYPE;
+
+ if (!init_cfg()) {
+ return SUBTYPE_INVALID;
+ }
+
+ ed = evb_data(port->ifname);
+
+ if (!ed)
+ return SUBTYPE_INVALID;
+
+ if (tlv->type == TYPE_127) {
+ /* check for length */
+ if (tlv->length < (OUI_SUB_SIZE)) {
+ return TLV_ERR;
+ }
+
+ /* check for oui */
+ if (memcmp(tlv->info, &oui_subtype, OUI_SUB_SIZE)) {
+ return SUBTYPE_INVALID;
+ }
+
+ /* disable rx if tx has been disabled by administrator */
+ if (!is_tlv_txenabled(ed->ifname, TLVID_8021Qbg(LLDP_EVB_SUBTYPE))) {
+ LLDPAD_WARN("%s:%s:EVB Config disabled\n",
+ __func__, ed->ifname);
+ return TLV_OK;
+ }
+
+ LLDPAD_DBG("%s(%i): received tlv:", __func__, __LINE__);
+ evb_dump_tlv(tlv);
+ evb_print_tlvinfo(tie);
+
+ /* change state */
+ evb_statemachine(ed, tie);
+
+ /* check which values have been taken over */
+ LLDPAD_DBG("%s(%i): new tlv:", __func__, __LINE__);
+ evb_print_tlvinfo(ed->tie);
+ }
+
+ return TLV_OK;
+}
+
+void evb_ifdown(char *ifname)
+{
+ struct evb_data *ed;
+
+ LLDPAD_DBG("%s called !\n", __func__);
+
+ ed = evb_data(ifname);
+ if (!ed)
+ goto out_err;
+
+ free(ed->policy);
+ free(ed->tie);
+ LIST_REMOVE(ed, entry);
+ evb_free_tlv(ed);
+ free(ed);
+ LLDPAD_INFO("%s:port %s removed\n", __func__, ifname);
+ return;
+out_err:
+ LLDPAD_ERR("%s:port %s remove failed\n", __func__, ifname);
+
+ return;
+}
+
+void evb_ifup(char *ifname)
+{
+ struct evb_data *ed;
+ struct evb_user_data *ud;
+
+ ed = evb_data(ifname);
+ if (ed) {
+ LLDPAD_DBG("%s:%s exists\n", __func__, ifname);
+ goto out_err;
+ }
+
+ /* not found, alloc/init per-port tlv data */
+ ed = (struct evb_data *) calloc(1, sizeof(struct evb_data));
+ if (!ed) {
+ LLDPAD_ERR("%s:%s malloc %ld failed\n",
+ __func__, ifname, sizeof(*ed));
+ goto out_err;
+ }
+ strncpy(ed->ifname, ifname, IFNAMSIZ);
+
+ if (!init_cfg()) {
+ LLDPAD_ERR("%s:%s init_cfg failed\n", __func__, ifname);
+ goto out_free;
+ }
+
+ if (evb_init_cfg_tlv(ed)) {
+ LLDPAD_ERR("%s:%s evb_init_cfg_tlv failed\n", __func__, ifname);
+ goto out_free;
+ }
+
+ ed->state = EVB_OFFER_CAPABILITIES;
+
+ if (evb_bld_tlv(ed)) {
+ LLDPAD_ERR("%s:%s evb_bld_tlv failed\n", __func__, ifname);
+ goto out_free;
+ }
+
+ ud = find_module_user_data_by_if(ifname, &lldp_head, LLDP_MOD_EVB);
+ LIST_INSERT_HEAD(&ud->head, ed, entry);
+ LLDPAD_DBG("%s:port %s added\n", __func__, ifname);
+ return;
+
+out_free:
+ free(ed->tie);
+ free(ed->policy);
+ free(ed);
+
+out_err:
+ return;
+}
+
+static const struct lldp_mod_ops evb_ops = {
+ .lldp_mod_register = evb_register,
+ .lldp_mod_unregister = evb_unregister,
+ .lldp_mod_gettlv = evb_gettlv,
+ .lldp_mod_rchange = evb_rchange,
+ .lldp_mod_ifup = evb_ifup,
+ .lldp_mod_ifdown = evb_ifdown,
+ .get_arg_handler = evb_get_arg_handlers,
+};
+
+struct lldp_module *evb_register(void)
+{
+ struct lldp_module *mod;
+ struct evb_user_data *ud;
+
+ mod = malloc(sizeof(*mod));
+ if (!mod) {
+ LLDPAD_ERR("failed to malloc module data\n");
+ log_message(MSG_ERR_SERVICE_START_FAILURE,
+ "%s", "failed to malloc module data");
+ goto out_err;
+ }
+ ud = malloc(sizeof(struct evb_user_data));
+ if (!ud) {
+ free(mod);
+ LLDPAD_ERR("failed to malloc module user data\n");
+ log_message(MSG_ERR_SERVICE_START_FAILURE,
+ "%s", "failed to malloc module user data");
+ goto out_err;
+ }
+ LIST_INIT(&ud->head);
+ mod->id = LLDP_MOD_EVB;
+ mod->ops = &evb_ops;
+ mod->data = ud;
+ LLDPAD_DBG("%s:done\n", __func__);
+ return mod;
+
+out_err:
+ LLDPAD_ERR("%s:failed\n", __func__);
+ return NULL;
+}
+
+void evb_unregister(struct lldp_module *mod)
+{
+ if (mod->data) {
+ evb_free_data((struct evb_user_data *) mod->data);
+ free(mod->data);
+ }
+ free(mod);
+ LLDPAD_DBG("%s:done", __func__);
+}
diff --git a/lldp_evb_clif.c b/lldp_evb_clif.c
new file mode 100644
index 0000000..7479d74
--- /dev/null
+++ b/lldp_evb_clif.c
@@ -0,0 +1,226 @@
+/*******************************************************************************
+
+ implementation of EVB TLVs for LLDP
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include "includes.h"
+#include "common.h"
+#include <stdio.h>
+#include <syslog.h>
+#include <sys/un.h>
+#include <sys/stat.h>
+#include "lldp_mod.h"
+#include "lldptool.h"
+#include "lldp.h"
+#include "lldp_evb.h"
+#include "lldp_evb_clif.h"
+
+void evb_print_cfg_tlv(u16, char *info);
+int evb_print_help();
+
+u32 evb_lookup_tlv_name(char *tlvid_str);
+
+static const struct lldp_mod_ops evb_ops_clif = {
+ .lldp_mod_register = evb_cli_register,
+ .lldp_mod_unregister = evb_cli_unregister,
+ .print_tlv = evb_print_tlv,
+ .lookup_tlv_name = evb_lookup_tlv_name,
+ .print_help = evb_print_help,
+};
+
+struct type_name_info evb_tlv_names[] = {
+ { (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE,
+ "EVB Configuration TLV",
+ "evbCfg", evb_print_cfg_tlv },
+ { INVALID_TLVID, NULL, NULL }
+};
+
+int evb_print_help()
+{
+ struct type_name_info *tn = &evb_tlv_names[0];
+
+ while (tn->type != INVALID_TLVID) {
+ if (tn->key && strlen(tn->key) && tn->name) {
+ printf(" %s", tn->key);
+ if (strlen(tn->key)+3 <= 8)
+ printf("\t");
+ printf("\t: %s\n", tn->name);
+ }
+ tn++;
+ }
+
+ return 0;
+}
+
+struct lldp_module *evb_cli_register(void)
+{
+ struct lldp_module *mod;
+
+ mod = malloc(sizeof(*mod));
+ if (!mod) {
+ fprintf(stderr, "failed to malloc module data\n");
+ return NULL;
+ }
+ mod->id = LLDP_MOD_EVB;
+ mod->ops = &evb_ops_clif;
+
+ return mod;
+}
+
+void evb_cli_unregister(struct lldp_module *mod)
+{
+ free(mod);
+}
+
+void evb_print_cfg_tlv(u16 len, char *info)
+{
+ u8 smode;
+ u8 scap;
+ u8 cmode;
+ u8 ccap;
+ u16 svsi;
+ u16 cvsi;
+ u8 rte;
+
+ if (len != 9) {
+ printf("Bad Cfg TLV: %s\n", info);
+ return;
+ }
+
+ if (!hexstr2bin(info, &smode, sizeof(smode))) {
+ printf("supported forwarding mode: (0x%02hhx)", smode);
+
+ if (smode & LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY)
+ printf(" reflective relay");
+
+ if (smode & LLDP_EVB_CAPABILITY_FORWARD_STANDARD)
+ printf(" standard 802.1Q");
+
+ printf("\n");
+ } else {
+ printf("Unable to decode smode !\n");
+ }
+
+ if (!hexstr2bin(info+2, &scap, sizeof(scap))) {
+ printf("\tsupported capabilities: (0x%02hhx)", scap);
+
+ if ( scap & LLDP_EVB_CAPABILITY_PROTOCOL_RTE)
+ printf(" RTE");
+
+ if ( scap & LLDP_EVB_CAPABILITY_PROTOCOL_ECP)
+ printf(" ECP");
+
+ if ( scap & LLDP_EVB_CAPABILITY_PROTOCOL_VDP)
+ printf(" VDP");
+
+ printf("\n");
+ } else {
+ printf("Unable to decode scap !\n");
+ }
+
+ if (!hexstr2bin(info+4, &cmode, sizeof(cmode))) {
+ printf("\tconfigured forwarding mode: (0x%02hhx)", cmode);
+
+ if (cmode & LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY)
+ printf(" reflective relay");
+
+ if (cmode & LLDP_EVB_CAPABILITY_FORWARD_STANDARD)
+ printf(" standard 802.1Q");
+
+ printf("\n");
+ } else {
+ printf("Unable to decode cmode !\n");
+ }
+
+ if (!hexstr2bin(info+6, &ccap, sizeof(ccap))) {
+ printf("\tconfigured capabilities: (0x%02hhx)", ccap);
+
+ if ( ccap & LLDP_EVB_CAPABILITY_PROTOCOL_RTE)
+ printf(" RTE");
+
+ if ( ccap & LLDP_EVB_CAPABILITY_PROTOCOL_ECP)
+ printf(" ECP");
+
+ if ( ccap & LLDP_EVB_CAPABILITY_PROTOCOL_VDP)
+ printf(" VDP");
+
+ printf("\n");
+ } else {
+ printf("Unable to decode ccap !\n");
+ }
+
+ if (!hexstr2bin(info+8, (u8 *)&svsi, sizeof(svsi))) {
+ printf("\tno. of supported VSIs: %04i\n",svsi);
+ } else {
+ printf("Unable to decode svsi !\n");
+ }
+
+ if (!hexstr2bin(info+12, (u8 *)&cvsi, sizeof(cvsi))) {
+ printf("\tno. of configured VSIs: %04i\n",cvsi);
+ } else {
+ printf("Unable to decode cvsi !\n");
+ }
+
+ if (!hexstr2bin(info+16, &rte, sizeof(rte))) {
+ printf("\tRTE: %i\n",rte);
+ } else {
+ printf("Unable to decode cvsi !\n");
+ }
+
+ printf("\n");
+}
+
+/* return 1: if it printed the TLV
+ * 0: if it did not
+ */
+int evb_print_tlv(u32 tlvid, u16 len, char *info)
+{
+ struct type_name_info *tn = &evb_tlv_names[0];
+
+ while (tn->type != INVALID_TLVID) {
+ if (tlvid == tn->type) {
+ printf("%s\n", tn->name);
+ if (tn->print_info) {
+ printf("\t");
+ tn->print_info(len-4, info);
+ }
+ return 1;
+ }
+ tn++;
+ }
+
+ return 0;
+}
+
+u32 evb_lookup_tlv_name(char *tlvid_str)
+{
+ struct type_name_info *tn = &evb_tlv_names[0];
+
+ while (tn->type != INVALID_TLVID) {
+ if (!strcasecmp(tn->key, tlvid_str))
+ return tn->type;
+ tn++;
+ }
+ return INVALID_TLVID;
+}
+
diff --git a/lldp_evb_cmds.c b/lldp_evb_cmds.c
new file mode 100644
index 0000000..bee0ec8
--- /dev/null
+++ b/lldp_evb_cmds.c
@@ -0,0 +1,522 @@
+/*******************************************************************************
+
+ implementation of EVB TLVs for LLDP
+ (c) Copyright IBM Corp. 2010
+
+ Author(s): Jens Osterkamp <jens@linux.vnet.ibm.com>
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms and conditions of the GNU General Public License,
+ version 2, as published by the Free Software Foundation.
+
+ This program is distributed in the hope it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc.,
+ 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+
+ The full GNU General Public License is included in this distribution in
+ the file called "COPYING".
+
+*******************************************************************************/
+
+#include "includes.h"
+#include "common.h"
+#include <stdio.h>
+#include <syslog.h>
+#include <sys/un.h>
+#include <sys/stat.h>
+#include <arpa/inet.h>
+#include <string.h>
+#include "lldpad.h"
+#include "ctrl_iface.h"
+#include "lldp.h"
+#include "lldp_evb.h"
+#include "lldp_mand_clif.h"
+#include "lldp_evb_clif.h"
+#include "lldp/ports.h"
+#include "libconfig.h"
+#include "config.h"
+#include "clif_msgs.h"
+#include "lldp/states.h"
+
+static int get_arg_tlvtxenable(struct cmd *, char *, char *, char *);
+static int set_arg_tlvtxenable(struct cmd *, char *, char *, char *);
+
+static int get_arg_fmode(struct cmd *, char *, char *, char *);
+static int set_arg_fmode(struct cmd *, char *, char *, char *);
+
+static int get_arg_rte(struct cmd *, char *, char *, char *);
+static int set_arg_rte(struct cmd *, char *, char *, char *);
+
+static int get_arg_vsis(struct cmd *, char *, char *, char *);
+static int set_arg_vsis(struct cmd *, char *, char *, char *);
+
+static int get_arg_capabilities(struct cmd *, char *, char *, char *);
+static int set_arg_capabilities(struct cmd *, char *, char *, char *);
+
+static struct arg_handlers arg_handlers[] = {
+ { ARG_EVB_FORWARDING_MODE, get_arg_fmode, set_arg_fmode },
+ { ARG_EVB_CAPABILITIES, get_arg_capabilities, set_arg_capabilities },
+ { ARG_EVB_VSIS, get_arg_vsis, set_arg_vsis },
+ { ARG_EVB_RTE, get_arg_rte, set_arg_rte },
+ { ARG_TLVTXENABLE, get_arg_tlvtxenable, set_arg_tlvtxenable },
+ { NULL }
+};
+
+static int get_arg_tlvtxenable(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ int value;
+ char *s;
+ char arg_path[EVB_BUF_SIZE];
+
+ if (cmd->cmd != cmd_gettlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ snprintf(arg_path, sizeof(arg_path), "%s%08x.%s",
+ TLVID_PREFIX, cmd->tlvid, arg);
+
+ if (get_cfg(cmd->ifname, arg_path, (void *)&value,
+ CONFIG_TYPE_BOOL))
+ value = false;
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ if (value)
+ s = VAL_YES;
+ else
+ s = VAL_NO;
+
+ sprintf(obuf, "%02x%s%04x%s", (unsigned int) strlen(arg), arg,
+ (unsigned int) strlen(s), s);
+
+ return cmd_success;
+}
+
+static int set_arg_tlvtxenable(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ int value;
+ char arg_path[EVB_BUF_SIZE];
+
+ if (cmd->cmd != cmd_settlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ if (!strcasecmp(argvalue, VAL_YES))
+ value = 1;
+ else if (!strcasecmp(argvalue, VAL_NO))
+ value = 0;
+ else
+ return cmd_invalid;
+
+ snprintf(arg_path, sizeof(arg_path), "%s%08x.%s",
+ TLVID_PREFIX, cmd->tlvid, arg);
+
+ if (set_cfg(cmd->ifname, arg_path, (void *)&value, CONFIG_TYPE_BOOL))
+ return cmd_failed;
+
+ somethingChangedLocal(cmd->ifname);
+
+ return cmd_success;
+}
+
+static int get_arg_fmode(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ char *s;
+ struct evb_data *ed;
+
+ if (cmd->cmd != cmd_gettlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ ed = evb_data((char *) &cmd->ifname);
+
+ if (!ed)
+ return cmd_invalid;
+
+ if (ed->policy->smode & LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY)
+ s = VAL_EVB_FMODE_REFLECTIVE_RELAY;
+ else
+ s = VAL_EVB_FMODE_BRIDGE;
+
+ sprintf(obuf, "%02x%s%04x%s", (unsigned int) strlen(arg), arg,
+ (unsigned int) strlen(s), s);
+
+ return cmd_success;
+}
+
+static int set_arg_fmode(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ u8 smode;
+ char arg_path[EVB_BUF_SIZE];
+ struct evb_data *ed;
+
+ if (cmd->cmd != cmd_settlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ ed = evb_data((char *) &cmd->ifname);
+
+ if (!ed)
+ return cmd_invalid;
+
+ smode = 0;
+
+ if (!strcasecmp(argvalue, VAL_EVB_FMODE_BRIDGE)) {
+ smode = LLDP_EVB_CAPABILITY_FORWARD_STANDARD;
+ }
+
+ if (!strcasecmp(argvalue, VAL_EVB_FMODE_REFLECTIVE_RELAY)) {
+ smode = LLDP_EVB_CAPABILITY_FORWARD_REFLECTIVE_RELAY;
+ }
+
+ if (smode == 0) {
+ return cmd_invalid;
+ } else {
+ ed->policy->smode = smode;
+ }
+
+ snprintf(arg_path, sizeof(arg_path), "%s%08x.fmode",
+ TLVID_PREFIX, cmd->tlvid);
+
+ if (set_cfg(ed->ifname, arg_path, (void *) &argvalue, CONFIG_TYPE_STRING)) {
+ printf("%s:%s: saving EVB forwarding mode failed.\n",
+ __func__, ed->ifname);
+ return cmd_invalid;
+ }
+
+ somethingChangedLocal(cmd->ifname);
+
+ return cmd_success;
+}
+
+static int get_arg_capabilities(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ int c;
+ char *s, *t;
+ struct evb_data *ed;
+
+ printf("%s(%i): arg %s, argvalue %s !\n", __func__, __LINE__, arg, argvalue);
+
+ s = t = malloc(EVB_BUF_SIZE);
+
+ if (!s)
+ goto out_free;
+
+ memset(s, 0, EVB_BUF_SIZE);
+
+ if (cmd->cmd != cmd_gettlv)
+ goto out_free;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ goto out_free;
+ default:
+ free(t);
+ return cmd_not_applicable;
+ }
+
+ ed = evb_data((char *) &cmd->ifname);
+ if (!ed)
+ goto out_free;
+
+ if (ed->policy->scap & LLDP_EVB_CAPABILITY_PROTOCOL_RTE) {
+ c = sprintf(s, VAL_EVB_CAPA_RTE " ");
+ if (c <= 0)
+ goto out_free;
+ s += c;
+ }
+
+ if (ed->policy->scap & LLDP_EVB_CAPABILITY_PROTOCOL_ECP) {
+ c = sprintf(s, VAL_EVB_CAPA_ECP " ");
+ if (c <= 0)
+ goto out_free;
+ s += c;
+ }
+
+ if (ed->policy->scap & LLDP_EVB_CAPABILITY_PROTOCOL_VDP) {
+ c = sprintf(s, VAL_EVB_CAPA_VDP " ");
+ if (c <= 0)
+ goto out_free;
+ s += c;
+ }
+
+ sprintf(obuf, "%02x%s%04x%s", (unsigned int) strlen(arg), arg,
+ (unsigned int) strlen(t), t);
+
+ free(t);
+ return cmd_success;
+
+out_free:
+ free(t);
+ return cmd_invalid;
+}
+
+static int set_arg_capabilities(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ u8 scap = 0;
+ char arg_path[EVB_BUF_SIZE];
+ struct evb_data *ed;
+
+ if (cmd->cmd != cmd_settlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ ed = evb_data((char *) &cmd->ifname);
+
+ if (!ed)
+ return cmd_invalid;
+
+ if (strcasestr(argvalue, VAL_EVB_CAPA_RTE))
+ scap |= LLDP_EVB_CAPABILITY_PROTOCOL_RTE;
+
+ if (strcasestr(argvalue, VAL_EVB_CAPA_ECP))
+ scap |= LLDP_EVB_CAPABILITY_PROTOCOL_ECP;
+
+ if (strcasestr(argvalue, VAL_EVB_CAPA_VDP))
+ scap |= LLDP_EVB_CAPABILITY_PROTOCOL_VDP;
+
+ ed->policy->scap = scap;
+
+ snprintf(arg_path, sizeof(arg_path), "%s%08x.capabilities",
+ TLVID_PREFIX, cmd->tlvid);
+
+ if (set_cfg(ed->ifname, arg_path, (void *) &argvalue, CONFIG_TYPE_STRING)) {
+ printf("%s:%s: saving EVB capabilities failed.\n",
+ __func__, ed->ifname);
+ return cmd_invalid;
+ }
+
+ somethingChangedLocal(cmd->ifname);
+
+ return cmd_success;
+}
+
+static int get_arg_rte(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ char s[EVB_BUF_SIZE];
+ struct evb_data *ed;
+
+ if (cmd->cmd != cmd_gettlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ ed = evb_data((char *) &cmd->ifname);
+ if (!ed)
+ return cmd_invalid;
+
+ if (sprintf(s, "%i", ed->policy->rte) <= 0)
+ return cmd_invalid;
+
+ sprintf(obuf, "%02x%s%04x%s", (unsigned int) strlen(arg), arg,
+ (unsigned int) strlen(s), s);
+
+ return cmd_success;
+}
+
+static int set_arg_rte(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ int value, err;
+ char arg_path[EVB_BUF_SIZE];
+ char svalue[10];
+ struct evb_data *ed = NULL;
+
+ if (cmd->cmd != cmd_settlv)
+ goto out_err;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ goto out_err;
+ default:
+ return cmd_not_applicable;
+ }
+
+ ed = evb_data((char *) &cmd->ifname);
+
+ if (!ed)
+ return cmd_invalid;
+
+ value = atoi(argvalue);
+
+ if ((value < 0))
+ goto out_err;
+
+ ed->policy->rte = value;
+
+ err = snprintf(arg_path, sizeof(arg_path), "%s%08x.rte",
+ TLVID_PREFIX, cmd->tlvid);
+
+ if (err < 0)
+ goto out_err;
+
+ err = snprintf(svalue, sizeof(svalue), "%i", value);
+
+ if (err < 0)
+ goto out_err;
+
+ if (set_cfg(ed->ifname, arg_path, (void *) svalue, CONFIG_TYPE_STRING))
+ goto out_err;
+
+ somethingChangedLocal(cmd->ifname);
+
+ return cmd_success;
+
+out_err:
+ printf("%s:%s: saving EVB rte failed.\n", __func__, ed->ifname);
+ return cmd_invalid;
+}
+
+static int get_arg_vsis(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ char s[EVB_BUF_SIZE];
+ struct evb_data *ed;
+
+ if (cmd->cmd != cmd_gettlv)
+ return cmd_invalid;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ return cmd_invalid;
+ default:
+ return cmd_not_applicable;
+ }
+
+ ed = evb_data((char *) &cmd->ifname);
+ if (!ed)
+ return cmd_invalid;
+
+ if (sprintf(s, "%04i", ed->policy->svsi) <= 0)
+ return cmd_invalid;
+
+ sprintf(obuf, "%02x%s%04x%s", (unsigned int) strlen(arg), arg,
+ (unsigned int) strlen(s), s);
+
+ return cmd_success;
+}
+
+static int set_arg_vsis(struct cmd *cmd, char *arg, char *argvalue,
+ char *obuf)
+{
+ int value, err;
+ char arg_path[EVB_BUF_SIZE];
+ char svalue[10];
+ char *sv;
+ struct evb_data *ed = NULL;
+
+ ed = evb_data((char *) &cmd->ifname);
+
+ if (!ed)
+ return cmd_invalid;
+
+ if (cmd->cmd != cmd_settlv)
+ goto out_err;
+
+ switch (cmd->tlvid) {
+ case (LLDP_MOD_EVB << 8) | LLDP_EVB_SUBTYPE:
+ break;
+ case INVALID_TLVID:
+ goto out_err;
+ default:
+ return cmd_not_applicable;
+ }
+
+ value = atoi(argvalue);
+
+ if ((value < 0) || (value > LLDP_EVB_DEFAULT_MAX_VSI))
+ goto out_err;
+
+ ed->policy->svsi = value;
+
+ err = snprintf(arg_path, sizeof(arg_path), "%s%08x.vsis",
+ TLVID_PREFIX, cmd->tlvid);
+
+ if (err < 0)
+ goto out_err;
+
+ err = snprintf(svalue, sizeof(svalue), "%i", value);
+
+ if (err < 0)
+ goto out_err;
+
+ sv = &svalue[0];
+
+ if (set_cfg(ed->ifname, arg_path, (void *) &sv, CONFIG_TYPE_STRING))
+ goto out_err;
+
+ somethingChangedLocal(cmd->ifname);
+
+ return cmd_success;
+
+out_err:
+ printf("%s:%s: saving EVB vsis failed.\n", __func__, ed->ifname);
+ return cmd_invalid;
+}
+
+struct arg_handlers *evb_get_arg_handlers()
+{
+ return &arg_handlers[0];
+}
diff --git a/lldpad.c b/lldpad.c
index a89b5a4..571da31 100644
--- a/lldpad.c
+++ b/lldpad.c
@@ -49,6 +49,7 @@
#include "lldp_dcbx.h"
#include "lldp_med.h"
#include "lldp_8023.h"
+#include "lldp_evb.h"
#include "config.h"
#include "lldpad_shm.h"
#include "clif.h"
@@ -63,6 +64,7 @@ struct lldp_module *(*register_tlv_table[])(void) = {
dcbx_register,
med_register,
ieee8023_register,
+ evb_register,
NULL,
};
diff --git a/lldptool.c b/lldptool.c
index 5cf2846..7e166fe 100644
--- a/lldptool.c
+++ b/lldptool.c
@@ -39,6 +39,7 @@
#include "lldp_med_clif.h"
#include "lldp_8023_clif.h"
#include "lldp_dcbx_clif.h"
+#include "lldp_evb_clif.h"
#include "lldptool.h"
#include "lldptool_cli.h"
#include "lldp_mod.h"
@@ -156,6 +157,7 @@ struct lldp_module *(*register_tlv_table[])(void) = {
ieee8023_cli_register,
med_cli_register,
dcbx_cli_register,
+ evb_cli_register,
NULL,
};
--
1.7.1
^ permalink raw reply related
* [PATCH 1/9] consolidation of MIN and MAX macros in common.h
From: Jens Osterkamp @ 2010-10-22 11:26 UTC (permalink / raw)
To: e1000-eedc, virtualization, evb; +Cc: chrisw, Jens Osterkamp
In-Reply-To: <1287746820-6036-1-git-send-email-jens@linux.vnet.ibm.com>
This patch consolidates a modified version of the already existing MIN
macro in lldpad to include/common.h and add a MAX macro.
Signed-off-by: Jens Osterkamp <jens@linux.vnet.ibm.com>
---
include/common.h | 14 ++++++++++++++
include/dcb_protocol.h | 4 ----
2 files changed, 14 insertions(+), 4 deletions(-)
diff --git a/include/common.h b/include/common.h
index 01746ea..ab07dc2 100644
--- a/include/common.h
+++ b/include/common.h
@@ -387,4 +387,18 @@ typedef int socklen_t;
const char * wpa_ssid_txt(u8 *ssid, size_t ssid_len);
+#define MIN(x,y) \
+ ({ \
+ typeof (x) __x = (x); \
+ typeof (y) __y = (y); \
+ __x < __y ? __x : __y; \
+ })
+
+#define MAX(x,y) \
+ ({ \
+ typeof (x) __x = (x); \
+ typeof (y) __y = (y); \
+ __x > __y ? __x : __y; \
+ })
+
#endif /* COMMON_H */
diff --git a/include/dcb_protocol.h b/include/dcb_protocol.h
index 63e2d3c..d7127f5 100644
--- a/include/dcb_protocol.h
+++ b/include/dcb_protocol.h
@@ -58,10 +58,6 @@ typedef enum {
#define DUP_DCBX_TLV_LLINK 0x0020
#define TOO_MANY_NGHBRS 0x0040
-//#ifndef min /*todo: change to min()
- #define MIN(x, y) ((x) < (y) ? x : y)
-//#endif
-
#define INIT_DCB_OUI {0x00,0x1b,0x21}
int add_adapter(char *device_name);
--
1.7.1
^ permalink raw reply related
* implementation of IEEE 802.1Qbg in lldpad
From: Jens Osterkamp @ 2010-10-22 11:26 UTC (permalink / raw)
To: e1000-eedc, virtualization, evb; +Cc: chrisw
Hi,
this set of patches contains the initial implementation of the IEEE 802.1Qbg
standard: code for the exchange of EVB TLVs in LLDP frames to negotiate VSI
capabalities as well as VDP VSI TLVs between a host with virtual machines
and an adjacent switch.
It supports setting the parameters of the TLV exchange from the command
line using lldptool.
VDP profiles consisting of mode,mgrid,typeid,typeidversion,instanceid,mac,vlan
can be given to lldpad with lldptool or sent to lldpad via netlink messages from
the kernel or another program, e.g. libvirt.
VDP profiles are processed through the VDP/VSI and ECP state machines and
sent out in ECP frames.
ACK frames are received and processed through ECP and VDP/VSI state machines.
It implements a VDP bridge role for a port together with a lldptool command to
switch a port to the bridge role.
These patches apply against lldpad 0.9.38. Protocol debug messages are now logged
to syslog via LLDPAD_* macros and no longer to console. If you want to see them,
configure your syslog daemon to collect debug messages and start lldpad with -V 7.
- improved cleanup in case of interface down
- improved checking if a port really exists
- converted all our code to use LLDPAD_ log messages
- addressed Johns comments
- modified EVB TLV defaults
- increase max number of VSIs to 64k
- fixed a number of bugs and memleaks
For more information about lldpad take a look at
http://sourceforge.net/projects/e1000/files/DCB%20Tools/lldpad/
Please review and comment.
Thank you !
Jens
^ permalink raw reply
* Re: xen PV on HVM and initial domain merge in linux-next
From: Stefano Stabellini @ 2010-10-21 13:44 UTC (permalink / raw)
To: Stephen Rothwell
Cc: Stefano Stabellini, Konrad Rzeszutek Wilk,
linux-kernel@vger.kernel.org, linux-next@vger.kernel.org,
Andrew Morton, Jeremy Fitzhardinge, Chris Wright,
virtualization@lists.osdl.org, xen-devel@lists.xensource.com
In-Reply-To: <20101021104541.8e10159a.sfr@canb.auug.org.au>
On Thu, 21 Oct 2010, Stephen Rothwell wrote:
> Hi Stefano,
>
> On Wed, 20 Oct 2010 16:47:51 +0100 Stefano Stabellini <stefano.stabellini@eu.citrix.com> wrote:
> >
> > Good idea, I have created a link called linux-next to
> > 2.6.36-rc8-initial-domain-v5 so the following should also work:
> >
> > git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git linux-next
>
> I have added that branch from today with just you listed as the contact
> (if you ant anyone else (even a mailing list) listed as a contact, just
> let me know.
>
> Thanks for adding your subsystem tree as a participant of linux-next. As
> you may know, this is not a judgment of your code. The purpose of
> linux-next is for integration testing and to lower the impact of
> conflicts between subsystems in the next merge window.
>
Thank you!
^ permalink raw reply
* Re: [GIT PULL net-2.6] vhost-net: access_ok fix
From: David Miller @ 2010-10-21 8:08 UTC (permalink / raw)
To: mst; +Cc: kvm, virtualization, netdev, linux-kernel
In-Reply-To: <20101019145901.GA16025@redhat.com>
From: "Michael S. Tsirkin" <mst@redhat.com>
Date: Tue, 19 Oct 2010 16:59:01 +0200
> David,
> Not sure if it's too late for 2.6.36 - in case it's not, the following tree
> includes a last minute bugfix for vhost-net, found by code inspection.
> It is on top of net-2.6.
> Thanks!
>
> The following changes since commit b0057c51db66c5f0f38059f242c57d61c4741d89:
>
> tg3: restore rx_dropped accounting (2010-10-11 16:06:24 -0700)
>
> are available in the git repository at:
> git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost.git vhost-net
Even though it's too late, I've pulled this.
^ permalink raw reply
* [PULL] virtio (console and one block)
From: Rusty Russell @ 2010-10-21 7:22 UTC (permalink / raw)
To: Linus Torvalds; +Cc: Amit Shah, Christoph Hellwig, virtualization
The following changes since commit f6f94e2ab1b33f0082ac22d71f66385a60d8157f:
Linux 2.6.36 (2010-10-20 13:30:22 -0700)
are available in the git repository at:
ssh://master.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus.git virtio
Amit Shah (18):
virtio: console: Reset vdev before removing device
virtio: console: Remove control vq data only if using multiport support
virtio: console: Check if portdev is valid in send_control_msg()
virtio: console: Un-block reads on chardev close
virtio: console: Unblock poll on port hot-unplug
virtio: console: Make read() return -ENODEV on hot-unplug
virtio: console: Make write() return -ENODEV on hot-unplug
virtio: console: remove_port() should return void
virtio: console: open: Use a common path for error handling
virtio: console: Add a list of portdevs that are active
virtio: console: Add a find_port_by_devt() function
virtio: console: Use cdev_alloc() instead of cdev_init()
virtio: console: Add reference counting for port struct
virtio: console: Reference counting portdev structs is not needed
virtio: console: Send SIGIO to processes that request it for host events
virtio: console: Send SIGIO on new data arrival on ports
virtio: console: Send SIGIO in case of port unplug
virtio: console: Disable lseek(2) for port file operations
Christoph Hellwig (1):
virtio_blk: remove BKL leftovers
drivers/block/virtio_blk.c | 17 +---
drivers/char/virtio_console.c | 240 +++++++++++++++++++++++++++++++++++------
2 files changed, 209 insertions(+), 48 deletions(-)
^ permalink raw reply
* Re: xen PV on HVM and initial domain merge in linux-next
From: Stephen Rothwell @ 2010-10-20 23:45 UTC (permalink / raw)
To: Stefano Stabellini
Cc: Jeremy Fitzhardinge, Stefano Stabellini, Konrad Rzeszutek Wilk,
linux-kernel@vger.kernel.org, Chris Wright,
virtualization@lists.osdl.org, linux-next@vger.kernel.org,
Andrew Morton, xen-devel@lists.xensource.com
In-Reply-To: <alpine.DEB.2.00.1010201644380.10348@kaball-desktop>
[-- Attachment #1.1: Type: text/plain, Size: 2019 bytes --]
Hi Stefano,
On Wed, 20 Oct 2010 16:47:51 +0100 Stefano Stabellini <stefano.stabellini@eu.citrix.com> wrote:
>
> Good idea, I have created a link called linux-next to
> 2.6.36-rc8-initial-domain-v5 so the following should also work:
>
> git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git linux-next
I have added that branch from today with just you listed as the contact
(if you ant anyone else (even a mailing list) listed as a contact, just
let me know.
Thanks for adding your subsystem tree as a participant of linux-next. As
you may know, this is not a judgment of your code. The purpose of
linux-next is for integration testing and to lower the impact of
conflicts between subsystems in the next merge window.
You will need to ensure that the patches/commits in your tree/series have
been:
* submitted under GPL v2 (or later) and include the Contributor's
Signed-off-by,
* posted to the relevant mailing list,
* reviewed by you (or another maintainer of your subsystem tree),
* successfully unit tested, and
* destined for the current or next Linux merge window.
Basically, this should be just what you would send to Linus (or ask him
to fetch). It is allowed to be rebased if you deem it necessary.
--
Cheers,
Stephen Rothwell
sfr@canb.auug.org.au
Legal Stuff:
By participating in linux-next, your subsystem tree contributions are
public and will be included in the linux-next trees. You may be sent
e-mail messages indicating errors or other issues when the
patches/commits from your subsystem tree are merged and tested in
linux-next. These messages may also be cross-posted to the linux-next
mailing list, the linux-kernel mailing list, etc. The linux-next tree
project and IBM (my employer) make no warranties regarding the linux-next
project, the testing procedures, the results, the e-mails, etc. If you
don't agree to these ground rules, let me know and I'll remove your tree
from participation in linux-next.
[-- Attachment #1.2: Type: application/pgp-signature, Size: 490 bytes --]
[-- Attachment #2: Type: text/plain, Size: 138 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
^ permalink raw reply
* Re: [GIT PULL net-2.6] vhost-net: access_ok fix
From: Michael S. Tsirkin @ 2010-10-20 22:43 UTC (permalink / raw)
To: David Miller; +Cc: kvm, virtualization, netdev, linux-kernel
In-Reply-To: <20101019145901.GA16025@redhat.com>
On Tue, Oct 19, 2010 at 04:59:01PM +0200, Michael S. Tsirkin wrote:
> David,
> Not sure if it's too late for 2.6.36 - in case it's not, the following tree
> includes a last minute bugfix for vhost-net, found by code inspection.
Um, I see it was too late :) Never mind ...
--
MST
^ permalink raw reply
* Re: [Xen-devel] Re: xen PV on HVM and initial domain merge in linux-next
From: Chris Wright @ 2010-10-20 16:43 UTC (permalink / raw)
To: Dan Magenheimer
Cc: Stephen Rothwell, Stefano Stabellini, xen-devel,
Jeremy Fitzhardinge, Konrad Wilk, linux-kernel, Chris Wright,
virtualization, linux-next, Andrew Morton
In-Reply-To: <5d143f14-26a1-4452-8b44-dae2a8b6df11@default>
* Dan Magenheimer (dan.magenheimer@oracle.com) wrote:
> > Not following the Xen develpment at all, I would like to have a
> > positive reply from the listed Xen contacts, please,
>
> I am not officially listed as a maintainer for Xen, but fwiw:
>
> Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
>
> And, Stephen, I think Chris Wright and virtualization@lists.osdl.org
> are stale entries in the MAINTAINERS file for Xen development,
> so you are unlikely to receive replies from him/them.
> (Chris, virtualization@lists.osdl.org ... please feel free
> to correct me if I am wrong.)
Yeah, I'm not really doing Xen pv stuff these days.
The virtualization list itself is always open, and it's useful for
things that cross over (e.g. virtio, pv clock, pv spinlocks).
thanks,
-chris
^ permalink raw reply
* Re: xen PV on HVM and initial domain merge in linux-next
From: Stefano Stabellini @ 2010-10-20 15:47 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk
Cc: Stefano Stabellini, Stephen Rothwell,
linux-kernel@vger.kernel.org, linux-next@vger.kernel.org,
Andrew Morton, Jeremy Fitzhardinge, Chris Wright,
virtualization@lists.osdl.org, xen-devel@lists.xensource.com
In-Reply-To: <20101020152808.GA22328@dumpdata.com>
On Wed, 20 Oct 2010, Konrad Rzeszutek Wilk wrote:
> The patches touch the Xen PCI components, so:
>
> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>
> And yeah, we need to update the MAINTAINERS file. Let me spin one out
> with the folks who are currently involved in this.
>
Thanks
> > All right.
> > I merged Jeremy's and Konrad's branches (the ones you just merged on
> > linux-next) on top of linux 2.6.36 rc8, then I rebased my series on top
> > of the result.
> > Please checkout this branch:
> >
> > git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git 2.6.36-rc8-initial-domain-v5
>
> Just as a note, you might want to create a branch titled
>
> #linux-next in your git tree and link the reviewed/acked/tested branch to it. That way
> Stephen does not have to change his entries in the linux-next tree every release.
>
Good idea, I have created a link called linux-next to
2.6.36-rc8-initial-domain-v5 so the following should also work:
git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git linux-next
> >
> > and let me know if it is suitable, it shouldn't have any merge
> > conflicts.
>
> /me nods. Works nicely. Thank you for doing this on such a short-time frame.
>
Great, thanks!
^ permalink raw reply
* Re: xen PV on HVM and initial domain merge in linux-next
From: Jeremy Fitzhardinge @ 2010-10-20 15:45 UTC (permalink / raw)
To: Stefano Stabellini, Stephen Rothwell
Cc: Jeremy Fitzhardinge, Rzeszutek Wilk, Stefano Stabellini,
linux-kernel@vger.kernel.org, Chris Wright,
virtualization@lists.osdl.org, linux-next@vger.kernel.org, Konrad,
Andrew Morton, xen-devel@lists.xensource.com
In-Reply-To: <alpine.DEB.2.00.1010201455160.10348@kaball-desktop>
Yes all ok.
J
"Stefano Stabellini" <stefano.stabellini@eu.citrix.com> wrote:
>On Wed, 20 Oct 2010, Stephen Rothwell wrote:
>> Hi Stefano,
>>
>> [just casting the net a bit wider ...]
>>
>> On Tue, 19 Oct 2010 18:51:47 +0100 Stefano Stabellini
><stefano.stabellini@eu.citrix.com> wrote:
>> >
>> > I forgot to CC the LKML and linux-next...
>> >
>> > On Tue, 19 Oct 2010, Stefano Stabellini wrote:
>> > > Stephen,
>> > > I have two patch series to merge in linux-next:
>> > >
>> > > PV on HVM: receive interrupts as xen events
>> > > xen: initial domain support
>> > >
>> > > they have all the acked-by needed and are both stable since
>several
>> > > weeks, however they depend on Konrad's xen-pcifront series and
>for this
>> > > reason I waited until now to ask for a merge in linux-next.
>> > >
>> > > Could you please pull:
>> > >
>> > > git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git
>linux-next-initial-domain-v4
>> > >
>> > > it contains both series rebased on Konrad's pcifront series
>merged on
>> > > linux-next (warning: it still contains the merge commit of
>> > > xen-pcifront-0.8.2 in linux-next).
>> > > Let me know if you have any conflicts or if you need me to change
>the
>> > > branch somehow.
>>
>> Not following the Xen develpment at all, I would like to have a
>positive
>> reply from the listed Xen contacts, please,
>>
>
>Sure.
>Jeremy?
>
>
>> I do have concerns that this is turning up so late, but I realise
>that
>> that is mainly due to a misunderstanding on the part of some of the
>Xen
>> community.
>>
>
>Thank you very much for understanding!
>
>
>> Also, the above tree is based on next-20101019 which means that I
>cannot
>> use it as is. All the trees merged into linux-next must be base on
>some
>> other stable tree (almost always Linus' tree). linux-next is rebuilt
>> from scratch every day, so I cannot ever include a previous day's
>version.
>>
>> Merging in other stable trees is OK (as long as the other maintainer
>is
>> aware of that and makes sure that their tree does not reabse).
>>
>> Basically what you send to me should be what you intend to send to
>Linus
>> during the next merge window.
>
>All right.
>I merged Jeremy's and Konrad's branches (the ones you just merged on
>linux-next) on top of linux 2.6.36 rc8, then I rebased my series on top
>of the result.
>Please checkout this branch:
>
>git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git
>2.6.36-rc8-initial-domain-v5
>
>and let me know if it is suitable, it shouldn't have any merge
>conflicts.
>
>Cheers,
>
>Stefano
>
>_______________________________________________
>Virtualization mailing list
>Virtualization@lists.linux-foundation.org
>https://lists.linux-foundation.org/mailman/listinfo/virtualization
--
Sent from my Android phone with K-9 Mail. Please excuse my brevity.
^ permalink raw reply
* Re: xen PV on HVM and initial domain merge in linux-next
From: Konrad Rzeszutek Wilk @ 2010-10-20 15:28 UTC (permalink / raw)
To: Stefano Stabellini
Cc: Stephen Rothwell, linux-kernel@vger.kernel.org,
linux-next@vger.kernel.org, Andrew Morton, Jeremy Fitzhardinge,
Chris Wright, virtualization@lists.osdl.org,
xen-devel@lists.xensource.com
In-Reply-To: <alpine.DEB.2.00.1010201455160.10348@kaball-desktop>
On Wed, Oct 20, 2010 at 04:16:17PM +0100, Stefano Stabellini wrote:
> On Wed, 20 Oct 2010, Stephen Rothwell wrote:
> > Hi Stefano,
> >
> > [just casting the net a bit wider ...]
> >
> > On Tue, 19 Oct 2010 18:51:47 +0100 Stefano Stabellini <stefano.stabellini@eu.citrix.com> wrote:
> > >
> > > I forgot to CC the LKML and linux-next...
> > >
> > > On Tue, 19 Oct 2010, Stefano Stabellini wrote:
> > > > Stephen,
> > > > I have two patch series to merge in linux-next:
> > > >
> > > > PV on HVM: receive interrupts as xen events
> > > > xen: initial domain support
> > > >
> > > > they have all the acked-by needed and are both stable since several
> > > > weeks, however they depend on Konrad's xen-pcifront series and for this
> > > > reason I waited until now to ask for a merge in linux-next.
> > > >
> > > > Could you please pull:
> > > >
> > > > git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git linux-next-initial-domain-v4
> > > >
> > > > it contains both series rebased on Konrad's pcifront series merged on
> > > > linux-next (warning: it still contains the merge commit of
> > > > xen-pcifront-0.8.2 in linux-next).
> > > > Let me know if you have any conflicts or if you need me to change the
> > > > branch somehow.
> >
> > Not following the Xen develpment at all, I would like to have a positive
> > reply from the listed Xen contacts, please,
> >
>
> Sure.
> Jeremy?
The patches touch the Xen PCI components, so:
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
And yeah, we need to update the MAINTAINERS file. Let me spin one out
with the folks who are currently involved in this.
>
>
> > I do have concerns that this is turning up so late, but I realise that
> > that is mainly due to a misunderstanding on the part of some of the Xen
> > community.
> >
>
> Thank you very much for understanding!
>
>
> > Also, the above tree is based on next-20101019 which means that I cannot
> > use it as is. All the trees merged into linux-next must be base on some
> > other stable tree (almost always Linus' tree). linux-next is rebuilt
> > from scratch every day, so I cannot ever include a previous day's version.
> >
> > Merging in other stable trees is OK (as long as the other maintainer is
> > aware of that and makes sure that their tree does not reabse).
> >
> > Basically what you send to me should be what you intend to send to Linus
> > during the next merge window.
>
> All right.
> I merged Jeremy's and Konrad's branches (the ones you just merged on
> linux-next) on top of linux 2.6.36 rc8, then I rebased my series on top
> of the result.
> Please checkout this branch:
>
> git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git 2.6.36-rc8-initial-domain-v5
Just as a note, you might want to create a branch titled
#linux-next in your git tree and link the reviewed/acked/tested branch to it. That way
Stephen does not have to change his entries in the linux-next tree every release.
>
> and let me know if it is suitable, it shouldn't have any merge
> conflicts.
/me nods. Works nicely. Thank you for doing this on such a short-time frame.
^ permalink raw reply
* Re: xen PV on HVM and initial domain merge in linux-next
From: Stefano Stabellini @ 2010-10-20 15:16 UTC (permalink / raw)
To: Stephen Rothwell
Cc: Stefano Stabellini, linux-kernel@vger.kernel.org,
Konrad Rzeszutek Wilk, linux-next@vger.kernel.org, Andrew Morton,
Jeremy Fitzhardinge, Chris Wright, virtualization@lists.osdl.org,
xen-devel@lists.xensource.com
In-Reply-To: <20101020113256.3647b1d9.sfr@canb.auug.org.au>
On Wed, 20 Oct 2010, Stephen Rothwell wrote:
> Hi Stefano,
>
> [just casting the net a bit wider ...]
>
> On Tue, 19 Oct 2010 18:51:47 +0100 Stefano Stabellini <stefano.stabellini@eu.citrix.com> wrote:
> >
> > I forgot to CC the LKML and linux-next...
> >
> > On Tue, 19 Oct 2010, Stefano Stabellini wrote:
> > > Stephen,
> > > I have two patch series to merge in linux-next:
> > >
> > > PV on HVM: receive interrupts as xen events
> > > xen: initial domain support
> > >
> > > they have all the acked-by needed and are both stable since several
> > > weeks, however they depend on Konrad's xen-pcifront series and for this
> > > reason I waited until now to ask for a merge in linux-next.
> > >
> > > Could you please pull:
> > >
> > > git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git linux-next-initial-domain-v4
> > >
> > > it contains both series rebased on Konrad's pcifront series merged on
> > > linux-next (warning: it still contains the merge commit of
> > > xen-pcifront-0.8.2 in linux-next).
> > > Let me know if you have any conflicts or if you need me to change the
> > > branch somehow.
>
> Not following the Xen develpment at all, I would like to have a positive
> reply from the listed Xen contacts, please,
>
Sure.
Jeremy?
> I do have concerns that this is turning up so late, but I realise that
> that is mainly due to a misunderstanding on the part of some of the Xen
> community.
>
Thank you very much for understanding!
> Also, the above tree is based on next-20101019 which means that I cannot
> use it as is. All the trees merged into linux-next must be base on some
> other stable tree (almost always Linus' tree). linux-next is rebuilt
> from scratch every day, so I cannot ever include a previous day's version.
>
> Merging in other stable trees is OK (as long as the other maintainer is
> aware of that and makes sure that their tree does not reabse).
>
> Basically what you send to me should be what you intend to send to Linus
> during the next merge window.
All right.
I merged Jeremy's and Konrad's branches (the ones you just merged on
linux-next) on top of linux 2.6.36 rc8, then I rebased my series on top
of the result.
Please checkout this branch:
git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git 2.6.36-rc8-initial-domain-v5
and let me know if it is suitable, it shouldn't have any merge
conflicts.
Cheers,
Stefano
^ permalink raw reply
* RE: [Xen-devel] Re: xen PV on HVM and initial domain merge in linux-next
From: Dan Magenheimer @ 2010-10-20 15:11 UTC (permalink / raw)
To: Stephen Rothwell, Stefano Stabellini
Cc: xen-devel, Jeremy Fitzhardinge, Konrad Wilk, linux-kernel,
Chris Wright, virtualization, linux-next, Andrew Morton
In-Reply-To: <20101020113256.3647b1d9.sfr@canb.auug.org.au>
> Not following the Xen develpment at all, I would like to have a
> positive reply from the listed Xen contacts, please,
I am not officially listed as a maintainer for Xen, but fwiw:
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
And, Stephen, I think Chris Wright and virtualization@lists.osdl.org
are stale entries in the MAINTAINERS file for Xen development,
so you are unlikely to receive replies from him/them.
(Chris, virtualization@lists.osdl.org ... please feel free
to correct me if I am wrong.)
> -----Original Message-----
> From: Stephen Rothwell [mailto:sfr@canb.auug.org.au]
> Sent: Tuesday, October 19, 2010 6:33 PM
> To: Stefano Stabellini
> Cc: xen-devel@lists.xensource.com; Jeremy Fitzhardinge; Konrad
> Rzeszutek Wilk; linux-kernel@vger.kernel.org; Chris Wright;
> virtualization@lists.osdl.org; linux-next@vger.kernel.org; Andrew
> Morton
> Subject: [Xen-devel] Re: xen PV on HVM and initial domain merge in
> linux-next
>
> Hi Stefano,
>
> [just casting the net a bit wider ...]
>
> On Tue, 19 Oct 2010 18:51:47 +0100 Stefano Stabellini
> <stefano.stabellini@eu.citrix.com> wrote:
> >
> > I forgot to CC the LKML and linux-next...
> >
> > On Tue, 19 Oct 2010, Stefano Stabellini wrote:
> > > Stephen,
> > > I have two patch series to merge in linux-next:
> > >
> > > PV on HVM: receive interrupts as xen events
> > > xen: initial domain support
> > >
> > > they have all the acked-by needed and are both stable since several
> > > weeks, however they depend on Konrad's xen-pcifront series and for
> this
> > > reason I waited until now to ask for a merge in linux-next.
> > >
> > > Could you please pull:
> > >
> > > git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git linux-
> next-initial-domain-v4
> > >
> > > it contains both series rebased on Konrad's pcifront series merged
> on
> > > linux-next (warning: it still contains the merge commit of
> > > xen-pcifront-0.8.2 in linux-next).
> > > Let me know if you have any conflicts or if you need me to change
> the
> > > branch somehow.
>
> Not following the Xen develpment at all, I would like to have a
> positive
> reply from the listed Xen contacts, please,
>
> I do have concerns that this is turning up so late, but I realise that
> that is mainly due to a misunderstanding on the part of some of the Xen
> community.
>
> Also, the above tree is based on next-20101019 which means that I
> cannot
> use it as is. All the trees merged into linux-next must be base on
> some
> other stable tree (almost always Linus' tree). linux-next is rebuilt
> from scratch every day, so I cannot ever include a previous day's
> version.
>
> Merging in other stable trees is OK (as long as the other maintainer is
> aware of that and makes sure that their tree does not reabse).
>
> Basically what you send to me should be what you intend to send to
> Linus
> during the next merge window.
>
> --
> Cheers,
> Stephen Rothwell sfr@canb.auug.org.au
> http://www.canb.auug.org.au/~sfr/
^ permalink raw reply
* virtio: console: Don't block entire guest if host doesn't read data
From: Rusty Russell @ 2010-10-20 3:15 UTC (permalink / raw)
To: Linus Torvalds; +Cc: Amit Shah, Hans de Goede, virtualization
From: Amit Shah <amit.shah@redhat.com>
If the host is slow in reading data or doesn't read data at all,
blocking write calls not only blocked the program that called write()
but the entire guest itself.
To overcome this, let's not block till the host signals it has given
back the virtio ring element we passed it. Instead, send the buffer to
the host and return to userspace. This operation then becomes similar
to how non-blocking writes work, so let's use the existing code for this
path as well.
This code change also ensures blocking write calls do get blocked if
there's not enough room in the virtio ring as well as they don't return
-EAGAIN to userspace.
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
CC: stable@kernel.org
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
---
drivers/char/virtio_console.c | 17 ++++++++++++++---
1 file changed, 14 insertions(+), 3 deletions(-)
diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
index c810481..0f69c5e 100644
--- a/drivers/char/virtio_console.c
+++ b/drivers/char/virtio_console.c
@@ -459,9 +459,12 @@ static ssize_t send_buf(struct port *port, void *in_buf, size_t in_count,
/*
* Wait till the host acknowledges it pushed out the data we
- * sent. This is done for ports in blocking mode or for data
- * from the hvc_console; the tty operations are performed with
- * spinlocks held so we can't sleep here.
+ * sent. This is done for data from the hvc_console; the tty
+ * operations are performed with spinlocks held so we can't
+ * sleep here. An alternative would be to copy the data to a
+ * buffer and relax the spinning requirement. The downside is
+ * we need to kmalloc a GFP_ATOMIC buffer each time the
+ * console driver writes something out.
*/
while (!virtqueue_get_buf(out_vq, &len))
cpu_relax();
@@ -626,6 +629,14 @@ static ssize_t port_fops_write(struct file *filp, const char __user *ubuf,
goto free_buf;
}
+ /*
+ * We now ask send_buf() to not spin for generic ports -- we
+ * can re-use the same code path that non-blocking file
+ * descriptors take for blocking file descriptors since the
+ * wait is already done and we're certain the write will go
+ * through to the host.
+ */
+ nonblock = true;
ret = send_buf(port, buf, count, nonblock);
if (nonblock && ret > 0)
^ permalink raw reply related
* Re: xen PV on HVM and initial domain merge in linux-next
From: Stephen Rothwell @ 2010-10-20 0:32 UTC (permalink / raw)
To: Stefano Stabellini
Cc: xen-devel, Jeremy Fitzhardinge, Konrad Rzeszutek Wilk,
linux-kernel, Chris Wright, virtualization, linux-next,
Andrew Morton
In-Reply-To: <alpine.DEB.2.00.1010191847290.10348@kaball-desktop>
[-- Attachment #1.1: Type: text/plain, Size: 1976 bytes --]
Hi Stefano,
[just casting the net a bit wider ...]
On Tue, 19 Oct 2010 18:51:47 +0100 Stefano Stabellini <stefano.stabellini@eu.citrix.com> wrote:
>
> I forgot to CC the LKML and linux-next...
>
> On Tue, 19 Oct 2010, Stefano Stabellini wrote:
> > Stephen,
> > I have two patch series to merge in linux-next:
> >
> > PV on HVM: receive interrupts as xen events
> > xen: initial domain support
> >
> > they have all the acked-by needed and are both stable since several
> > weeks, however they depend on Konrad's xen-pcifront series and for this
> > reason I waited until now to ask for a merge in linux-next.
> >
> > Could you please pull:
> >
> > git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git linux-next-initial-domain-v4
> >
> > it contains both series rebased on Konrad's pcifront series merged on
> > linux-next (warning: it still contains the merge commit of
> > xen-pcifront-0.8.2 in linux-next).
> > Let me know if you have any conflicts or if you need me to change the
> > branch somehow.
Not following the Xen develpment at all, I would like to have a positive
reply from the listed Xen contacts, please,
I do have concerns that this is turning up so late, but I realise that
that is mainly due to a misunderstanding on the part of some of the Xen
community.
Also, the above tree is based on next-20101019 which means that I cannot
use it as is. All the trees merged into linux-next must be base on some
other stable tree (almost always Linus' tree). linux-next is rebuilt
from scratch every day, so I cannot ever include a previous day's version.
Merging in other stable trees is OK (as long as the other maintainer is
aware of that and makes sure that their tree does not reabse).
Basically what you send to me should be what you intend to send to Linus
during the next merge window.
--
Cheers,
Stephen Rothwell sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/
[-- Attachment #1.2: Type: application/pgp-signature, Size: 490 bytes --]
[-- Attachment #2: Type: text/plain, Size: 138 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
^ permalink raw reply
* KVM Forum 2010: videos online [was Re: KVM Forum 2010: presentations online]
From: Chris Wright @ 2010-10-19 17:14 UTC (permalink / raw)
To: kvm, qemu-devel; +Cc: libvir-list, Chris Wright, virtualization
In-Reply-To: <20100816215039.GI17935@x200.localdomain>
* Chris Wright (chrisw@redhat.com) wrote:
> We were also able to video the speakers, and will send a note when the
> videos are available.
> (and thanks again to Andrew Cathrow for making this happen)
I don't think a note went out yet. The videos are available as well.
thanks,
-chris
^ permalink raw reply
* [GIT PULL net-2.6] vhost-net: access_ok fix
From: Michael S. Tsirkin @ 2010-10-19 14:59 UTC (permalink / raw)
To: David Miller; +Cc: kvm, virtualization, netdev, linux-kernel
David,
Not sure if it's too late for 2.6.36 - in case it's not, the following tree
includes a last minute bugfix for vhost-net, found by code inspection.
It is on top of net-2.6.
Thanks!
The following changes since commit b0057c51db66c5f0f38059f242c57d61c4741d89:
tg3: restore rx_dropped accounting (2010-10-11 16:06:24 -0700)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost.git vhost-net
Dan Carpenter (1):
vhost: fix return code for log_access_ok()
drivers/vhost/vhost.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
--
MST
^ permalink raw reply
* Re: [PATCH] virtio: console: Don't block entire guest if host doesn't read data
From: Hans de Goede @ 2010-10-19 8:03 UTC (permalink / raw)
To: Amit Shah; +Cc: stable, Virtualization List
In-Reply-To: <20101019073216.GA8896@amit-laptop.redhat.com>
Hi,
On 10/19/2010 09:32 AM, Amit Shah wrote:
> On (Tue) Oct 19 2010 [09:23:00], Hans de Goede wrote:
>>>> 3) This patch will cause processes filling the virtqueue fast enough to block
>>>> to never wake up again, due to a missing waitqueue wakeup, see:
>>>> https://bugzilla.redhat.com/show_bug.cgi?id=643750
>>>
>>> Doesn't happen in my testcase, but this patch shouldn't cause that
>>> problem if it exists -- it's a problem that exists even now for
>>> nonblocking ports. So if such a bug exists, it needs to be fixed
>>> independently.
>>
>> First of all lets agree that this is a real problem,
>
> Sure, got a testcase for the test-virtserial or kvm-autotest projects?
> ;-)
>
> I did try it and POLLOUT gets set for me immediately when I read one
> buffer from the host.
>
Ok, weird but ok :)
<snip>
>> I agree it is an independent problem, and should be fixed in a separate
>> patch, but that patch should be part of the same set and become *before*
>> this one, as this patch now extends the problem to ports opened in blocking
>> mode too.
>
> Strongly disagree. This patch fixes a problem wherein blocking-mode
> writes to a port freeze the entire guest. That's a much uglier problem
> to have than poll not indicating a port is writable again.
Ack, you're right.
Acked-by: Hans de Goede <hdegoede@redhat.com>
Regards,
Hans
^ permalink raw reply
* Re: [PATCH] virtio: console: Don't block entire guest if host doesn't read data
From: Amit Shah @ 2010-10-19 7:32 UTC (permalink / raw)
To: Hans de Goede; +Cc: stable, Virtualization List
In-Reply-To: <4CBD4754.5030504@redhat.com>
On (Tue) Oct 19 2010 [09:23:00], Hans de Goede wrote:
> >>3) This patch will cause processes filling the virtqueue fast enough to block
> >> to never wake up again, due to a missing waitqueue wakeup, see:
> >> https://bugzilla.redhat.com/show_bug.cgi?id=643750
> >
> >Doesn't happen in my testcase, but this patch shouldn't cause that
> >problem if it exists -- it's a problem that exists even now for
> >nonblocking ports. So if such a bug exists, it needs to be fixed
> >independently.
>
> First of all lets agree that this is a real problem,
Sure, got a testcase for the test-virtserial or kvm-autotest projects?
;-)
I did try it and POLLOUT gets set for me immediately when I read one
buffer from the host.
> there is simply nothing
> waking the waitqueue were fops_write (or poll) block on when buffers become
> available in out_vq, it may be hard to come up with a test case which fills
> the queue fast enough to hit this scenario, but it is very real.
Not at all.
Connect guest
Connect host
On guest, check for POLLOUT. As long as it's set, write buffers.
When POLLOUT goes off, read one buffer from host. See if POLLOUT is set
again.
Also, as I mentioned in a private chat, the fix for that problem is easy
enough.
> I agree it is an independent problem, and should be fixed in a separate
> patch, but that patch should be part of the same set and become *before*
> this one, as this patch now extends the problem to ports opened in blocking
> mode too.
Strongly disagree. This patch fixes a problem wherein blocking-mode
writes to a port freeze the entire guest. That's a much uglier problem
to have than poll not indicating a port is writable again.
> BTW, many thanks for working on this, it is appreciated :)
Sure, thanks :-)
Amit
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox