summaryrefslogtreecommitdiff
path: root/src/plugins/open/open.c
diff options
context:
space:
mode:
Diffstat (limited to 'src/plugins/open/open.c')
-rw-r--r--src/plugins/open/open.c439
1 files changed, 439 insertions, 0 deletions
diff --git a/src/plugins/open/open.c b/src/plugins/open/open.c
new file mode 100644
index 0000000..92b6f37
--- /dev/null
+++ b/src/plugins/open/open.c
@@ -0,0 +1,439 @@
+/*
+ * Copyright (c) 2003 Sun Microsystems, Inc. All Rights Reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * Redistribution of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *
+ * Redistribution in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * Neither the name of Sun Microsystems, Inc. or the names of
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * This software is provided "AS IS," without a warranty of any kind.
+ * ALL EXPRESS OR IMPLIED CONDITIONS, REPRESENTATIONS AND WARRANTIES,
+ * INCLUDING ANY IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR A
+ * PARTICULAR PURPOSE OR NON-INFRINGEMENT, ARE HEREBY EXCLUDED.
+ * SUN MICROSYSTEMS, INC. ("SUN") AND ITS LICENSORS SHALL NOT BE LIABLE
+ * FOR ANY DAMAGES SUFFERED BY LICENSEE AS A RESULT OF USING, MODIFYING
+ * OR DISTRIBUTING THIS SOFTWARE OR ITS DERIVATIVES. IN NO EVENT WILL
+ * SUN OR ITS LICENSORS BE LIABLE FOR ANY LOST REVENUE, PROFIT OR DATA,
+ * OR FOR DIRECT, INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL OR
+ * PUNITIVE DAMAGES, HOWEVER CAUSED AND REGARDLESS OF THE THEORY OF
+ * LIABILITY, ARISING OUT OF THE USE OF OR INABILITY TO USE THIS SOFTWARE,
+ * EVEN IF SUN HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
+ */
+
+#include <stdio.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <sys/ioctl.h>
+#include <errno.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+
+#include <ipmitool/ipmi.h>
+#include <ipmitool/ipmi_intf.h>
+#include <ipmitool/helper.h>
+#include <ipmitool/log.h>
+
+#if defined(HAVE_CONFIG_H)
+# include <config.h>
+#endif
+
+#if defined(HAVE_SYS_IOCCOM_H)
+# include <sys/ioccom.h>
+#endif
+
+#if defined(HAVE_OPENIPMI_H)
+# if defined(HAVE_LINUX_COMPILER_H)
+# include <linux/compiler.h>
+# endif
+# include <linux/ipmi.h>
+#elif defined(HAVE_FREEBSD_IPMI_H)
+/* FreeBSD OpenIPMI-compatible header */
+# include <sys/ipmi.h>
+#else
+# include "open.h"
+#endif
+
+/**
+ * Maximum input message size for KCS/SMIC is 40 with 2 utility bytes and
+ * 38 bytes of data.
+ * Maximum input message size for BT is 42 with 4 utility bytes and
+ * 38 bytes of data.
+ */
+#define IPMI_OPENIPMI_MAX_RQ_DATA_SIZE 38
+
+/**
+ * Maximum output message size for KCS/SMIC is 38 with 2 utility bytes, a byte
+ * for completion code and 35 bytes of data.
+ * Maximum output message size for BT is 40 with 4 utility bytes, a byte
+ * for completion code and 35 bytes of data.
+ */
+#define IPMI_OPENIPMI_MAX_RS_DATA_SIZE 35
+
+extern int verbose;
+
+static int
+ipmi_openipmi_open(struct ipmi_intf * intf)
+{
+ int i = 0;
+
+ char ipmi_dev[16];
+ char ipmi_devfs[16];
+ char ipmi_devfs2[16];
+ int devnum = 0;
+
+ devnum = intf->devnum;
+
+ sprintf(ipmi_dev, "/dev/ipmi%d", devnum);
+ sprintf(ipmi_devfs, "/dev/ipmi/%d", devnum);
+ sprintf(ipmi_devfs2, "/dev/ipmidev/%d", devnum);
+ lprintf(LOG_DEBUG, "Using ipmi device %d", devnum);
+
+ intf->fd = open(ipmi_dev, O_RDWR);
+
+ if (intf->fd < 0) {
+ intf->fd = open(ipmi_devfs, O_RDWR);
+ if (intf->fd < 0) {
+ intf->fd = open(ipmi_devfs2, O_RDWR);
+ }
+ if (intf->fd < 0) {
+ lperror(LOG_ERR, "Could not open device at %s or %s or %s",
+ ipmi_dev, ipmi_devfs , ipmi_devfs2);
+ return -1;
+ }
+ }
+
+ if (ioctl(intf->fd, IPMICTL_SET_GETS_EVENTS_CMD, &i) < 0) {
+ lperror(LOG_ERR, "Could not enable event receiver");
+ return -1;
+ }
+
+ intf->opened = 1;
+
+ /* This is never set to 0, the default is IPMI_BMC_SLAVE_ADDR */
+ if (intf->my_addr != 0) {
+ if (intf->set_my_addr(intf, intf->my_addr) < 0) {
+ lperror(LOG_ERR, "Could not set IPMB address");
+ return -1;
+ }
+ lprintf(LOG_DEBUG, "Set IPMB address to 0x%x",
+ intf->my_addr );
+ }
+
+ intf->manufacturer_id = ipmi_get_oem(intf);
+ return intf->fd;
+}
+static int
+ipmi_openipmi_set_my_addr(struct ipmi_intf *intf, uint8_t addr)
+{
+ unsigned int a = addr;
+ if (ioctl(intf->fd, IPMICTL_SET_MY_ADDRESS_CMD, &a) < 0) {
+ lperror(LOG_ERR, "Could not set IPMB address");
+ return -1;
+ }
+ intf->my_addr = addr;
+ return 0;
+}
+
+static void
+ipmi_openipmi_close(struct ipmi_intf * intf)
+{
+ if (intf->fd >= 0) {
+ close(intf->fd);
+ intf->fd = -1;
+ }
+
+ intf->opened = 0;
+ intf->manufacturer_id = IPMI_OEM_UNKNOWN;
+}
+
+static struct ipmi_rs *
+ipmi_openipmi_send_cmd(struct ipmi_intf * intf, struct ipmi_rq * req)
+{
+ struct ipmi_recv recv;
+ struct ipmi_addr addr;
+ struct ipmi_system_interface_addr bmc_addr = {
+ addr_type: IPMI_SYSTEM_INTERFACE_ADDR_TYPE,
+ channel: IPMI_BMC_CHANNEL,
+ };
+ struct ipmi_ipmb_addr ipmb_addr = {
+ addr_type: IPMI_IPMB_ADDR_TYPE,
+ };
+ struct ipmi_req _req;
+ static struct ipmi_rs rsp;
+ static int curr_seq = 0;
+ fd_set rset;
+
+ uint8_t * data = NULL;
+ int data_len = 0;
+
+
+ if (intf == NULL || req == NULL)
+ return NULL;
+
+ ipmb_addr.channel = intf->target_channel & 0x0f;
+
+ if (intf->opened == 0 && intf->open != NULL)
+ if (intf->open(intf) < 0)
+ return NULL;
+
+ if (verbose > 2)
+ printbuf(req->msg.data, req->msg.data_len,
+ "OpenIPMI Request Message");
+
+ /*
+ * setup and send message
+ */
+
+ memset(&_req, 0, sizeof(struct ipmi_req));
+
+ if (intf->target_addr != 0 &&
+ intf->target_addr != intf->my_addr) {
+ /* use IPMB address if needed */
+ ipmb_addr.slave_addr = intf->target_addr;
+ ipmb_addr.lun = req->msg.lun;
+ lprintf(LOG_DEBUG, "Sending request 0x%x to "
+ "IPMB target @ 0x%x:0x%x (from 0x%x)",
+ req->msg.cmd,
+ intf->target_addr,intf->target_channel, intf->my_addr);
+
+ if(intf->transit_addr != 0 && intf->transit_addr != intf->my_addr) {
+ uint8_t index = 0;
+
+ lprintf(LOG_DEBUG, "Encapsulating data sent to "
+ "end target [0x%02x,0x%02x] using transit [0x%02x,0x%02x] from 0x%x ",
+ (0x40 | intf->target_channel),
+ intf->target_addr,
+ intf->transit_channel,
+ intf->transit_addr,
+ intf->my_addr
+ );
+
+ /* Convert Message to 'Send Message' */
+ /* Supplied req : req , internal req : _req */
+
+ if (verbose > 4) {
+ fprintf(stderr, "Converting message:\n");
+ fprintf(stderr, " netfn = 0x%x\n", req->msg.netfn );
+ fprintf(stderr, " cmd = 0x%x\n", req->msg.cmd);
+ if (recv.msg.data && recv.msg.data_len) {
+ fprintf(stderr, " data_len = %d\n", req->msg.data_len);
+ fprintf(stderr, " data = %s\n",
+ buf2str(req->msg.data,req->msg.data_len));
+ }
+ }
+
+ /* Modify target address to use 'transit' instead */
+ ipmb_addr.slave_addr = intf->transit_addr;
+ ipmb_addr.channel = intf->transit_channel;
+
+ /* FIXME backup "My address" */
+ data_len = req->msg.data_len + 8;
+ data = malloc(data_len);
+ if (data == NULL) {
+ lprintf(LOG_ERR, "ipmitool: malloc failure");
+ return NULL;
+ }
+
+ memset(data, 0, data_len);
+
+ data[index++] = (0x40|intf->target_channel);
+ data[index++] = intf->target_addr;
+ data[index++] = ( req->msg.netfn << 2 ) | req->msg.lun ;
+ data[index++] = ipmi_csum(data+1, 2);
+ data[index++] = 0xFF; /* normally 0x20 , overwritten by IPMC */
+ data[index++] = ( (0) << 2) | 0 ; /* FIXME */
+ data[index++] = req->msg.cmd;
+ memcpy( (data+index) , req->msg.data, req->msg.data_len);
+ index += req->msg.data_len;
+ data[index++] = ipmi_csum( (data+4),(req->msg.data_len + 3) );
+
+ if (verbose > 4) {
+ fprintf(stderr, "Encapsulated message:\n");
+ fprintf(stderr, " netfn = 0x%x\n", IPMI_NETFN_APP );
+ fprintf(stderr, " cmd = 0x%x\n", 0x34 );
+ if (data && data_len) {
+ fprintf(stderr, " data_len = %d\n", data_len);
+ fprintf(stderr, " data = %s\n",
+ buf2str(data,data_len));
+ }
+ }
+ }
+ _req.addr = (unsigned char *) &ipmb_addr;
+ _req.addr_len = sizeof(ipmb_addr);
+ } else {
+ /* otherwise use system interface */
+ lprintf(LOG_DEBUG+2, "Sending request 0x%x to "
+ "System Interface", req->msg.cmd);
+ bmc_addr.lun = req->msg.lun;
+ _req.addr = (unsigned char *) &bmc_addr;
+ _req.addr_len = sizeof(bmc_addr);
+ }
+
+ _req.msgid = curr_seq++;
+
+ /* In case of a bridge request */
+ if( data != NULL && data_len != 0 ) {
+ _req.msg.data = data;
+ _req.msg.data_len = data_len;
+ _req.msg.netfn = IPMI_NETFN_APP;
+ _req.msg.cmd = 0x34;
+
+ } else {
+ _req.msg.data = req->msg.data;
+ _req.msg.data_len = req->msg.data_len;
+ _req.msg.netfn = req->msg.netfn;
+ _req.msg.cmd = req->msg.cmd;
+ }
+
+ if (ioctl(intf->fd, IPMICTL_SEND_COMMAND, &_req) < 0) {
+ lperror(LOG_ERR, "Unable to send command");
+ if (data != NULL) {
+ free(data);
+ data = NULL;
+ }
+ return NULL;
+ }
+
+ /*
+ * wait for and retrieve response
+ */
+
+ if (intf->noanswer) {
+ if (data != NULL) {
+ free(data);
+ data = NULL;
+ }
+ return NULL;
+ }
+
+ FD_ZERO(&rset);
+ FD_SET(intf->fd, &rset);
+
+ if (select(intf->fd+1, &rset, NULL, NULL, NULL) < 0) {
+ lperror(LOG_ERR, "I/O Error");
+ if (data != NULL) {
+ free(data);
+ data = NULL;
+ }
+ return NULL;
+ }
+ if (FD_ISSET(intf->fd, &rset) == 0) {
+ lprintf(LOG_ERR, "No data available");
+ if (data != NULL) {
+ free(data);
+ data = NULL;
+ }
+ return NULL;
+ }
+
+ recv.addr = (unsigned char *) &addr;
+ recv.addr_len = sizeof(addr);
+ recv.msg.data = rsp.data;
+ recv.msg.data_len = sizeof(rsp.data);
+
+ /* get data */
+ if (ioctl(intf->fd, IPMICTL_RECEIVE_MSG_TRUNC, &recv) < 0) {
+ lperror(LOG_ERR, "Error receiving message");
+ if (errno != EMSGSIZE) {
+ if (data != NULL) {
+ free(data);
+ data = NULL;
+ }
+ return NULL;
+ }
+ }
+
+ if (verbose > 4) {
+ fprintf(stderr, "Got message:");
+ fprintf(stderr, " type = %d\n", recv.recv_type);
+ fprintf(stderr, " channel = 0x%x\n", addr.channel);
+ fprintf(stderr, " msgid = %ld\n", recv.msgid);
+ fprintf(stderr, " netfn = 0x%x\n", recv.msg.netfn);
+ fprintf(stderr, " cmd = 0x%x\n", recv.msg.cmd);
+ if (recv.msg.data && recv.msg.data_len) {
+ fprintf(stderr, " data_len = %d\n", recv.msg.data_len);
+ fprintf(stderr, " data = %s\n",
+ buf2str(recv.msg.data, recv.msg.data_len));
+ }
+ }
+
+ if(intf->transit_addr != 0 && intf->transit_addr != intf->my_addr) {
+ uint8_t index = 0;
+
+ /* ipmb_addr.transit_slave_addr = intf->transit_addr; */
+ lprintf(LOG_DEBUG, "Decapsulating data received from transit "
+ "IPMB target @ 0x%x", intf->transit_addr);
+
+ /* comp code */
+ /* Check data */
+
+ if( recv.msg.data[0] == 0 ) {
+ recv.msg.netfn = recv.msg.data[2] >> 2;
+ recv.msg.cmd = recv.msg.data[6];
+
+ recv.msg.data = memmove(recv.msg.data ,recv.msg.data+7 , recv.msg.data_len - 7);
+ recv.msg.data_len -=8;
+
+ if (verbose > 4) {
+ fprintf(stderr, "Decapsulated message:\n");
+ fprintf(stderr, " netfn = 0x%x\n", recv.msg.netfn );
+ fprintf(stderr, " cmd = 0x%x\n", recv.msg.cmd);
+ if (recv.msg.data && recv.msg.data_len) {
+ fprintf(stderr, " data_len = %d\n", recv.msg.data_len);
+ fprintf(stderr, " data = %s\n",
+ buf2str(recv.msg.data,recv.msg.data_len));
+ }
+ }
+ }
+ }
+
+ /* save completion code */
+ rsp.ccode = recv.msg.data[0];
+ rsp.data_len = recv.msg.data_len - 1;
+
+ /* save response data for caller */
+ if (rsp.ccode == 0 && rsp.data_len > 0) {
+ memmove(rsp.data, rsp.data + 1, rsp.data_len);
+ rsp.data[rsp.data_len] = 0;
+ }
+
+ if (data != NULL) {
+ free(data);
+ data = NULL;
+ }
+
+ return &rsp;
+}
+
+int ipmi_openipmi_setup(struct ipmi_intf * intf)
+{
+ /* set default payload size */
+ intf->max_request_data_size = IPMI_OPENIPMI_MAX_RQ_DATA_SIZE;
+ intf->max_response_data_size = IPMI_OPENIPMI_MAX_RS_DATA_SIZE;
+
+ return 0;
+}
+
+struct ipmi_intf ipmi_open_intf = {
+ name: "open",
+ desc: "Linux OpenIPMI Interface",
+ setup: ipmi_openipmi_setup,
+ open: ipmi_openipmi_open,
+ close: ipmi_openipmi_close,
+ sendrecv: ipmi_openipmi_send_cmd,
+ set_my_addr: ipmi_openipmi_set_my_addr,
+ my_addr: IPMI_BMC_SLAVE_ADDR,
+ target_addr: 0, /* init so -m local_addr does not cause bridging */
+};