linux/drivers/scsi/libfc/fc_lport.c

1839 lines
48 KiB
C
Raw Normal View History

/*
* Copyright(c) 2007 Intel Corporation. All rights reserved.
*
* 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.
*
* Maintained at www.Open-FCoE.org
*/
/*
* PORT LOCKING NOTES
*
* These comments only apply to the 'port code' which consists of the lport,
* disc and rport blocks.
*
* MOTIVATION
*
* The lport, disc and rport blocks all have mutexes that are used to protect
* those objects. The main motivation for these locks is to prevent from
* having an lport reset just before we send a frame. In that scenario the
* lport's FID would get set to zero and then we'd send a frame with an
* invalid SID. We also need to ensure that states don't change unexpectedly
* while processing another state.
*
* HEIRARCHY
*
* The following heirarchy defines the locking rules. A greater lock
* may be held before acquiring a lesser lock, but a lesser lock should never
* be held while attempting to acquire a greater lock. Here is the heirarchy-
*
* lport > disc, lport > rport, disc > rport
*
* CALLBACKS
*
* The callbacks cause complications with this scheme. There is a callback
* from the rport (to either lport or disc) and a callback from disc
* (to the lport).
*
* As rports exit the rport state machine a callback is made to the owner of
* the rport to notify success or failure. Since the callback is likely to
* cause the lport or disc to grab its lock we cannot hold the rport lock
* while making the callback. To ensure that the rport is not free'd while
* processing the callback the rport callbacks are serialized through a
* single-threaded workqueue. An rport would never be free'd while in a
* callback handler becuase no other rport work in this queue can be executed
* at the same time.
*
* When discovery succeeds or fails a callback is made to the lport as
* notification. Currently, successful discovery causes the lport to take no
* action. A failure will cause the lport to reset. There is likely a circular
* locking problem with this implementation.
*/
/*
* LPORT LOCKING
*
* The critical sections protected by the lport's mutex are quite broad and
* may be improved upon in the future. The lport code and its locking doesn't
* influence the I/O path, so excessive locking doesn't penalize I/O
* performance.
*
* The strategy is to lock whenever processing a request or response. Note
* that every _enter_* function corresponds to a state change. They generally
* change the lports state and then send a request out on the wire. We lock
* before calling any of these functions to protect that state change. This
* means that the entry points into the lport block manage the locks while
* the state machine can transition between states (i.e. _enter_* functions)
* while always staying protected.
*
* When handling responses we also hold the lport mutex broadly. When the
* lport receives the response frame it locks the mutex and then calls the
* appropriate handler for the particuar response. Generally a response will
* trigger a state change and so the lock must already be held.
*
* Retries also have to consider the locking. The retries occur from a work
* context and the work function will lock the lport and then retry the state
* (i.e. _enter_* function).
*/
#include <linux/timer.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <asm/unaligned.h>
#include <scsi/fc/fc_gs.h>
#include <scsi/libfc.h>
#include <scsi/fc_encode.h>
#include <linux/scatterlist.h>
#include "fc_libfc.h"
/* Fabric IDs to use for point-to-point mode, chosen on whims. */
#define FC_LOCAL_PTP_FID_LO 0x010101
#define FC_LOCAL_PTP_FID_HI 0x010102
#define DNS_DELAY 3 /* Discovery delay after RSCN (in seconds)*/
static void fc_lport_error(struct fc_lport *, struct fc_frame *);
static void fc_lport_enter_reset(struct fc_lport *);
static void fc_lport_enter_flogi(struct fc_lport *);
static void fc_lport_enter_dns(struct fc_lport *);
static void fc_lport_enter_ns(struct fc_lport *, enum fc_lport_state);
static void fc_lport_enter_scr(struct fc_lport *);
static void fc_lport_enter_ready(struct fc_lport *);
static void fc_lport_enter_logo(struct fc_lport *);
static const char *fc_lport_state_names[] = {
[LPORT_ST_DISABLED] = "disabled",
[LPORT_ST_FLOGI] = "FLOGI",
[LPORT_ST_DNS] = "dNS",
[LPORT_ST_RNN_ID] = "RNN_ID",
[LPORT_ST_RSNN_NN] = "RSNN_NN",
[LPORT_ST_RSPN_ID] = "RSPN_ID",
[LPORT_ST_RFT_ID] = "RFT_ID",
[LPORT_ST_RFF_ID] = "RFF_ID",
[LPORT_ST_SCR] = "SCR",
[LPORT_ST_READY] = "Ready",
[LPORT_ST_LOGO] = "LOGO",
[LPORT_ST_RESET] = "reset",
};
/**
* struct fc_bsg_info - FC Passthrough managemet structure
* @job: The passthrough job
* @lport: The local port to pass through a command
* @rsp_code: The expected response code
* @sg: job->reply_payload.sg_list
* @nents: job->reply_payload.sg_cnt
* @offset: The offset into the response data
*/
struct fc_bsg_info {
struct fc_bsg_job *job;
struct fc_lport *lport;
u16 rsp_code;
struct scatterlist *sg;
u32 nents;
size_t offset;
};
/**
* fc_frame_drop() - Dummy frame handler
* @lport: The local port the frame was received on
* @fp: The received frame
*/
static int fc_frame_drop(struct fc_lport *lport, struct fc_frame *fp)
{
fc_frame_free(fp);
return 0;
}
/**
* fc_lport_rport_callback() - Event handler for rport events
* @lport: The lport which is receiving the event
* @rdata: private remote port data
* @event: The event that occured
*
* Locking Note: The rport lock should not be held when calling
* this function.
*/
static void fc_lport_rport_callback(struct fc_lport *lport,
struct fc_rport_priv *rdata,
enum fc_rport_event event)
{
FC_LPORT_DBG(lport, "Received a %d event for port (%6.6x)\n", event,
rdata->ids.port_id);
mutex_lock(&lport->lp_mutex);
switch (event) {
case RPORT_EV_READY:
if (lport->state == LPORT_ST_DNS) {
lport->dns_rdata = rdata;
fc_lport_enter_ns(lport, LPORT_ST_RNN_ID);
} else {
FC_LPORT_DBG(lport, "Received an READY event "
"on port (%6.6x) for the directory "
"server, but the lport is not "
"in the DNS state, it's in the "
"%d state", rdata->ids.port_id,
lport->state);
lport->tt.rport_logoff(rdata);
}
break;
case RPORT_EV_LOGO:
case RPORT_EV_FAILED:
case RPORT_EV_STOP:
lport->dns_rdata = NULL;
break;
case RPORT_EV_NONE:
break;
}
mutex_unlock(&lport->lp_mutex);
}
/**
* fc_lport_state() - Return a string which represents the lport's state
* @lport: The lport whose state is to converted to a string
*/
static const char *fc_lport_state(struct fc_lport *lport)
{
const char *cp;
cp = fc_lport_state_names[lport->state];
if (!cp)
cp = "unknown";
return cp;
}
/**
* fc_lport_ptp_setup() - Create an rport for point-to-point mode
* @lport: The lport to attach the ptp rport to
* @remote_fid: The FID of the ptp rport
* @remote_wwpn: The WWPN of the ptp rport
* @remote_wwnn: The WWNN of the ptp rport
*/
static void fc_lport_ptp_setup(struct fc_lport *lport,
u32 remote_fid, u64 remote_wwpn,
u64 remote_wwnn)
{
mutex_lock(&lport->disc.disc_mutex);
if (lport->ptp_rdata) {
lport->tt.rport_logoff(lport->ptp_rdata);
kref_put(&lport->ptp_rdata->kref, lport->tt.rport_destroy);
}
lport->ptp_rdata = lport->tt.rport_create(lport, remote_fid);
kref_get(&lport->ptp_rdata->kref);
lport->ptp_rdata->ids.port_name = remote_wwpn;
lport->ptp_rdata->ids.node_name = remote_wwnn;
mutex_unlock(&lport->disc.disc_mutex);
lport->tt.rport_login(lport->ptp_rdata);
fc_lport_enter_ready(lport);
}
/**
* fc_get_host_port_state() - Return the port state of the given Scsi_Host
* @shost: The SCSI host whose port state is to be determined
*/
void fc_get_host_port_state(struct Scsi_Host *shost)
{
struct fc_lport *lport = shost_priv(shost);
mutex_lock(&lport->lp_mutex);
if (!lport->link_up)
fc_host_port_state(shost) = FC_PORTSTATE_LINKDOWN;
else
switch (lport->state) {
case LPORT_ST_READY:
fc_host_port_state(shost) = FC_PORTSTATE_ONLINE;
break;
default:
fc_host_port_state(shost) = FC_PORTSTATE_OFFLINE;
}
mutex_unlock(&lport->lp_mutex);
}
EXPORT_SYMBOL(fc_get_host_port_state);
/**
* fc_get_host_speed() - Return the speed of the given Scsi_Host
* @shost: The SCSI host whose port speed is to be determined
*/
void fc_get_host_speed(struct Scsi_Host *shost)
{
struct fc_lport *lport = shost_priv(shost);
fc_host_speed(shost) = lport->link_speed;
}
EXPORT_SYMBOL(fc_get_host_speed);
/**
* fc_get_host_stats() - Return the Scsi_Host's statistics
* @shost: The SCSI host whose statistics are to be returned
*/
struct fc_host_statistics *fc_get_host_stats(struct Scsi_Host *shost)
{
struct fc_host_statistics *fcoe_stats;
struct fc_lport *lport = shost_priv(shost);
struct timespec v0, v1;
unsigned int cpu;
fcoe_stats = &lport->host_stats;
memset(fcoe_stats, 0, sizeof(struct fc_host_statistics));
jiffies_to_timespec(jiffies, &v0);
jiffies_to_timespec(lport->boot_time, &v1);
fcoe_stats->seconds_since_last_reset = (v0.tv_sec - v1.tv_sec);
for_each_possible_cpu(cpu) {
struct fcoe_dev_stats *stats;
stats = per_cpu_ptr(lport->dev_stats, cpu);
fcoe_stats->tx_frames += stats->TxFrames;
fcoe_stats->tx_words += stats->TxWords;
fcoe_stats->rx_frames += stats->RxFrames;
fcoe_stats->rx_words += stats->RxWords;
fcoe_stats->error_frames += stats->ErrorFrames;
fcoe_stats->invalid_crc_count += stats->InvalidCRCCount;
fcoe_stats->fcp_input_requests += stats->InputRequests;
fcoe_stats->fcp_output_requests += stats->OutputRequests;
fcoe_stats->fcp_control_requests += stats->ControlRequests;
fcoe_stats->fcp_input_megabytes += stats->InputMegabytes;
fcoe_stats->fcp_output_megabytes += stats->OutputMegabytes;
fcoe_stats->link_failure_count += stats->LinkFailureCount;
}
fcoe_stats->lip_count = -1;
fcoe_stats->nos_count = -1;
fcoe_stats->loss_of_sync_count = -1;
fcoe_stats->loss_of_signal_count = -1;
fcoe_stats->prim_seq_protocol_err_count = -1;
fcoe_stats->dumped_frames = -1;
return fcoe_stats;
}
EXPORT_SYMBOL(fc_get_host_stats);
/**
* fc_lport_flogi_fill() - Fill in FLOGI command for request
* @lport: The local port the FLOGI is for
* @flogi: The FLOGI command
* @op: The opcode
*/
static void fc_lport_flogi_fill(struct fc_lport *lport,
struct fc_els_flogi *flogi,
unsigned int op)
{
struct fc_els_csp *sp;
struct fc_els_cssp *cp;
memset(flogi, 0, sizeof(*flogi));
flogi->fl_cmd = (u8) op;
put_unaligned_be64(lport->wwpn, &flogi->fl_wwpn);
put_unaligned_be64(lport->wwnn, &flogi->fl_wwnn);
sp = &flogi->fl_csp;
sp->sp_hi_ver = 0x20;
sp->sp_lo_ver = 0x20;
sp->sp_bb_cred = htons(10); /* this gets set by gateway */
sp->sp_bb_data = htons((u16) lport->mfs);
cp = &flogi->fl_cssp[3 - 1]; /* class 3 parameters */
cp->cp_class = htons(FC_CPC_VALID | FC_CPC_SEQ);
if (op != ELS_FLOGI) {
sp->sp_features = htons(FC_SP_FT_CIRO);
sp->sp_tot_seq = htons(255); /* seq. we accept */
sp->sp_rel_off = htons(0x1f);
sp->sp_e_d_tov = htonl(lport->e_d_tov);
cp->cp_rdfs = htons((u16) lport->mfs);
cp->cp_con_seq = htons(255);
cp->cp_open_seq = 1;
}
}
/**
* fc_lport_add_fc4_type() - Add a supported FC-4 type to a local port
* @lport: The local port to add a new FC-4 type to
* @type: The new FC-4 type
*/
static void fc_lport_add_fc4_type(struct fc_lport *lport, enum fc_fh_type type)
{
__be32 *mp;
mp = &lport->fcts.ff_type_map[type / FC_NS_BPW];
*mp = htonl(ntohl(*mp) | 1UL << (type % FC_NS_BPW));
}
/**
* fc_lport_recv_rlir_req() - Handle received Registered Link Incident Report.
* @lport: Fibre Channel local port recieving the RLIR
* @fp: The RLIR request frame
*
* Locking Note: The lport lock is expected to be held before calling
* this function.
*/
static void fc_lport_recv_rlir_req(struct fc_lport *lport, struct fc_frame *fp)
{
FC_LPORT_DBG(lport, "Received RLIR request while in state %s\n",
fc_lport_state(lport));
lport->tt.seq_els_rsp_send(fp, ELS_LS_ACC, NULL);
fc_frame_free(fp);
}
/**
* fc_lport_recv_echo_req() - Handle received ECHO request
* @lport: The local port recieving the ECHO
* @fp: ECHO request frame
*
* Locking Note: The lport lock is expected to be held before calling
* this function.
*/
static void fc_lport_recv_echo_req(struct fc_lport *lport,
struct fc_frame *in_fp)
{
struct fc_frame *fp;
unsigned int len;
void *pp;
void *dp;
FC_LPORT_DBG(lport, "Received ECHO request while in state %s\n",
fc_lport_state(lport));
len = fr_len(in_fp) - sizeof(struct fc_frame_header);
pp = fc_frame_payload_get(in_fp, len);
if (len < sizeof(__be32))
len = sizeof(__be32);
fp = fc_frame_alloc(lport, len);
if (fp) {
dp = fc_frame_payload_get(fp, len);
memcpy(dp, pp, len);
*((__be32 *)dp) = htonl(ELS_LS_ACC << 24);
fc_fill_reply_hdr(fp, in_fp, FC_RCTL_ELS_REP, 0);
lport->tt.frame_send(lport, fp);
}
fc_frame_free(in_fp);
}
/**
* fc_lport_recv_rnid_req() - Handle received Request Node ID data request
* @lport: The local port recieving the RNID
* @fp: The RNID request frame
*
* Locking Note: The lport lock is expected to be held before calling
* this function.
*/
static void fc_lport_recv_rnid_req(struct fc_lport *lport,
struct fc_frame *in_fp)
{
struct fc_frame *fp;
struct fc_els_rnid *req;
struct {
struct fc_els_rnid_resp rnid;
struct fc_els_rnid_cid cid;
struct fc_els_rnid_gen gen;
} *rp;
struct fc_seq_els_data rjt_data;
u8 fmt;
size_t len;
FC_LPORT_DBG(lport, "Received RNID request while in state %s\n",
fc_lport_state(lport));
req = fc_frame_payload_get(in_fp, sizeof(*req));
if (!req) {
rjt_data.reason = ELS_RJT_LOGIC;
rjt_data.explan = ELS_EXPL_NONE;
lport->tt.seq_els_rsp_send(in_fp, ELS_LS_RJT, &rjt_data);
} else {
fmt = req->rnid_fmt;
len = sizeof(*rp);
if (fmt != ELS_RNIDF_GEN ||
ntohl(lport->rnid_gen.rnid_atype) == 0) {
fmt = ELS_RNIDF_NONE; /* nothing to provide */
len -= sizeof(rp->gen);
}
fp = fc_frame_alloc(lport, len);
if (fp) {
rp = fc_frame_payload_get(fp, len);
memset(rp, 0, len);
rp->rnid.rnid_cmd = ELS_LS_ACC;
rp->rnid.rnid_fmt = fmt;
rp->rnid.rnid_cid_len = sizeof(rp->cid);
rp->cid.rnid_wwpn = htonll(lport->wwpn);
rp->cid.rnid_wwnn = htonll(lport->wwnn);
if (fmt == ELS_RNIDF_GEN) {
rp->rnid.rnid_sid_len = sizeof(rp->gen);
memcpy(&rp->gen, &lport->rnid_gen,
sizeof(rp->gen));
}
fc_fill_reply_hdr(fp, in_fp, FC_RCTL_ELS_REP, 0);
lport->tt.frame_send(lport, fp);
}
}
fc_frame_free(in_fp);
}
/**
* fc_lport_recv_logo_req() - Handle received fabric LOGO request
* @lport: The local port recieving the LOGO
* @fp: The LOGO request frame
*
* Locking Note: The lport lock is exected to be held before calling
* this function.
*/
static void fc_lport_recv_logo_req(struct fc_lport *lport, struct fc_frame *fp)
{
lport->tt.seq_els_rsp_send(fp, ELS_LS_ACC, NULL);
fc_lport_enter_reset(lport);
fc_frame_free(fp);
}
/**
* fc_fabric_login() - Start the lport state machine
* @lport: The local port that should log into the fabric
*
* Locking Note: This function should not be called
* with the lport lock held.
*/
int fc_fabric_login(struct fc_lport *lport)
{
int rc = -1;
mutex_lock(&lport->lp_mutex);
if (lport->state == LPORT_ST_DISABLED ||
lport->state == LPORT_ST_LOGO) {
fc_lport_state_enter(lport, LPORT_ST_RESET);
fc_lport_enter_reset(lport);
rc = 0;
}
mutex_unlock(&lport->lp_mutex);
return rc;
}
EXPORT_SYMBOL(fc_fabric_login);
/**
* __fc_linkup() - Handler for transport linkup events
* @lport: The lport whose link is up
*
* Locking: must be called with the lp_mutex held
*/
void __fc_linkup(struct fc_lport *lport)
{
[SCSI] libfc, fcoe: fixed locking issues with lport->lp_mutex around lport->link_status The fcoe_xmit could call fc_pause in case the pending skb queue len is larger than FCOE_MAX_QUEUE_DEPTH, the fc_pause was trying to grab lport->lp_muex to change lport->link_status and that had these issues :- 1. The fcoe_xmit was getting called with bh disabled, thus causing "BUG: scheduling while atomic" when grabbing lport->lp_muex with bh disabled. 2. fc_linkup and fc_linkdown function calls lport_enter function with lport->lp_mutex held and these enter function in turn calls fcoe_xmit to send lport related FC frame, e.g. fc_linkup => fc_lport_enter_flogi to send flogi req. In this case grabbing the same lport->lp_mutex again in fc_puase from fcoe_xmit would cause deadlock. The lport->lp_mutex was used for setting FC_PAUSE in fcoe_xmit path but FC_PAUSE bit was not used anywhere beside just setting and clear this bit in lport->link_status, instead used a separate field qfull in fc_lport to eliminate need for lport->lp_mutex to track pending queue full condition and in turn avoid above described two locking issues. Also added check for lp->qfull in fc_fcp_lport_queue_ready to trigger SCSI_MLQUEUE_HOST_BUSY when lp->qfull is set to prevent more scsi-ml cmds while lp->qfull is set. This patch eliminated FC_LINK_UP and FC_PAUSE and instead used dedicated fields in fc_lport for this, this simplified all related conditional code. Also removed fc_pause and fc_unpause functions and instead used newly added lport->qfull directly in fcoe. Signed-off-by: Vasu Dev <vasu.dev@intel.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2009-02-27 18:54:57 +00:00
if (!lport->link_up) {
lport->link_up = 1;
if (lport->state == LPORT_ST_RESET)
fc_lport_enter_flogi(lport);
}
}
/**
* fc_linkup() - Handler for transport linkup events
* @lport: The local port whose link is up
*/
void fc_linkup(struct fc_lport *lport)
{
printk(KERN_INFO "host%d: libfc: Link up on port (%6.6x)\n",
lport->host->host_no, lport->port_id);
mutex_lock(&lport->lp_mutex);
__fc_linkup(lport);
mutex_unlock(&lport->lp_mutex);
}
EXPORT_SYMBOL(fc_linkup);
/**
* __fc_linkdown() - Handler for transport linkdown events
* @lport: The lport whose link is down
*
* Locking: must be called with the lp_mutex held
*/
void __fc_linkdown(struct fc_lport *lport)
{
[SCSI] libfc, fcoe: fixed locking issues with lport->lp_mutex around lport->link_status The fcoe_xmit could call fc_pause in case the pending skb queue len is larger than FCOE_MAX_QUEUE_DEPTH, the fc_pause was trying to grab lport->lp_muex to change lport->link_status and that had these issues :- 1. The fcoe_xmit was getting called with bh disabled, thus causing "BUG: scheduling while atomic" when grabbing lport->lp_muex with bh disabled. 2. fc_linkup and fc_linkdown function calls lport_enter function with lport->lp_mutex held and these enter function in turn calls fcoe_xmit to send lport related FC frame, e.g. fc_linkup => fc_lport_enter_flogi to send flogi req. In this case grabbing the same lport->lp_mutex again in fc_puase from fcoe_xmit would cause deadlock. The lport->lp_mutex was used for setting FC_PAUSE in fcoe_xmit path but FC_PAUSE bit was not used anywhere beside just setting and clear this bit in lport->link_status, instead used a separate field qfull in fc_lport to eliminate need for lport->lp_mutex to track pending queue full condition and in turn avoid above described two locking issues. Also added check for lp->qfull in fc_fcp_lport_queue_ready to trigger SCSI_MLQUEUE_HOST_BUSY when lp->qfull is set to prevent more scsi-ml cmds while lp->qfull is set. This patch eliminated FC_LINK_UP and FC_PAUSE and instead used dedicated fields in fc_lport for this, this simplified all related conditional code. Also removed fc_pause and fc_unpause functions and instead used newly added lport->qfull directly in fcoe. Signed-off-by: Vasu Dev <vasu.dev@intel.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2009-02-27 18:54:57 +00:00
if (lport->link_up) {
lport->link_up = 0;
fc_lport_enter_reset(lport);
lport->tt.fcp_cleanup(lport);
}
}
/**
* fc_linkdown() - Handler for transport linkdown events
* @lport: The local port whose link is down
*/
void fc_linkdown(struct fc_lport *lport)
{
printk(KERN_INFO "host%d: libfc: Link down on port (%6.6x)\n",
lport->host->host_no, lport->port_id);
mutex_lock(&lport->lp_mutex);
__fc_linkdown(lport);
mutex_unlock(&lport->lp_mutex);
}
EXPORT_SYMBOL(fc_linkdown);
/**
* fc_fabric_logoff() - Logout of the fabric
* @lport: The local port to logoff the fabric
*
* Return value:
* 0 for success, -1 for failure
*/
int fc_fabric_logoff(struct fc_lport *lport)
{
lport->tt.disc_stop_final(lport);
mutex_lock(&lport->lp_mutex);
if (lport->dns_rdata)
lport->tt.rport_logoff(lport->dns_rdata);
mutex_unlock(&lport->lp_mutex);
lport->tt.rport_flush_queue();
mutex_lock(&lport->lp_mutex);
fc_lport_enter_logo(lport);
mutex_unlock(&lport->lp_mutex);
cancel_delayed_work_sync(&lport->retry_work);
return 0;
}
EXPORT_SYMBOL(fc_fabric_logoff);
/**
* fc_lport_destroy() - Unregister a fc_lport
* @lport: The local port to unregister
*
* Note:
* exit routine for fc_lport instance
* clean-up all the allocated memory
* and free up other system resources.
*
*/
int fc_lport_destroy(struct fc_lport *lport)
{
mutex_lock(&lport->lp_mutex);
lport->state = LPORT_ST_DISABLED;
lport->link_up = 0;
lport->tt.frame_send = fc_frame_drop;
mutex_unlock(&lport->lp_mutex);
lport->tt.fcp_abort_io(lport);
lport->tt.disc_stop_final(lport);
lport->tt.exch_mgr_reset(lport, 0, 0);
return 0;
}
EXPORT_SYMBOL(fc_lport_destroy);
/**
* fc_set_mfs() - Set the maximum frame size for a local port
* @lport: The local port to set the MFS for
* @mfs: The new MFS
*/
int fc_set_mfs(struct fc_lport *lport, u32 mfs)
{
unsigned int old_mfs;
int rc = -EINVAL;
mutex_lock(&lport->lp_mutex);
old_mfs = lport->mfs;
if (mfs >= FC_MIN_MAX_FRAME) {
mfs &= ~3;
if (mfs > FC_MAX_FRAME)
mfs = FC_MAX_FRAME;
mfs -= sizeof(struct fc_frame_header);
lport->mfs = mfs;
rc = 0;
}
if (!rc && mfs < old_mfs)
fc_lport_enter_reset(lport);
mutex_unlock(&lport->lp_mutex);
return rc;
}
EXPORT_SYMBOL(fc_set_mfs);
/**
* fc_lport_disc_callback() - Callback for discovery events
* @lport: The local port receiving the event
* @event: The discovery event
*/
void fc_lport_disc_callback(struct fc_lport *lport, enum fc_disc_event event)
{
switch (event) {
case DISC_EV_SUCCESS:
FC_LPORT_DBG(lport, "Discovery succeeded\n");
break;
case DISC_EV_FAILED:
printk(KERN_ERR "host%d: libfc: "
"Discovery failed for port (%6.6x)\n",
lport->host->host_no, lport->port_id);
mutex_lock(&lport->lp_mutex);
fc_lport_enter_reset(lport);
mutex_unlock(&lport->lp_mutex);
break;
case DISC_EV_NONE:
WARN_ON(1);
break;
}
}
/**
* fc_rport_enter_ready() - Enter the ready state and start discovery
* @lport: The local port that is ready
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static void fc_lport_enter_ready(struct fc_lport *lport)
{
FC_LPORT_DBG(lport, "Entered READY from state %s\n",
fc_lport_state(lport));
fc_lport_state_enter(lport, LPORT_ST_READY);
if (lport->vport)
fc_vport_set_state(lport->vport, FC_VPORT_ACTIVE);
fc_vports_linkchange(lport);
if (!lport->ptp_rdata)
lport->tt.disc_start(fc_lport_disc_callback, lport);
}
/**
* fc_lport_set_port_id() - set the local port Port ID
* @lport: The local port which will have its Port ID set.
* @port_id: The new port ID.
* @fp: The frame containing the incoming request, or NULL.
*
* Locking Note: The lport lock is expected to be held before calling
* this function.
*/
static void fc_lport_set_port_id(struct fc_lport *lport, u32 port_id,
struct fc_frame *fp)
{
if (port_id)
printk(KERN_INFO "host%d: Assigned Port ID %6.6x\n",
lport->host->host_no, port_id);
lport->port_id = port_id;
/* Update the fc_host */
fc_host_port_id(lport->host) = port_id;
if (lport->tt.lport_set_port_id)
lport->tt.lport_set_port_id(lport, port_id, fp);
}
/**
* fc_lport_set_port_id() - set the local port Port ID for point-to-multipoint
* @lport: The local port which will have its Port ID set.
* @port_id: The new port ID.
*
* Called by the lower-level driver when transport sets the local port_id.
* This is used in VN_port to VN_port mode for FCoE, and causes FLOGI and
* discovery to be skipped.
*/
void fc_lport_set_local_id(struct fc_lport *lport, u32 port_id)
{
mutex_lock(&lport->lp_mutex);
fc_lport_set_port_id(lport, port_id, NULL);
switch (lport->state) {
case LPORT_ST_RESET:
case LPORT_ST_FLOGI:
if (port_id)
fc_lport_enter_ready(lport);
break;
default:
break;
}
mutex_unlock(&lport->lp_mutex);
}
EXPORT_SYMBOL(fc_lport_set_local_id);
/**
* fc_lport_recv_flogi_req() - Receive a FLOGI request
* @lport: The local port that recieved the request
* @rx_fp: The FLOGI frame
*
* A received FLOGI request indicates a point-to-point connection.
* Accept it with the common service parameters indicating our N port.
* Set up to do a PLOGI if we have the higher-number WWPN.
*
* Locking Note: The lport lock is expected to be held before calling
* this function.
*/
static void fc_lport_recv_flogi_req(struct fc_lport *lport,
struct fc_frame *rx_fp)
{
struct fc_frame *fp;
struct fc_frame_header *fh;
struct fc_els_flogi *flp;
struct fc_els_flogi *new_flp;
u64 remote_wwpn;
u32 remote_fid;
u32 local_fid;
FC_LPORT_DBG(lport, "Received FLOGI request while in state %s\n",
fc_lport_state(lport));
remote_fid = fc_frame_sid(rx_fp);
flp = fc_frame_payload_get(rx_fp, sizeof(*flp));
if (!flp)
goto out;
remote_wwpn = get_unaligned_be64(&flp->fl_wwpn);
if (remote_wwpn == lport->wwpn) {
printk(KERN_WARNING "host%d: libfc: Received FLOGI from port "
"with same WWPN %16.16llx\n",
lport->host->host_no, remote_wwpn);
goto out;
}
FC_LPORT_DBG(lport, "FLOGI from port WWPN %16.16llx\n", remote_wwpn);
/*
* XXX what is the right thing to do for FIDs?
* The originator might expect our S_ID to be 0xfffffe.
* But if so, both of us could end up with the same FID.
*/
local_fid = FC_LOCAL_PTP_FID_LO;
if (remote_wwpn < lport->wwpn) {
local_fid = FC_LOCAL_PTP_FID_HI;
if (!remote_fid || remote_fid == local_fid)
remote_fid = FC_LOCAL_PTP_FID_LO;
} else if (!remote_fid) {
remote_fid = FC_LOCAL_PTP_FID_HI;
}
fc_lport_set_port_id(lport, local_fid, rx_fp);
fp = fc_frame_alloc(lport, sizeof(*flp));
if (fp) {
new_flp = fc_frame_payload_get(fp, sizeof(*flp));
fc_lport_flogi_fill(lport, new_flp, ELS_FLOGI);
new_flp->fl_cmd = (u8) ELS_LS_ACC;
/*
* Send the response. If this fails, the originator should
* repeat the sequence.
*/
fc_fill_reply_hdr(fp, rx_fp, FC_RCTL_ELS_REP, 0);
fh = fc_frame_header_get(fp);
hton24(fh->fh_s_id, local_fid);
hton24(fh->fh_d_id, remote_fid);
lport->tt.frame_send(lport, fp);
} else {
fc_lport_error(lport, fp);
}
fc_lport_ptp_setup(lport, remote_fid, remote_wwpn,
get_unaligned_be64(&flp->fl_wwnn));
out:
fc_frame_free(rx_fp);
}
/**
* fc_lport_recv_req() - The generic lport request handler
* @lport: The local port that received the request
* @fp: The request frame
*
* This function will see if the lport handles the request or
* if an rport should handle the request.
*
* Locking Note: This function should not be called with the lport
* lock held becuase it will grab the lock.
*/
static void fc_lport_recv_req(struct fc_lport *lport, struct fc_frame *fp)
{
struct fc_frame_header *fh = fc_frame_header_get(fp);
void (*recv)(struct fc_lport *, struct fc_frame *);
mutex_lock(&lport->lp_mutex);
/*
* Handle special ELS cases like FLOGI, LOGO, and
* RSCN here. These don't require a session.
* Even if we had a session, it might not be ready.
*/
if (!lport->link_up)
fc_frame_free(fp);
else if (fh->fh_type == FC_TYPE_ELS &&
fh->fh_r_ctl == FC_RCTL_ELS_REQ) {
/*
* Check opcode.
*/
recv = lport->tt.rport_recv_req;
switch (fc_frame_payload_op(fp)) {
case ELS_FLOGI:
if (!lport->point_to_multipoint)
recv = fc_lport_recv_flogi_req;
break;
case ELS_LOGO:
if (fc_frame_sid(fp) == FC_FID_FLOGI)
recv = fc_lport_recv_logo_req;
break;
case ELS_RSCN:
recv = lport->tt.disc_recv_req;
break;
case ELS_ECHO:
recv = fc_lport_recv_echo_req;
break;
case ELS_RLIR:
recv = fc_lport_recv_rlir_req;
break;
case ELS_RNID:
recv = fc_lport_recv_rnid_req;
break;
}
recv(lport, fp);
} else {
FC_LPORT_DBG(lport, "dropping invalid frame (eof %x)\n",
fr_eof(fp));
fc_frame_free(fp);
}
mutex_unlock(&lport->lp_mutex);
}
/**
* fc_lport_reset() - Reset a local port
* @lport: The local port which should be reset
*
* Locking Note: This functions should not be called with the
* lport lock held.
*/
int fc_lport_reset(struct fc_lport *lport)
{
cancel_delayed_work_sync(&lport->retry_work);
mutex_lock(&lport->lp_mutex);
fc_lport_enter_reset(lport);
mutex_unlock(&lport->lp_mutex);
return 0;
}
EXPORT_SYMBOL(fc_lport_reset);
/**
* fc_lport_reset_locked() - Reset the local port w/ the lport lock held
* @lport: The local port to be reset
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static void fc_lport_reset_locked(struct fc_lport *lport)
{
if (lport->dns_rdata)
lport->tt.rport_logoff(lport->dns_rdata);
if (lport->ptp_rdata) {
lport->tt.rport_logoff(lport->ptp_rdata);
kref_put(&lport->ptp_rdata->kref, lport->tt.rport_destroy);
lport->ptp_rdata = NULL;
}
lport->tt.disc_stop(lport);
lport->tt.exch_mgr_reset(lport, 0, 0);
fc_host_fabric_name(lport->host) = 0;
if (lport->port_id && (!lport->point_to_multipoint || !lport->link_up))
fc_lport_set_port_id(lport, 0, NULL);
}
/**
* fc_lport_enter_reset() - Reset the local port
* @lport: The local port to be reset
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static void fc_lport_enter_reset(struct fc_lport *lport)
{
FC_LPORT_DBG(lport, "Entered RESET state from %s state\n",
fc_lport_state(lport));
if (lport->state == LPORT_ST_DISABLED || lport->state == LPORT_ST_LOGO)
return;
if (lport->vport) {
if (lport->link_up)
fc_vport_set_state(lport->vport, FC_VPORT_INITIALIZING);
else
fc_vport_set_state(lport->vport, FC_VPORT_LINKDOWN);
}
fc_lport_state_enter(lport, LPORT_ST_RESET);
fc_vports_linkchange(lport);
fc_lport_reset_locked(lport);
[SCSI] libfc, fcoe: fixed locking issues with lport->lp_mutex around lport->link_status The fcoe_xmit could call fc_pause in case the pending skb queue len is larger than FCOE_MAX_QUEUE_DEPTH, the fc_pause was trying to grab lport->lp_muex to change lport->link_status and that had these issues :- 1. The fcoe_xmit was getting called with bh disabled, thus causing "BUG: scheduling while atomic" when grabbing lport->lp_muex with bh disabled. 2. fc_linkup and fc_linkdown function calls lport_enter function with lport->lp_mutex held and these enter function in turn calls fcoe_xmit to send lport related FC frame, e.g. fc_linkup => fc_lport_enter_flogi to send flogi req. In this case grabbing the same lport->lp_mutex again in fc_puase from fcoe_xmit would cause deadlock. The lport->lp_mutex was used for setting FC_PAUSE in fcoe_xmit path but FC_PAUSE bit was not used anywhere beside just setting and clear this bit in lport->link_status, instead used a separate field qfull in fc_lport to eliminate need for lport->lp_mutex to track pending queue full condition and in turn avoid above described two locking issues. Also added check for lp->qfull in fc_fcp_lport_queue_ready to trigger SCSI_MLQUEUE_HOST_BUSY when lp->qfull is set to prevent more scsi-ml cmds while lp->qfull is set. This patch eliminated FC_LINK_UP and FC_PAUSE and instead used dedicated fields in fc_lport for this, this simplified all related conditional code. Also removed fc_pause and fc_unpause functions and instead used newly added lport->qfull directly in fcoe. Signed-off-by: Vasu Dev <vasu.dev@intel.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2009-02-27 18:54:57 +00:00
if (lport->link_up)
fc_lport_enter_flogi(lport);
}
/**
* fc_lport_enter_disabled() - Disable the local port
* @lport: The local port to be reset
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static void fc_lport_enter_disabled(struct fc_lport *lport)
{
FC_LPORT_DBG(lport, "Entered disabled state from %s state\n",
fc_lport_state(lport));
fc_lport_state_enter(lport, LPORT_ST_DISABLED);
fc_vports_linkchange(lport);
fc_lport_reset_locked(lport);
}
/**
* fc_lport_error() - Handler for any errors
* @lport: The local port that the error was on
* @fp: The error code encoded in a frame pointer
*
* If the error was caused by a resource allocation failure
* then wait for half a second and retry, otherwise retry
* after the e_d_tov time.
*/
static void fc_lport_error(struct fc_lport *lport, struct fc_frame *fp)
{
unsigned long delay = 0;
FC_LPORT_DBG(lport, "Error %ld in state %s, retries %d\n",
PTR_ERR(fp), fc_lport_state(lport),
lport->retry_count);
if (PTR_ERR(fp) == -FC_EX_CLOSED)
return;
/*
* Memory allocation failure, or the exchange timed out
* or we received LS_RJT.
* Retry after delay
*/
if (lport->retry_count < lport->max_retry_count) {
lport->retry_count++;
if (!fp)
delay = msecs_to_jiffies(500);
else
delay = msecs_to_jiffies(lport->e_d_tov);
schedule_delayed_work(&lport->retry_work, delay);
} else
fc_lport_enter_reset(lport);
}
/**
* fc_lport_ns_resp() - Handle response to a name server
* registration exchange
* @sp: current sequence in exchange
* @fp: response frame
* @lp_arg: Fibre Channel host port instance
*
* Locking Note: This function will be called without the lport lock
* held, but it will lock, call an _enter_* function or fc_lport_error()
* and then unlock the lport.
*/
static void fc_lport_ns_resp(struct fc_seq *sp, struct fc_frame *fp,
void *lp_arg)
{
struct fc_lport *lport = lp_arg;
struct fc_frame_header *fh;
struct fc_ct_hdr *ct;
FC_LPORT_DBG(lport, "Received a ns %s\n", fc_els_resp_type(fp));
if (fp == ERR_PTR(-FC_EX_CLOSED))
return;
mutex_lock(&lport->lp_mutex);
if (lport->state < LPORT_ST_RNN_ID || lport->state > LPORT_ST_RFF_ID) {
FC_LPORT_DBG(lport, "Received a name server response, "
"but in state %s\n", fc_lport_state(lport));
if (IS_ERR(fp))
goto err;
goto out;
}
if (IS_ERR(fp)) {
fc_lport_error(lport, fp);
goto err;
}
fh = fc_frame_header_get(fp);
ct = fc_frame_payload_get(fp, sizeof(*ct));
if (fh && ct && fh->fh_type == FC_TYPE_CT &&
ct->ct_fs_type == FC_FST_DIR &&
ct->ct_fs_subtype == FC_NS_SUBTYPE &&
ntohs(ct->ct_cmd) == FC_FS_ACC)
switch (lport->state) {
case LPORT_ST_RNN_ID:
fc_lport_enter_ns(lport, LPORT_ST_RSNN_NN);
break;
case LPORT_ST_RSNN_NN:
fc_lport_enter_ns(lport, LPORT_ST_RSPN_ID);
break;
case LPORT_ST_RSPN_ID:
fc_lport_enter_ns(lport, LPORT_ST_RFT_ID);
break;
case LPORT_ST_RFT_ID:
fc_lport_enter_ns(lport, LPORT_ST_RFF_ID);
break;
case LPORT_ST_RFF_ID:
fc_lport_enter_scr(lport);
break;
default:
/* should have already been caught by state checks */
break;
}
else
fc_lport_error(lport, fp);
out:
fc_frame_free(fp);
err:
mutex_unlock(&lport->lp_mutex);
}
/**
* fc_lport_scr_resp() - Handle response to State Change Register (SCR) request
* @sp: current sequence in SCR exchange
* @fp: response frame
* @lp_arg: Fibre Channel lport port instance that sent the registration request
*
* Locking Note: This function will be called without the lport lock
* held, but it will lock, call an _enter_* function or fc_lport_error
* and then unlock the lport.
*/
static void fc_lport_scr_resp(struct fc_seq *sp, struct fc_frame *fp,
void *lp_arg)
{
struct fc_lport *lport = lp_arg;
u8 op;
FC_LPORT_DBG(lport, "Received a SCR %s\n", fc_els_resp_type(fp));
if (fp == ERR_PTR(-FC_EX_CLOSED))
return;
mutex_lock(&lport->lp_mutex);
if (lport->state != LPORT_ST_SCR) {
FC_LPORT_DBG(lport, "Received a SCR response, but in state "
"%s\n", fc_lport_state(lport));
if (IS_ERR(fp))
goto err;
goto out;
}
if (IS_ERR(fp)) {
fc_lport_error(lport, fp);
goto err;
}
op = fc_frame_payload_op(fp);
if (op == ELS_LS_ACC)
fc_lport_enter_ready(lport);
else
fc_lport_error(lport, fp);
out:
fc_frame_free(fp);
err:
mutex_unlock(&lport->lp_mutex);
}
/**
* fc_lport_enter_scr() - Send a SCR (State Change Register) request
* @lport: The local port to register for state changes
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static void fc_lport_enter_scr(struct fc_lport *lport)
{
struct fc_frame *fp;
FC_LPORT_DBG(lport, "Entered SCR state from %s state\n",
fc_lport_state(lport));
fc_lport_state_enter(lport, LPORT_ST_SCR);
fp = fc_frame_alloc(lport, sizeof(struct fc_els_scr));
if (!fp) {
fc_lport_error(lport, fp);
return;
}
if (!lport->tt.elsct_send(lport, FC_FID_FCTRL, fp, ELS_SCR,
fc_lport_scr_resp, lport,
2 * lport->r_a_tov))
[SCSI] libfc: fix memory corruption caused by double frees and bad error handling I was running into several different panics under stress, which I traced down to a few different possible slab corruption issues in error handling paths. I have not yet looked into why these exchange sends fail, but with these fixes my test system is much more stable under stress than before. fc_elsct_send() could fail and either leave the passed in frame intact (failure in fc_ct/els_fill) or the frame could have been freed if the failure was is fc_exch_seq_send(). The caller had no way of knowing, and there was a potential double free in the error handling in fc_fcp_rec(). Make fc_elsct_send() always free the frame before returning, and remove the fc_frame_free() call in fc_fcp_rec(). While fc_exch_seq_send() did always consume the frame, there were double free bugs in the error handling of fc_fcp_cmd_send() and fc_fcp_srr() as well. Numerous calls to error handling routines (fc_disc_error(), fc_lport_error(), fc_rport_error_retry() ) were passing in a frame pointer that had already been freed in the case of an error. I have changed the call sites to pass in a NULL pointer, but there may be more appropriate error codes to use. Question: Why do these error routines take a frame pointer anyway? I understand passing in a pointer encoded error to the response handlers, but the error routines take no action on a valid pointer and should never be called that way. Signed-off-by: Chris Leech <christopher.leech@intel.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-21 23:28:09 +00:00
fc_lport_error(lport, NULL);
}
/**
* fc_lport_enter_ns() - register some object with the name server
* @lport: Fibre Channel local port to register
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static void fc_lport_enter_ns(struct fc_lport *lport, enum fc_lport_state state)
{
struct fc_frame *fp;
enum fc_ns_req cmd;
int size = sizeof(struct fc_ct_hdr);
size_t len;
FC_LPORT_DBG(lport, "Entered %s state from %s state\n",
fc_lport_state_names[state],
fc_lport_state(lport));
fc_lport_state_enter(lport, state);
switch (state) {
case LPORT_ST_RNN_ID:
cmd = FC_NS_RNN_ID;
size += sizeof(struct fc_ns_rn_id);
break;
case LPORT_ST_RSNN_NN:
len = strnlen(fc_host_symbolic_name(lport->host), 255);
/* if there is no symbolic name, skip to RFT_ID */
if (!len)
return fc_lport_enter_ns(lport, LPORT_ST_RFT_ID);
cmd = FC_NS_RSNN_NN;
size += sizeof(struct fc_ns_rsnn) + len;
break;
case LPORT_ST_RSPN_ID:
len = strnlen(fc_host_symbolic_name(lport->host), 255);
/* if there is no symbolic name, skip to RFT_ID */
if (!len)
return fc_lport_enter_ns(lport, LPORT_ST_RFT_ID);
cmd = FC_NS_RSPN_ID;
size += sizeof(struct fc_ns_rspn) + len;
break;
case LPORT_ST_RFT_ID:
cmd = FC_NS_RFT_ID;
size += sizeof(struct fc_ns_rft);
break;
case LPORT_ST_RFF_ID:
cmd = FC_NS_RFF_ID;
size += sizeof(struct fc_ns_rff_id);
break;
default:
fc_lport_error(lport, NULL);
return;
}
fp = fc_frame_alloc(lport, size);
if (!fp) {
fc_lport_error(lport, fp);
return;
}
if (!lport->tt.elsct_send(lport, FC_FID_DIR_SERV, fp, cmd,
fc_lport_ns_resp,
lport, 3 * lport->r_a_tov))
fc_lport_error(lport, fp);
}
static struct fc_rport_operations fc_lport_rport_ops = {
.event_callback = fc_lport_rport_callback,
};
/**
* fc_rport_enter_dns() - Create a fc_rport for the name server
* @lport: The local port requesting a remote port for the name server
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static void fc_lport_enter_dns(struct fc_lport *lport)
{
struct fc_rport_priv *rdata;
FC_LPORT_DBG(lport, "Entered DNS state from %s state\n",
fc_lport_state(lport));
fc_lport_state_enter(lport, LPORT_ST_DNS);
mutex_lock(&lport->disc.disc_mutex);
rdata = lport->tt.rport_create(lport, FC_FID_DIR_SERV);
mutex_unlock(&lport->disc.disc_mutex);
if (!rdata)
goto err;
rdata->ops = &fc_lport_rport_ops;
lport->tt.rport_login(rdata);
return;
err:
fc_lport_error(lport, NULL);
}
/**
* fc_lport_timeout() - Handler for the retry_work timer
* @work: The work struct of the local port
*/
static void fc_lport_timeout(struct work_struct *work)
{
struct fc_lport *lport =
container_of(work, struct fc_lport,
retry_work.work);
mutex_lock(&lport->lp_mutex);
switch (lport->state) {
case LPORT_ST_DISABLED:
WARN_ON(1);
break;
case LPORT_ST_READY:
WARN_ON(1);
break;
case LPORT_ST_RESET:
break;
case LPORT_ST_FLOGI:
fc_lport_enter_flogi(lport);
break;
case LPORT_ST_DNS:
fc_lport_enter_dns(lport);
break;
case LPORT_ST_RNN_ID:
case LPORT_ST_RSNN_NN:
case LPORT_ST_RSPN_ID:
case LPORT_ST_RFT_ID:
case LPORT_ST_RFF_ID:
fc_lport_enter_ns(lport, lport->state);
break;
case LPORT_ST_SCR:
fc_lport_enter_scr(lport);
break;
case LPORT_ST_LOGO:
fc_lport_enter_logo(lport);
break;
}
mutex_unlock(&lport->lp_mutex);
}
/**
* fc_lport_logo_resp() - Handle response to LOGO request
* @sp: The sequence that the LOGO was on
* @fp: The LOGO frame
* @lp_arg: The lport port that received the LOGO request
*
* Locking Note: This function will be called without the lport lock
* held, but it will lock, call an _enter_* function or fc_lport_error()
* and then unlock the lport.
*/
void fc_lport_logo_resp(struct fc_seq *sp, struct fc_frame *fp,
void *lp_arg)
{
struct fc_lport *lport = lp_arg;
u8 op;
FC_LPORT_DBG(lport, "Received a LOGO %s\n", fc_els_resp_type(fp));
if (fp == ERR_PTR(-FC_EX_CLOSED))
return;
mutex_lock(&lport->lp_mutex);
if (lport->state != LPORT_ST_LOGO) {
FC_LPORT_DBG(lport, "Received a LOGO response, but in state "
"%s\n", fc_lport_state(lport));
if (IS_ERR(fp))
goto err;
goto out;
}
if (IS_ERR(fp)) {
fc_lport_error(lport, fp);
goto err;
}
op = fc_frame_payload_op(fp);
if (op == ELS_LS_ACC)
fc_lport_enter_disabled(lport);
else
fc_lport_error(lport, fp);
out:
fc_frame_free(fp);
err:
mutex_unlock(&lport->lp_mutex);
}
EXPORT_SYMBOL(fc_lport_logo_resp);
/**
* fc_rport_enter_logo() - Logout of the fabric
* @lport: The local port to be logged out
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static void fc_lport_enter_logo(struct fc_lport *lport)
{
struct fc_frame *fp;
struct fc_els_logo *logo;
FC_LPORT_DBG(lport, "Entered LOGO state from %s state\n",
fc_lport_state(lport));
fc_lport_state_enter(lport, LPORT_ST_LOGO);
fc_vports_linkchange(lport);
fp = fc_frame_alloc(lport, sizeof(*logo));
if (!fp) {
fc_lport_error(lport, fp);
return;
}
if (!lport->tt.elsct_send(lport, FC_FID_FLOGI, fp, ELS_LOGO,
fc_lport_logo_resp, lport,
2 * lport->r_a_tov))
[SCSI] libfc: fix memory corruption caused by double frees and bad error handling I was running into several different panics under stress, which I traced down to a few different possible slab corruption issues in error handling paths. I have not yet looked into why these exchange sends fail, but with these fixes my test system is much more stable under stress than before. fc_elsct_send() could fail and either leave the passed in frame intact (failure in fc_ct/els_fill) or the frame could have been freed if the failure was is fc_exch_seq_send(). The caller had no way of knowing, and there was a potential double free in the error handling in fc_fcp_rec(). Make fc_elsct_send() always free the frame before returning, and remove the fc_frame_free() call in fc_fcp_rec(). While fc_exch_seq_send() did always consume the frame, there were double free bugs in the error handling of fc_fcp_cmd_send() and fc_fcp_srr() as well. Numerous calls to error handling routines (fc_disc_error(), fc_lport_error(), fc_rport_error_retry() ) were passing in a frame pointer that had already been freed in the case of an error. I have changed the call sites to pass in a NULL pointer, but there may be more appropriate error codes to use. Question: Why do these error routines take a frame pointer anyway? I understand passing in a pointer encoded error to the response handlers, but the error routines take no action on a valid pointer and should never be called that way. Signed-off-by: Chris Leech <christopher.leech@intel.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-21 23:28:09 +00:00
fc_lport_error(lport, NULL);
}
/**
* fc_lport_flogi_resp() - Handle response to FLOGI request
* @sp: The sequence that the FLOGI was on
* @fp: The FLOGI response frame
* @lp_arg: The lport port that received the FLOGI response
*
* Locking Note: This function will be called without the lport lock
* held, but it will lock, call an _enter_* function or fc_lport_error()
* and then unlock the lport.
*/
void fc_lport_flogi_resp(struct fc_seq *sp, struct fc_frame *fp,
void *lp_arg)
{
struct fc_lport *lport = lp_arg;
struct fc_els_flogi *flp;
u32 did;
u16 csp_flags;
unsigned int r_a_tov;
unsigned int e_d_tov;
u16 mfs;
FC_LPORT_DBG(lport, "Received a FLOGI %s\n", fc_els_resp_type(fp));
if (fp == ERR_PTR(-FC_EX_CLOSED))
return;
mutex_lock(&lport->lp_mutex);
if (lport->state != LPORT_ST_FLOGI) {
FC_LPORT_DBG(lport, "Received a FLOGI response, but in state "
"%s\n", fc_lport_state(lport));
if (IS_ERR(fp))
goto err;
goto out;
}
if (IS_ERR(fp)) {
fc_lport_error(lport, fp);
goto err;
}
did = fc_frame_did(fp);
if (!did) {
FC_LPORT_DBG(lport, "Bad FLOGI response\n");
goto out;
}
if (fc_frame_payload_op(fp) == ELS_LS_ACC) {
flp = fc_frame_payload_get(fp, sizeof(*flp));
if (flp) {
mfs = ntohs(flp->fl_csp.sp_bb_data) &
FC_SP_BB_DATA_MASK;
if (mfs >= FC_SP_MIN_MAX_PAYLOAD &&
mfs < lport->mfs)
lport->mfs = mfs;
csp_flags = ntohs(flp->fl_csp.sp_features);
r_a_tov = ntohl(flp->fl_csp.sp_r_a_tov);
e_d_tov = ntohl(flp->fl_csp.sp_e_d_tov);
if (csp_flags & FC_SP_FT_EDTR)
e_d_tov /= 1000000;
lport->npiv_enabled = !!(csp_flags & FC_SP_FT_NPIV_ACC);
if ((csp_flags & FC_SP_FT_FPORT) == 0) {
if (e_d_tov > lport->e_d_tov)
lport->e_d_tov = e_d_tov;
lport->r_a_tov = 2 * e_d_tov;
fc_lport_set_port_id(lport, did, fp);
printk(KERN_INFO "host%d: libfc: "
"Port (%6.6x) entered "
"point-to-point mode\n",
lport->host->host_no, did);
fc_lport_ptp_setup(lport, fc_frame_sid(fp),
get_unaligned_be64(
&flp->fl_wwpn),
get_unaligned_be64(
&flp->fl_wwnn));
} else {
lport->e_d_tov = e_d_tov;
lport->r_a_tov = r_a_tov;
fc_host_fabric_name(lport->host) =
get_unaligned_be64(&flp->fl_wwnn);
fc_lport_set_port_id(lport, did, fp);
fc_lport_enter_dns(lport);
}
}
} else
fc_lport_error(lport, fp);
out:
fc_frame_free(fp);
err:
mutex_unlock(&lport->lp_mutex);
}
EXPORT_SYMBOL(fc_lport_flogi_resp);
/**
* fc_rport_enter_flogi() - Send a FLOGI request to the fabric manager
* @lport: Fibre Channel local port to be logged in to the fabric
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
void fc_lport_enter_flogi(struct fc_lport *lport)
{
struct fc_frame *fp;
FC_LPORT_DBG(lport, "Entered FLOGI state from %s state\n",
fc_lport_state(lport));
fc_lport_state_enter(lport, LPORT_ST_FLOGI);
if (lport->point_to_multipoint) {
if (lport->port_id)
fc_lport_enter_ready(lport);
return;
}
fp = fc_frame_alloc(lport, sizeof(struct fc_els_flogi));
if (!fp)
return fc_lport_error(lport, fp);
if (!lport->tt.elsct_send(lport, FC_FID_FLOGI, fp,
lport->vport ? ELS_FDISC : ELS_FLOGI,
fc_lport_flogi_resp, lport,
lport->vport ? 2 * lport->r_a_tov :
lport->e_d_tov))
[SCSI] libfc: fix memory corruption caused by double frees and bad error handling I was running into several different panics under stress, which I traced down to a few different possible slab corruption issues in error handling paths. I have not yet looked into why these exchange sends fail, but with these fixes my test system is much more stable under stress than before. fc_elsct_send() could fail and either leave the passed in frame intact (failure in fc_ct/els_fill) or the frame could have been freed if the failure was is fc_exch_seq_send(). The caller had no way of knowing, and there was a potential double free in the error handling in fc_fcp_rec(). Make fc_elsct_send() always free the frame before returning, and remove the fc_frame_free() call in fc_fcp_rec(). While fc_exch_seq_send() did always consume the frame, there were double free bugs in the error handling of fc_fcp_cmd_send() and fc_fcp_srr() as well. Numerous calls to error handling routines (fc_disc_error(), fc_lport_error(), fc_rport_error_retry() ) were passing in a frame pointer that had already been freed in the case of an error. I have changed the call sites to pass in a NULL pointer, but there may be more appropriate error codes to use. Question: Why do these error routines take a frame pointer anyway? I understand passing in a pointer encoded error to the response handlers, but the error routines take no action on a valid pointer and should never be called that way. Signed-off-by: Chris Leech <christopher.leech@intel.com> Signed-off-by: Robert Love <robert.w.love@intel.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-21 23:28:09 +00:00
fc_lport_error(lport, NULL);
}
/**
* fc_lport_config() - Configure a fc_lport
* @lport: The local port to be configured
*/
int fc_lport_config(struct fc_lport *lport)
{
INIT_DELAYED_WORK(&lport->retry_work, fc_lport_timeout);
mutex_init(&lport->lp_mutex);
fc_lport_state_enter(lport, LPORT_ST_DISABLED);
fc_lport_add_fc4_type(lport, FC_TYPE_FCP);
fc_lport_add_fc4_type(lport, FC_TYPE_CT);
return 0;
}
EXPORT_SYMBOL(fc_lport_config);
/**
* fc_lport_init() - Initialize the lport layer for a local port
* @lport: The local port to initialize the exchange layer for
*/
int fc_lport_init(struct fc_lport *lport)
{
if (!lport->tt.lport_recv)
lport->tt.lport_recv = fc_lport_recv_req;
if (!lport->tt.lport_reset)
lport->tt.lport_reset = fc_lport_reset;
fc_host_port_type(lport->host) = FC_PORTTYPE_NPORT;
fc_host_node_name(lport->host) = lport->wwnn;
fc_host_port_name(lport->host) = lport->wwpn;
fc_host_supported_classes(lport->host) = FC_COS_CLASS3;
memset(fc_host_supported_fc4s(lport->host), 0,
sizeof(fc_host_supported_fc4s(lport->host)));
fc_host_supported_fc4s(lport->host)[2] = 1;
fc_host_supported_fc4s(lport->host)[7] = 1;
/* This value is also unchanging */
memset(fc_host_active_fc4s(lport->host), 0,
sizeof(fc_host_active_fc4s(lport->host)));
fc_host_active_fc4s(lport->host)[2] = 1;
fc_host_active_fc4s(lport->host)[7] = 1;
fc_host_maxframe_size(lport->host) = lport->mfs;
fc_host_supported_speeds(lport->host) = 0;
if (lport->link_supported_speeds & FC_PORTSPEED_1GBIT)
fc_host_supported_speeds(lport->host) |= FC_PORTSPEED_1GBIT;
if (lport->link_supported_speeds & FC_PORTSPEED_10GBIT)
fc_host_supported_speeds(lport->host) |= FC_PORTSPEED_10GBIT;
return 0;
}
EXPORT_SYMBOL(fc_lport_init);
/**
* fc_lport_bsg_resp() - The common response handler for FC Passthrough requests
* @sp: The sequence for the FC Passthrough response
* @fp: The response frame
* @info_arg: The BSG info that the response is for
*/
static void fc_lport_bsg_resp(struct fc_seq *sp, struct fc_frame *fp,
void *info_arg)
{
struct fc_bsg_info *info = info_arg;
struct fc_bsg_job *job = info->job;
struct fc_lport *lport = info->lport;
struct fc_frame_header *fh;
size_t len;
void *buf;
if (IS_ERR(fp)) {
job->reply->result = (PTR_ERR(fp) == -FC_EX_CLOSED) ?
-ECONNABORTED : -ETIMEDOUT;
job->reply_len = sizeof(uint32_t);
job->state_flags |= FC_RQST_STATE_DONE;
job->job_done(job);
kfree(info);
return;
}
mutex_lock(&lport->lp_mutex);
fh = fc_frame_header_get(fp);
len = fr_len(fp) - sizeof(*fh);
buf = fc_frame_payload_get(fp, 0);
if (fr_sof(fp) == FC_SOF_I3 && !ntohs(fh->fh_seq_cnt)) {
/* Get the response code from the first frame payload */
unsigned short cmd = (info->rsp_code == FC_FS_ACC) ?
ntohs(((struct fc_ct_hdr *)buf)->ct_cmd) :
(unsigned short)fc_frame_payload_op(fp);
/* Save the reply status of the job */
job->reply->reply_data.ctels_reply.status =
(cmd == info->rsp_code) ?
FC_CTELS_STATUS_OK : FC_CTELS_STATUS_REJECT;
}
job->reply->reply_payload_rcv_len +=
fc_copy_buffer_to_sglist(buf, len, info->sg, &info->nents,
&info->offset, KM_BIO_SRC_IRQ, NULL);
if (fr_eof(fp) == FC_EOF_T &&
(ntoh24(fh->fh_f_ctl) & (FC_FC_LAST_SEQ | FC_FC_END_SEQ)) ==
(FC_FC_LAST_SEQ | FC_FC_END_SEQ)) {
if (job->reply->reply_payload_rcv_len >
job->reply_payload.payload_len)
job->reply->reply_payload_rcv_len =
job->reply_payload.payload_len;
job->reply->result = 0;
job->state_flags |= FC_RQST_STATE_DONE;
job->job_done(job);
kfree(info);
}
fc_frame_free(fp);
mutex_unlock(&lport->lp_mutex);
}
/**
* fc_lport_els_request() - Send ELS passthrough request
* @job: The BSG Passthrough job
* @lport: The local port sending the request
* @did: The destination port id
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static int fc_lport_els_request(struct fc_bsg_job *job,
struct fc_lport *lport,
u32 did, u32 tov)
{
struct fc_bsg_info *info;
struct fc_frame *fp;
struct fc_frame_header *fh;
char *pp;
int len;
fp = fc_frame_alloc(lport, job->request_payload.payload_len);
if (!fp)
return -ENOMEM;
len = job->request_payload.payload_len;
pp = fc_frame_payload_get(fp, len);
sg_copy_to_buffer(job->request_payload.sg_list,
job->request_payload.sg_cnt,
pp, len);
fh = fc_frame_header_get(fp);
fh->fh_r_ctl = FC_RCTL_ELS_REQ;
hton24(fh->fh_d_id, did);
hton24(fh->fh_s_id, lport->port_id);
fh->fh_type = FC_TYPE_ELS;
hton24(fh->fh_f_ctl, FC_FCTL_REQ);
fh->fh_cs_ctl = 0;
fh->fh_df_ctl = 0;
fh->fh_parm_offset = 0;
info = kzalloc(sizeof(struct fc_bsg_info), GFP_KERNEL);
if (!info) {
fc_frame_free(fp);
return -ENOMEM;
}
info->job = job;
info->lport = lport;
info->rsp_code = ELS_LS_ACC;
info->nents = job->reply_payload.sg_cnt;
info->sg = job->reply_payload.sg_list;
if (!lport->tt.exch_seq_send(lport, fp, fc_lport_bsg_resp,
NULL, info, tov))
return -ECOMM;
return 0;
}
/**
* fc_lport_ct_request() - Send CT Passthrough request
* @job: The BSG Passthrough job
* @lport: The local port sending the request
* @did: The destination FC-ID
* @tov: The timeout period to wait for the response
*
* Locking Note: The lport lock is expected to be held before calling
* this routine.
*/
static int fc_lport_ct_request(struct fc_bsg_job *job,
struct fc_lport *lport, u32 did, u32 tov)
{
struct fc_bsg_info *info;
struct fc_frame *fp;
struct fc_frame_header *fh;
struct fc_ct_req *ct;
size_t len;
fp = fc_frame_alloc(lport, sizeof(struct fc_ct_hdr) +
job->request_payload.payload_len);
if (!fp)
return -ENOMEM;
len = job->request_payload.payload_len;
ct = fc_frame_payload_get(fp, len);
sg_copy_to_buffer(job->request_payload.sg_list,
job->request_payload.sg_cnt,
ct, len);
fh = fc_frame_header_get(fp);
fh->fh_r_ctl = FC_RCTL_DD_UNSOL_CTL;
hton24(fh->fh_d_id, did);
hton24(fh->fh_s_id, lport->port_id);
fh->fh_type = FC_TYPE_CT;
hton24(fh->fh_f_ctl, FC_FCTL_REQ);
fh->fh_cs_ctl = 0;
fh->fh_df_ctl = 0;
fh->fh_parm_offset = 0;
info = kzalloc(sizeof(struct fc_bsg_info), GFP_KERNEL);
if (!info) {
fc_frame_free(fp);
return -ENOMEM;
}
info->job = job;
info->lport = lport;
info->rsp_code = FC_FS_ACC;
info->nents = job->reply_payload.sg_cnt;
info->sg = job->reply_payload.sg_list;
if (!lport->tt.exch_seq_send(lport, fp, fc_lport_bsg_resp,
NULL, info, tov))
return -ECOMM;
return 0;
}
/**
* fc_lport_bsg_request() - The common entry point for sending
* FC Passthrough requests
* @job: The BSG passthrough job
*/
int fc_lport_bsg_request(struct fc_bsg_job *job)
{
struct request *rsp = job->req->next_rq;
struct Scsi_Host *shost = job->shost;
struct fc_lport *lport = shost_priv(shost);
struct fc_rport *rport;
struct fc_rport_priv *rdata;
int rc = -EINVAL;
u32 did;
job->reply->reply_payload_rcv_len = 0;
if (rsp)
rsp->resid_len = job->reply_payload.payload_len;
mutex_lock(&lport->lp_mutex);
switch (job->request->msgcode) {
case FC_BSG_RPT_ELS:
rport = job->rport;
if (!rport)
break;
rdata = rport->dd_data;
rc = fc_lport_els_request(job, lport, rport->port_id,
rdata->e_d_tov);
break;
case FC_BSG_RPT_CT:
rport = job->rport;
if (!rport)
break;
rdata = rport->dd_data;
rc = fc_lport_ct_request(job, lport, rport->port_id,
rdata->e_d_tov);
break;
case FC_BSG_HST_CT:
did = ntoh24(job->request->rqst_data.h_ct.port_id);
if (did == FC_FID_DIR_SERV)
rdata = lport->dns_rdata;
else
rdata = lport->tt.rport_lookup(lport, did);
if (!rdata)
break;
rc = fc_lport_ct_request(job, lport, did, rdata->e_d_tov);
break;
case FC_BSG_HST_ELS_NOLOGIN:
did = ntoh24(job->request->rqst_data.h_els.port_id);
rc = fc_lport_els_request(job, lport, did, lport->e_d_tov);
break;
}
mutex_unlock(&lport->lp_mutex);
return rc;
}
EXPORT_SYMBOL(fc_lport_bsg_request);