linux/fs/dlm/lowcomms.c

2047 lines
47 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/******************************************************************************
*******************************************************************************
**
** Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
** Copyright (C) 2004-2009 Red Hat, Inc. All rights reserved.
**
**
*******************************************************************************
******************************************************************************/
/*
* lowcomms.c
*
* This is the "low-level" comms layer.
*
* It is responsible for sending/receiving messages
* from other nodes in the cluster.
*
* Cluster nodes are referred to by their nodeids. nodeids are
* simply 32 bit numbers to the locking module - if they need to
* be expanded for the cluster infrastructure then that is its
* responsibility. It is this layer's
* responsibility to resolve these into IP address or
* whatever it needs for inter-node communication.
*
* The comms level is two kernel threads that deal mainly with
* the receiving of messages from other nodes and passing them
* up to the mid-level comms layer (which understands the
* message format) for execution by the locking core, and
* a send thread which does all the setting up of connections
* to remote nodes and the sending of data. Threads are not allowed
* to send their own data because it may cause them to wait in times
* of high load. Also, this way, the sending thread can collect together
* messages bound for one node and send them in one block.
*
* lowcomms will choose to use either TCP or SCTP as its transport layer
* depending on the configuration variable 'protocol'. This should be set
* to 0 (default) for TCP or 1 for SCTP. It should be configured using a
* cluster-wide mechanism as it must be the same on all nodes of the cluster
* for the DLM to function.
*
*/
#include <asm/ioctls.h>
#include <net/sock.h>
#include <net/tcp.h>
#include <linux/pagemap.h>
#include <linux/file.h>
#include <linux/mutex.h>
#include <linux/sctp.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 <net/sctp/sctp.h>
#include <net/ipv6.h>
#include <trace/events/dlm.h>
#include "dlm_internal.h"
#include "lowcomms.h"
#include "midcomms.h"
#include "memory.h"
#include "config.h"
#define NEEDED_RMEM (4*1024*1024)
/* Number of messages to send before rescheduling */
#define MAX_SEND_MSG_COUNT 25
#define DLM_SHUTDOWN_WAIT_TIMEOUT msecs_to_jiffies(10000)
struct connection {
struct socket *sock; /* NULL if not connected */
uint32_t nodeid; /* So we know who we are in the list */
struct mutex sock_mutex;
unsigned long flags;
#define CF_READ_PENDING 1
#define CF_WRITE_PENDING 2
#define CF_INIT_PENDING 4
#define CF_IS_OTHERCON 5
#define CF_CLOSE 6
#define CF_APP_LIMITED 7
#define CF_CLOSING 8
#define CF_SHUTDOWN 9
#define CF_CONNECTED 10
#define CF_RECONNECT 11
#define CF_DELAY_CONNECT 12
#define CF_EOF 13
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
struct list_head writequeue; /* List of outgoing writequeue_entries */
spinlock_t writequeue_lock;
atomic_t writequeue_cnt;
int retries;
#define MAX_CONNECT_RETRIES 3
struct hlist_node list;
struct connection *othercon;
struct connection *sendcon;
struct work_struct rwork; /* Receive workqueue */
struct work_struct swork; /* Send workqueue */
wait_queue_head_t shutdown_wait; /* wait for graceful shutdown */
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
unsigned char *rx_buf;
int rx_buflen;
int rx_leftover;
struct rcu_head rcu;
};
#define sock2con(x) ((struct connection *)(x)->sk_user_data)
struct listen_connection {
struct socket *sock;
struct work_struct rwork;
};
#define DLM_WQ_REMAIN_BYTES(e) (PAGE_SIZE - e->end)
#define DLM_WQ_LENGTH_BYTES(e) (e->end - e->offset)
/* An entry waiting to be sent */
struct writequeue_entry {
struct list_head list;
struct page *page;
int offset;
int len;
int end;
int users;
bool dirty;
struct connection *con;
struct list_head msgs;
struct kref ref;
};
struct dlm_msg {
struct writequeue_entry *entry;
struct dlm_msg *orig_msg;
bool retransmit;
void *ppc;
int len;
int idx; /* new()/commit() idx exchange */
struct list_head list;
struct kref ref;
};
struct dlm_node_addr {
struct list_head list;
int nodeid;
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
int mark;
int addr_count;
int curr_addr_index;
struct sockaddr_storage *addr[DLM_MAX_ADDR_COUNT];
};
struct dlm_proto_ops {
bool try_new_addr;
const char *name;
int proto;
int (*connect)(struct connection *con, struct socket *sock,
struct sockaddr *addr, int addr_len);
void (*sockopts)(struct socket *sock);
int (*bind)(struct socket *sock);
int (*listen_validate)(void);
void (*listen_sockopts)(struct socket *sock);
int (*listen_bind)(struct socket *sock);
/* What to do to shutdown */
void (*shutdown_action)(struct connection *con);
/* What to do to eof check */
bool (*eof_condition)(struct connection *con);
};
static struct listen_sock_callbacks {
void (*sk_error_report)(struct sock *);
void (*sk_data_ready)(struct sock *);
void (*sk_state_change)(struct sock *);
void (*sk_write_space)(struct sock *);
} listen_sock;
static LIST_HEAD(dlm_node_addrs);
static DEFINE_SPINLOCK(dlm_node_addrs_spin);
static struct listen_connection listen_con;
static struct sockaddr_storage *dlm_local_addr[DLM_MAX_ADDR_COUNT];
static int dlm_local_count;
int dlm_allow_conn;
/* Work queues */
static struct workqueue_struct *recv_workqueue;
static struct workqueue_struct *send_workqueue;
static struct hlist_head connection_hash[CONN_HASH_SIZE];
static DEFINE_SPINLOCK(connections_lock);
DEFINE_STATIC_SRCU(connections_srcu);
static const struct dlm_proto_ops *dlm_proto_ops;
static void process_recv_sockets(struct work_struct *work);
static void process_send_sockets(struct work_struct *work);
static void writequeue_entry_ctor(void *data)
{
struct writequeue_entry *entry = data;
INIT_LIST_HEAD(&entry->msgs);
}
struct kmem_cache *dlm_lowcomms_writequeue_cache_create(void)
{
return kmem_cache_create("dlm_writequeue", sizeof(struct writequeue_entry),
0, 0, writequeue_entry_ctor);
}
struct kmem_cache *dlm_lowcomms_msg_cache_create(void)
{
return kmem_cache_create("dlm_msg", sizeof(struct dlm_msg), 0, 0, NULL);
}
/* need to held writequeue_lock */
static struct writequeue_entry *con_next_wq(struct connection *con)
{
struct writequeue_entry *e;
if (list_empty(&con->writequeue))
return NULL;
e = list_first_entry(&con->writequeue, struct writequeue_entry,
list);
/* if len is zero nothing is to send, if there are users filling
* buffers we wait until the users are done so we can send more.
*/
if (e->users || e->len == 0)
return NULL;
return e;
}
static struct connection *__find_con(int nodeid, int r)
{
struct connection *con;
hlist_for_each_entry_rcu(con, &connection_hash[r], list) {
if (con->nodeid == nodeid)
return con;
}
return NULL;
}
static bool tcp_eof_condition(struct connection *con)
{
return atomic_read(&con->writequeue_cnt);
}
static int dlm_con_init(struct connection *con, int nodeid)
{
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
con->rx_buflen = dlm_config.ci_buffer_size;
con->rx_buf = kmalloc(con->rx_buflen, GFP_NOFS);
if (!con->rx_buf)
return -ENOMEM;
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
con->nodeid = nodeid;
mutex_init(&con->sock_mutex);
INIT_LIST_HEAD(&con->writequeue);
spin_lock_init(&con->writequeue_lock);
atomic_set(&con->writequeue_cnt, 0);
INIT_WORK(&con->swork, process_send_sockets);
INIT_WORK(&con->rwork, process_recv_sockets);
init_waitqueue_head(&con->shutdown_wait);
return 0;
}
/*
* If 'allocation' is zero then we don't attempt to create a new
* connection structure for this node.
*/
static struct connection *nodeid2con(int nodeid, gfp_t alloc)
{
struct connection *con, *tmp;
int r, ret;
r = nodeid_hash(nodeid);
con = __find_con(nodeid, r);
if (con || !alloc)
return con;
con = kzalloc(sizeof(*con), alloc);
if (!con)
return NULL;
ret = dlm_con_init(con, nodeid);
if (ret) {
kfree(con);
return NULL;
}
spin_lock(&connections_lock);
/* Because multiple workqueues/threads calls this function it can
* race on multiple cpu's. Instead of locking hot path __find_con()
* we just check in rare cases of recently added nodes again
* under protection of connections_lock. If this is the case we
* abort our connection creation and return the existing connection.
*/
tmp = __find_con(nodeid, r);
if (tmp) {
spin_unlock(&connections_lock);
kfree(con->rx_buf);
kfree(con);
return tmp;
}
hlist_add_head_rcu(&con->list, &connection_hash[r]);
spin_unlock(&connections_lock);
return con;
}
/* Loop round all connections */
static void foreach_conn(void (*conn_func)(struct connection *c))
{
int i;
struct connection *con;
for (i = 0; i < CONN_HASH_SIZE; i++) {
hlist_for_each_entry_rcu(con, &connection_hash[i], list)
conn_func(con);
}
}
static struct dlm_node_addr *find_node_addr(int nodeid)
{
struct dlm_node_addr *na;
list_for_each_entry(na, &dlm_node_addrs, list) {
if (na->nodeid == nodeid)
return na;
}
return NULL;
}
static int addr_compare(const struct sockaddr_storage *x,
const struct sockaddr_storage *y)
{
switch (x->ss_family) {
case AF_INET: {
struct sockaddr_in *sinx = (struct sockaddr_in *)x;
struct sockaddr_in *siny = (struct sockaddr_in *)y;
if (sinx->sin_addr.s_addr != siny->sin_addr.s_addr)
return 0;
if (sinx->sin_port != siny->sin_port)
return 0;
break;
}
case AF_INET6: {
struct sockaddr_in6 *sinx = (struct sockaddr_in6 *)x;
struct sockaddr_in6 *siny = (struct sockaddr_in6 *)y;
if (!ipv6_addr_equal(&sinx->sin6_addr, &siny->sin6_addr))
return 0;
if (sinx->sin6_port != siny->sin6_port)
return 0;
break;
}
default:
return 0;
}
return 1;
}
static int nodeid_to_addr(int nodeid, struct sockaddr_storage *sas_out,
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
struct sockaddr *sa_out, bool try_new_addr,
unsigned int *mark)
{
struct sockaddr_storage sas;
struct dlm_node_addr *na;
if (!dlm_local_count)
return -1;
spin_lock(&dlm_node_addrs_spin);
na = find_node_addr(nodeid);
if (na && na->addr_count) {
memcpy(&sas, na->addr[na->curr_addr_index],
sizeof(struct sockaddr_storage));
if (try_new_addr) {
na->curr_addr_index++;
if (na->curr_addr_index == na->addr_count)
na->curr_addr_index = 0;
}
}
spin_unlock(&dlm_node_addrs_spin);
if (!na)
return -EEXIST;
if (!na->addr_count)
return -ENOENT;
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
*mark = na->mark;
if (sas_out)
memcpy(sas_out, &sas, sizeof(struct sockaddr_storage));
if (!sa_out)
return 0;
if (dlm_local_addr[0]->ss_family == AF_INET) {
struct sockaddr_in *in4 = (struct sockaddr_in *) &sas;
struct sockaddr_in *ret4 = (struct sockaddr_in *) sa_out;
ret4->sin_addr.s_addr = in4->sin_addr.s_addr;
} else {
struct sockaddr_in6 *in6 = (struct sockaddr_in6 *) &sas;
struct sockaddr_in6 *ret6 = (struct sockaddr_in6 *) sa_out;
ret6->sin6_addr = in6->sin6_addr;
}
return 0;
}
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
static int addr_to_nodeid(struct sockaddr_storage *addr, int *nodeid,
unsigned int *mark)
{
struct dlm_node_addr *na;
int rv = -EEXIST;
int addr_i;
spin_lock(&dlm_node_addrs_spin);
list_for_each_entry(na, &dlm_node_addrs, list) {
if (!na->addr_count)
continue;
for (addr_i = 0; addr_i < na->addr_count; addr_i++) {
if (addr_compare(na->addr[addr_i], addr)) {
*nodeid = na->nodeid;
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
*mark = na->mark;
rv = 0;
goto unlock;
}
}
}
unlock:
spin_unlock(&dlm_node_addrs_spin);
return rv;
}
/* caller need to held dlm_node_addrs_spin lock */
static bool dlm_lowcomms_na_has_addr(const struct dlm_node_addr *na,
const struct sockaddr_storage *addr)
{
int i;
for (i = 0; i < na->addr_count; i++) {
if (addr_compare(na->addr[i], addr))
return true;
}
return false;
}
int dlm_lowcomms_addr(int nodeid, struct sockaddr_storage *addr, int len)
{
struct sockaddr_storage *new_addr;
struct dlm_node_addr *new_node, *na;
bool ret;
new_node = kzalloc(sizeof(struct dlm_node_addr), GFP_NOFS);
if (!new_node)
return -ENOMEM;
new_addr = kzalloc(sizeof(struct sockaddr_storage), GFP_NOFS);
if (!new_addr) {
kfree(new_node);
return -ENOMEM;
}
memcpy(new_addr, addr, len);
spin_lock(&dlm_node_addrs_spin);
na = find_node_addr(nodeid);
if (!na) {
new_node->nodeid = nodeid;
new_node->addr[0] = new_addr;
new_node->addr_count = 1;
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
new_node->mark = dlm_config.ci_mark;
list_add(&new_node->list, &dlm_node_addrs);
spin_unlock(&dlm_node_addrs_spin);
return 0;
}
ret = dlm_lowcomms_na_has_addr(na, addr);
if (ret) {
spin_unlock(&dlm_node_addrs_spin);
kfree(new_addr);
kfree(new_node);
return -EEXIST;
}
if (na->addr_count >= DLM_MAX_ADDR_COUNT) {
spin_unlock(&dlm_node_addrs_spin);
kfree(new_addr);
kfree(new_node);
return -ENOSPC;
}
na->addr[na->addr_count++] = new_addr;
spin_unlock(&dlm_node_addrs_spin);
kfree(new_node);
return 0;
}
/* Data available on socket or listen socket received a connect */
static void lowcomms_data_ready(struct sock *sk)
{
struct connection *con;
con = sock2con(sk);
if (con && !test_and_set_bit(CF_READ_PENDING, &con->flags))
queue_work(recv_workqueue, &con->rwork);
}
static void lowcomms_listen_data_ready(struct sock *sk)
{
if (!dlm_allow_conn)
return;
queue_work(recv_workqueue, &listen_con.rwork);
}
static void lowcomms_write_space(struct sock *sk)
{
struct connection *con;
con = sock2con(sk);
if (!con)
return;
if (!test_and_set_bit(CF_CONNECTED, &con->flags)) {
log_print("connected to node %d", con->nodeid);
queue_work(send_workqueue, &con->swork);
return;
}
clear_bit(SOCK_NOSPACE, &con->sock->flags);
if (test_and_clear_bit(CF_APP_LIMITED, &con->flags)) {
con->sock->sk->sk_write_pending--;
clear_bit(SOCKWQ_ASYNC_NOSPACE, &con->sock->flags);
}
queue_work(send_workqueue, &con->swork);
}
static inline void lowcomms_connect_sock(struct connection *con)
{
if (test_bit(CF_CLOSE, &con->flags))
return;
queue_work(send_workqueue, &con->swork);
cond_resched();
}
static void lowcomms_state_change(struct sock *sk)
{
/* SCTP layer is not calling sk_data_ready when the connection
* is done, so we catch the signal through here. Also, it
* doesn't switch socket state when entering shutdown, so we
* skip the write in that case.
*/
if (sk->sk_shutdown) {
if (sk->sk_shutdown == RCV_SHUTDOWN)
lowcomms_data_ready(sk);
} else if (sk->sk_state == TCP_ESTABLISHED) {
lowcomms_write_space(sk);
}
}
int dlm_lowcomms_connect_node(int nodeid)
{
struct connection *con;
int idx;
if (nodeid == dlm_our_nodeid())
return 0;
idx = srcu_read_lock(&connections_srcu);
con = nodeid2con(nodeid, GFP_NOFS);
if (!con) {
srcu_read_unlock(&connections_srcu, idx);
return -ENOMEM;
}
lowcomms_connect_sock(con);
srcu_read_unlock(&connections_srcu, idx);
return 0;
}
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
int dlm_lowcomms_nodes_set_mark(int nodeid, unsigned int mark)
{
struct dlm_node_addr *na;
spin_lock(&dlm_node_addrs_spin);
na = find_node_addr(nodeid);
if (!na) {
spin_unlock(&dlm_node_addrs_spin);
return -ENOENT;
}
na->mark = mark;
spin_unlock(&dlm_node_addrs_spin);
return 0;
}
static void lowcomms_error_report(struct sock *sk)
{
struct connection *con;
void (*orig_report)(struct sock *) = NULL;
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
struct inet_sock *inet;
con = sock2con(sk);
if (con == NULL)
goto out;
orig_report = listen_sock.sk_error_report;
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
inet = inet_sk(sk);
switch (sk->sk_family) {
case AF_INET:
printk_ratelimited(KERN_ERR "dlm: node %d: socket error "
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
"sending to node %d at %pI4, dport %d, "
"sk_err=%d/%d\n", dlm_our_nodeid(),
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
con->nodeid, &inet->inet_daddr,
ntohs(inet->inet_dport), sk->sk_err,
sk->sk_err_soft);
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
break;
#if IS_ENABLED(CONFIG_IPV6)
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
case AF_INET6:
printk_ratelimited(KERN_ERR "dlm: node %d: socket error "
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
"sending to node %d at %pI6c, "
"dport %d, sk_err=%d/%d\n", dlm_our_nodeid(),
con->nodeid, &sk->sk_v6_daddr,
ntohs(inet->inet_dport), sk->sk_err,
sk->sk_err_soft);
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
break;
#endif
fs: dlm: don't call kernel_getpeername() in error_report() In some cases kernel_getpeername() will held the socket lock which is already held when the socket layer calls error_report() callback. Since commit 9dfc685e0262 ("inet: remove races in inet{6}_getname()") this problem becomes more likely because the socket lock will be held always. You will see something like: bob9-u5 login: [ 562.316860] BUG: spinlock recursion on CPU#7, swapper/7/0 [ 562.318562] lock: 0xffff8f2284720088, .magic: dead4ead, .owner: swapper/7/0, .owner_cpu: 7 [ 562.319522] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 5.15.0+ #135 [ 562.320346] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 [ 562.321277] Call Trace: [ 562.321529] <IRQ> [ 562.321734] dump_stack_lvl+0x33/0x42 [ 562.322282] do_raw_spin_lock+0x8b/0xc0 [ 562.322674] lock_sock_nested+0x1e/0x50 [ 562.323057] inet_getname+0x39/0x110 [ 562.323425] ? sock_def_readable+0x80/0x80 [ 562.323838] lowcomms_error_report+0x63/0x260 [dlm] [ 562.324338] ? wait_for_completion_interruptible_timeout+0xd2/0x120 [ 562.324949] ? lock_timer_base+0x67/0x80 [ 562.325330] ? do_raw_spin_unlock+0x49/0xc0 [ 562.325735] ? _raw_spin_unlock_irqrestore+0x1e/0x40 [ 562.326218] ? del_timer+0x54/0x80 [ 562.326549] sk_error_report+0x12/0x70 [ 562.326919] tcp_validate_incoming+0x3c8/0x530 [ 562.327347] ? kvm_clock_read+0x14/0x30 [ 562.327718] ? ktime_get+0x3b/0xa0 [ 562.328055] tcp_rcv_established+0x121/0x660 [ 562.328466] tcp_v4_do_rcv+0x132/0x260 [ 562.328835] tcp_v4_rcv+0xcea/0xe20 [ 562.329173] ip_protocol_deliver_rcu+0x35/0x1f0 [ 562.329615] ip_local_deliver_finish+0x54/0x60 [ 562.330050] ip_local_deliver+0xf7/0x110 [ 562.330431] ? inet_rtm_getroute+0x211/0x840 [ 562.330848] ? ip_protocol_deliver_rcu+0x1f0/0x1f0 [ 562.331310] ip_rcv+0xe1/0xf0 [ 562.331603] ? ip_local_deliver+0x110/0x110 [ 562.332011] __netif_receive_skb_core+0x46a/0x1040 [ 562.332476] ? inet_gro_receive+0x263/0x2e0 [ 562.332885] __netif_receive_skb_list_core+0x13b/0x2c0 [ 562.333383] netif_receive_skb_list_internal+0x1c8/0x2f0 [ 562.333896] ? update_load_avg+0x7e/0x5e0 [ 562.334285] gro_normal_list.part.149+0x19/0x40 [ 562.334722] napi_complete_done+0x67/0x160 [ 562.335134] virtnet_poll+0x2ad/0x408 [virtio_net] [ 562.335644] __napi_poll+0x28/0x140 [ 562.336012] net_rx_action+0x23d/0x300 [ 562.336414] __do_softirq+0xf2/0x2ea [ 562.336803] irq_exit_rcu+0xc1/0xf0 [ 562.337173] common_interrupt+0xb9/0xd0 It is and was always forbidden to call kernel_getpeername() in context of error_report(). To get rid of the problem we access the destination address for the peer over the socket structure. While on it we fix to print out the destination port of the inet socket. Fixes: 1a31833d085a ("DLM: Replace nodeid_to_addr with kernel_getpeername") Reported-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-11-15 13:57:05 +00:00
default:
printk_ratelimited(KERN_ERR "dlm: node %d: socket error "
"invalid socket family %d set, "
"sk_err=%d/%d\n", dlm_our_nodeid(),
sk->sk_family, sk->sk_err, sk->sk_err_soft);
goto out;
}
/* below sendcon only handling */
if (test_bit(CF_IS_OTHERCON, &con->flags))
con = con->sendcon;
switch (sk->sk_err) {
case ECONNREFUSED:
set_bit(CF_DELAY_CONNECT, &con->flags);
break;
default:
break;
}
if (!test_and_set_bit(CF_RECONNECT, &con->flags))
queue_work(send_workqueue, &con->swork);
out:
if (orig_report)
orig_report(sk);
}
/* Note: sk_callback_lock must be locked before calling this function. */
static void save_listen_callbacks(struct socket *sock)
{
struct sock *sk = sock->sk;
listen_sock.sk_data_ready = sk->sk_data_ready;
listen_sock.sk_state_change = sk->sk_state_change;
listen_sock.sk_write_space = sk->sk_write_space;
listen_sock.sk_error_report = sk->sk_error_report;
}
static void restore_callbacks(struct socket *sock)
{
struct sock *sk = sock->sk;
lock_sock(sk);
sk->sk_user_data = NULL;
sk->sk_data_ready = listen_sock.sk_data_ready;
sk->sk_state_change = listen_sock.sk_state_change;
sk->sk_write_space = listen_sock.sk_write_space;
sk->sk_error_report = listen_sock.sk_error_report;
release_sock(sk);
}
static void add_listen_sock(struct socket *sock, struct listen_connection *con)
{
struct sock *sk = sock->sk;
lock_sock(sk);
save_listen_callbacks(sock);
con->sock = sock;
sk->sk_user_data = con;
sk->sk_allocation = GFP_NOFS;
/* Install a data_ready callback */
sk->sk_data_ready = lowcomms_listen_data_ready;
release_sock(sk);
}
/* Make a socket active */
static void add_sock(struct socket *sock, struct connection *con)
{
struct sock *sk = sock->sk;
lock_sock(sk);
con->sock = sock;
sk->sk_user_data = con;
/* Install a data_ready callback */
sk->sk_data_ready = lowcomms_data_ready;
sk->sk_write_space = lowcomms_write_space;
sk->sk_state_change = lowcomms_state_change;
sk->sk_allocation = GFP_NOFS;
sk->sk_error_report = lowcomms_error_report;
release_sock(sk);
}
/* Add the port number to an IPv6 or 4 sockaddr and return the address
length */
static void make_sockaddr(struct sockaddr_storage *saddr, uint16_t port,
int *addr_len)
{
saddr->ss_family = dlm_local_addr[0]->ss_family;
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
if (saddr->ss_family == AF_INET) {
struct sockaddr_in *in4_addr = (struct sockaddr_in *)saddr;
in4_addr->sin_port = cpu_to_be16(port);
*addr_len = sizeof(struct sockaddr_in);
memset(&in4_addr->sin_zero, 0, sizeof(in4_addr->sin_zero));
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
} else {
struct sockaddr_in6 *in6_addr = (struct sockaddr_in6 *)saddr;
in6_addr->sin6_port = cpu_to_be16(port);
*addr_len = sizeof(struct sockaddr_in6);
}
memset((char *)saddr + *addr_len, 0, sizeof(struct sockaddr_storage) - *addr_len);
}
static void dlm_page_release(struct kref *kref)
{
struct writequeue_entry *e = container_of(kref, struct writequeue_entry,
ref);
__free_page(e->page);
dlm_free_writequeue(e);
}
static void dlm_msg_release(struct kref *kref)
{
struct dlm_msg *msg = container_of(kref, struct dlm_msg, ref);
kref_put(&msg->entry->ref, dlm_page_release);
dlm_free_msg(msg);
}
static void free_entry(struct writequeue_entry *e)
{
struct dlm_msg *msg, *tmp;
list_for_each_entry_safe(msg, tmp, &e->msgs, list) {
if (msg->orig_msg) {
msg->orig_msg->retransmit = false;
kref_put(&msg->orig_msg->ref, dlm_msg_release);
}
list_del(&msg->list);
kref_put(&msg->ref, dlm_msg_release);
}
list_del(&e->list);
atomic_dec(&e->con->writequeue_cnt);
kref_put(&e->ref, dlm_page_release);
}
static void dlm_close_sock(struct socket **sock)
{
if (*sock) {
restore_callbacks(*sock);
sock_release(*sock);
*sock = NULL;
}
}
/* Close a remote connection and tidy up */
static void close_connection(struct connection *con, bool and_other,
bool tx, bool rx)
{
bool closing = test_and_set_bit(CF_CLOSING, &con->flags);
struct writequeue_entry *e;
if (tx && !closing && cancel_work_sync(&con->swork)) {
log_print("canceled swork for node %d", con->nodeid);
clear_bit(CF_WRITE_PENDING, &con->flags);
}
if (rx && !closing && cancel_work_sync(&con->rwork)) {
log_print("canceled rwork for node %d", con->nodeid);
clear_bit(CF_READ_PENDING, &con->flags);
}
mutex_lock(&con->sock_mutex);
dlm_close_sock(&con->sock);
if (con->othercon && and_other) {
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
/* Will only re-enter once. */
close_connection(con->othercon, false, tx, rx);
}
/* if we send a writequeue entry only a half way, we drop the
* whole entry because reconnection and that we not start of the
* middle of a msg which will confuse the other end.
*
* we can always drop messages because retransmits, but what we
* cannot allow is to transmit half messages which may be processed
* at the other side.
*
* our policy is to start on a clean state when disconnects, we don't
* know what's send/received on transport layer in this case.
*/
spin_lock(&con->writequeue_lock);
if (!list_empty(&con->writequeue)) {
e = list_first_entry(&con->writequeue, struct writequeue_entry,
list);
if (e->dirty)
free_entry(e);
}
spin_unlock(&con->writequeue_lock);
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
con->rx_leftover = 0;
con->retries = 0;
clear_bit(CF_APP_LIMITED, &con->flags);
clear_bit(CF_CONNECTED, &con->flags);
clear_bit(CF_DELAY_CONNECT, &con->flags);
clear_bit(CF_RECONNECT, &con->flags);
clear_bit(CF_EOF, &con->flags);
mutex_unlock(&con->sock_mutex);
clear_bit(CF_CLOSING, &con->flags);
}
static void shutdown_connection(struct connection *con)
{
int ret;
flush_work(&con->swork);
mutex_lock(&con->sock_mutex);
/* nothing to shutdown */
if (!con->sock) {
mutex_unlock(&con->sock_mutex);
return;
}
set_bit(CF_SHUTDOWN, &con->flags);
ret = kernel_sock_shutdown(con->sock, SHUT_WR);
mutex_unlock(&con->sock_mutex);
if (ret) {
log_print("Connection %p failed to shutdown: %d will force close",
con, ret);
goto force_close;
} else {
ret = wait_event_timeout(con->shutdown_wait,
!test_bit(CF_SHUTDOWN, &con->flags),
DLM_SHUTDOWN_WAIT_TIMEOUT);
if (ret == 0) {
log_print("Connection %p shutdown timed out, will force close",
con);
goto force_close;
}
}
return;
force_close:
clear_bit(CF_SHUTDOWN, &con->flags);
close_connection(con, false, true, true);
}
static void dlm_tcp_shutdown(struct connection *con)
{
if (con->othercon)
shutdown_connection(con->othercon);
shutdown_connection(con);
}
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
static int con_realloc_receive_buf(struct connection *con, int newlen)
{
unsigned char *newbuf;
newbuf = kmalloc(newlen, GFP_NOFS);
if (!newbuf)
return -ENOMEM;
/* copy any leftover from last receive */
if (con->rx_leftover)
memmove(newbuf, con->rx_buf, con->rx_leftover);
/* swap to new buffer space */
kfree(con->rx_buf);
con->rx_buflen = newlen;
con->rx_buf = newbuf;
return 0;
}
/* Data received from remote end */
static int receive_from_sock(struct connection *con)
{
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
struct msghdr msg;
struct kvec iov;
int ret, buflen;
mutex_lock(&con->sock_mutex);
if (con->sock == NULL) {
ret = -EAGAIN;
goto out_close;
}
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
/* realloc if we get new buffer size to read out */
buflen = dlm_config.ci_buffer_size;
if (con->rx_buflen != buflen && con->rx_leftover <= buflen) {
ret = con_realloc_receive_buf(con, buflen);
if (ret < 0)
goto out_resched;
}
for (;;) {
/* calculate new buffer parameter regarding last receive and
* possible leftover bytes
*/
iov.iov_base = con->rx_buf + con->rx_leftover;
iov.iov_len = con->rx_buflen - con->rx_leftover;
memset(&msg, 0, sizeof(msg));
msg.msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL;
ret = kernel_recvmsg(con->sock, &msg, &iov, 1, iov.iov_len,
msg.msg_flags);
trace_dlm_recv(con->nodeid, ret);
if (ret == -EAGAIN)
break;
else if (ret <= 0)
goto out_close;
/* new buflen according readed bytes and leftover from last receive */
buflen = ret + con->rx_leftover;
ret = dlm_process_incoming_buffer(con->nodeid, con->rx_buf, buflen);
if (ret < 0)
goto out_close;
/* calculate leftover bytes from process and put it into begin of
* the receive buffer, so next receive we have the full message
* at the start address of the receive buffer.
*/
con->rx_leftover = buflen - ret;
if (con->rx_leftover) {
memmove(con->rx_buf, con->rx_buf + ret,
con->rx_leftover);
}
}
dlm_midcomms_receive_done(con->nodeid);
mutex_unlock(&con->sock_mutex);
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
return 0;
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
out_resched:
if (!test_and_set_bit(CF_READ_PENDING, &con->flags))
queue_work(recv_workqueue, &con->rwork);
mutex_unlock(&con->sock_mutex);
return -EAGAIN;
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
out_close:
if (ret == 0) {
log_print("connection %p got EOF from %d",
con, con->nodeid);
if (dlm_proto_ops->eof_condition &&
dlm_proto_ops->eof_condition(con)) {
set_bit(CF_EOF, &con->flags);
mutex_unlock(&con->sock_mutex);
} else {
mutex_unlock(&con->sock_mutex);
close_connection(con, false, true, false);
/* handling for tcp shutdown */
clear_bit(CF_SHUTDOWN, &con->flags);
wake_up(&con->shutdown_wait);
}
/* signal to breaking receive worker */
ret = -1;
} else {
mutex_unlock(&con->sock_mutex);
}
return ret;
}
/* Listening socket is busy, accept a connection */
static int accept_from_sock(struct listen_connection *con)
{
int result;
struct sockaddr_storage peeraddr;
struct socket *newsock;
int len, idx;
int nodeid;
struct connection *newcon;
struct connection *addcon;
unsigned int mark;
if (!con->sock)
return -ENOTCONN;
result = kernel_accept(con->sock, &newsock, O_NONBLOCK);
if (result < 0)
goto accept_err;
/* Get the connected socket's peer */
memset(&peeraddr, 0, sizeof(peeraddr));
net: make getname() functions return length rather than use int* parameter Changes since v1: Added changes in these files: drivers/infiniband/hw/usnic/usnic_transport.c drivers/staging/lustre/lnet/lnet/lib-socket.c drivers/target/iscsi/iscsi_target_login.c drivers/vhost/net.c fs/dlm/lowcomms.c fs/ocfs2/cluster/tcp.c security/tomoyo/network.c Before: All these functions either return a negative error indicator, or store length of sockaddr into "int *socklen" parameter and return zero on success. "int *socklen" parameter is awkward. For example, if caller does not care, it still needs to provide on-stack storage for the value it does not need. None of the many FOO_getname() functions of various protocols ever used old value of *socklen. They always just overwrite it. This change drops this parameter, and makes all these functions, on success, return length of sockaddr. It's always >= 0 and can be differentiated from an error. Tests in callers are changed from "if (err)" to "if (err < 0)", where needed. rpc_sockname() lost "int buflen" parameter, since its only use was to be passed to kernel_getsockname() as &buflen and subsequently not used in any way. Userspace API is not changed. text data bss dec hex filename 30108430 2633624 873672 33615726 200ef6e vmlinux.before.o 30108109 2633612 873672 33615393 200ee21 vmlinux.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com> CC: David S. Miller <davem@davemloft.net> CC: linux-kernel@vger.kernel.org CC: netdev@vger.kernel.org CC: linux-bluetooth@vger.kernel.org CC: linux-decnet-user@lists.sourceforge.net CC: linux-wireless@vger.kernel.org CC: linux-rdma@vger.kernel.org CC: linux-sctp@vger.kernel.org CC: linux-nfs@vger.kernel.org CC: linux-x25@vger.kernel.org Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-12 19:00:20 +00:00
len = newsock->ops->getname(newsock, (struct sockaddr *)&peeraddr, 2);
if (len < 0) {
result = -ECONNABORTED;
goto accept_err;
}
/* Get the new node's NODEID */
make_sockaddr(&peeraddr, 0, &len);
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
if (addr_to_nodeid(&peeraddr, &nodeid, &mark)) {
switch (peeraddr.ss_family) {
case AF_INET: {
struct sockaddr_in *sin = (struct sockaddr_in *)&peeraddr;
log_print("connect from non cluster IPv4 node %pI4",
&sin->sin_addr);
break;
}
#if IS_ENABLED(CONFIG_IPV6)
case AF_INET6: {
struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)&peeraddr;
log_print("connect from non cluster IPv6 node %pI6c",
&sin6->sin6_addr);
break;
}
#endif
default:
log_print("invalid family from non cluster node");
break;
}
sock_release(newsock);
return -1;
}
log_print("got connection from %d", nodeid);
/* Check to see if we already have a connection to this node. This
* could happen if the two nodes initiate a connection at roughly
* the same time and the connections cross on the wire.
* In this case we store the incoming one in "othercon"
*/
idx = srcu_read_lock(&connections_srcu);
newcon = nodeid2con(nodeid, GFP_NOFS);
if (!newcon) {
srcu_read_unlock(&connections_srcu, idx);
result = -ENOMEM;
goto accept_err;
}
fs: dlm: fix mark setting deadlock This patch fixes an deadlock issue when dlm_lowcomms_close() is called. When dlm_lowcomms_close() is called the clusters_root.subsys.su_mutex is held to remove configfs items. At this time we flushing (e.g. cancel_work_sync()) the workers of send and recv workqueue. Due the fact that we accessing configfs items (mark values), these workers will lock clusters_root.subsys.su_mutex as well which are already hold by dlm_lowcomms_close() and ends in a deadlock situation. [67170.703046] ====================================================== [67170.703965] WARNING: possible circular locking dependency detected [67170.704758] 5.11.0-rc4+ #22 Tainted: G W [67170.705433] ------------------------------------------------------ [67170.706228] dlm_controld/280 is trying to acquire lock: [67170.706915] ffff9f2f475a6948 ((wq_completion)dlm_recv){+.+.}-{0:0}, at: __flush_work+0x203/0x4c0 [67170.708026] but task is already holding lock: [67170.708758] ffffffffa132f878 (&clusters_root.subsys.su_mutex){+.+.}-{3:3}, at: configfs_rmdir+0x29b/0x310 [67170.710016] which lock already depends on the new lock. The new behaviour adds the mark value to the node address configuration which doesn't require to held the clusters_root.subsys.su_mutex by accessing mark values in a separate datastructure. However the mark values can be set now only after a node address was set which is the case when the user is using dlm_controld. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:09 +00:00
sock_set_mark(newsock->sk, mark);
mutex_lock(&newcon->sock_mutex);
if (newcon->sock) {
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
struct connection *othercon = newcon->othercon;
if (!othercon) {
othercon = kzalloc(sizeof(*othercon), GFP_NOFS);
if (!othercon) {
log_print("failed to allocate incoming socket");
mutex_unlock(&newcon->sock_mutex);
srcu_read_unlock(&connections_srcu, idx);
result = -ENOMEM;
goto accept_err;
}
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
result = dlm_con_init(othercon, nodeid);
if (result < 0) {
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
kfree(othercon);
mutex_unlock(&newcon->sock_mutex);
srcu_read_unlock(&connections_srcu, idx);
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
goto accept_err;
}
lockdep_set_subclass(&othercon->sock_mutex, 1);
set_bit(CF_IS_OTHERCON, &othercon->flags);
newcon->othercon = othercon;
othercon->sendcon = newcon;
} else {
/* close other sock con if we have something new */
close_connection(othercon, false, true, false);
}
mutex_lock(&othercon->sock_mutex);
add_sock(newsock, othercon);
addcon = othercon;
mutex_unlock(&othercon->sock_mutex);
}
else {
/* accept copies the sk after we've saved the callbacks, so we
don't want to save them a second time or comm errors will
result in calling sk_error_report recursively. */
add_sock(newsock, newcon);
addcon = newcon;
}
set_bit(CF_CONNECTED, &addcon->flags);
mutex_unlock(&newcon->sock_mutex);
/*
* Add it to the active queue in case we got data
* between processing the accept adding the socket
* to the read_sockets list
*/
if (!test_and_set_bit(CF_READ_PENDING, &addcon->flags))
queue_work(recv_workqueue, &addcon->rwork);
srcu_read_unlock(&connections_srcu, idx);
return 0;
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
accept_err:
if (newsock)
sock_release(newsock);
if (result != -EAGAIN)
log_print("error accepting connection from node: %d", result);
return result;
}
/*
* writequeue_entry_complete - try to delete and free write queue entry
* @e: write queue entry to try to delete
* @completed: bytes completed
*
* writequeue_lock must be held.
*/
static void writequeue_entry_complete(struct writequeue_entry *e, int completed)
{
e->offset += completed;
e->len -= completed;
/* signal that page was half way transmitted */
e->dirty = true;
if (e->len == 0 && e->users == 0)
free_entry(e);
}
/*
* sctp_bind_addrs - bind a SCTP socket to all our addresses
*/
static int sctp_bind_addrs(struct socket *sock, uint16_t port)
{
struct sockaddr_storage localaddr;
struct sockaddr *addr = (struct sockaddr *)&localaddr;
int i, addr_len, result = 0;
for (i = 0; i < dlm_local_count; i++) {
memcpy(&localaddr, dlm_local_addr[i], sizeof(localaddr));
make_sockaddr(&localaddr, port, &addr_len);
if (!i)
result = kernel_bind(sock, addr, addr_len);
else
result = sock_bind_add(sock->sk, addr, addr_len);
if (result < 0) {
log_print("Can't bind to %d addr number %d, %d.\n",
port, i + 1, result);
break;
}
}
return result;
}
/* Get local addresses */
static void init_local(void)
{
struct sockaddr_storage sas, *addr;
int i;
dlm_local_count = 0;
for (i = 0; i < DLM_MAX_ADDR_COUNT; i++) {
if (dlm_our_addr(&sas, i))
break;
addr = kmemdup(&sas, sizeof(*addr), GFP_NOFS);
if (!addr)
break;
dlm_local_addr[dlm_local_count++] = addr;
}
}
static void deinit_local(void)
{
int i;
for (i = 0; i < dlm_local_count; i++)
kfree(dlm_local_addr[i]);
}
static struct writequeue_entry *new_writequeue_entry(struct connection *con)
{
struct writequeue_entry *entry;
entry = dlm_allocate_writequeue();
if (!entry)
return NULL;
entry->page = alloc_page(GFP_ATOMIC | __GFP_ZERO);
if (!entry->page) {
dlm_free_writequeue(entry);
return NULL;
}
entry->offset = 0;
entry->len = 0;
entry->end = 0;
entry->dirty = false;
entry->con = con;
entry->users = 1;
kref_init(&entry->ref);
return entry;
}
static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
char **ppc, void (*cb)(void *data),
void *data)
{
struct writequeue_entry *e;
spin_lock(&con->writequeue_lock);
if (!list_empty(&con->writequeue)) {
e = list_last_entry(&con->writequeue, struct writequeue_entry, list);
if (DLM_WQ_REMAIN_BYTES(e) >= len) {
kref_get(&e->ref);
*ppc = page_address(e->page) + e->end;
if (cb)
cb(data);
e->end += len;
e->users++;
goto out;
}
}
e = new_writequeue_entry(con);
if (!e)
goto out;
kref_get(&e->ref);
*ppc = page_address(e->page);
e->end += len;
atomic_inc(&con->writequeue_cnt);
if (cb)
cb(data);
list_add_tail(&e->list, &con->writequeue);
out:
spin_unlock(&con->writequeue_lock);
return e;
};
static struct dlm_msg *dlm_lowcomms_new_msg_con(struct connection *con, int len,
gfp_t allocation, char **ppc,
void (*cb)(void *data),
void *data)
{
struct writequeue_entry *e;
struct dlm_msg *msg;
msg = dlm_allocate_msg(allocation);
if (!msg)
return NULL;
kref_init(&msg->ref);
e = new_wq_entry(con, len, ppc, cb, data);
if (!e) {
dlm_free_msg(msg);
return NULL;
}
msg->retransmit = false;
msg->orig_msg = NULL;
msg->ppc = *ppc;
msg->len = len;
msg->entry = e;
return msg;
}
/* avoid false positive for nodes_srcu, unlock happens in
* dlm_lowcomms_commit_msg which is a must call if success
*/
#ifndef __CHECKER__
struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation,
char **ppc, void (*cb)(void *data),
void *data)
{
struct connection *con;
struct dlm_msg *msg;
int idx;
if (len > DLM_MAX_SOCKET_BUFSIZE ||
len < sizeof(struct dlm_header)) {
BUILD_BUG_ON(PAGE_SIZE < DLM_MAX_SOCKET_BUFSIZE);
log_print("failed to allocate a buffer of size %d", len);
WARN_ON(1);
return NULL;
}
idx = srcu_read_lock(&connections_srcu);
con = nodeid2con(nodeid, allocation);
if (!con) {
srcu_read_unlock(&connections_srcu, idx);
return NULL;
}
msg = dlm_lowcomms_new_msg_con(con, len, allocation, ppc, cb, data);
if (!msg) {
srcu_read_unlock(&connections_srcu, idx);
return NULL;
}
fs: dlm: fix race in lowcomms This patch fixes a race between queue_work() in _dlm_lowcomms_commit_msg() and srcu_read_unlock(). The queue_work() can take the final reference of a dlm_msg and so msg->idx can contain garbage which is signaled by the following warning: [ 676.237050] ------------[ cut here ]------------ [ 676.237052] WARNING: CPU: 0 PID: 1060 at include/linux/srcu.h:189 dlm_lowcomms_commit_msg+0x41/0x50 [ 676.238945] Modules linked in: dlm_locktorture torture rpcsec_gss_krb5 intel_rapl_msr intel_rapl_common iTCO_wdt iTCO_vendor_support qxl kvm_intel drm_ttm_helper vmw_vsock_virtio_transport kvm vmw_vsock_virtio_transport_common ttm irqbypass crc32_pclmul joydev crc32c_intel serio_raw drm_kms_helper vsock virtio_scsi virtio_console virtio_balloon snd_pcm drm syscopyarea sysfillrect sysimgblt snd_timer fb_sys_fops i2c_i801 lpc_ich snd i2c_smbus soundcore pcspkr [ 676.244227] CPU: 0 PID: 1060 Comm: lock_torture_wr Not tainted 5.19.0-rc3+ #1546 [ 676.245216] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.16.0-2.module+el8.7.0+15506+033991b0 04/01/2014 [ 676.246460] RIP: 0010:dlm_lowcomms_commit_msg+0x41/0x50 [ 676.247132] Code: fe ff ff ff 75 24 48 c7 c6 bd 0f 49 bb 48 c7 c7 38 7c 01 bd e8 00 e7 ca ff 89 de 48 c7 c7 60 78 01 bd e8 42 3d cd ff 5b 5d c3 <0f> 0b eb d8 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 [ 676.249253] RSP: 0018:ffffa401c18ffc68 EFLAGS: 00010282 [ 676.249855] RAX: 0000000000000001 RBX: 00000000ffff8b76 RCX: 0000000000000006 [ 676.250713] RDX: 0000000000000000 RSI: ffffffffbccf3a10 RDI: ffffffffbcc7b62e [ 676.251610] RBP: ffffa401c18ffc70 R08: 0000000000000001 R09: 0000000000000001 [ 676.252481] R10: 0000000000000001 R11: 0000000000000001 R12: 0000000000000005 [ 676.253421] R13: ffff8b76786ec370 R14: ffff8b76786ec370 R15: ffff8b76786ec480 [ 676.254257] FS: 0000000000000000(0000) GS:ffff8b7777800000(0000) knlGS:0000000000000000 [ 676.255239] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 676.255897] CR2: 00005590205d88b8 CR3: 000000017656c003 CR4: 0000000000770ee0 [ 676.256734] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 676.257567] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 676.258397] PKRU: 55555554 [ 676.258729] Call Trace: [ 676.259063] <TASK> [ 676.259354] dlm_midcomms_commit_mhandle+0xcc/0x110 [ 676.259964] queue_bast+0x8b/0xb0 [ 676.260423] grant_pending_locks+0x166/0x1b0 [ 676.261007] _unlock_lock+0x75/0x90 [ 676.261469] unlock_lock.isra.57+0x62/0xa0 [ 676.262009] dlm_unlock+0x21e/0x330 [ 676.262457] ? lock_torture_stats+0x80/0x80 [dlm_locktorture] [ 676.263183] torture_unlock+0x5a/0x90 [dlm_locktorture] [ 676.263815] ? preempt_count_sub+0xba/0x100 [ 676.264361] ? complete+0x1d/0x60 [ 676.264777] lock_torture_writer+0xb8/0x150 [dlm_locktorture] [ 676.265555] kthread+0x10a/0x130 [ 676.266007] ? kthread_complete_and_exit+0x20/0x20 [ 676.266616] ret_from_fork+0x22/0x30 [ 676.267097] </TASK> [ 676.267381] irq event stamp: 9579855 [ 676.267824] hardirqs last enabled at (9579863): [<ffffffffbb14e6f8>] __up_console_sem+0x58/0x60 [ 676.268896] hardirqs last disabled at (9579872): [<ffffffffbb14e6dd>] __up_console_sem+0x3d/0x60 [ 676.270008] softirqs last enabled at (9579798): [<ffffffffbc200349>] __do_softirq+0x349/0x4c7 [ 676.271438] softirqs last disabled at (9579897): [<ffffffffbb0d54c0>] irq_exit_rcu+0xb0/0xf0 [ 676.272796] ---[ end trace 0000000000000000 ]--- I reproduced this warning with dlm_locktorture test which is currently not upstream. However this patch fix the issue by make a additional refcount between dlm_lowcomms_new_msg() and dlm_lowcomms_commit_msg(). In case of the race the kref_put() in dlm_lowcomms_commit_msg() will be the final put. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2022-08-15 19:43:13 +00:00
/* for dlm_lowcomms_commit_msg() */
kref_get(&msg->ref);
/* we assume if successful commit must called */
msg->idx = idx;
return msg;
}
#endif
static void _dlm_lowcomms_commit_msg(struct dlm_msg *msg)
{
struct writequeue_entry *e = msg->entry;
struct connection *con = e->con;
int users;
spin_lock(&con->writequeue_lock);
kref_get(&msg->ref);
list_add(&msg->list, &e->msgs);
users = --e->users;
if (users)
goto out;
e->len = DLM_WQ_LENGTH_BYTES(e);
spin_unlock(&con->writequeue_lock);
queue_work(send_workqueue, &con->swork);
return;
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
out:
spin_unlock(&con->writequeue_lock);
return;
}
/* avoid false positive for nodes_srcu, lock was happen in
* dlm_lowcomms_new_msg
*/
#ifndef __CHECKER__
void dlm_lowcomms_commit_msg(struct dlm_msg *msg)
{
_dlm_lowcomms_commit_msg(msg);
srcu_read_unlock(&connections_srcu, msg->idx);
fs: dlm: fix race in lowcomms This patch fixes a race between queue_work() in _dlm_lowcomms_commit_msg() and srcu_read_unlock(). The queue_work() can take the final reference of a dlm_msg and so msg->idx can contain garbage which is signaled by the following warning: [ 676.237050] ------------[ cut here ]------------ [ 676.237052] WARNING: CPU: 0 PID: 1060 at include/linux/srcu.h:189 dlm_lowcomms_commit_msg+0x41/0x50 [ 676.238945] Modules linked in: dlm_locktorture torture rpcsec_gss_krb5 intel_rapl_msr intel_rapl_common iTCO_wdt iTCO_vendor_support qxl kvm_intel drm_ttm_helper vmw_vsock_virtio_transport kvm vmw_vsock_virtio_transport_common ttm irqbypass crc32_pclmul joydev crc32c_intel serio_raw drm_kms_helper vsock virtio_scsi virtio_console virtio_balloon snd_pcm drm syscopyarea sysfillrect sysimgblt snd_timer fb_sys_fops i2c_i801 lpc_ich snd i2c_smbus soundcore pcspkr [ 676.244227] CPU: 0 PID: 1060 Comm: lock_torture_wr Not tainted 5.19.0-rc3+ #1546 [ 676.245216] Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.16.0-2.module+el8.7.0+15506+033991b0 04/01/2014 [ 676.246460] RIP: 0010:dlm_lowcomms_commit_msg+0x41/0x50 [ 676.247132] Code: fe ff ff ff 75 24 48 c7 c6 bd 0f 49 bb 48 c7 c7 38 7c 01 bd e8 00 e7 ca ff 89 de 48 c7 c7 60 78 01 bd e8 42 3d cd ff 5b 5d c3 <0f> 0b eb d8 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 [ 676.249253] RSP: 0018:ffffa401c18ffc68 EFLAGS: 00010282 [ 676.249855] RAX: 0000000000000001 RBX: 00000000ffff8b76 RCX: 0000000000000006 [ 676.250713] RDX: 0000000000000000 RSI: ffffffffbccf3a10 RDI: ffffffffbcc7b62e [ 676.251610] RBP: ffffa401c18ffc70 R08: 0000000000000001 R09: 0000000000000001 [ 676.252481] R10: 0000000000000001 R11: 0000000000000001 R12: 0000000000000005 [ 676.253421] R13: ffff8b76786ec370 R14: ffff8b76786ec370 R15: ffff8b76786ec480 [ 676.254257] FS: 0000000000000000(0000) GS:ffff8b7777800000(0000) knlGS:0000000000000000 [ 676.255239] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 676.255897] CR2: 00005590205d88b8 CR3: 000000017656c003 CR4: 0000000000770ee0 [ 676.256734] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 676.257567] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 676.258397] PKRU: 55555554 [ 676.258729] Call Trace: [ 676.259063] <TASK> [ 676.259354] dlm_midcomms_commit_mhandle+0xcc/0x110 [ 676.259964] queue_bast+0x8b/0xb0 [ 676.260423] grant_pending_locks+0x166/0x1b0 [ 676.261007] _unlock_lock+0x75/0x90 [ 676.261469] unlock_lock.isra.57+0x62/0xa0 [ 676.262009] dlm_unlock+0x21e/0x330 [ 676.262457] ? lock_torture_stats+0x80/0x80 [dlm_locktorture] [ 676.263183] torture_unlock+0x5a/0x90 [dlm_locktorture] [ 676.263815] ? preempt_count_sub+0xba/0x100 [ 676.264361] ? complete+0x1d/0x60 [ 676.264777] lock_torture_writer+0xb8/0x150 [dlm_locktorture] [ 676.265555] kthread+0x10a/0x130 [ 676.266007] ? kthread_complete_and_exit+0x20/0x20 [ 676.266616] ret_from_fork+0x22/0x30 [ 676.267097] </TASK> [ 676.267381] irq event stamp: 9579855 [ 676.267824] hardirqs last enabled at (9579863): [<ffffffffbb14e6f8>] __up_console_sem+0x58/0x60 [ 676.268896] hardirqs last disabled at (9579872): [<ffffffffbb14e6dd>] __up_console_sem+0x3d/0x60 [ 676.270008] softirqs last enabled at (9579798): [<ffffffffbc200349>] __do_softirq+0x349/0x4c7 [ 676.271438] softirqs last disabled at (9579897): [<ffffffffbb0d54c0>] irq_exit_rcu+0xb0/0xf0 [ 676.272796] ---[ end trace 0000000000000000 ]--- I reproduced this warning with dlm_locktorture test which is currently not upstream. However this patch fix the issue by make a additional refcount between dlm_lowcomms_new_msg() and dlm_lowcomms_commit_msg(). In case of the race the kref_put() in dlm_lowcomms_commit_msg() will be the final put. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2022-08-15 19:43:13 +00:00
/* because dlm_lowcomms_new_msg() */
kref_put(&msg->ref, dlm_msg_release);
}
#endif
void dlm_lowcomms_put_msg(struct dlm_msg *msg)
{
kref_put(&msg->ref, dlm_msg_release);
}
/* does not held connections_srcu, usage workqueue only */
int dlm_lowcomms_resend_msg(struct dlm_msg *msg)
{
struct dlm_msg *msg_resend;
char *ppc;
if (msg->retransmit)
return 1;
msg_resend = dlm_lowcomms_new_msg_con(msg->entry->con, msg->len,
GFP_ATOMIC, &ppc, NULL, NULL);
if (!msg_resend)
return -ENOMEM;
msg->retransmit = true;
kref_get(&msg->ref);
msg_resend->orig_msg = msg;
memcpy(ppc, msg->ppc, msg->len);
_dlm_lowcomms_commit_msg(msg_resend);
dlm_lowcomms_put_msg(msg_resend);
return 0;
}
/* Send a message */
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
static void send_to_sock(struct connection *con)
{
const int msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL;
struct writequeue_entry *e;
int len, offset, ret;
int count = 0;
mutex_lock(&con->sock_mutex);
if (con->sock == NULL)
goto out_connect;
spin_lock(&con->writequeue_lock);
for (;;) {
e = con_next_wq(con);
if (!e)
break;
len = e->len;
offset = e->offset;
BUG_ON(len == 0 && e->users == 0);
spin_unlock(&con->writequeue_lock);
ret = kernel_sendpage(con->sock, e->page, offset, len,
msg_flags);
trace_dlm_send(con->nodeid, ret);
if (ret == -EAGAIN || ret == 0) {
if (ret == -EAGAIN &&
test_bit(SOCKWQ_ASYNC_NOSPACE, &con->sock->flags) &&
!test_and_set_bit(CF_APP_LIMITED, &con->flags)) {
/* Notify TCP that we're limited by the
* application window size.
*/
set_bit(SOCK_NOSPACE, &con->sock->flags);
con->sock->sk->sk_write_pending++;
}
cond_resched();
goto out;
} else if (ret < 0)
goto out;
/* Don't starve people filling buffers */
if (++count >= MAX_SEND_MSG_COUNT) {
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
cond_resched();
count = 0;
}
spin_lock(&con->writequeue_lock);
writequeue_entry_complete(e, ret);
}
spin_unlock(&con->writequeue_lock);
/* close if we got EOF */
if (test_and_clear_bit(CF_EOF, &con->flags)) {
mutex_unlock(&con->sock_mutex);
close_connection(con, false, false, true);
/* handling for tcp shutdown */
clear_bit(CF_SHUTDOWN, &con->flags);
wake_up(&con->shutdown_wait);
} else {
mutex_unlock(&con->sock_mutex);
}
return;
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
out:
mutex_unlock(&con->sock_mutex);
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
return;
[DLM] Clean up lowcomms This fixes up most of the things pointed out by akpm and Pavel Machek with comments below indicating why some things have been left: Andrew Morton wrote: > >> +static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc) >> +{ >> + struct nodeinfo *ni; >> + int r; >> + int n; >> + >> + down_read(&nodeinfo_lock); > > Given that this function can sleep, I wonder if `alloc' is useful. > > I see lots of callers passing in a literal "0" for `alloc'. That's in fact > a secret (GFP_ATOMIC & ~__GFP_HIGH). I doubt if that's what you really > meant. Particularly as the code could at least have used __GFP_WAIT (aka > GFP_NOIO) which is much, much more reliable than "0". In fact "0" is the > least reliable mode possible. > > IOW, this is all bollixed up. When 0 is passed into nodeid2nodeinfo the function does not try to allocate a new structure at all. it's an indication that the caller only wants the nodeinfo struct for that nodeid if there actually is one in existance. I've tidied the function itself so it's more obvious, (and tidier!) >> +/* Data received from remote end */ >> +static int receive_from_sock(void) >> +{ >> + int ret = 0; >> + struct msghdr msg; >> + struct kvec iov[2]; >> + unsigned len; >> + int r; >> + struct sctp_sndrcvinfo *sinfo; >> + struct cmsghdr *cmsg; >> + struct nodeinfo *ni; >> + >> + /* These two are marginally too big for stack allocation, but this >> + * function is (currently) only called by dlm_recvd so static should be >> + * OK. >> + */ >> + static struct sockaddr_storage msgname; >> + static char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > whoa. This is globally singly-threaded code?? Yes. it is only ever run in the context of dlm_recvd. >> >> +static void initiate_association(int nodeid) >> +{ >> + struct sockaddr_storage rem_addr; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Another static buffer to worry about. Globally singly-threaded code? Yes. Only ever called by dlm_sendd. >> + >> +/* Send a message */ >> +static int send_to_sock(struct nodeinfo *ni) >> +{ >> + int ret = 0; >> + struct writequeue_entry *e; >> + int len, offset; >> + struct msghdr outmsg; >> + static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))]; > > Singly-threaded? Yep. >> >> +static void dealloc_nodeinfo(void) >> +{ >> + int i; >> + >> + for (i=1; i<=max_nodeid; i++) { >> + struct nodeinfo *ni = nodeid2nodeinfo(i, 0); >> + if (ni) { >> + idr_remove(&nodeinfo_idr, i); > > Didn't that need locking? Not. it's only ever called at DLM shutdown after all the other threads have been stopped. >> >> +static int write_list_empty(void) >> +{ >> + int status; >> + >> + spin_lock_bh(&write_nodes_lock); >> + status = list_empty(&write_nodes); >> + spin_unlock_bh(&write_nodes_lock); >> + >> + return status; >> +} > > This function's return value is meaningless. As soon as the lock gets > dropped, the return value can get out of sync with reality. > > Looking at the caller, this _might_ happen to be OK, but it's a nasty and > dangerous thing. Really the locking should be moved into the caller. It's just an optimisation to allow the caller to schedule if there is no work to do. if something arrives immediately afterwards then it will get picked up when the process re-awakes (and it will be woken by that arrival). The 'accepting' atomic has gone completely. as Andrew pointed out it didn't really achieve much anyway. I suspect it was a plaster over some other startup or shutdown bug to be honest. Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com> Signed-off-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Andrew Morton <akpm@osdl.org> Cc: Pavel Machek <pavel@ucw.cz>
2006-12-06 15:10:37 +00:00
out_connect:
mutex_unlock(&con->sock_mutex);
queue_work(send_workqueue, &con->swork);
cond_resched();
}
static void clean_one_writequeue(struct connection *con)
{
struct writequeue_entry *e, *safe;
spin_lock(&con->writequeue_lock);
list_for_each_entry_safe(e, safe, &con->writequeue, list) {
free_entry(e);
}
spin_unlock(&con->writequeue_lock);
}
/* Called from recovery when it knows that a node has
left the cluster */
int dlm_lowcomms_close(int nodeid)
{
struct connection *con;
struct dlm_node_addr *na;
int idx;
log_print("closing connection to node %d", nodeid);
idx = srcu_read_lock(&connections_srcu);
con = nodeid2con(nodeid, 0);
if (con) {
set_bit(CF_CLOSE, &con->flags);
close_connection(con, true, true, true);
clean_one_writequeue(con);
if (con->othercon)
clean_one_writequeue(con->othercon);
}
srcu_read_unlock(&connections_srcu, idx);
spin_lock(&dlm_node_addrs_spin);
na = find_node_addr(nodeid);
if (na) {
list_del(&na->list);
while (na->addr_count--)
kfree(na->addr[na->addr_count]);
kfree(na);
}
spin_unlock(&dlm_node_addrs_spin);
return 0;
}
/* Receive workqueue function */
static void process_recv_sockets(struct work_struct *work)
{
struct connection *con = container_of(work, struct connection, rwork);
clear_bit(CF_READ_PENDING, &con->flags);
receive_from_sock(con);
}
static void process_listen_recv_socket(struct work_struct *work)
{
accept_from_sock(&listen_con);
}
static void dlm_connect(struct connection *con)
{
struct sockaddr_storage addr;
int result, addr_len;
struct socket *sock;
unsigned int mark;
/* Some odd races can cause double-connects, ignore them */
if (con->retries++ > MAX_CONNECT_RETRIES)
return;
if (con->sock) {
log_print("node %d already connected.", con->nodeid);
return;
}
memset(&addr, 0, sizeof(addr));
result = nodeid_to_addr(con->nodeid, &addr, NULL,
dlm_proto_ops->try_new_addr, &mark);
if (result < 0) {
log_print("no address for nodeid %d", con->nodeid);
return;
}
/* Create a socket to communicate with */
result = sock_create_kern(&init_net, dlm_local_addr[0]->ss_family,
SOCK_STREAM, dlm_proto_ops->proto, &sock);
if (result < 0)
goto socket_err;
sock_set_mark(sock->sk, mark);
dlm_proto_ops->sockopts(sock);
add_sock(sock, con);
result = dlm_proto_ops->bind(sock);
if (result < 0)
goto add_sock_err;
log_print_ratelimited("connecting to %d", con->nodeid);
make_sockaddr(&addr, dlm_config.ci_tcp_port, &addr_len);
result = dlm_proto_ops->connect(con, sock, (struct sockaddr *)&addr,
addr_len);
if (result < 0)
goto add_sock_err;
return;
add_sock_err:
dlm_close_sock(&con->sock);
socket_err:
/*
* Some errors are fatal and this list might need adjusting. For other
* errors we try again until the max number of retries is reached.
*/
if (result != -EHOSTUNREACH &&
result != -ENETUNREACH &&
result != -ENETDOWN &&
result != -EINVAL &&
result != -EPROTONOSUPPORT) {
log_print("connect %d try %d error %d", con->nodeid,
con->retries, result);
msleep(1000);
lowcomms_connect_sock(con);
}
}
/* Send workqueue function */
static void process_send_sockets(struct work_struct *work)
{
struct connection *con = container_of(work, struct connection, swork);
WARN_ON(test_bit(CF_IS_OTHERCON, &con->flags));
clear_bit(CF_WRITE_PENDING, &con->flags);
fs: dlm: add reliable connection if reconnect This patch introduce to make a tcp lowcomms connection reliable even if reconnects occurs. This is done by an application layer re-transmission handling and sequence numbers in dlm protocols. There are three new dlm commands: DLM_OPTS: This will encapsulate an existing dlm message (and rcom message if they don't have an own application side re-transmission handling). As optional handling additional tlv's (type length fields) can be appended. This can be for example a sequence number field. However because in DLM_OPTS the lockspace field is unused and a sequence number is a mandatory field it isn't made as a tlv and we put the sequence number inside the lockspace id. The possibility to add optional options are still there for future purposes. DLM_ACK: Just a dlm header to acknowledge the receive of a DLM_OPTS message to it's sender. DLM_FIN: This provides a 4 way handshake for connection termination inclusive support for half-closed connections. It's provided on application layer because SCTP doesn't support half-closed sockets, the shutdown() call can interrupted by e.g. TCP resets itself and a hard logic to implement it because the othercon paradigm in lowcomms. The 4-way termination handshake also solve problems to synchronize peer EOF arrival and that the cluster manager removes the peer in the node membership handling of DLM. In some cases messages can be still transmitted in this time and we need to wait for the node membership event. To provide a reliable connection the node will retransmit all unacknowledges message to it's peer on reconnect. The receiver will then filtering out the next received message and drop all messages which are duplicates. As RCOM_STATUS and RCOM_NAMES messages are the first messages which are exchanged and they have they own re-transmission handling, there exists logic that these messages must be first. If these messages arrives we store the dlm version field. This handling is on DLM 3.1 and after this patch 3.2 the same. A backwards compatibility handling has been added which seems to work on tests without tcpkill, however it's not recommended to use DLM 3.1 and 3.2 at the same time, because DLM 3.2 tries to fix long term bugs in the DLM protocol. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-05-21 19:08:46 +00:00
if (test_and_clear_bit(CF_RECONNECT, &con->flags)) {
close_connection(con, false, false, true);
fs: dlm: add reliable connection if reconnect This patch introduce to make a tcp lowcomms connection reliable even if reconnects occurs. This is done by an application layer re-transmission handling and sequence numbers in dlm protocols. There are three new dlm commands: DLM_OPTS: This will encapsulate an existing dlm message (and rcom message if they don't have an own application side re-transmission handling). As optional handling additional tlv's (type length fields) can be appended. This can be for example a sequence number field. However because in DLM_OPTS the lockspace field is unused and a sequence number is a mandatory field it isn't made as a tlv and we put the sequence number inside the lockspace id. The possibility to add optional options are still there for future purposes. DLM_ACK: Just a dlm header to acknowledge the receive of a DLM_OPTS message to it's sender. DLM_FIN: This provides a 4 way handshake for connection termination inclusive support for half-closed connections. It's provided on application layer because SCTP doesn't support half-closed sockets, the shutdown() call can interrupted by e.g. TCP resets itself and a hard logic to implement it because the othercon paradigm in lowcomms. The 4-way termination handshake also solve problems to synchronize peer EOF arrival and that the cluster manager removes the peer in the node membership handling of DLM. In some cases messages can be still transmitted in this time and we need to wait for the node membership event. To provide a reliable connection the node will retransmit all unacknowledges message to it's peer on reconnect. The receiver will then filtering out the next received message and drop all messages which are duplicates. As RCOM_STATUS and RCOM_NAMES messages are the first messages which are exchanged and they have they own re-transmission handling, there exists logic that these messages must be first. If these messages arrives we store the dlm version field. This handling is on DLM 3.1 and after this patch 3.2 the same. A backwards compatibility handling has been added which seems to work on tests without tcpkill, however it's not recommended to use DLM 3.1 and 3.2 at the same time, because DLM 3.2 tries to fix long term bugs in the DLM protocol. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-05-21 19:08:46 +00:00
dlm_midcomms_unack_msg_resend(con->nodeid);
}
if (con->sock == NULL) {
if (test_and_clear_bit(CF_DELAY_CONNECT, &con->flags))
msleep(1000);
mutex_lock(&con->sock_mutex);
dlm_connect(con);
mutex_unlock(&con->sock_mutex);
}
if (!list_empty(&con->writequeue))
send_to_sock(con);
}
static void work_stop(void)
{
if (recv_workqueue) {
destroy_workqueue(recv_workqueue);
recv_workqueue = NULL;
}
if (send_workqueue) {
destroy_workqueue(send_workqueue);
send_workqueue = NULL;
}
}
static int work_start(void)
{
recv_workqueue = alloc_ordered_workqueue("dlm_recv", WQ_MEM_RECLAIM);
if (!recv_workqueue) {
log_print("can't start dlm_recv");
return -ENOMEM;
}
send_workqueue = alloc_ordered_workqueue("dlm_send", WQ_MEM_RECLAIM);
if (!send_workqueue) {
log_print("can't start dlm_send");
destroy_workqueue(recv_workqueue);
recv_workqueue = NULL;
return -ENOMEM;
}
return 0;
}
fs: dlm: add shutdown hook This patch fixes issues which occurs when dlm lowcomms synchronize their workqueues but dlm application layer already released the lockspace. In such cases messages like: dlm: gfs2: release_lockspace final free dlm: invalid lockspace 3841231384 from 1 cmd 1 type 11 are printed on the kernel log. This patch is solving this issue by introducing a new "shutdown" hook before calling "stop" hook when the lockspace is going to be released finally. This should pretend any dlm messages sitting in the workqueues during or after lockspace removal. It's necessary to call dlm_scand_stop() as I instrumented dlm_lowcomms_get_buffer() code to report a warning after it's called after dlm_midcomms_shutdown() functionality, see below: WARNING: CPU: 1 PID: 3794 at fs/dlm/midcomms.c:1003 dlm_midcomms_get_buffer+0x167/0x180 Modules linked in: joydev iTCO_wdt intel_pmc_bxt iTCO_vendor_support drm_ttm_helper ttm pcspkr serio_raw i2c_i801 i2c_smbus drm_kms_helper virtio_scsi lpc_ich virtio_balloon virtio_console xhci_pci xhci_pci_renesas cec qemu_fw_cfg drm [last unloaded: qxl] CPU: 1 PID: 3794 Comm: dlm_scand Tainted: G W 5.11.0+ #26 Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 RIP: 0010:dlm_midcomms_get_buffer+0x167/0x180 Code: 5d 41 5c 41 5d 41 5e 41 5f c3 0f 0b 45 31 e4 5b 5d 4c 89 e0 41 5c 41 5d 41 5e 41 5f c3 4c 89 e7 45 31 e4 e8 3b f1 ec ff eb 86 <0f> 0b 4c 89 e7 45 31 e4 e8 2c f1 ec ff e9 74 ff ff ff 0f 1f 80 00 RSP: 0018:ffffa81503f8fe60 EFLAGS: 00010202 RAX: 0000000000000008 RBX: ffff8f969827f200 RCX: 0000000000000001 RDX: 0000000000000000 RSI: ffffffffad1e89a0 RDI: ffff8f96a5294160 RBP: 0000000000000001 R08: 0000000000000000 R09: ffff8f96a250bc60 R10: 00000000000045d3 R11: 0000000000000000 R12: ffff8f96a250bc60 R13: ffffa81503f8fec8 R14: 0000000000000070 R15: 0000000000000c40 FS: 0000000000000000(0000) GS:ffff8f96fbc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000055aa3351c000 CR3: 000000010bf22000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: dlm_scan_rsbs+0x420/0x670 ? dlm_uevent+0x20/0x20 dlm_scand+0xbf/0xe0 kthread+0x13a/0x150 ? __kthread_bind_mask+0x60/0x60 ret_from_fork+0x22/0x30 To synchronize all dlm scand messages we stop it right before shutdown hook. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:20 +00:00
static void shutdown_conn(struct connection *con)
{
if (dlm_proto_ops->shutdown_action)
dlm_proto_ops->shutdown_action(con);
fs: dlm: add shutdown hook This patch fixes issues which occurs when dlm lowcomms synchronize their workqueues but dlm application layer already released the lockspace. In such cases messages like: dlm: gfs2: release_lockspace final free dlm: invalid lockspace 3841231384 from 1 cmd 1 type 11 are printed on the kernel log. This patch is solving this issue by introducing a new "shutdown" hook before calling "stop" hook when the lockspace is going to be released finally. This should pretend any dlm messages sitting in the workqueues during or after lockspace removal. It's necessary to call dlm_scand_stop() as I instrumented dlm_lowcomms_get_buffer() code to report a warning after it's called after dlm_midcomms_shutdown() functionality, see below: WARNING: CPU: 1 PID: 3794 at fs/dlm/midcomms.c:1003 dlm_midcomms_get_buffer+0x167/0x180 Modules linked in: joydev iTCO_wdt intel_pmc_bxt iTCO_vendor_support drm_ttm_helper ttm pcspkr serio_raw i2c_i801 i2c_smbus drm_kms_helper virtio_scsi lpc_ich virtio_balloon virtio_console xhci_pci xhci_pci_renesas cec qemu_fw_cfg drm [last unloaded: qxl] CPU: 1 PID: 3794 Comm: dlm_scand Tainted: G W 5.11.0+ #26 Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 RIP: 0010:dlm_midcomms_get_buffer+0x167/0x180 Code: 5d 41 5c 41 5d 41 5e 41 5f c3 0f 0b 45 31 e4 5b 5d 4c 89 e0 41 5c 41 5d 41 5e 41 5f c3 4c 89 e7 45 31 e4 e8 3b f1 ec ff eb 86 <0f> 0b 4c 89 e7 45 31 e4 e8 2c f1 ec ff e9 74 ff ff ff 0f 1f 80 00 RSP: 0018:ffffa81503f8fe60 EFLAGS: 00010202 RAX: 0000000000000008 RBX: ffff8f969827f200 RCX: 0000000000000001 RDX: 0000000000000000 RSI: ffffffffad1e89a0 RDI: ffff8f96a5294160 RBP: 0000000000000001 R08: 0000000000000000 R09: ffff8f96a250bc60 R10: 00000000000045d3 R11: 0000000000000000 R12: ffff8f96a250bc60 R13: ffffa81503f8fec8 R14: 0000000000000070 R15: 0000000000000c40 FS: 0000000000000000(0000) GS:ffff8f96fbc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000055aa3351c000 CR3: 000000010bf22000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: dlm_scan_rsbs+0x420/0x670 ? dlm_uevent+0x20/0x20 dlm_scand+0xbf/0xe0 kthread+0x13a/0x150 ? __kthread_bind_mask+0x60/0x60 ret_from_fork+0x22/0x30 To synchronize all dlm scand messages we stop it right before shutdown hook. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:20 +00:00
}
void dlm_lowcomms_shutdown(void)
{
int idx;
fs: dlm: add shutdown hook This patch fixes issues which occurs when dlm lowcomms synchronize their workqueues but dlm application layer already released the lockspace. In such cases messages like: dlm: gfs2: release_lockspace final free dlm: invalid lockspace 3841231384 from 1 cmd 1 type 11 are printed on the kernel log. This patch is solving this issue by introducing a new "shutdown" hook before calling "stop" hook when the lockspace is going to be released finally. This should pretend any dlm messages sitting in the workqueues during or after lockspace removal. It's necessary to call dlm_scand_stop() as I instrumented dlm_lowcomms_get_buffer() code to report a warning after it's called after dlm_midcomms_shutdown() functionality, see below: WARNING: CPU: 1 PID: 3794 at fs/dlm/midcomms.c:1003 dlm_midcomms_get_buffer+0x167/0x180 Modules linked in: joydev iTCO_wdt intel_pmc_bxt iTCO_vendor_support drm_ttm_helper ttm pcspkr serio_raw i2c_i801 i2c_smbus drm_kms_helper virtio_scsi lpc_ich virtio_balloon virtio_console xhci_pci xhci_pci_renesas cec qemu_fw_cfg drm [last unloaded: qxl] CPU: 1 PID: 3794 Comm: dlm_scand Tainted: G W 5.11.0+ #26 Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 RIP: 0010:dlm_midcomms_get_buffer+0x167/0x180 Code: 5d 41 5c 41 5d 41 5e 41 5f c3 0f 0b 45 31 e4 5b 5d 4c 89 e0 41 5c 41 5d 41 5e 41 5f c3 4c 89 e7 45 31 e4 e8 3b f1 ec ff eb 86 <0f> 0b 4c 89 e7 45 31 e4 e8 2c f1 ec ff e9 74 ff ff ff 0f 1f 80 00 RSP: 0018:ffffa81503f8fe60 EFLAGS: 00010202 RAX: 0000000000000008 RBX: ffff8f969827f200 RCX: 0000000000000001 RDX: 0000000000000000 RSI: ffffffffad1e89a0 RDI: ffff8f96a5294160 RBP: 0000000000000001 R08: 0000000000000000 R09: ffff8f96a250bc60 R10: 00000000000045d3 R11: 0000000000000000 R12: ffff8f96a250bc60 R13: ffffa81503f8fec8 R14: 0000000000000070 R15: 0000000000000c40 FS: 0000000000000000(0000) GS:ffff8f96fbc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000055aa3351c000 CR3: 000000010bf22000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: dlm_scan_rsbs+0x420/0x670 ? dlm_uevent+0x20/0x20 dlm_scand+0xbf/0xe0 kthread+0x13a/0x150 ? __kthread_bind_mask+0x60/0x60 ret_from_fork+0x22/0x30 To synchronize all dlm scand messages we stop it right before shutdown hook. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:20 +00:00
/* Set all the flags to prevent any
* socket activity.
*/
dlm_allow_conn = 0;
if (recv_workqueue)
flush_workqueue(recv_workqueue);
if (send_workqueue)
flush_workqueue(send_workqueue);
dlm_close_sock(&listen_con.sock);
idx = srcu_read_lock(&connections_srcu);
fs: dlm: add shutdown hook This patch fixes issues which occurs when dlm lowcomms synchronize their workqueues but dlm application layer already released the lockspace. In such cases messages like: dlm: gfs2: release_lockspace final free dlm: invalid lockspace 3841231384 from 1 cmd 1 type 11 are printed on the kernel log. This patch is solving this issue by introducing a new "shutdown" hook before calling "stop" hook when the lockspace is going to be released finally. This should pretend any dlm messages sitting in the workqueues during or after lockspace removal. It's necessary to call dlm_scand_stop() as I instrumented dlm_lowcomms_get_buffer() code to report a warning after it's called after dlm_midcomms_shutdown() functionality, see below: WARNING: CPU: 1 PID: 3794 at fs/dlm/midcomms.c:1003 dlm_midcomms_get_buffer+0x167/0x180 Modules linked in: joydev iTCO_wdt intel_pmc_bxt iTCO_vendor_support drm_ttm_helper ttm pcspkr serio_raw i2c_i801 i2c_smbus drm_kms_helper virtio_scsi lpc_ich virtio_balloon virtio_console xhci_pci xhci_pci_renesas cec qemu_fw_cfg drm [last unloaded: qxl] CPU: 1 PID: 3794 Comm: dlm_scand Tainted: G W 5.11.0+ #26 Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 RIP: 0010:dlm_midcomms_get_buffer+0x167/0x180 Code: 5d 41 5c 41 5d 41 5e 41 5f c3 0f 0b 45 31 e4 5b 5d 4c 89 e0 41 5c 41 5d 41 5e 41 5f c3 4c 89 e7 45 31 e4 e8 3b f1 ec ff eb 86 <0f> 0b 4c 89 e7 45 31 e4 e8 2c f1 ec ff e9 74 ff ff ff 0f 1f 80 00 RSP: 0018:ffffa81503f8fe60 EFLAGS: 00010202 RAX: 0000000000000008 RBX: ffff8f969827f200 RCX: 0000000000000001 RDX: 0000000000000000 RSI: ffffffffad1e89a0 RDI: ffff8f96a5294160 RBP: 0000000000000001 R08: 0000000000000000 R09: ffff8f96a250bc60 R10: 00000000000045d3 R11: 0000000000000000 R12: ffff8f96a250bc60 R13: ffffa81503f8fec8 R14: 0000000000000070 R15: 0000000000000c40 FS: 0000000000000000(0000) GS:ffff8f96fbc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000055aa3351c000 CR3: 000000010bf22000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: dlm_scan_rsbs+0x420/0x670 ? dlm_uevent+0x20/0x20 dlm_scand+0xbf/0xe0 kthread+0x13a/0x150 ? __kthread_bind_mask+0x60/0x60 ret_from_fork+0x22/0x30 To synchronize all dlm scand messages we stop it right before shutdown hook. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:20 +00:00
foreach_conn(shutdown_conn);
srcu_read_unlock(&connections_srcu, idx);
fs: dlm: add shutdown hook This patch fixes issues which occurs when dlm lowcomms synchronize their workqueues but dlm application layer already released the lockspace. In such cases messages like: dlm: gfs2: release_lockspace final free dlm: invalid lockspace 3841231384 from 1 cmd 1 type 11 are printed on the kernel log. This patch is solving this issue by introducing a new "shutdown" hook before calling "stop" hook when the lockspace is going to be released finally. This should pretend any dlm messages sitting in the workqueues during or after lockspace removal. It's necessary to call dlm_scand_stop() as I instrumented dlm_lowcomms_get_buffer() code to report a warning after it's called after dlm_midcomms_shutdown() functionality, see below: WARNING: CPU: 1 PID: 3794 at fs/dlm/midcomms.c:1003 dlm_midcomms_get_buffer+0x167/0x180 Modules linked in: joydev iTCO_wdt intel_pmc_bxt iTCO_vendor_support drm_ttm_helper ttm pcspkr serio_raw i2c_i801 i2c_smbus drm_kms_helper virtio_scsi lpc_ich virtio_balloon virtio_console xhci_pci xhci_pci_renesas cec qemu_fw_cfg drm [last unloaded: qxl] CPU: 1 PID: 3794 Comm: dlm_scand Tainted: G W 5.11.0+ #26 Hardware name: Red Hat KVM/RHEL-AV, BIOS 1.13.0-2.module+el8.3.0+7353+9de0a3cc 04/01/2014 RIP: 0010:dlm_midcomms_get_buffer+0x167/0x180 Code: 5d 41 5c 41 5d 41 5e 41 5f c3 0f 0b 45 31 e4 5b 5d 4c 89 e0 41 5c 41 5d 41 5e 41 5f c3 4c 89 e7 45 31 e4 e8 3b f1 ec ff eb 86 <0f> 0b 4c 89 e7 45 31 e4 e8 2c f1 ec ff e9 74 ff ff ff 0f 1f 80 00 RSP: 0018:ffffa81503f8fe60 EFLAGS: 00010202 RAX: 0000000000000008 RBX: ffff8f969827f200 RCX: 0000000000000001 RDX: 0000000000000000 RSI: ffffffffad1e89a0 RDI: ffff8f96a5294160 RBP: 0000000000000001 R08: 0000000000000000 R09: ffff8f96a250bc60 R10: 00000000000045d3 R11: 0000000000000000 R12: ffff8f96a250bc60 R13: ffffa81503f8fec8 R14: 0000000000000070 R15: 0000000000000c40 FS: 0000000000000000(0000) GS:ffff8f96fbc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000055aa3351c000 CR3: 000000010bf22000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: dlm_scan_rsbs+0x420/0x670 ? dlm_uevent+0x20/0x20 dlm_scand+0xbf/0xe0 kthread+0x13a/0x150 ? __kthread_bind_mask+0x60/0x60 ret_from_fork+0x22/0x30 To synchronize all dlm scand messages we stop it right before shutdown hook. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2021-03-01 22:05:20 +00:00
}
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
static void _stop_conn(struct connection *con, bool and_other)
{
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
mutex_lock(&con->sock_mutex);
set_bit(CF_CLOSE, &con->flags);
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
set_bit(CF_READ_PENDING, &con->flags);
set_bit(CF_WRITE_PENDING, &con->flags);
if (con->sock && con->sock->sk) {
lock_sock(con->sock->sk);
con->sock->sk->sk_user_data = NULL;
release_sock(con->sock->sk);
}
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
if (con->othercon && and_other)
_stop_conn(con->othercon, false);
mutex_unlock(&con->sock_mutex);
}
static void stop_conn(struct connection *con)
{
_stop_conn(con, true);
}
fs: dlm: rework receive handling This patch reworks the current receive handling of dlm. As I tried to change the send handling to fix reorder issues I took a look into the receive handling and simplified it, it works as the following: Each connection has a preallocated receive buffer with a minimum length of 4096. On receive, the upper layer protocol will process all dlm message until there is not enough data anymore. If there exists "leftover" data at the end of the receive buffer because the dlm message wasn't fully received it will be copied to the begin of the preallocated receive buffer. Next receive more data will be appended to the previous "leftover" data and processing will begin again. This will remove a lot of code of the current mechanism. Inside the processing functionality we will ensure with a memmove() that the dlm message should be memory aligned. To have a dlm message always started at the beginning of the buffer will reduce some amount of memmove() calls because src and dest pointers are the same. The cluster attribute "buffer_size" becomes a new meaning, it's now the size of application layer receive buffer size. If this is changed during runtime the receive buffer will be reallocated. It's important that the receive buffer size has at minimum the size of the maximum possible dlm message size otherwise the received message cannot be placed inside the receive buffer size. Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com>
2020-09-24 14:31:26 +00:00
static void connection_release(struct rcu_head *rcu)
{
struct connection *con = container_of(rcu, struct connection, rcu);
kfree(con->rx_buf);
kfree(con);
}
static void free_conn(struct connection *con)
{
close_connection(con, true, true, true);
spin_lock(&connections_lock);
hlist_del_rcu(&con->list);
spin_unlock(&connections_lock);
if (con->othercon) {
clean_one_writequeue(con->othercon);
call_srcu(&connections_srcu, &con->othercon->rcu,
connection_release);
}
clean_one_writequeue(con);
call_srcu(&connections_srcu, &con->rcu, connection_release);
}
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
static void work_flush(void)
{
int ok;
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
int i;
struct connection *con;
do {
ok = 1;
foreach_conn(stop_conn);
if (recv_workqueue)
flush_workqueue(recv_workqueue);
if (send_workqueue)
flush_workqueue(send_workqueue);
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
for (i = 0; i < CONN_HASH_SIZE && ok; i++) {
hlist_for_each_entry_rcu(con, &connection_hash[i],
list) {
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
ok &= test_bit(CF_READ_PENDING, &con->flags);
ok &= test_bit(CF_WRITE_PENDING, &con->flags);
if (con->othercon) {
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
ok &= test_bit(CF_READ_PENDING,
&con->othercon->flags);
ok &= test_bit(CF_WRITE_PENDING,
&con->othercon->flags);
}
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
}
}
} while (!ok);
}
void dlm_lowcomms_stop(void)
{
int idx;
idx = srcu_read_lock(&connections_srcu);
DLM: fix double list_del() dlm_lowcomms_stop() was not functioning properly. Correctly, we have to wait until all processing is finished with send_workqueue and recv_workqueue. This problem causes the following issue. Senario is 1. dlm_send thread: send_to_sock refers con->writequeue 2. main thread: dlm_lowcomms_stop calls list_del 3. dlm_send thread: send_to_sock calls list_del in writequeue_entry_complete [ 1925.770305] dlm: canceled swork for node 4 [ 1925.772374] general protection fault: 0000 [#1] SMP [ 1925.777930] Modules linked in: ocfs2_stack_user ocfs2 ocfs2_nodemanager ocfs2_stackglue dlm fmxnet(O) fmx_api(O) fmx_cu(O) igb(O) kvm_intel kvm irqbypass autofs4 [ 1925.794131] CPU: 3 PID: 6994 Comm: kworker/u8:0 Tainted: G O 4.4.39 #1 [ 1925.802684] Hardware name: TOSHIBA OX/OX, BIOS OX-P0015 12/03/2015 [ 1925.809595] Workqueue: dlm_send process_send_sockets [dlm] [ 1925.815714] task: ffff8804398d3c00 ti: ffff88046910c000 task.ti: ffff88046910c000 [ 1925.824072] RIP: 0010:[<ffffffffa04bd158>] [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1925.834480] RSP: 0018:ffff88046910fde0 EFLAGS: 00010246 [ 1925.840411] RAX: dead000000000200 RBX: 0000000000000001 RCX: 000000000000000a [ 1925.848372] RDX: ffff88046bd980c0 RSI: 0000000000000000 RDI: ffff8804673c5670 [ 1925.856341] RBP: ffff88046910fe20 R08: 00000000000000c9 R09: 0000000000000010 [ 1925.864311] R10: ffffffff81e22fc0 R11: 0000000000000000 R12: ffff8804673c56d8 [ 1925.872281] R13: ffff8804673c5660 R14: ffff88046bd98440 R15: 0000000000000058 [ 1925.880251] FS: 0000000000000000(0000) GS:ffff88047fd80000(0000) knlGS:0000000000000000 [ 1925.889280] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 1925.895694] CR2: 00007fff09eadf58 CR3: 00000004690f5000 CR4: 00000000001006e0 [ 1925.903663] Stack: [ 1925.905903] ffff8804673c5630 ffff8804673c5620 ffff8804673c5670 ffff88007d219b40 [ 1925.914181] ffff88046f095800 0000000000000100 ffff8800717a1400 ffff8804673c56d8 [ 1925.922459] ffff88046910fe60 ffffffff81073db2 00ff880400000000 ffff88007d219b40 [ 1925.930736] Call Trace: [ 1925.933468] [<ffffffff81073db2>] process_one_work+0x162/0x450 [ 1925.939983] [<ffffffff81074459>] worker_thread+0x69/0x4a0 [ 1925.946109] [<ffffffff810743f0>] ? rescuer_thread+0x350/0x350 [ 1925.952622] [<ffffffff8107956f>] kthread+0xef/0x110 [ 1925.958165] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.964283] [<ffffffff8186ab2f>] ret_from_fork+0x3f/0x70 [ 1925.970312] [<ffffffff81079480>] ? kthread_park+0x60/0x60 [ 1925.976436] Code: 01 00 00 48 8b 7d d0 e8 07 d3 3a e1 45 01 7e 18 45 29 7e 1c 75 ab 41 8b 46 24 85 c0 75 a3 49 8b 16 49 8b 46 08 31 f6 48 89 42 08 <48> 89 10 48 b8 00 01 00 00 00 00 ad de 49 8b 7e 10 49 89 06 66 [ 1925.997791] RIP [<ffffffffa04bd158>] process_send_sockets+0xf8/0x280 [dlm] [ 1926.005577] RSP <ffff88046910fde0> Signed-off-by: Tadashi Miyauchi <miyauchi@toshiba-tops.co.jp> Signed-off-by: Tsutomu Owa <tsutomu.owa@toshiba.co.jp> Signed-off-by: David Teigland <teigland@redhat.com>
2017-09-12 08:55:40 +00:00
work_flush();
foreach_conn(free_conn);
srcu_read_unlock(&connections_srcu, idx);
work_stop();
deinit_local();
dlm_proto_ops = NULL;
}
static int dlm_listen_for_all(void)
{
struct socket *sock;
int result;
log_print("Using %s for communications",
dlm_proto_ops->name);
result = dlm_proto_ops->listen_validate();
if (result < 0)
return result;
result = sock_create_kern(&init_net, dlm_local_addr[0]->ss_family,
SOCK_STREAM, dlm_proto_ops->proto, &sock);
if (result < 0) {
log_print("Can't create comms socket: %d", result);
return result;
}
sock_set_mark(sock->sk, dlm_config.ci_mark);
dlm_proto_ops->listen_sockopts(sock);
result = dlm_proto_ops->listen_bind(sock);
if (result < 0)
goto out;
save_listen_callbacks(sock);
add_listen_sock(sock, &listen_con);
INIT_WORK(&listen_con.rwork, process_listen_recv_socket);
result = sock->ops->listen(sock, 5);
if (result < 0) {
dlm_close_sock(&listen_con.sock);
goto out;
}
return 0;
out:
sock_release(sock);
return result;
}
static int dlm_tcp_bind(struct socket *sock)
{
struct sockaddr_storage src_addr;
int result, addr_len;
/* Bind to our cluster-known address connecting to avoid
* routing problems.
*/
memcpy(&src_addr, dlm_local_addr[0], sizeof(src_addr));
make_sockaddr(&src_addr, 0, &addr_len);
result = sock->ops->bind(sock, (struct sockaddr *)&src_addr,
addr_len);
if (result < 0) {
/* This *may* not indicate a critical error */
log_print("could not bind for connect: %d", result);
}
return 0;
}
static int dlm_tcp_connect(struct connection *con, struct socket *sock,
struct sockaddr *addr, int addr_len)
{
int ret;
ret = sock->ops->connect(sock, addr, addr_len, O_NONBLOCK);
switch (ret) {
case -EINPROGRESS:
fallthrough;
case 0:
return 0;
}
return ret;
}
static int dlm_tcp_listen_validate(void)
{
/* We don't support multi-homed hosts */
if (dlm_local_count > 1) {
log_print("TCP protocol can't handle multi-homed hosts, try SCTP");
return -EINVAL;
}
return 0;
}
static void dlm_tcp_sockopts(struct socket *sock)
{
/* Turn off Nagle's algorithm */
tcp_sock_set_nodelay(sock->sk);
}
static void dlm_tcp_listen_sockopts(struct socket *sock)
{
dlm_tcp_sockopts(sock);
sock_set_reuseaddr(sock->sk);
}
static int dlm_tcp_listen_bind(struct socket *sock)
{
int addr_len;
/* Bind to our port */
make_sockaddr(dlm_local_addr[0], dlm_config.ci_tcp_port, &addr_len);
return sock->ops->bind(sock, (struct sockaddr *)dlm_local_addr[0],
addr_len);
}
static const struct dlm_proto_ops dlm_tcp_ops = {
.name = "TCP",
.proto = IPPROTO_TCP,
.connect = dlm_tcp_connect,
.sockopts = dlm_tcp_sockopts,
.bind = dlm_tcp_bind,
.listen_validate = dlm_tcp_listen_validate,
.listen_sockopts = dlm_tcp_listen_sockopts,
.listen_bind = dlm_tcp_listen_bind,
.shutdown_action = dlm_tcp_shutdown,
.eof_condition = tcp_eof_condition,
};
static int dlm_sctp_bind(struct socket *sock)
{
return sctp_bind_addrs(sock, 0);
}
static int dlm_sctp_connect(struct connection *con, struct socket *sock,
struct sockaddr *addr, int addr_len)
{
int ret;
/*
* Make sock->ops->connect() function return in specified time,
* since O_NONBLOCK argument in connect() function does not work here,
* then, we should restore the default value of this attribute.
*/
sock_set_sndtimeo(sock->sk, 5);
ret = sock->ops->connect(sock, addr, addr_len, 0);
sock_set_sndtimeo(sock->sk, 0);
if (ret < 0)
return ret;
if (!test_and_set_bit(CF_CONNECTED, &con->flags))
log_print("connected to node %d", con->nodeid);
return 0;
}
static int dlm_sctp_listen_validate(void)
{
if (!IS_ENABLED(CONFIG_IP_SCTP)) {
log_print("SCTP is not enabled by this kernel");
return -EOPNOTSUPP;
}
request_module("sctp");
return 0;
}
static int dlm_sctp_bind_listen(struct socket *sock)
{
return sctp_bind_addrs(sock, dlm_config.ci_tcp_port);
}
static void dlm_sctp_sockopts(struct socket *sock)
{
/* Turn off Nagle's algorithm */
sctp_sock_set_nodelay(sock->sk);
sock_set_rcvbuf(sock->sk, NEEDED_RMEM);
}
static const struct dlm_proto_ops dlm_sctp_ops = {
.name = "SCTP",
.proto = IPPROTO_SCTP,
.try_new_addr = true,
.connect = dlm_sctp_connect,
.sockopts = dlm_sctp_sockopts,
.bind = dlm_sctp_bind,
.listen_validate = dlm_sctp_listen_validate,
.listen_sockopts = dlm_sctp_sockopts,
.listen_bind = dlm_sctp_bind_listen,
};
int dlm_lowcomms_start(void)
{
int error = -EINVAL;
int i;
for (i = 0; i < CONN_HASH_SIZE; i++)
INIT_HLIST_HEAD(&connection_hash[i]);
init_local();
if (!dlm_local_count) {
error = -ENOTCONN;
log_print("no local IP address has been set");
goto fail;
}
INIT_WORK(&listen_con.rwork, process_listen_recv_socket);
error = work_start();
if (error)
goto fail_local;
dlm_allow_conn = 1;
/* Start listening */
switch (dlm_config.ci_protocol) {
case DLM_PROTO_TCP:
dlm_proto_ops = &dlm_tcp_ops;
break;
case DLM_PROTO_SCTP:
dlm_proto_ops = &dlm_sctp_ops;
break;
default:
log_print("Invalid protocol identifier %d set",
dlm_config.ci_protocol);
error = -EINVAL;
goto fail_proto_ops;
}
error = dlm_listen_for_all();
if (error)
goto fail_listen;
return 0;
fail_listen:
dlm_proto_ops = NULL;
fail_proto_ops:
dlm_allow_conn = 0;
dlm_close_sock(&listen_con.sock);
work_stop();
fail_local:
deinit_local();
fail:
return error;
}
void dlm_lowcomms_exit(void)
{
struct dlm_node_addr *na, *safe;
spin_lock(&dlm_node_addrs_spin);
list_for_each_entry_safe(na, safe, &dlm_node_addrs, list) {
list_del(&na->list);
while (na->addr_count--)
kfree(na->addr[na->addr_count]);
kfree(na);
}
spin_unlock(&dlm_node_addrs_spin);
}