* [PATCH net-next 2/4] afs: Kill afs_wait_mode
From: David Howells @ 2017-01-09 14:53 UTC (permalink / raw)
To: netdev; +Cc: dhowells, linux-afs, linux-kernel
In-Reply-To: <148397356909.20445.15077871371099721338.stgit@warthog.procyon.org.uk>
The afs_wait_mode struct isn't really necessary. Client calls only use one
of a choice of two (synchronous or the asynchronous) and incoming calls
don't use the wait at all. Replace with a boolean parameter.
Signed-off-by: David Howells <dhowells@redhat.com>
---
fs/afs/callback.c | 2 +
fs/afs/fsclient.c | 80 ++++++++++++++++++++++++++--------------------------
fs/afs/internal.h | 68 ++++++++++++--------------------------------
fs/afs/rxrpc.c | 51 ++++++++-------------------------
fs/afs/vlclient.c | 8 +++--
fs/afs/vlocation.c | 4 +--
fs/afs/vnode.c | 26 ++++++++---------
7 files changed, 90 insertions(+), 149 deletions(-)
diff --git a/fs/afs/callback.c b/fs/afs/callback.c
index 1e9d2f84e5b5..b29447e03ede 100644
--- a/fs/afs/callback.c
+++ b/fs/afs/callback.c
@@ -343,7 +343,7 @@ void afs_dispatch_give_up_callbacks(struct work_struct *work)
* had callbacks entirely, and the server will call us later to break
* them
*/
- afs_fs_give_up_callbacks(server, &afs_async_call);
+ afs_fs_give_up_callbacks(server, true);
}
/*
diff --git a/fs/afs/fsclient.c b/fs/afs/fsclient.c
index 7dc1f6fb3661..ac8e766978dc 100644
--- a/fs/afs/fsclient.c
+++ b/fs/afs/fsclient.c
@@ -275,7 +275,7 @@ int afs_fs_fetch_file_status(struct afs_server *server,
struct key *key,
struct afs_vnode *vnode,
struct afs_volsync *volsync,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -300,7 +300,7 @@ int afs_fs_fetch_file_status(struct afs_server *server,
bp[2] = htonl(vnode->fid.vnode);
bp[3] = htonl(vnode->fid.unique);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -464,7 +464,7 @@ static int afs_fs_fetch_data64(struct afs_server *server,
struct key *key,
struct afs_vnode *vnode,
struct afs_read *req,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -495,7 +495,7 @@ static int afs_fs_fetch_data64(struct afs_server *server,
bp[7] = htonl(lower_32_bits(req->len));
atomic_inc(&req->usage);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -505,7 +505,7 @@ int afs_fs_fetch_data(struct afs_server *server,
struct key *key,
struct afs_vnode *vnode,
struct afs_read *req,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -513,7 +513,7 @@ int afs_fs_fetch_data(struct afs_server *server,
if (upper_32_bits(req->pos) ||
upper_32_bits(req->len) ||
upper_32_bits(req->pos + req->len))
- return afs_fs_fetch_data64(server, key, vnode, req, wait_mode);
+ return afs_fs_fetch_data64(server, key, vnode, req, async);
_enter("");
@@ -539,7 +539,7 @@ int afs_fs_fetch_data(struct afs_server *server,
bp[5] = htonl(lower_32_bits(req->len));
atomic_inc(&req->usage);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -568,7 +568,7 @@ static const struct afs_call_type afs_RXFSGiveUpCallBacks = {
* - the callbacks are held in the server->cb_break ring
*/
int afs_fs_give_up_callbacks(struct afs_server *server,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
size_t ncallbacks;
@@ -622,7 +622,7 @@ int afs_fs_give_up_callbacks(struct afs_server *server,
ASSERT(ncallbacks > 0);
wake_up_nr(&server->cb_break_waitq, ncallbacks);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -673,7 +673,7 @@ int afs_fs_create(struct afs_server *server,
struct afs_fid *newfid,
struct afs_file_status *newstatus,
struct afs_callback *newcb,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
size_t namesz, reqsz, padsz;
@@ -718,7 +718,7 @@ int afs_fs_create(struct afs_server *server,
*bp++ = htonl(mode & S_IALLUGO); /* unix mode */
*bp++ = 0; /* segment size */
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -763,7 +763,7 @@ int afs_fs_remove(struct afs_server *server,
struct afs_vnode *vnode,
const char *name,
bool isdir,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
size_t namesz, reqsz, padsz;
@@ -798,7 +798,7 @@ int afs_fs_remove(struct afs_server *server,
bp = (void *) bp + padsz;
}
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -844,7 +844,7 @@ int afs_fs_link(struct afs_server *server,
struct afs_vnode *dvnode,
struct afs_vnode *vnode,
const char *name,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
size_t namesz, reqsz, padsz;
@@ -883,7 +883,7 @@ int afs_fs_link(struct afs_server *server,
*bp++ = htonl(vnode->fid.vnode);
*bp++ = htonl(vnode->fid.unique);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -932,7 +932,7 @@ int afs_fs_symlink(struct afs_server *server,
const char *contents,
struct afs_fid *newfid,
struct afs_file_status *newstatus,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
size_t namesz, reqsz, padsz, c_namesz, c_padsz;
@@ -987,7 +987,7 @@ int afs_fs_symlink(struct afs_server *server,
*bp++ = htonl(S_IRWXUGO); /* unix mode */
*bp++ = 0; /* segment size */
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1036,7 +1036,7 @@ int afs_fs_rename(struct afs_server *server,
const char *orig_name,
struct afs_vnode *new_dvnode,
const char *new_name,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
size_t reqsz, o_namesz, o_padsz, n_namesz, n_padsz;
@@ -1090,7 +1090,7 @@ int afs_fs_rename(struct afs_server *server,
bp = (void *) bp + n_padsz;
}
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1145,7 +1145,7 @@ static int afs_fs_store_data64(struct afs_server *server,
pgoff_t first, pgoff_t last,
unsigned offset, unsigned to,
loff_t size, loff_t pos, loff_t i_size,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_vnode *vnode = wb->vnode;
struct afs_call *call;
@@ -1194,7 +1194,7 @@ static int afs_fs_store_data64(struct afs_server *server,
*bp++ = htonl(i_size >> 32);
*bp++ = htonl((u32) i_size);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1203,7 +1203,7 @@ static int afs_fs_store_data64(struct afs_server *server,
int afs_fs_store_data(struct afs_server *server, struct afs_writeback *wb,
pgoff_t first, pgoff_t last,
unsigned offset, unsigned to,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_vnode *vnode = wb->vnode;
struct afs_call *call;
@@ -1229,7 +1229,7 @@ int afs_fs_store_data(struct afs_server *server, struct afs_writeback *wb,
if (pos >> 32 || i_size >> 32 || size >> 32 || (pos + size) >> 32)
return afs_fs_store_data64(server, wb, first, last, offset, to,
- size, pos, i_size, wait_mode);
+ size, pos, i_size, async);
call = afs_alloc_flat_call(&afs_RXFSStoreData,
(4 + 6 + 3) * 4,
@@ -1268,7 +1268,7 @@ int afs_fs_store_data(struct afs_server *server, struct afs_writeback *wb,
*bp++ = htonl(size);
*bp++ = htonl(i_size);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1330,7 +1330,7 @@ static const struct afs_call_type afs_RXFSStoreData64_as_Status = {
*/
static int afs_fs_setattr_size64(struct afs_server *server, struct key *key,
struct afs_vnode *vnode, struct iattr *attr,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -1369,7 +1369,7 @@ static int afs_fs_setattr_size64(struct afs_server *server, struct key *key,
*bp++ = htonl(attr->ia_size >> 32); /* new file length */
*bp++ = htonl((u32) attr->ia_size);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1378,7 +1378,7 @@ static int afs_fs_setattr_size64(struct afs_server *server, struct key *key,
*/
static int afs_fs_setattr_size(struct afs_server *server, struct key *key,
struct afs_vnode *vnode, struct iattr *attr,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -1389,7 +1389,7 @@ static int afs_fs_setattr_size(struct afs_server *server, struct key *key,
ASSERT(attr->ia_valid & ATTR_SIZE);
if (attr->ia_size >> 32)
return afs_fs_setattr_size64(server, key, vnode, attr,
- wait_mode);
+ async);
call = afs_alloc_flat_call(&afs_RXFSStoreData_as_Status,
(4 + 6 + 3) * 4,
@@ -1417,7 +1417,7 @@ static int afs_fs_setattr_size(struct afs_server *server, struct key *key,
*bp++ = 0; /* size of write */
*bp++ = htonl(attr->ia_size); /* new file length */
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1426,14 +1426,14 @@ static int afs_fs_setattr_size(struct afs_server *server, struct key *key,
*/
int afs_fs_setattr(struct afs_server *server, struct key *key,
struct afs_vnode *vnode, struct iattr *attr,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
if (attr->ia_valid & ATTR_SIZE)
return afs_fs_setattr_size(server, key, vnode, attr,
- wait_mode);
+ async);
_enter(",%x,{%x:%u},,",
key_serial(key), vnode->fid.vid, vnode->fid.vnode);
@@ -1459,7 +1459,7 @@ int afs_fs_setattr(struct afs_server *server, struct key *key,
xdr_encode_AFS_StoreStatus(&bp, attr);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1661,7 +1661,7 @@ int afs_fs_get_volume_status(struct afs_server *server,
struct key *key,
struct afs_vnode *vnode,
struct afs_volume_status *vs,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -1691,7 +1691,7 @@ int afs_fs_get_volume_status(struct afs_server *server,
bp[0] = htonl(FSGETVOLUMESTATUS);
bp[1] = htonl(vnode->fid.vid);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1753,7 +1753,7 @@ int afs_fs_set_lock(struct afs_server *server,
struct key *key,
struct afs_vnode *vnode,
afs_lock_type_t type,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -1777,7 +1777,7 @@ int afs_fs_set_lock(struct afs_server *server,
*bp++ = htonl(vnode->fid.unique);
*bp++ = htonl(type);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1786,7 +1786,7 @@ int afs_fs_set_lock(struct afs_server *server,
int afs_fs_extend_lock(struct afs_server *server,
struct key *key,
struct afs_vnode *vnode,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -1809,7 +1809,7 @@ int afs_fs_extend_lock(struct afs_server *server,
*bp++ = htonl(vnode->fid.vnode);
*bp++ = htonl(vnode->fid.unique);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
/*
@@ -1818,7 +1818,7 @@ int afs_fs_extend_lock(struct afs_server *server,
int afs_fs_release_lock(struct afs_server *server,
struct key *key,
struct afs_vnode *vnode,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -1841,5 +1841,5 @@ int afs_fs_release_lock(struct afs_server *server,
*bp++ = htonl(vnode->fid.vnode);
*bp++ = htonl(vnode->fid.unique);
- return afs_make_call(&server->addr, call, GFP_NOFS, wait_mode);
+ return afs_make_call(&server->addr, call, GFP_NOFS, async);
}
diff --git a/fs/afs/internal.h b/fs/afs/internal.h
index f71e58fcc2f2..b411670d5f67 100644
--- a/fs/afs/internal.h
+++ b/fs/afs/internal.h
@@ -51,23 +51,6 @@ struct afs_mount_params {
struct key *key; /* key to use for secure mounting */
};
-/*
- * definition of how to wait for the completion of an operation
- */
-struct afs_wait_mode {
- /* RxRPC received message notification */
- rxrpc_notify_rx_t notify_rx;
-
- /* synchronous call waiter and call dispatched notification */
- int (*wait)(struct afs_call *call);
-
- /* asynchronous call completion */
- void (*async_complete)(void *reply, int error);
-};
-
-extern const struct afs_wait_mode afs_sync_call;
-extern const struct afs_wait_mode afs_async_call;
-
enum afs_call_state {
AFS_CALL_REQUESTING, /* request is being sent for outgoing call */
AFS_CALL_AWAIT_REPLY, /* awaiting reply to outgoing call */
@@ -82,7 +65,6 @@ enum afs_call_state {
*/
struct afs_call {
const struct afs_call_type *type; /* type of call */
- const struct afs_wait_mode *wait_mode; /* completion wait mode */
wait_queue_head_t waitq; /* processes awaiting completion */
struct work_struct async_work; /* asynchronous work processor */
struct work_struct work; /* actual work processor */
@@ -111,6 +93,7 @@ struct afs_call {
bool incoming; /* T if incoming call */
bool send_pages; /* T if data from mapping should be sent */
bool need_attention; /* T if RxRPC poked us */
+ bool async; /* T if asynchronous */
u16 service_id; /* RxRPC service ID to call */
__be16 port; /* target UDP port */
u32 operation_ID; /* operation ID for an incoming call */
@@ -527,50 +510,37 @@ extern int afs_flock(struct file *, int, struct file_lock *);
*/
extern int afs_fs_fetch_file_status(struct afs_server *, struct key *,
struct afs_vnode *, struct afs_volsync *,
- const struct afs_wait_mode *);
-extern int afs_fs_give_up_callbacks(struct afs_server *,
- const struct afs_wait_mode *);
+ bool);
+extern int afs_fs_give_up_callbacks(struct afs_server *, bool);
extern int afs_fs_fetch_data(struct afs_server *, struct key *,
- struct afs_vnode *, struct afs_read *,
- const struct afs_wait_mode *);
+ struct afs_vnode *, struct afs_read *, bool);
extern int afs_fs_create(struct afs_server *, struct key *,
struct afs_vnode *, const char *, umode_t,
struct afs_fid *, struct afs_file_status *,
- struct afs_callback *,
- const struct afs_wait_mode *);
+ struct afs_callback *, bool);
extern int afs_fs_remove(struct afs_server *, struct key *,
- struct afs_vnode *, const char *, bool,
- const struct afs_wait_mode *);
+ struct afs_vnode *, const char *, bool, bool);
extern int afs_fs_link(struct afs_server *, struct key *, struct afs_vnode *,
- struct afs_vnode *, const char *,
- const struct afs_wait_mode *);
+ struct afs_vnode *, const char *, bool);
extern int afs_fs_symlink(struct afs_server *, struct key *,
struct afs_vnode *, const char *, const char *,
- struct afs_fid *, struct afs_file_status *,
- const struct afs_wait_mode *);
+ struct afs_fid *, struct afs_file_status *, bool);
extern int afs_fs_rename(struct afs_server *, struct key *,
struct afs_vnode *, const char *,
- struct afs_vnode *, const char *,
- const struct afs_wait_mode *);
+ struct afs_vnode *, const char *, bool);
extern int afs_fs_store_data(struct afs_server *, struct afs_writeback *,
- pgoff_t, pgoff_t, unsigned, unsigned,
- const struct afs_wait_mode *);
+ pgoff_t, pgoff_t, unsigned, unsigned, bool);
extern int afs_fs_setattr(struct afs_server *, struct key *,
- struct afs_vnode *, struct iattr *,
- const struct afs_wait_mode *);
+ struct afs_vnode *, struct iattr *, bool);
extern int afs_fs_get_volume_status(struct afs_server *, struct key *,
struct afs_vnode *,
- struct afs_volume_status *,
- const struct afs_wait_mode *);
+ struct afs_volume_status *, bool);
extern int afs_fs_set_lock(struct afs_server *, struct key *,
- struct afs_vnode *, afs_lock_type_t,
- const struct afs_wait_mode *);
+ struct afs_vnode *, afs_lock_type_t, bool);
extern int afs_fs_extend_lock(struct afs_server *, struct key *,
- struct afs_vnode *,
- const struct afs_wait_mode *);
+ struct afs_vnode *, bool);
extern int afs_fs_release_lock(struct afs_server *, struct key *,
- struct afs_vnode *,
- const struct afs_wait_mode *);
+ struct afs_vnode *, bool);
/*
* inode.c
@@ -624,8 +594,7 @@ extern struct socket *afs_socket;
extern int afs_open_socket(void);
extern void afs_close_socket(void);
-extern int afs_make_call(struct in_addr *, struct afs_call *, gfp_t,
- const struct afs_wait_mode *);
+extern int afs_make_call(struct in_addr *, struct afs_call *, gfp_t, bool);
extern struct afs_call *afs_alloc_flat_call(const struct afs_call_type *,
size_t, size_t);
extern void afs_flat_call_destructor(struct afs_call *);
@@ -681,11 +650,10 @@ extern int afs_get_MAC_address(u8 *, size_t);
*/
extern int afs_vl_get_entry_by_name(struct in_addr *, struct key *,
const char *, struct afs_cache_vlocation *,
- const struct afs_wait_mode *);
+ bool);
extern int afs_vl_get_entry_by_id(struct in_addr *, struct key *,
afs_volid_t, afs_voltype_t,
- struct afs_cache_vlocation *,
- const struct afs_wait_mode *);
+ struct afs_cache_vlocation *, bool);
/*
* vlocation.c
diff --git a/fs/afs/rxrpc.c b/fs/afs/rxrpc.c
index f26344a8c029..ec1e41f929d1 100644
--- a/fs/afs/rxrpc.c
+++ b/fs/afs/rxrpc.c
@@ -25,29 +25,11 @@ static void afs_free_call(struct afs_call *);
static void afs_wake_up_call_waiter(struct sock *, struct rxrpc_call *, unsigned long);
static int afs_wait_for_call_to_complete(struct afs_call *);
static void afs_wake_up_async_call(struct sock *, struct rxrpc_call *, unsigned long);
-static int afs_dont_wait_for_call_to_complete(struct afs_call *);
static void afs_process_async_call(struct work_struct *);
static void afs_rx_new_call(struct sock *, struct rxrpc_call *, unsigned long);
static void afs_rx_discard_new_call(struct rxrpc_call *, unsigned long);
static int afs_deliver_cm_op_id(struct afs_call *);
-/* synchronous call management */
-const struct afs_wait_mode afs_sync_call = {
- .notify_rx = afs_wake_up_call_waiter,
- .wait = afs_wait_for_call_to_complete,
-};
-
-/* asynchronous call management */
-const struct afs_wait_mode afs_async_call = {
- .notify_rx = afs_wake_up_async_call,
- .wait = afs_dont_wait_for_call_to_complete,
-};
-
-/* asynchronous incoming call management */
-static const struct afs_wait_mode afs_async_incoming_call = {
- .notify_rx = afs_wake_up_async_call,
-};
-
/* asynchronous incoming call initial processing */
static const struct afs_call_type afs_RXCMxxxx = {
.name = "CB.xxxx",
@@ -315,7 +297,7 @@ static int afs_send_pages(struct afs_call *call, struct msghdr *msg,
* initiate a call
*/
int afs_make_call(struct in_addr *addr, struct afs_call *call, gfp_t gfp,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct sockaddr_rxrpc srx;
struct rxrpc_call *rxcall;
@@ -332,7 +314,7 @@ int afs_make_call(struct in_addr *addr, struct afs_call *call, gfp_t gfp,
call, call->type->name, key_serial(call->key),
atomic_read(&afs_outstanding_calls));
- call->wait_mode = wait_mode;
+ call->async = async;
INIT_WORK(&call->async_work, afs_process_async_call);
memset(&srx, 0, sizeof(srx));
@@ -347,7 +329,9 @@ int afs_make_call(struct in_addr *addr, struct afs_call *call, gfp_t gfp,
/* create a call */
rxcall = rxrpc_kernel_begin_call(afs_socket, &srx, call->key,
(unsigned long) call, gfp,
- wait_mode->notify_rx);
+ (async ?
+ afs_wake_up_async_call :
+ afs_wake_up_call_waiter));
call->key = NULL;
if (IS_ERR(rxcall)) {
ret = PTR_ERR(rxcall);
@@ -386,7 +370,10 @@ int afs_make_call(struct in_addr *addr, struct afs_call *call, gfp_t gfp,
/* at this point, an async call may no longer exist as it may have
* already completed */
- return wait_mode->wait(call);
+ if (call->async)
+ return -EINPROGRESS;
+
+ return afs_wait_for_call_to_complete(call);
error_do_abort:
rxrpc_kernel_abort_call(afs_socket, rxcall, RX_USER_ABORT, -ret, "KSD");
@@ -549,17 +536,6 @@ static void afs_wake_up_async_call(struct sock *sk, struct rxrpc_call *rxcall,
}
/*
- * put a call into asynchronous mode
- * - mustn't touch the call descriptor as the call my have completed by the
- * time we get here
- */
-static int afs_dont_wait_for_call_to_complete(struct afs_call *call)
-{
- _enter("");
- return -EINPROGRESS;
-}
-
-/*
* delete an asynchronous call
*/
static void afs_delete_async_call(struct work_struct *work)
@@ -587,10 +563,7 @@ static void afs_process_async_call(struct work_struct *work)
afs_deliver_to_call(call);
}
- if (call->state == AFS_CALL_COMPLETE && call->wait_mode) {
- if (call->wait_mode->async_complete)
- call->wait_mode->async_complete(call->reply,
- call->error);
+ if (call->state == AFS_CALL_COMPLETE) {
call->reply = NULL;
/* kill the call */
@@ -626,10 +599,10 @@ static void afs_charge_preallocation(struct work_struct *work)
break;
INIT_WORK(&call->async_work, afs_process_async_call);
- call->wait_mode = &afs_async_incoming_call;
call->type = &afs_RXCMxxxx;
- init_waitqueue_head(&call->waitq);
+ call->async = true;
call->state = AFS_CALL_AWAIT_OP_ID;
+ init_waitqueue_head(&call->waitq);
}
if (rxrpc_kernel_charge_accept(afs_socket,
diff --git a/fs/afs/vlclient.c b/fs/afs/vlclient.c
index 94bcd97d22b8..a5e4cc561b6c 100644
--- a/fs/afs/vlclient.c
+++ b/fs/afs/vlclient.c
@@ -147,7 +147,7 @@ int afs_vl_get_entry_by_name(struct in_addr *addr,
struct key *key,
const char *volname,
struct afs_cache_vlocation *entry,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
size_t volnamesz, reqsz, padsz;
@@ -177,7 +177,7 @@ int afs_vl_get_entry_by_name(struct in_addr *addr,
memset((void *) bp + volnamesz, 0, padsz);
/* initiate the call */
- return afs_make_call(addr, call, GFP_KERNEL, wait_mode);
+ return afs_make_call(addr, call, GFP_KERNEL, async);
}
/*
@@ -188,7 +188,7 @@ int afs_vl_get_entry_by_id(struct in_addr *addr,
afs_volid_t volid,
afs_voltype_t voltype,
struct afs_cache_vlocation *entry,
- const struct afs_wait_mode *wait_mode)
+ bool async)
{
struct afs_call *call;
__be32 *bp;
@@ -211,5 +211,5 @@ int afs_vl_get_entry_by_id(struct in_addr *addr,
*bp = htonl(voltype);
/* initiate the call */
- return afs_make_call(addr, call, GFP_KERNEL, wait_mode);
+ return afs_make_call(addr, call, GFP_KERNEL, async);
}
diff --git a/fs/afs/vlocation.c b/fs/afs/vlocation.c
index 45a86396fd2d..d7d8dd8c0b31 100644
--- a/fs/afs/vlocation.c
+++ b/fs/afs/vlocation.c
@@ -53,7 +53,7 @@ static int afs_vlocation_access_vl_by_name(struct afs_vlocation *vl,
/* attempt to access the VL server */
ret = afs_vl_get_entry_by_name(&addr, key, vl->vldb.name, vldb,
- &afs_sync_call);
+ false);
switch (ret) {
case 0:
goto out;
@@ -111,7 +111,7 @@ static int afs_vlocation_access_vl_by_id(struct afs_vlocation *vl,
/* attempt to access the VL server */
ret = afs_vl_get_entry_by_id(&addr, key, volid, voltype, vldb,
- &afs_sync_call);
+ false);
switch (ret) {
case 0:
goto out;
diff --git a/fs/afs/vnode.c b/fs/afs/vnode.c
index 45aa874f5d32..dcb956143c86 100644
--- a/fs/afs/vnode.c
+++ b/fs/afs/vnode.c
@@ -358,7 +358,7 @@ int afs_vnode_fetch_status(struct afs_vnode *vnode,
server, ntohl(server->addr.s_addr));
ret = afs_fs_fetch_file_status(server, key, vnode, NULL,
- &afs_sync_call);
+ false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -421,7 +421,7 @@ int afs_vnode_fetch_data(struct afs_vnode *vnode, struct key *key,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
ret = afs_fs_fetch_data(server, key, vnode, desc,
- &afs_sync_call);
+ false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -477,7 +477,7 @@ int afs_vnode_create(struct afs_vnode *vnode, struct key *key,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
ret = afs_fs_create(server, key, vnode, name, mode, newfid,
- newstatus, newcb, &afs_sync_call);
+ newstatus, newcb, false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -533,7 +533,7 @@ int afs_vnode_remove(struct afs_vnode *vnode, struct key *key, const char *name,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
ret = afs_fs_remove(server, key, vnode, name, isdir,
- &afs_sync_call);
+ false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -595,7 +595,7 @@ int afs_vnode_link(struct afs_vnode *dvnode, struct afs_vnode *vnode,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
ret = afs_fs_link(server, key, dvnode, vnode, name,
- &afs_sync_call);
+ false);
} while (!afs_volume_release_fileserver(dvnode, server, ret));
@@ -659,7 +659,7 @@ int afs_vnode_symlink(struct afs_vnode *vnode, struct key *key,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
ret = afs_fs_symlink(server, key, vnode, name, content,
- newfid, newstatus, &afs_sync_call);
+ newfid, newstatus, false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -729,7 +729,7 @@ int afs_vnode_rename(struct afs_vnode *orig_dvnode,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
ret = afs_fs_rename(server, key, orig_dvnode, orig_name,
- new_dvnode, new_name, &afs_sync_call);
+ new_dvnode, new_name, false);
} while (!afs_volume_release_fileserver(orig_dvnode, server, ret));
@@ -795,7 +795,7 @@ int afs_vnode_store_data(struct afs_writeback *wb, pgoff_t first, pgoff_t last,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
ret = afs_fs_store_data(server, wb, first, last, offset, to,
- &afs_sync_call);
+ false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -847,7 +847,7 @@ int afs_vnode_setattr(struct afs_vnode *vnode, struct key *key,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
- ret = afs_fs_setattr(server, key, vnode, attr, &afs_sync_call);
+ ret = afs_fs_setattr(server, key, vnode, attr, false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -894,7 +894,7 @@ int afs_vnode_get_volume_status(struct afs_vnode *vnode, struct key *key,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
- ret = afs_fs_get_volume_status(server, key, vnode, vs, &afs_sync_call);
+ ret = afs_fs_get_volume_status(server, key, vnode, vs, false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -933,7 +933,7 @@ int afs_vnode_set_lock(struct afs_vnode *vnode, struct key *key,
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
- ret = afs_fs_set_lock(server, key, vnode, type, &afs_sync_call);
+ ret = afs_fs_set_lock(server, key, vnode, type, false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -971,7 +971,7 @@ int afs_vnode_extend_lock(struct afs_vnode *vnode, struct key *key)
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
- ret = afs_fs_extend_lock(server, key, vnode, &afs_sync_call);
+ ret = afs_fs_extend_lock(server, key, vnode, false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
@@ -1009,7 +1009,7 @@ int afs_vnode_release_lock(struct afs_vnode *vnode, struct key *key)
_debug("USING SERVER: %08x\n", ntohl(server->addr.s_addr));
- ret = afs_fs_release_lock(server, key, vnode, &afs_sync_call);
+ ret = afs_fs_release_lock(server, key, vnode, false);
} while (!afs_volume_release_fileserver(vnode, server, ret));
^ permalink raw reply related
* [PATCH net-next 3/4] rxrpc: Allow listen(sock, 0) to be used to disable listening
From: David Howells @ 2017-01-09 14:53 UTC (permalink / raw)
To: netdev; +Cc: dhowells, linux-afs, linux-kernel
In-Reply-To: <148397356909.20445.15077871371099721338.stgit@warthog.procyon.org.uk>
Allow listen() with a backlog of 0 to be used to disable listening on an
AF_RXRPC socket. This also releases any preallocation, thereby making it
easier for a kernel service to account for all allocated call structures
when shutting down the service.
The socket cannot thereafter have listening reenabled, but must rather be
closed and reopened.
Signed-off-by: David Howells <dhowells@redhat.com>
---
net/rxrpc/af_rxrpc.c | 8 ++++++++
net/rxrpc/ar-internal.h | 1 +
net/rxrpc/call_accept.c | 3 ++-
3 files changed, 11 insertions(+), 1 deletion(-)
diff --git a/net/rxrpc/af_rxrpc.c b/net/rxrpc/af_rxrpc.c
index 5f63f6dcaabb..199b46e93e64 100644
--- a/net/rxrpc/af_rxrpc.c
+++ b/net/rxrpc/af_rxrpc.c
@@ -224,6 +224,14 @@ static int rxrpc_listen(struct socket *sock, int backlog)
else
sk->sk_max_ack_backlog = old;
break;
+ case RXRPC_SERVER_LISTENING:
+ if (backlog == 0) {
+ rx->sk.sk_state = RXRPC_SERVER_LISTEN_DISABLED;
+ sk->sk_max_ack_backlog = 0;
+ rxrpc_discard_prealloc(rx);
+ ret = 0;
+ break;
+ }
default:
ret = -EBUSY;
break;
diff --git a/net/rxrpc/ar-internal.h b/net/rxrpc/ar-internal.h
index 84927c7b5fdf..12be432be9b2 100644
--- a/net/rxrpc/ar-internal.h
+++ b/net/rxrpc/ar-internal.h
@@ -60,6 +60,7 @@ enum {
RXRPC_CLIENT_BOUND, /* client local address bound */
RXRPC_SERVER_BOUND, /* server local address bound */
RXRPC_SERVER_LISTENING, /* server listening for connections */
+ RXRPC_SERVER_LISTEN_DISABLED, /* server listening disabled */
RXRPC_CLOSE, /* socket is being closed */
};
diff --git a/net/rxrpc/call_accept.c b/net/rxrpc/call_accept.c
index 832d854c2d5c..7c4c64ab8da2 100644
--- a/net/rxrpc/call_accept.c
+++ b/net/rxrpc/call_accept.c
@@ -349,7 +349,8 @@ struct rxrpc_call *rxrpc_new_incoming_call(struct rxrpc_local *local,
found_service:
spin_lock(&rx->incoming_lock);
- if (rx->sk.sk_state == RXRPC_CLOSE) {
+ if (rx->sk.sk_state == RXRPC_SERVER_LISTEN_DISABLED ||
+ rx->sk.sk_state == RXRPC_CLOSE) {
trace_rxrpc_abort("CLS", sp->hdr.cid, sp->hdr.callNumber,
sp->hdr.seq, RX_INVALID_OPERATION, ESHUTDOWN);
skb->mark = RXRPC_SKB_MARK_LOCAL_ABORT;
^ permalink raw reply related
* [PATCH net-next 4/4] afs: Refcount the afs_call struct
From: David Howells @ 2017-01-09 14:53 UTC (permalink / raw)
To: netdev; +Cc: dhowells, linux-afs, linux-kernel
In-Reply-To: <148397356909.20445.15077871371099721338.stgit@warthog.procyon.org.uk>
A static checker warning occurs in the AFS filesystem:
fs/afs/cmservice.c:155 SRXAFSCB_CallBack()
error: dereferencing freed memory 'call'
due to the reply being sent before we access the server it points to. The
act of sending the reply causes the call to be freed if an error occurs
(but not if it doesn't).
On top of this, the lifetime handling of afs_call structs is fragile
because they get passed around through workqueues without any sort of
refcounting.
Deal with the issues by:
(1) Fix the maybe/maybe not nature of the reply sending functions with
regards to whether they release the call struct.
(2) Refcount the afs_call struct and sort out places that need to get/put
references.
(3) Pass a ref through the work queue and release (or pass on) that ref in
the work function. Care has to be taken because a work queue may
already own a ref to the call.
(4) Do the cleaning up in the put function only.
(5) Simplify module cleanup by always incrementing afs_outstanding_calls
whenever a call is allocated.
(6) Set the backlog to 0 with kernel_listen() at the beginning of the
process of closing the socket to prevent new incoming calls from
occurring and to remove the contribution of preallocated calls from
afs_outstanding_calls before we wait on it.
A tracepoint is also added to monitor the afs_call refcount and lifetime.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Fixes: 08e0e7c82eea: "[AF_RXRPC]: Make the in-kernel AFS filesystem use AF_RXRPC."
---
fs/afs/cmservice.c | 41 ++++++------
fs/afs/internal.h | 9 ++-
fs/afs/rxrpc.c | 153 +++++++++++++++++++++++++++-----------------
include/trace/events/afs.h | 75 ++++++++++++++++++++++
4 files changed, 199 insertions(+), 79 deletions(-)
diff --git a/fs/afs/cmservice.c b/fs/afs/cmservice.c
index a2e1e02005f6..e349a3316303 100644
--- a/fs/afs/cmservice.c
+++ b/fs/afs/cmservice.c
@@ -24,6 +24,11 @@ static int afs_deliver_cb_callback(struct afs_call *);
static int afs_deliver_cb_probe_uuid(struct afs_call *);
static int afs_deliver_cb_tell_me_about_yourself(struct afs_call *);
static void afs_cm_destructor(struct afs_call *);
+static void SRXAFSCB_CallBack(struct work_struct *);
+static void SRXAFSCB_InitCallBackState(struct work_struct *);
+static void SRXAFSCB_Probe(struct work_struct *);
+static void SRXAFSCB_ProbeUuid(struct work_struct *);
+static void SRXAFSCB_TellMeAboutYourself(struct work_struct *);
#define CM_NAME(name) \
const char afs_SRXCB##name##_name[] __tracepoint_string = \
@@ -38,6 +43,7 @@ static const struct afs_call_type afs_SRXCBCallBack = {
.deliver = afs_deliver_cb_callback,
.abort_to_error = afs_abort_to_error,
.destructor = afs_cm_destructor,
+ .work = SRXAFSCB_CallBack,
};
/*
@@ -49,6 +55,7 @@ static const struct afs_call_type afs_SRXCBInitCallBackState = {
.deliver = afs_deliver_cb_init_call_back_state,
.abort_to_error = afs_abort_to_error,
.destructor = afs_cm_destructor,
+ .work = SRXAFSCB_InitCallBackState,
};
/*
@@ -60,6 +67,7 @@ static const struct afs_call_type afs_SRXCBInitCallBackState3 = {
.deliver = afs_deliver_cb_init_call_back_state3,
.abort_to_error = afs_abort_to_error,
.destructor = afs_cm_destructor,
+ .work = SRXAFSCB_InitCallBackState,
};
/*
@@ -71,6 +79,7 @@ static const struct afs_call_type afs_SRXCBProbe = {
.deliver = afs_deliver_cb_probe,
.abort_to_error = afs_abort_to_error,
.destructor = afs_cm_destructor,
+ .work = SRXAFSCB_Probe,
};
/*
@@ -82,6 +91,7 @@ static const struct afs_call_type afs_SRXCBProbeUuid = {
.deliver = afs_deliver_cb_probe_uuid,
.abort_to_error = afs_abort_to_error,
.destructor = afs_cm_destructor,
+ .work = SRXAFSCB_ProbeUuid,
};
/*
@@ -93,6 +103,7 @@ static const struct afs_call_type afs_SRXCBTellMeAboutYourself = {
.deliver = afs_deliver_cb_tell_me_about_yourself,
.abort_to_error = afs_abort_to_error,
.destructor = afs_cm_destructor,
+ .work = SRXAFSCB_TellMeAboutYourself,
};
/*
@@ -163,6 +174,7 @@ static void SRXAFSCB_CallBack(struct work_struct *work)
afs_send_empty_reply(call);
afs_break_callbacks(call->server, call->count, call->request);
+ afs_put_call(call);
_leave("");
}
@@ -284,9 +296,7 @@ static int afs_deliver_cb_callback(struct afs_call *call)
return -ENOTCONN;
call->server = server;
- INIT_WORK(&call->work, SRXAFSCB_CallBack);
- queue_work(afs_wq, &call->work);
- return 0;
+ return afs_queue_call_work(call);
}
/*
@@ -300,6 +310,7 @@ static void SRXAFSCB_InitCallBackState(struct work_struct *work)
afs_init_callback_state(call->server);
afs_send_empty_reply(call);
+ afs_put_call(call);
_leave("");
}
@@ -330,9 +341,7 @@ static int afs_deliver_cb_init_call_back_state(struct afs_call *call)
return -ENOTCONN;
call->server = server;
- INIT_WORK(&call->work, SRXAFSCB_InitCallBackState);
- queue_work(afs_wq, &call->work);
- return 0;
+ return afs_queue_call_work(call);
}
/*
@@ -404,9 +413,7 @@ static int afs_deliver_cb_init_call_back_state3(struct afs_call *call)
return -ENOTCONN;
call->server = server;
- INIT_WORK(&call->work, SRXAFSCB_InitCallBackState);
- queue_work(afs_wq, &call->work);
- return 0;
+ return afs_queue_call_work(call);
}
/*
@@ -418,6 +425,7 @@ static void SRXAFSCB_Probe(struct work_struct *work)
_enter("");
afs_send_empty_reply(call);
+ afs_put_call(call);
_leave("");
}
@@ -437,9 +445,7 @@ static int afs_deliver_cb_probe(struct afs_call *call)
/* no unmarshalling required */
call->state = AFS_CALL_REPLYING;
- INIT_WORK(&call->work, SRXAFSCB_Probe);
- queue_work(afs_wq, &call->work);
- return 0;
+ return afs_queue_call_work(call);
}
/*
@@ -462,6 +468,7 @@ static void SRXAFSCB_ProbeUuid(struct work_struct *work)
reply.match = htonl(1);
afs_send_simple_reply(call, &reply, sizeof(reply));
+ afs_put_call(call);
_leave("");
}
@@ -520,9 +527,7 @@ static int afs_deliver_cb_probe_uuid(struct afs_call *call)
call->state = AFS_CALL_REPLYING;
- INIT_WORK(&call->work, SRXAFSCB_ProbeUuid);
- queue_work(afs_wq, &call->work);
- return 0;
+ return afs_queue_call_work(call);
}
/*
@@ -584,7 +589,7 @@ static void SRXAFSCB_TellMeAboutYourself(struct work_struct *work)
reply.cap.capcount = htonl(1);
reply.cap.caps[0] = htonl(AFS_CAP_ERROR_TRANSLATION);
afs_send_simple_reply(call, &reply, sizeof(reply));
-
+ afs_put_call(call);
_leave("");
}
@@ -604,7 +609,5 @@ static int afs_deliver_cb_tell_me_about_yourself(struct afs_call *call)
/* no unmarshalling required */
call->state = AFS_CALL_REPLYING;
- INIT_WORK(&call->work, SRXAFSCB_TellMeAboutYourself);
- queue_work(afs_wq, &call->work);
- return 0;
+ return afs_queue_call_work(call);
}
diff --git a/fs/afs/internal.h b/fs/afs/internal.h
index b411670d5f67..65504e218d35 100644
--- a/fs/afs/internal.h
+++ b/fs/afs/internal.h
@@ -66,7 +66,7 @@ enum afs_call_state {
struct afs_call {
const struct afs_call_type *type; /* type of call */
wait_queue_head_t waitq; /* processes awaiting completion */
- struct work_struct async_work; /* asynchronous work processor */
+ struct work_struct async_work; /* async I/O processor */
struct work_struct work; /* actual work processor */
struct rxrpc_call *rxcall; /* RxRPC call handle */
struct key *key; /* security for this call */
@@ -82,6 +82,7 @@ struct afs_call {
pgoff_t first; /* first page in mapping to deal with */
pgoff_t last; /* last page in mapping to deal with */
size_t offset; /* offset into received data store */
+ atomic_t usage;
enum afs_call_state state;
int error; /* error code */
u32 abort_code; /* Remote abort ID or 0 */
@@ -115,6 +116,9 @@ struct afs_call_type {
/* clean up a call */
void (*destructor)(struct afs_call *call);
+
+ /* Work function */
+ void (*work)(struct work_struct *work);
};
/*
@@ -591,9 +595,12 @@ extern void afs_proc_cell_remove(struct afs_cell *);
* rxrpc.c
*/
extern struct socket *afs_socket;
+extern atomic_t afs_outstanding_calls;
extern int afs_open_socket(void);
extern void afs_close_socket(void);
+extern void afs_put_call(struct afs_call *);
+extern int afs_queue_call_work(struct afs_call *);
extern int afs_make_call(struct in_addr *, struct afs_call *, gfp_t, bool);
extern struct afs_call *afs_alloc_flat_call(const struct afs_call_type *,
size_t, size_t);
diff --git a/fs/afs/rxrpc.c b/fs/afs/rxrpc.c
index ec1e41f929d1..95f42872b787 100644
--- a/fs/afs/rxrpc.c
+++ b/fs/afs/rxrpc.c
@@ -19,9 +19,8 @@
struct socket *afs_socket; /* my RxRPC socket */
static struct workqueue_struct *afs_async_calls;
static struct afs_call *afs_spare_incoming_call;
-static atomic_t afs_outstanding_calls;
+atomic_t afs_outstanding_calls;
-static void afs_free_call(struct afs_call *);
static void afs_wake_up_call_waiter(struct sock *, struct rxrpc_call *, unsigned long);
static int afs_wait_for_call_to_complete(struct afs_call *);
static void afs_wake_up_async_call(struct sock *, struct rxrpc_call *, unsigned long);
@@ -112,9 +111,11 @@ void afs_close_socket(void)
{
_enter("");
+ kernel_listen(afs_socket, 0);
+ flush_workqueue(afs_async_calls);
+
if (afs_spare_incoming_call) {
- atomic_inc(&afs_outstanding_calls);
- afs_free_call(afs_spare_incoming_call);
+ afs_put_call(afs_spare_incoming_call);
afs_spare_incoming_call = NULL;
}
@@ -123,7 +124,6 @@ void afs_close_socket(void)
TASK_UNINTERRUPTIBLE);
_debug("no outstanding calls");
- flush_workqueue(afs_async_calls);
kernel_sock_shutdown(afs_socket, SHUT_RDWR);
flush_workqueue(afs_async_calls);
sock_release(afs_socket);
@@ -134,44 +134,79 @@ void afs_close_socket(void)
}
/*
- * free a call
+ * Allocate a call.
*/
-static void afs_free_call(struct afs_call *call)
+static struct afs_call *afs_alloc_call(const struct afs_call_type *type,
+ gfp_t gfp)
{
- _debug("DONE %p{%s} [%d]",
- call, call->type->name, atomic_read(&afs_outstanding_calls));
+ struct afs_call *call;
+ int o;
- ASSERTCMP(call->rxcall, ==, NULL);
- ASSERT(!work_pending(&call->async_work));
- ASSERT(call->type->name != NULL);
+ call = kzalloc(sizeof(*call), gfp);
+ if (!call)
+ return NULL;
- kfree(call->request);
- kfree(call);
+ call->type = type;
+ atomic_set(&call->usage, 1);
+ INIT_WORK(&call->async_work, afs_process_async_call);
+ init_waitqueue_head(&call->waitq);
- if (atomic_dec_and_test(&afs_outstanding_calls))
- wake_up_atomic_t(&afs_outstanding_calls);
+ o = atomic_inc_return(&afs_outstanding_calls);
+ trace_afs_call(call, afs_call_trace_alloc, 1, o,
+ __builtin_return_address(0));
+ return call;
}
/*
- * End a call but do not free it
+ * Dispose of a reference on a call.
*/
-static void afs_end_call_nofree(struct afs_call *call)
+void afs_put_call(struct afs_call *call)
{
- if (call->rxcall) {
- rxrpc_kernel_end_call(afs_socket, call->rxcall);
- call->rxcall = NULL;
+ int n = atomic_dec_return(&call->usage);
+ int o = atomic_read(&afs_outstanding_calls);
+
+ trace_afs_call(call, afs_call_trace_put, n + 1, o,
+ __builtin_return_address(0));
+
+ ASSERTCMP(n, >=, 0);
+ if (n == 0) {
+ ASSERT(!work_pending(&call->async_work));
+ ASSERT(call->type->name != NULL);
+
+ if (call->rxcall) {
+ rxrpc_kernel_end_call(afs_socket, call->rxcall);
+ call->rxcall = NULL;
+ }
+ if (call->type->destructor)
+ call->type->destructor(call);
+
+ kfree(call->request);
+ kfree(call);
+
+ o = atomic_dec_return(&afs_outstanding_calls);
+ trace_afs_call(call, afs_call_trace_free, 0, o,
+ __builtin_return_address(0));
+ if (o == 0)
+ wake_up_atomic_t(&afs_outstanding_calls);
}
- if (call->type->destructor)
- call->type->destructor(call);
}
/*
- * End a call and free it
+ * Queue the call for actual work. Returns 0 unconditionally for convenience.
*/
-static void afs_end_call(struct afs_call *call)
+int afs_queue_call_work(struct afs_call *call)
{
- afs_end_call_nofree(call);
- afs_free_call(call);
+ int u = atomic_inc_return(&call->usage);
+
+ trace_afs_call(call, afs_call_trace_work, u,
+ atomic_read(&afs_outstanding_calls),
+ __builtin_return_address(0));
+
+ INIT_WORK(&call->work, call->type->work);
+
+ if (!queue_work(afs_wq, &call->work))
+ afs_put_call(call);
+ return 0;
}
/*
@@ -182,25 +217,19 @@ struct afs_call *afs_alloc_flat_call(const struct afs_call_type *type,
{
struct afs_call *call;
- call = kzalloc(sizeof(*call), GFP_NOFS);
+ call = afs_alloc_call(type, GFP_NOFS);
if (!call)
goto nomem_call;
- _debug("CALL %p{%s} [%d]",
- call, type->name, atomic_read(&afs_outstanding_calls));
- atomic_inc(&afs_outstanding_calls);
-
- call->type = type;
- call->request_size = request_size;
- call->reply_max = reply_max;
-
if (request_size) {
+ call->request_size = request_size;
call->request = kmalloc(request_size, GFP_NOFS);
if (!call->request)
goto nomem_free;
}
if (reply_max) {
+ call->reply_max = reply_max;
call->buffer = kmalloc(reply_max, GFP_NOFS);
if (!call->buffer)
goto nomem_free;
@@ -210,7 +239,7 @@ struct afs_call *afs_alloc_flat_call(const struct afs_call_type *type,
return call;
nomem_free:
- afs_free_call(call);
+ afs_put_call(call);
nomem_call:
return NULL;
}
@@ -315,7 +344,6 @@ int afs_make_call(struct in_addr *addr, struct afs_call *call, gfp_t gfp,
atomic_read(&afs_outstanding_calls));
call->async = async;
- INIT_WORK(&call->async_work, afs_process_async_call);
memset(&srx, 0, sizeof(srx));
srx.srx_family = AF_RXRPC;
@@ -378,7 +406,7 @@ int afs_make_call(struct in_addr *addr, struct afs_call *call, gfp_t gfp,
error_do_abort:
rxrpc_kernel_abort_call(afs_socket, rxcall, RX_USER_ABORT, -ret, "KSD");
error_kill_call:
- afs_end_call(call);
+ afs_put_call(call);
_leave(" = %d", ret);
return ret;
}
@@ -448,7 +476,7 @@ static void afs_deliver_to_call(struct afs_call *call)
done:
if (call->state == AFS_CALL_COMPLETE && call->incoming)
- afs_end_call(call);
+ afs_put_call(call);
out:
_leave("");
return;
@@ -505,7 +533,7 @@ static int afs_wait_for_call_to_complete(struct afs_call *call)
}
_debug("call complete");
- afs_end_call(call);
+ afs_put_call(call);
_leave(" = %d", ret);
return ret;
}
@@ -529,14 +557,25 @@ static void afs_wake_up_async_call(struct sock *sk, struct rxrpc_call *rxcall,
unsigned long call_user_ID)
{
struct afs_call *call = (struct afs_call *)call_user_ID;
+ int u;
trace_afs_notify_call(rxcall, call);
call->need_attention = true;
- queue_work(afs_async_calls, &call->async_work);
+
+ u = __atomic_add_unless(&call->usage, 1, 0);
+ if (u != 0) {
+ trace_afs_call(call, afs_call_trace_wake, u,
+ atomic_read(&afs_outstanding_calls),
+ __builtin_return_address(0));
+
+ if (!queue_work(afs_async_calls, &call->async_work))
+ afs_put_call(call);
+ }
}
/*
- * delete an asynchronous call
+ * Delete an asynchronous call. The work item carries a ref to the call struct
+ * that we need to release.
*/
static void afs_delete_async_call(struct work_struct *work)
{
@@ -544,13 +583,14 @@ static void afs_delete_async_call(struct work_struct *work)
_enter("");
- afs_free_call(call);
+ afs_put_call(call);
_leave("");
}
/*
- * perform processing on an asynchronous call
+ * Perform I/O processing on an asynchronous call. The work item carries a ref
+ * to the call struct that we either need to release or to pass on.
*/
static void afs_process_async_call(struct work_struct *work)
{
@@ -566,15 +606,16 @@ static void afs_process_async_call(struct work_struct *work)
if (call->state == AFS_CALL_COMPLETE) {
call->reply = NULL;
- /* kill the call */
- afs_end_call_nofree(call);
-
- /* we can't just delete the call because the work item may be
- * queued */
+ /* We have two refs to release - one from the alloc and one
+ * queued with the work item - and we can't just deallocate the
+ * call because the work item may be queued again.
+ */
call->async_work.func = afs_delete_async_call;
- queue_work(afs_async_calls, &call->async_work);
+ if (!queue_work(afs_async_calls, &call->async_work))
+ afs_put_call(call);
}
+ afs_put_call(call);
_leave("");
}
@@ -594,12 +635,10 @@ static void afs_charge_preallocation(struct work_struct *work)
for (;;) {
if (!call) {
- call = kzalloc(sizeof(struct afs_call), GFP_KERNEL);
+ call = afs_alloc_call(&afs_RXCMxxxx, GFP_KERNEL);
if (!call)
break;
- INIT_WORK(&call->async_work, afs_process_async_call);
- call->type = &afs_RXCMxxxx;
call->async = true;
call->state = AFS_CALL_AWAIT_OP_ID;
init_waitqueue_head(&call->waitq);
@@ -624,9 +663,8 @@ static void afs_rx_discard_new_call(struct rxrpc_call *rxcall,
{
struct afs_call *call = (struct afs_call *)user_call_ID;
- atomic_inc(&afs_outstanding_calls);
call->rxcall = NULL;
- afs_free_call(call);
+ afs_put_call(call);
}
/*
@@ -635,7 +673,6 @@ static void afs_rx_discard_new_call(struct rxrpc_call *rxcall,
static void afs_rx_new_call(struct sock *sk, struct rxrpc_call *rxcall,
unsigned long user_call_ID)
{
- atomic_inc(&afs_outstanding_calls);
queue_work(afs_wq, &afs_charge_preallocation_work);
}
@@ -699,7 +736,6 @@ void afs_send_empty_reply(struct afs_call *call)
rxrpc_kernel_abort_call(afs_socket, call->rxcall,
RX_USER_ABORT, ENOMEM, "KOO");
default:
- afs_end_call(call);
_leave(" [error]");
return;
}
@@ -738,7 +774,6 @@ void afs_send_simple_reply(struct afs_call *call, const void *buf, size_t len)
rxrpc_kernel_abort_call(afs_socket, call->rxcall,
RX_USER_ABORT, ENOMEM, "KOO");
}
- afs_end_call(call);
_leave(" [error]");
}
diff --git a/include/trace/events/afs.h b/include/trace/events/afs.h
index 845907b04ff4..8b95c16b7045 100644
--- a/include/trace/events/afs.h
+++ b/include/trace/events/afs.h
@@ -16,6 +16,51 @@
#include <linux/tracepoint.h>
+/*
+ * Define enums for tracing information.
+ */
+#ifndef __AFS_DECLARE_TRACE_ENUMS_ONCE_ONLY
+#define __AFS_DECLARE_TRACE_ENUMS_ONCE_ONLY
+
+enum afs_call_trace {
+ afs_call_trace_alloc,
+ afs_call_trace_free,
+ afs_call_trace_put,
+ afs_call_trace_wake,
+ afs_call_trace_work,
+};
+
+#endif /* end __AFS_DECLARE_TRACE_ENUMS_ONCE_ONLY */
+
+/*
+ * Declare tracing information enums and their string mappings for display.
+ */
+#define afs_call_traces \
+ EM(afs_call_trace_alloc, "ALLOC") \
+ EM(afs_call_trace_free, "FREE ") \
+ EM(afs_call_trace_put, "PUT ") \
+ EM(afs_call_trace_wake, "WAKE ") \
+ E_(afs_call_trace_work, "WORK ")
+
+/*
+ * Export enum symbols via userspace.
+ */
+#undef EM
+#undef E_
+#define EM(a, b) TRACE_DEFINE_ENUM(a);
+#define E_(a, b) TRACE_DEFINE_ENUM(a);
+
+afs_call_traces;
+
+/*
+ * Now redefine the EM() and E_() macros to map the enums to the strings that
+ * will be printed in the output.
+ */
+#undef EM
+#undef E_
+#define EM(a, b) { a, b },
+#define E_(a, b) { a, b }
+
TRACE_EVENT(afs_recv_data,
TP_PROTO(struct afs_call *call, unsigned count, unsigned offset,
bool want_more, int ret),
@@ -103,6 +148,36 @@ TRACE_EVENT(afs_cb_call,
__entry->op)
);
+TRACE_EVENT(afs_call,
+ TP_PROTO(struct afs_call *call, enum afs_call_trace op,
+ int usage, int outstanding, const void *where),
+
+ TP_ARGS(call, op, usage, outstanding, where),
+
+ TP_STRUCT__entry(
+ __field(struct afs_call *, call )
+ __field(int, op )
+ __field(int, usage )
+ __field(int, outstanding )
+ __field(const void *, where )
+ ),
+
+ TP_fast_assign(
+ __entry->call = call;
+ __entry->op = op;
+ __entry->usage = usage;
+ __entry->outstanding = outstanding;
+ __entry->where = where;
+ ),
+
+ TP_printk("c=%p %s u=%d o=%d sp=%pSR",
+ __entry->call,
+ __print_symbolic(__entry->op, afs_call_traces),
+ __entry->usage,
+ __entry->outstanding,
+ __entry->where)
+ );
+
#endif /* _TRACE_AFS_H */
/* This part must be outside protection */
^ permalink raw reply related
* [net-next PATCH 0/3] net: optimize ICMP-reply code path
From: Jesper Dangaard Brouer @ 2017-01-09 15:03 UTC (permalink / raw)
To: netdev, Eric Dumazet, Jesper Dangaard Brouer; +Cc: xiyou.wangcong
This patchset is optimizing the ICMP-reply code path, for ICMP packets
that gets rate limited. A remote party can easily trigger this code
path by sending packets to port number with no listening service.
Generally the patchset moves the sysctl_icmp_msgs_per_sec ratelimit
checking to earlier in the code path and removes an allocation.
Use-case: The specific case I experienced this being a bottleneck is,
sending UDP packets to a port with no listener, which obviously result
in kernel replying with ICMP Destination Unreachable (type:3), Port
Unreachable (code:3), which cause the bottleneck.
After Eric and Paolo optimized the UDP socket code, the kernels PPS
processing capabilities is lower for no-listen ports, than normal UDP
sockets. This is bad for capacity planning when restarting a service.
UDP no-listen benchmark 8xCPUs using pktgen_sample04_many_flows.sh:
Baseline: 6.6 Mpps
Patch: 14.7 Mpps
Driver mlx5 at 50Gbit/s.
---
Jesper Dangaard Brouer (3):
Revert "icmp: avoid allocating large struct on stack"
net: reduce cycles spend on ICMP replies that gets rate limited
net: for rate-limited ICMP replies save one atomic operation
net/ipv4/icmp.c | 125 +++++++++++++++++++++++++++++++++----------------------
net/ipv6/icmp.c | 68 +++++++++++++++++++++---------
2 files changed, 123 insertions(+), 70 deletions(-)
^ permalink raw reply
* [net-next PATCH 1/3] Revert "icmp: avoid allocating large struct on stack"
From: Jesper Dangaard Brouer @ 2017-01-09 15:04 UTC (permalink / raw)
To: netdev, Eric Dumazet, Jesper Dangaard Brouer; +Cc: xiyou.wangcong
In-Reply-To: <20170109150246.30215.63371.stgit@firesoul>
This reverts commit 9a99d4a50cb8 ("icmp: avoid allocating large struct
on stack"), because struct icmp_bxm no really a large struct, and
allocating and free of this small 112 bytes hurts performance.
Fixes: 9a99d4a50cb8 ("icmp: avoid allocating large struct on stack")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
---
net/ipv4/icmp.c | 40 +++++++++++++++++-----------------------
1 file changed, 17 insertions(+), 23 deletions(-)
diff --git a/net/ipv4/icmp.c b/net/ipv4/icmp.c
index 0777ea949223..b4b9807329a7 100644
--- a/net/ipv4/icmp.c
+++ b/net/ipv4/icmp.c
@@ -571,7 +571,7 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
{
struct iphdr *iph;
int room;
- struct icmp_bxm *icmp_param;
+ struct icmp_bxm icmp_param;
struct rtable *rt = skb_rtable(skb_in);
struct ipcm_cookie ipc;
struct flowi4 fl4;
@@ -648,13 +648,9 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
}
}
- icmp_param = kmalloc(sizeof(*icmp_param), GFP_ATOMIC);
- if (!icmp_param)
- return;
-
sk = icmp_xmit_lock(net);
if (!sk)
- goto out_free;
+ return;
/*
* Construct source address and options.
@@ -681,7 +677,7 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
iph->tos;
mark = IP4_REPLY_MARK(net, skb_in->mark);
- if (ip_options_echo(&icmp_param->replyopts.opt.opt, skb_in))
+ if (ip_options_echo(&icmp_param.replyopts.opt.opt, skb_in))
goto out_unlock;
@@ -689,22 +685,22 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
* Prepare data for ICMP header.
*/
- icmp_param->data.icmph.type = type;
- icmp_param->data.icmph.code = code;
- icmp_param->data.icmph.un.gateway = info;
- icmp_param->data.icmph.checksum = 0;
- icmp_param->skb = skb_in;
- icmp_param->offset = skb_network_offset(skb_in);
+ icmp_param.data.icmph.type = type;
+ icmp_param.data.icmph.code = code;
+ icmp_param.data.icmph.un.gateway = info;
+ icmp_param.data.icmph.checksum = 0;
+ icmp_param.skb = skb_in;
+ icmp_param.offset = skb_network_offset(skb_in);
inet_sk(sk)->tos = tos;
sk->sk_mark = mark;
ipc.addr = iph->saddr;
- ipc.opt = &icmp_param->replyopts.opt;
+ ipc.opt = &icmp_param.replyopts.opt;
ipc.tx_flags = 0;
ipc.ttl = 0;
ipc.tos = -1;
rt = icmp_route_lookup(net, &fl4, skb_in, iph, saddr, tos, mark,
- type, code, icmp_param);
+ type, code, &icmp_param);
if (IS_ERR(rt))
goto out_unlock;
@@ -716,21 +712,19 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
room = dst_mtu(&rt->dst);
if (room > 576)
room = 576;
- room -= sizeof(struct iphdr) + icmp_param->replyopts.opt.opt.optlen;
+ room -= sizeof(struct iphdr) + icmp_param.replyopts.opt.opt.optlen;
room -= sizeof(struct icmphdr);
- icmp_param->data_len = skb_in->len - icmp_param->offset;
- if (icmp_param->data_len > room)
- icmp_param->data_len = room;
- icmp_param->head_len = sizeof(struct icmphdr);
+ icmp_param.data_len = skb_in->len - icmp_param.offset;
+ if (icmp_param.data_len > room)
+ icmp_param.data_len = room;
+ icmp_param.head_len = sizeof(struct icmphdr);
- icmp_push_reply(icmp_param, &fl4, &ipc, &rt);
+ icmp_push_reply(&icmp_param, &fl4, &ipc, &rt);
ende:
ip_rt_put(rt);
out_unlock:
icmp_xmit_unlock(sk);
-out_free:
- kfree(icmp_param);
out:;
}
EXPORT_SYMBOL(icmp_send);
^ permalink raw reply related
* [net-next PATCH 2/3] net: reduce cycles spend on ICMP replies that gets rate limited
From: Jesper Dangaard Brouer @ 2017-01-09 15:04 UTC (permalink / raw)
To: netdev, Eric Dumazet, Jesper Dangaard Brouer; +Cc: xiyou.wangcong
In-Reply-To: <20170109150246.30215.63371.stgit@firesoul>
This patch split the global and per (inet)peer ICMP-reply limiter
code, and moves the global limit check to earlier in the packet
processing path. Thus, avoid spending cycles on ICMP replies that
gets limited/suppressed anyhow.
The global ICMP rate limiter icmp_global_allow() is a good solution,
it just happens too late in the process. The kernel goes through the
full route lookup (return path) for the ICMP message, before taking
the rate limit decision of not sending the ICMP reply.
Details: The kernels global rate limiter for ICMP messages got added
in commit 4cdf507d5452 ("icmp: add a global rate limitation"). It is
a token bucket limiter with a global lock. It brilliantly avoids
locking congestion by only updating when 20ms (HZ/50) were elapsed. It
can then avoids taking lock when credit is exhausted (when under
pressure) and time constraint for refill is not yet meet.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
---
net/ipv4/icmp.c | 71 +++++++++++++++++++++++++++++++++++++------------------
net/ipv6/icmp.c | 49 ++++++++++++++++++++++++++------------
2 files changed, 82 insertions(+), 38 deletions(-)
diff --git a/net/ipv4/icmp.c b/net/ipv4/icmp.c
index b4b9807329a7..58d75ca58b83 100644
--- a/net/ipv4/icmp.c
+++ b/net/ipv4/icmp.c
@@ -282,6 +282,33 @@ bool icmp_global_allow(void)
}
EXPORT_SYMBOL(icmp_global_allow);
+static bool icmpv4_mask_allow(struct net *net, int type, int code)
+{
+ if (type > NR_ICMP_TYPES)
+ return true;
+
+ /* Don't limit PMTU discovery. */
+ if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED)
+ return true;
+
+ /* Limit if icmp type is enabled in ratemask. */
+ if (!((1 << type) & net->ipv4.sysctl_icmp_ratemask))
+ return true;
+
+ return false;
+}
+
+static bool icmpv4_global_allow(struct net *net, int type, int code)
+{
+ if (icmpv4_mask_allow(net, type, code))
+ return true;
+
+ if (icmp_global_allow())
+ return true;
+
+ return false;
+}
+
/*
* Send an ICMP frame.
*/
@@ -290,34 +317,22 @@ static bool icmpv4_xrlim_allow(struct net *net, struct rtable *rt,
struct flowi4 *fl4, int type, int code)
{
struct dst_entry *dst = &rt->dst;
+ struct inet_peer *peer;
bool rc = true;
+ int vif;
- if (type > NR_ICMP_TYPES)
- goto out;
-
- /* Don't limit PMTU discovery. */
- if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED)
+ if (icmpv4_mask_allow(net, type, code))
goto out;
/* No rate limit on loopback */
if (dst->dev && (dst->dev->flags&IFF_LOOPBACK))
goto out;
- /* Limit if icmp type is enabled in ratemask. */
- if (!((1 << type) & net->ipv4.sysctl_icmp_ratemask))
- goto out;
-
- rc = false;
- if (icmp_global_allow()) {
- int vif = l3mdev_master_ifindex(dst->dev);
- struct inet_peer *peer;
-
- peer = inet_getpeer_v4(net->ipv4.peers, fl4->daddr, vif, 1);
- rc = inet_peer_xrlim_allow(peer,
- net->ipv4.sysctl_icmp_ratelimit);
- if (peer)
- inet_putpeer(peer);
- }
+ vif = l3mdev_master_ifindex(dst->dev);
+ peer = inet_getpeer_v4(net->ipv4.peers, fl4->daddr, vif, 1);
+ rc = inet_peer_xrlim_allow(peer, net->ipv4.sysctl_icmp_ratelimit);
+ if (peer)
+ inet_putpeer(peer);
out:
return rc;
}
@@ -396,6 +411,8 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
struct inet_sock *inet;
__be32 daddr, saddr;
u32 mark = IP4_REPLY_MARK(net, skb->mark);
+ int type = icmp_param->data.icmph.type;
+ int code = icmp_param->data.icmph.code;
if (ip_options_echo(&icmp_param->replyopts.opt.opt, skb))
return;
@@ -405,6 +422,10 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
return;
inet = inet_sk(sk);
+ /* global icmp_msgs_per_sec */
+ if (!icmpv4_global_allow(net, type, code))
+ goto out_unlock;
+
icmp_param->data.icmph.checksum = 0;
inet->tos = ip_hdr(skb)->tos;
@@ -433,8 +454,7 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
rt = ip_route_output_key(net, &fl4);
if (IS_ERR(rt))
goto out_unlock;
- if (icmpv4_xrlim_allow(net, rt, &fl4, icmp_param->data.icmph.type,
- icmp_param->data.icmph.code))
+ if (icmpv4_xrlim_allow(net, rt, &fl4, type, code))
icmp_push_reply(icmp_param, &fl4, &ipc, &rt);
ip_rt_put(rt);
out_unlock:
@@ -650,7 +670,11 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
sk = icmp_xmit_lock(net);
if (!sk)
- return;
+ goto out;
+
+ /* Check global sysctl_icmp_msgs_per_sec ratelimit */
+ if (!icmpv4_global_allow(net, type, code))
+ goto out_unlock;
/*
* Construct source address and options.
@@ -704,6 +728,7 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
if (IS_ERR(rt))
goto out_unlock;
+ /* peer icmp_ratelimit */
if (!icmpv4_xrlim_allow(net, rt, &fl4, type, code))
goto ende;
diff --git a/net/ipv6/icmp.c b/net/ipv6/icmp.c
index 3036f665e6c8..b26ae8b5c1ce 100644
--- a/net/ipv6/icmp.c
+++ b/net/ipv6/icmp.c
@@ -168,6 +168,30 @@ static bool is_ineligible(const struct sk_buff *skb)
return false;
}
+static bool icmpv6_mask_allow(int type)
+{
+ /* Informational messages are not limited. */
+ if (type & ICMPV6_INFOMSG_MASK)
+ return true;
+
+ /* Do not limit pmtu discovery, it would break it. */
+ if (type == ICMPV6_PKT_TOOBIG)
+ return true;
+
+ return false;
+}
+
+static bool icmpv6_global_allow(int type)
+{
+ if (icmpv6_mask_allow(type))
+ return true;
+
+ if (icmp_global_allow())
+ return true;
+
+ return false;
+}
+
/*
* Check the ICMP output rate limit
*/
@@ -178,12 +202,7 @@ static bool icmpv6_xrlim_allow(struct sock *sk, u8 type,
struct dst_entry *dst;
bool res = false;
- /* Informational messages are not limited. */
- if (type & ICMPV6_INFOMSG_MASK)
- return true;
-
- /* Do not limit pmtu discovery, it would break it. */
- if (type == ICMPV6_PKT_TOOBIG)
+ if (icmpv6_mask_allow(type))
return true;
/*
@@ -200,20 +219,16 @@ static bool icmpv6_xrlim_allow(struct sock *sk, u8 type,
} else {
struct rt6_info *rt = (struct rt6_info *)dst;
int tmo = net->ipv6.sysctl.icmpv6_time;
+ struct inet_peer *peer;
/* Give more bandwidth to wider prefixes. */
if (rt->rt6i_dst.plen < 128)
tmo >>= ((128 - rt->rt6i_dst.plen)>>5);
- if (icmp_global_allow()) {
- struct inet_peer *peer;
-
- peer = inet_getpeer_v6(net->ipv6.peers,
- &fl6->daddr, 1);
- res = inet_peer_xrlim_allow(peer, tmo);
- if (peer)
- inet_putpeer(peer);
- }
+ peer = inet_getpeer_v6(net->ipv6.peers, &fl6->daddr, 1);
+ res = inet_peer_xrlim_allow(peer, tmo);
+ if (peer)
+ inet_putpeer(peer);
}
dst_release(dst);
return res;
@@ -493,6 +508,10 @@ static void icmp6_send(struct sk_buff *skb, u8 type, u8 code, __u32 info,
sk = icmpv6_xmit_lock(net);
if (!sk)
return;
+
+ if (!icmpv6_global_allow(type))
+ goto out;
+
sk->sk_mark = mark;
np = inet6_sk(sk);
^ permalink raw reply related
* [net-next PATCH 3/3] net: for rate-limited ICMP replies save one atomic operation
From: Jesper Dangaard Brouer @ 2017-01-09 15:04 UTC (permalink / raw)
To: netdev, Eric Dumazet, Jesper Dangaard Brouer; +Cc: xiyou.wangcong
In-Reply-To: <20170109150246.30215.63371.stgit@firesoul>
It is possible to avoid the atomic operation in icmp{v6,}_xmit_lock,
by checking the sysctl_icmp_msgs_per_sec ratelimit before these calls,
as pointed out by Eric Dumazet, but the BH disabled state must be correct.
The icmp_global_allow() call states it must be called with BH
disabled. This protection was given by the calls icmp_xmit_lock and
icmpv6_xmit_lock. Thus, split out local_bh_disable/enable from these
functions and maintain it explicitly at callers.
Suggested-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
---
net/ipv4/icmp.c | 34 +++++++++++++++++++++-------------
net/ipv6/icmp.c | 25 ++++++++++++++++---------
2 files changed, 37 insertions(+), 22 deletions(-)
diff --git a/net/ipv4/icmp.c b/net/ipv4/icmp.c
index 58d75ca58b83..fc310db2708b 100644
--- a/net/ipv4/icmp.c
+++ b/net/ipv4/icmp.c
@@ -209,19 +209,17 @@ static struct sock *icmp_sk(struct net *net)
return *this_cpu_ptr(net->ipv4.icmp_sk);
}
+/* Called with BH disabled */
static inline struct sock *icmp_xmit_lock(struct net *net)
{
struct sock *sk;
- local_bh_disable();
-
sk = icmp_sk(net);
if (unlikely(!spin_trylock(&sk->sk_lock.slock))) {
/* This can happen if the output path signals a
* dst_link_failure() for an outgoing ICMP packet.
*/
- local_bh_enable();
return NULL;
}
return sk;
@@ -229,7 +227,7 @@ static inline struct sock *icmp_xmit_lock(struct net *net)
static inline void icmp_xmit_unlock(struct sock *sk)
{
- spin_unlock_bh(&sk->sk_lock.slock);
+ spin_unlock(&sk->sk_lock.slock);
}
int sysctl_icmp_msgs_per_sec __read_mostly = 1000;
@@ -417,14 +415,17 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
if (ip_options_echo(&icmp_param->replyopts.opt.opt, skb))
return;
- sk = icmp_xmit_lock(net);
- if (!sk)
- return;
- inet = inet_sk(sk);
+ /* Needed by both icmp_global_allow and icmp_xmit_lock */
+ local_bh_disable();
/* global icmp_msgs_per_sec */
if (!icmpv4_global_allow(net, type, code))
- goto out_unlock;
+ goto out_bh_enable;
+
+ sk = icmp_xmit_lock(net);
+ if (!sk)
+ goto out_bh_enable;
+ inet = inet_sk(sk);
icmp_param->data.icmph.checksum = 0;
@@ -459,6 +460,8 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
ip_rt_put(rt);
out_unlock:
icmp_xmit_unlock(sk);
+out_bh_enable:
+ local_bh_enable();
}
#ifdef CONFIG_IP_ROUTE_MULTIPATH
@@ -668,13 +671,16 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
}
}
- sk = icmp_xmit_lock(net);
- if (!sk)
- goto out;
+ /* Needed by both icmp_global_allow and icmp_xmit_lock */
+ local_bh_disable();
/* Check global sysctl_icmp_msgs_per_sec ratelimit */
if (!icmpv4_global_allow(net, type, code))
- goto out_unlock;
+ goto out_bh_enable;
+
+ sk = icmp_xmit_lock(net);
+ if (!sk)
+ goto out_bh_enable;
/*
* Construct source address and options.
@@ -750,6 +756,8 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
ip_rt_put(rt);
out_unlock:
icmp_xmit_unlock(sk);
+out_bh_enable:
+ local_bh_enable();
out:;
}
EXPORT_SYMBOL(icmp_send);
diff --git a/net/ipv6/icmp.c b/net/ipv6/icmp.c
index b26ae8b5c1ce..230b5aac9f03 100644
--- a/net/ipv6/icmp.c
+++ b/net/ipv6/icmp.c
@@ -110,19 +110,17 @@ static const struct inet6_protocol icmpv6_protocol = {
.flags = INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
};
+/* Called with BH disabled */
static __inline__ struct sock *icmpv6_xmit_lock(struct net *net)
{
struct sock *sk;
- local_bh_disable();
-
sk = icmpv6_sk(net);
if (unlikely(!spin_trylock(&sk->sk_lock.slock))) {
/* This can happen if the output path (f.e. SIT or
* ip6ip6 tunnel) signals dst_link_failure() for an
* outgoing ICMP6 packet.
*/
- local_bh_enable();
return NULL;
}
return sk;
@@ -130,7 +128,7 @@ static __inline__ struct sock *icmpv6_xmit_lock(struct net *net)
static __inline__ void icmpv6_xmit_unlock(struct sock *sk)
{
- spin_unlock_bh(&sk->sk_lock.slock);
+ spin_unlock(&sk->sk_lock.slock);
}
/*
@@ -489,6 +487,13 @@ static void icmp6_send(struct sk_buff *skb, u8 type, u8 code, __u32 info,
return;
}
+ /* Needed by both icmp_global_allow and icmpv6_xmit_lock */
+ local_bh_disable();
+
+ /* Check global sysctl_icmp_msgs_per_sec ratelimit */
+ if (!icmpv6_global_allow(type))
+ goto out_bh_enable;
+
mip6_addr_swap(skb);
memset(&fl6, 0, sizeof(fl6));
@@ -507,10 +512,7 @@ static void icmp6_send(struct sk_buff *skb, u8 type, u8 code, __u32 info,
sk = icmpv6_xmit_lock(net);
if (!sk)
- return;
-
- if (!icmpv6_global_allow(type))
- goto out;
+ goto out_bh_enable;
sk->sk_mark = mark;
np = inet6_sk(sk);
@@ -571,6 +573,8 @@ static void icmp6_send(struct sk_buff *skb, u8 type, u8 code, __u32 info,
dst_release(dst);
out:
icmpv6_xmit_unlock(sk);
+out_bh_enable:
+ local_bh_enable();
}
/* Slightly more convenient version of icmp6_send.
@@ -684,9 +688,10 @@ static void icmpv6_echo_reply(struct sk_buff *skb)
fl6.flowi6_uid = sock_net_uid(net, NULL);
security_skb_classify_flow(skb, flowi6_to_flowi(&fl6));
+ local_bh_disable();
sk = icmpv6_xmit_lock(net);
if (!sk)
- return;
+ goto out_bh_enable;
sk->sk_mark = mark;
np = inet6_sk(sk);
@@ -728,6 +733,8 @@ static void icmpv6_echo_reply(struct sk_buff *skb)
dst_release(dst);
out:
icmpv6_xmit_unlock(sk);
+out_bh_enable:
+ local_bh_enable();
}
void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info)
^ permalink raw reply related
* Re: [PATCH net-next v2] net: dsa: make "label" property optional for dsa2
From: Vivien Didelot @ 2017-01-09 15:04 UTC (permalink / raw)
To: Jiri Pirko
Cc: netdev, linux-kernel, kernel, David S. Miller, Florian Fainelli,
Andrew Lunn, Uwe Kleine-König, Andrey Smirnov
In-Reply-To: <20170109073236.GA1862@nanopsycho>
Hi Jiri,
Jiri Pirko <jiri@resnulli.us> writes:
>> # cat /etc/udev/rules.d/90-net-dsa.rules
>> SUBSYSTEM=="net", ACTION=="add", ENV{DEVTYPE}=="dsa", PROGRAM="/lib/udev/dsanitizer $attr{phys_switch_id} $attr{phys_port_id}", NAME="$result"
>
> I know this is kind of confusing, but phys_port_id is to be used to
> indicate same physical port that is shared by multiple netdevices- for
> example sr-iov usecase. For switchdev usecase, you should use
> phys_port_name.
>
> I will add some documentation to kernel regarding this. But I see that
> net/dsa/slave.c already implements .ndo_get_phys_port_id :(
>
> I recently made changes in udev so it names the switch ports according
> to phys_port_name, out of the box, without need for any rules:
> https://github.com/systemd/systemd/pull/4506/commits/c960caa0c2a620fc506c6f0f7b6c40eeace48e4d
Thanks for the details. So if I understand correctly, what will be found
in phys_port_name will be used as is by udev to name the interface?
Extra question: shouldn't phys_port_{id,name} be switchdev attributes in
addition to SWITCHDEV_ATTR_ID_PORT_PARENT_ID?
> I guess that it should be enough for you to implement
> ndo_get_phys_port_name.
Well, if this name must be unique on a system, it's not likely to happen
until we agree that we use an ugly tXsYpZ template where X is a tree ID,
or we assign system-wide unique IDs to switches, which requires a bit of
changes.
I'm thinking, since DSA slaves are switchdev users, can't we all use a
switchdev helper to optionally get a system-wide unique name for a given
switch port? e.g. a template such as "swp%d"? I'd prefer that switchdev
and DSA do not diverge much when it comes to implement such attributes.
But again, this is not related to this patch ;-)
Thanks,
Vivien
^ permalink raw reply
* Re: [PATCH net-next 0/4] net: dsa: Make dsa_switch_ops const
From: Vivien Didelot @ 2017-01-09 15:11 UTC (permalink / raw)
To: Florian Fainelli, netdev; +Cc: davem, andrew, Florian Fainelli
In-Reply-To: <20170108225208.15431-1-f.fainelli@gmail.com>
Hi Florian,
Florian Fainelli <f.fainelli@gmail.com> writes:
> This patch series allows us to annotate dsa_switch_ops with a const
> qualifier.
>
> Florian Fainelli (4):
> net: dsa: b53: Export most operations to other drivers
> net: dsa: bcm_sf2: Declare our own dsa_switch_ops
> net: dsa: Encapsulate legacy switch drivers into dsa_switch_driver
> net: dsa: Make dsa_switch_ops const
Reviewed-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
This is much cleaner now, thanks!
Vivien
^ permalink raw reply
* Re: [PATCH net-next v2] net: dsa: make "label" property optional for dsa2
From: Jiri Pirko @ 2017-01-09 15:11 UTC (permalink / raw)
To: Vivien Didelot
Cc: netdev, linux-kernel, kernel, David S. Miller, Florian Fainelli,
Andrew Lunn, Uwe Kleine-König, Andrey Smirnov
In-Reply-To: <877f6446lp.fsf@weeman.i-did-not-set--mail-host-address--so-tickle-me>
Mon, Jan 09, 2017 at 04:04:50PM CET, vivien.didelot@savoirfairelinux.com wrote:
>Hi Jiri,
>
>Jiri Pirko <jiri@resnulli.us> writes:
>
>>> # cat /etc/udev/rules.d/90-net-dsa.rules
>>> SUBSYSTEM=="net", ACTION=="add", ENV{DEVTYPE}=="dsa", PROGRAM="/lib/udev/dsanitizer $attr{phys_switch_id} $attr{phys_port_id}", NAME="$result"
>>
>> I know this is kind of confusing, but phys_port_id is to be used to
>> indicate same physical port that is shared by multiple netdevices- for
>> example sr-iov usecase. For switchdev usecase, you should use
>> phys_port_name.
>>
>> I will add some documentation to kernel regarding this. But I see that
>> net/dsa/slave.c already implements .ndo_get_phys_port_id :(
>>
>> I recently made changes in udev so it names the switch ports according
>> to phys_port_name, out of the box, without need for any rules:
>> https://github.com/systemd/systemd/pull/4506/commits/c960caa0c2a620fc506c6f0f7b6c40eeace48e4d
>
>Thanks for the details. So if I understand correctly, what will be found
>in phys_port_name will be used as is by udev to name the interface?
Yes.
>
>Extra question: shouldn't phys_port_{id,name} be switchdev attributes in
Again, phys_port_id has nothing to do with switches. Should be removed
from dsa because its use there is incorrect.
>addition to SWITCHDEV_ATTR_ID_PORT_PARENT_ID?
Perhaps it could. Now it is a netdev op. I thinks it works nicely.
>
>> I guess that it should be enough for you to implement
>> ndo_get_phys_port_name.
>
>Well, if this name must be unique on a system, it's not likely to happen
>until we agree that we use an ugly tXsYpZ template where X is a tree ID,
>or we assign system-wide unique IDs to switches, which requires a bit of
>changes.
No. That should be unique within one switch. In mlxsw we name it "p1",
"p2", ...
The final netdev names are:
enp3s0np1, enp3s0np2, ...
>
>I'm thinking, since DSA slaves are switchdev users, can't we all use a
>switchdev helper to optionally get a system-wide unique name for a given
>switch port? e.g. a template such as "swp%d"? I'd prefer that switchdev
>and DSA do not diverge much when it comes to implement such attributes.
Again, not system-wide, just switch-wide.
>
>But again, this is not related to this patch ;-)
It is! You are using phys_port_id, which is completely wrong. You should
not use it.
^ permalink raw reply
* [PATCH] vhost: scsi: constify target_core_fabric_ops structures
From: Bhumika Goyal @ 2017-01-09 15:21 UTC (permalink / raw)
To: julia.lawall, mst, jasowang, kvm, virtualization, netdev,
linux-kernel
Cc: Bhumika Goyal
Declare target_core_fabric_ops strucrues as const as they are only
passed as an argument to the functions target_register_template and
target_unregister_template. The arguments are of type const struct
target_core_fabric_ops *, so target_core_fabric_ops structures having
this property can be declared const.
Done using Coccinelle:
@r disable optional_qualifier@
identifier i;
position p;
@@
static struct target_core_fabric_ops i@p={...};
@ok@
position p;
identifier r.i;
@@
(
target_register_template(&i@p)
|
target_unregister_template(&i@p)
)
@bad@
position p!={r.p,ok.p};
identifier r.i;
@@
i@p
@depends on !bad disable optional_qualifier@
identifier r.i;
@@
+const
struct target_core_fabric_ops i;
File size before: drivers/vhost/scsi.o
text data bss dec hex filename
18063 2985 40 21088 5260 drivers/vhost/scsi.o
File size after: drivers/vhost/scsi.o
text data bss dec hex filename
18479 2601 40 21120 5280 drivers/vhost/scsi.o
Signed-off-by: Bhumika Goyal <bhumirks@gmail.com>
---
drivers/vhost/scsi.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/vhost/scsi.c b/drivers/vhost/scsi.c
index 253310c..620366d 100644
--- a/drivers/vhost/scsi.c
+++ b/drivers/vhost/scsi.c
@@ -2087,7 +2087,7 @@ static void vhost_scsi_drop_tport(struct se_wwn *wwn)
NULL,
};
-static struct target_core_fabric_ops vhost_scsi_ops = {
+static const struct target_core_fabric_ops vhost_scsi_ops = {
.module = THIS_MODULE,
.name = "vhost",
.get_fabric_name = vhost_scsi_get_fabric_name,
--
1.9.1
^ permalink raw reply related
* Re: [PATCH net-next] bridge: multicast to unicast
From: michael-dev @ 2017-01-09 15:25 UTC (permalink / raw)
To: Johannes Berg
Cc: Linus Lüssing, Felix Fietkau, netdev, David S . Miller,
Stephen Hemminger, bridge, linux-kernel, linux-wireless
In-Reply-To: <1483964159.17582.34.camel@sipsolutions.net>
Am 09.01.2017 13:15, schrieb Johannes Berg:
>> That is bridge fdb entries (need to) expire so the bridge might
>> "forget" a still-connected station not sending but only consuming
>> broadcast traffic.
>
> Ok, that I don't know. Somehow if you address a unicast packet there
> the bridge has to make a decision - so it really should know?
If the bridge has not learned the unicast destination mac address on any
port, it will flood the packet on all ports except the port it received
the packet on.
Regards,
M. Braun
^ permalink raw reply
* [PATCHv3 1/6] sh_eth: use correct name for ECMR_MPDE bit
From: Niklas Söderlund @ 2017-01-09 15:34 UTC (permalink / raw)
To: Sergei Shtylyov, Simon Horman, netdev, linux-renesas-soc
Cc: Geert Uytterhoeven, linux-pm, Niklas Söderlund
In-Reply-To: <20170109153409.13956-1-niklas.soderlund+renesas@ragnatech.se>
This bit was wrongly named due to a typo, Sergei checked the SH7734/63
manuals and this bit should be named MPDE.
Suggested-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
---
drivers/net/ethernet/renesas/sh_eth.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/renesas/sh_eth.h b/drivers/net/ethernet/renesas/sh_eth.h
index 9eec1e185adf..11798d5c4a1d 100644
--- a/drivers/net/ethernet/renesas/sh_eth.h
+++ b/drivers/net/ethernet/renesas/sh_eth.h
@@ -339,7 +339,7 @@ enum FELIC_MODE_BIT {
ECMR_DPAD = 0x00200000, ECMR_RZPF = 0x00100000,
ECMR_ZPF = 0x00080000, ECMR_PFR = 0x00040000, ECMR_RXF = 0x00020000,
ECMR_TXF = 0x00010000, ECMR_MCT = 0x00002000, ECMR_PRCEF = 0x00001000,
- ECMR_PMDE = 0x00000200, ECMR_RE = 0x00000040, ECMR_TE = 0x00000020,
+ ECMR_MPDE = 0x00000200, ECMR_RE = 0x00000040, ECMR_TE = 0x00000020,
ECMR_RTM = 0x00000010, ECMR_ILB = 0x00000008, ECMR_ELB = 0x00000004,
ECMR_DM = 0x00000002, ECMR_PRM = 0x00000001,
};
--
2.11.0
^ permalink raw reply related
* [PATCHv3 2/6] sh_eth: add generic wake-on-lan support via magic packet
From: Niklas Söderlund @ 2017-01-09 15:34 UTC (permalink / raw)
To: Sergei Shtylyov, Simon Horman, netdev, linux-renesas-soc
Cc: Geert Uytterhoeven, linux-pm, Niklas Söderlund
In-Reply-To: <20170109153409.13956-1-niklas.soderlund+renesas@ragnatech.se>
Add generic functionality to support Wake-on-LAN using MagicPacket which
are supported by at least a few versions of sh_eth. Only add
functionality for WoL, no specific sh_eth versions are marked to support
WoL yet.
WoL is enabled in the suspend callback by setting MagicPacket detection
and disabling all interrupts expect MagicPacket. In the resume path the
driver needs to reset the hardware to rearm the WoL logic, this prevents
the driver from simply restoring the registers and to take advantage of
that sh_eth was not suspended to reduce resume time. To reset the
hardware the driver closes and reopens the device just like it would do
in a normal suspend/resume scenario without WoL enabled, but it both
closes and opens the device in the resume callback since the device
needs to be open for WoL to work.
One quirk needed for WoL is that the module clock needs to be prevented
from being switched off by Runtime PM. To keep the clock alive the
suspend callback need to call clk_enable() directly to increase the
usage count of the clock. Then when Runtime PM decreases the clock usage
count it won't reach 0 and be switched off.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
---
drivers/net/ethernet/renesas/sh_eth.c | 114 +++++++++++++++++++++++++++++++---
drivers/net/ethernet/renesas/sh_eth.h | 3 +
2 files changed, 109 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c
index 8a784dce45fa..542c92b57b35 100644
--- a/drivers/net/ethernet/renesas/sh_eth.c
+++ b/drivers/net/ethernet/renesas/sh_eth.c
@@ -1552,6 +1552,8 @@ static void sh_eth_emac_interrupt(struct net_device *ndev)
sh_eth_rcv_snd_enable(ndev);
}
}
+ if (felic_stat & ECSR_MPD)
+ pm_wakeup_event(&mdp->pdev->dev, 0);
}
/* error control function */
@@ -2201,6 +2203,33 @@ static int sh_eth_set_ringparam(struct net_device *ndev,
return 0;
}
+static void sh_eth_get_wol(struct net_device *ndev, struct ethtool_wolinfo *wol)
+{
+ struct sh_eth_private *mdp = netdev_priv(ndev);
+
+ wol->supported = 0;
+ wol->wolopts = 0;
+
+ if (mdp->cd->magic && mdp->clk) {
+ wol->supported = WAKE_MAGIC;
+ wol->wolopts = mdp->wol_enabled ? WAKE_MAGIC : 0;
+ }
+}
+
+static int sh_eth_set_wol(struct net_device *ndev, struct ethtool_wolinfo *wol)
+{
+ struct sh_eth_private *mdp = netdev_priv(ndev);
+
+ if (!mdp->cd->magic || !mdp->clk || wol->wolopts & ~WAKE_MAGIC)
+ return -EOPNOTSUPP;
+
+ mdp->wol_enabled = !!(wol->wolopts & WAKE_MAGIC);
+
+ device_set_wakeup_enable(&mdp->pdev->dev, mdp->wol_enabled);
+
+ return 0;
+}
+
static const struct ethtool_ops sh_eth_ethtool_ops = {
.get_regs_len = sh_eth_get_regs_len,
.get_regs = sh_eth_get_regs,
@@ -2215,6 +2244,8 @@ static const struct ethtool_ops sh_eth_ethtool_ops = {
.set_ringparam = sh_eth_set_ringparam,
.get_link_ksettings = sh_eth_get_link_ksettings,
.set_link_ksettings = sh_eth_set_link_ksettings,
+ .get_wol = sh_eth_get_wol,
+ .set_wol = sh_eth_set_wol,
};
/* network device open function */
@@ -3017,6 +3048,11 @@ static int sh_eth_drv_probe(struct platform_device *pdev)
goto out_release;
}
+ /* Get clock, if not found that's OK but Wake-On-Lan is unavailable */
+ mdp->clk = devm_clk_get(&pdev->dev, NULL);
+ if (IS_ERR(mdp->clk))
+ mdp->clk = NULL;
+
ndev->base_addr = res->start;
spin_lock_init(&mdp->lock);
@@ -3111,6 +3147,9 @@ static int sh_eth_drv_probe(struct platform_device *pdev)
if (ret)
goto out_napi_del;
+ if (mdp->cd->magic && mdp->clk)
+ device_set_wakeup_capable(&pdev->dev, 1);
+
/* print device information */
netdev_info(ndev, "Base address at 0x%x, %pM, IRQ %d.\n",
(u32)ndev->base_addr, ndev->dev_addr, ndev->irq);
@@ -3150,15 +3189,67 @@ static int sh_eth_drv_remove(struct platform_device *pdev)
#ifdef CONFIG_PM
#ifdef CONFIG_PM_SLEEP
+static int sh_eth_wol_setup(struct net_device *ndev)
+{
+ struct sh_eth_private *mdp = netdev_priv(ndev);
+
+ /* Only allow ECI interrupts */
+ synchronize_irq(ndev->irq);
+ napi_disable(&mdp->napi);
+ sh_eth_write(ndev, DMAC_M_ECI, EESIPR);
+
+ /* Enable MagicPacket */
+ sh_eth_modify(ndev, ECMR, 0, ECMR_MPDE);
+
+ /* Increased clock usage so device won't be suspended */
+ clk_enable(mdp->clk);
+
+ return enable_irq_wake(ndev->irq);
+}
+
+static int sh_eth_wol_restore(struct net_device *ndev)
+{
+ struct sh_eth_private *mdp = netdev_priv(ndev);
+ int ret;
+
+ napi_enable(&mdp->napi);
+
+ /* Disable MagicPacket */
+ sh_eth_modify(ndev, ECMR, ECMR_MPDE, 0);
+
+ /* The device needs to be reset to restore MagicPacket logic
+ * for next wakeup. If we close and open the device it will
+ * both be reset and all registers restored. This is what
+ * happens during suspend and resume without WoL enabled.
+ */
+ ret = sh_eth_close(ndev);
+ if (ret < 0)
+ return ret;
+ ret = sh_eth_open(ndev);
+ if (ret < 0)
+ return ret;
+
+ /* Restore clock usage count */
+ clk_disable(mdp->clk);
+
+ return disable_irq_wake(ndev->irq);
+}
+
static int sh_eth_suspend(struct device *dev)
{
struct net_device *ndev = dev_get_drvdata(dev);
+ struct sh_eth_private *mdp = netdev_priv(ndev);
int ret = 0;
- if (netif_running(ndev)) {
- netif_device_detach(ndev);
+ if (!netif_running(ndev))
+ return 0;
+
+ netif_device_detach(ndev);
+
+ if (mdp->wol_enabled)
+ ret = sh_eth_wol_setup(ndev);
+ else
ret = sh_eth_close(ndev);
- }
return ret;
}
@@ -3166,14 +3257,21 @@ static int sh_eth_suspend(struct device *dev)
static int sh_eth_resume(struct device *dev)
{
struct net_device *ndev = dev_get_drvdata(dev);
+ struct sh_eth_private *mdp = netdev_priv(ndev);
int ret = 0;
- if (netif_running(ndev)) {
+ if (!netif_running(ndev))
+ return 0;
+
+ if (mdp->wol_enabled)
+ ret = sh_eth_wol_restore(ndev);
+ else
ret = sh_eth_open(ndev);
- if (ret < 0)
- return ret;
- netif_device_attach(ndev);
- }
+
+ if (ret < 0)
+ return ret;
+
+ netif_device_attach(ndev);
return ret;
}
diff --git a/drivers/net/ethernet/renesas/sh_eth.h b/drivers/net/ethernet/renesas/sh_eth.h
index 11798d5c4a1d..96b5459d131f 100644
--- a/drivers/net/ethernet/renesas/sh_eth.h
+++ b/drivers/net/ethernet/renesas/sh_eth.h
@@ -493,6 +493,7 @@ struct sh_eth_cpu_data {
unsigned shift_rd0:1; /* shift Rx descriptor word 0 right by 16 */
unsigned rmiimode:1; /* EtherC has RMIIMODE register */
unsigned rtrate:1; /* EtherC has RTRATE register */
+ unsigned magic:1; /* EtherC has ECMR.MPDE and ECSR.MPD */
};
struct sh_eth_private {
@@ -501,6 +502,7 @@ struct sh_eth_private {
const u16 *reg_offset;
void __iomem *addr;
void __iomem *tsu_addr;
+ struct clk *clk;
u32 num_rx_ring;
u32 num_tx_ring;
dma_addr_t rx_desc_dma;
@@ -529,6 +531,7 @@ struct sh_eth_private {
unsigned no_ether_link:1;
unsigned ether_link_active_low:1;
unsigned is_opened:1;
+ unsigned wol_enabled:1;
};
static inline void sh_eth_soft_swap(char *src, int len)
--
2.11.0
^ permalink raw reply related
* [PATCHv3 5/6] sh_eth: enable wake-on-lan for sh7734
From: Niklas Söderlund @ 2017-01-09 15:34 UTC (permalink / raw)
To: Sergei Shtylyov, Simon Horman, netdev, linux-renesas-soc
Cc: Geert Uytterhoeven, linux-pm, Niklas Söderlund
In-Reply-To: <20170109153409.13956-1-niklas.soderlund+renesas@ragnatech.se>
This is based on public datasheet for sh7734 which shows it has the
same behavior and registers for WoL as other versions of sh_eth.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
---
drivers/net/ethernet/renesas/sh_eth.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c
index 67cb6b3793ca..5292022fe6b1 100644
--- a/drivers/net/ethernet/renesas/sh_eth.c
+++ b/drivers/net/ethernet/renesas/sh_eth.c
@@ -819,6 +819,7 @@ static struct sh_eth_cpu_data sh7734_data = {
.hw_crc = 1,
.select_mii = 1,
.shift_rd0 = 1,
+ .magic = 1,
};
/* SH7763 */
--
2.11.0
^ permalink raw reply related
* [PATCHv3 6/6] sh_eth: enable wake-on-lan for sh7763
From: Niklas Söderlund @ 2017-01-09 15:34 UTC (permalink / raw)
To: Sergei Shtylyov, Simon Horman, netdev, linux-renesas-soc
Cc: Geert Uytterhoeven, linux-pm, Niklas Söderlund
In-Reply-To: <20170109153409.13956-1-niklas.soderlund+renesas@ragnatech.se>
This is based on public datasheet for sh7763 which shows it has the
same behavior and registers for WoL as other versions of sh_eth.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
---
drivers/net/ethernet/renesas/sh_eth.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c
index 5292022fe6b1..ecf0f7997a72 100644
--- a/drivers/net/ethernet/renesas/sh_eth.c
+++ b/drivers/net/ethernet/renesas/sh_eth.c
@@ -847,6 +847,7 @@ static struct sh_eth_cpu_data sh7763_data = {
.no_ade = 1,
.tsu = 1,
.irq_flags = IRQF_SHARED,
+ .magic = 1,
};
static struct sh_eth_cpu_data sh7619_data = {
--
2.11.0
^ permalink raw reply related
* [PATCHv3 0/6] sh_eth: add wake-on-lan support via magic packet
From: Niklas Söderlund @ 2017-01-09 15:34 UTC (permalink / raw)
To: Sergei Shtylyov, Simon Horman, netdev, linux-renesas-soc
Cc: Geert Uytterhoeven, linux-pm, Niklas Söderlund
Hi,
This series adds support for Wake-on-Lan using Magic Packet for a few
models of the sh_eth driver. Patch 1/6 fix a naming error, patch 2/6
adds generic support to control and support WoL while patches 3/6 - 6/6
enable different models.
Based ontop of net-next master.
Changes since v2.
- Fix bookkeeping for "active_count" and "event_count" reported in
/sys/kernel/debug/wakeup_sources. Thanks Geert for noticing this.
- Add new patch 1/6 which corrects the name of ECMR_MPDE bit, suggested
by Sergei.
- s/sh7743/sh7734/ in patch 5/6. Thanks Geert for spotting this.
- Spelling improvements suggested by Sergei and Geert.
- Add Tested-by to 3/6 and 4/6.
Changes since v1.
- Split generic WoL functionality and device enablement to different
patches.
- Enable more devices then Gen2 after feedback from Geert and
datasheets.
- Do not set mdp->irq_enabled = false and remove specific MagicPacket
interrupt clearing, instead let sh_eth_error() clear the interrupt as
for other EMAC interrupts, thanks Sergei for the suggestion.
- Use the original return logic in sh_eth_resume().
- Moved sh_eth_private variable *clk to top of data structure to avoid
possible gaps due to alignment restrictions.
- Make wol_enabled in sh_eth_private part of the already existing
bitfield instead of a bool.
- Do not initiate mdp->wol_enabled to 0, the struct is kzalloc'ed so
it's already set to 0.
Niklas Söderlund (6):
sh_eth: use correct name for ECMR_MPDE bit
sh_eth: add generic wake-on-lan support via magic packet
sh_eth: enable wake-on-lan for R-Car Gen2 devices
sh_eth: enable wake-on-lan for r8a7740/armadillo
sh_eth: enable wake-on-lan for sh7734
sh_eth: enable wake-on-lan for sh7763
drivers/net/ethernet/renesas/sh_eth.c | 123 +++++++++++++++++++++++++++++++---
drivers/net/ethernet/renesas/sh_eth.h | 5 +-
2 files changed, 117 insertions(+), 11 deletions(-)
--
2.11.0
^ permalink raw reply
* [PATCHv3 4/6] sh_eth: enable wake-on-lan for r8a7740/armadillo
From: Niklas Söderlund @ 2017-01-09 15:34 UTC (permalink / raw)
To: Sergei Shtylyov, Simon Horman, netdev, linux-renesas-soc
Cc: Geert Uytterhoeven, linux-pm, Niklas Söderlund
In-Reply-To: <20170109153409.13956-1-niklas.soderlund+renesas@ragnatech.se>
Geert Uytterhoeven reported WoL worked on his Armadillo board.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
---
drivers/net/ethernet/renesas/sh_eth.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c
index afaf8785ed72..67cb6b3793ca 100644
--- a/drivers/net/ethernet/renesas/sh_eth.c
+++ b/drivers/net/ethernet/renesas/sh_eth.c
@@ -577,6 +577,7 @@ static struct sh_eth_cpu_data r8a7740_data = {
.tsu = 1,
.select_mii = 1,
.shift_rd0 = 1,
+ .magic = 1,
};
/* There is CPU dependent code */
--
2.11.0
^ permalink raw reply related
* [PATCHv3 3/6] sh_eth: enable wake-on-lan for R-Car Gen2 devices
From: Niklas Söderlund @ 2017-01-09 15:34 UTC (permalink / raw)
To: Sergei Shtylyov, Simon Horman, netdev, linux-renesas-soc
Cc: Geert Uytterhoeven, linux-pm, Niklas Söderlund
In-Reply-To: <20170109153409.13956-1-niklas.soderlund+renesas@ragnatech.se>
Tested on Gen2 r8a7791/Koelsch.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
---
drivers/net/ethernet/renesas/sh_eth.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c
index 542c92b57b35..afaf8785ed72 100644
--- a/drivers/net/ethernet/renesas/sh_eth.c
+++ b/drivers/net/ethernet/renesas/sh_eth.c
@@ -623,8 +623,9 @@ static struct sh_eth_cpu_data r8a779x_data = {
.register_type = SH_ETH_REG_FAST_RCAR,
- .ecsr_value = ECSR_PSRTO | ECSR_LCHNG | ECSR_ICD,
- .ecsipr_value = ECSIPR_PSRTOIP | ECSIPR_LCHNGIP | ECSIPR_ICDIP,
+ .ecsr_value = ECSR_PSRTO | ECSR_LCHNG | ECSR_ICD | ECSR_MPD,
+ .ecsipr_value = ECSIPR_PSRTOIP | ECSIPR_LCHNGIP | ECSIPR_ICDIP |
+ ECSIPR_MPDIP,
.eesipr_value = 0x01ff009f,
.tx_check = EESR_FTC | EESR_CND | EESR_DLC | EESR_CD | EESR_RTO,
@@ -639,6 +640,7 @@ static struct sh_eth_cpu_data r8a779x_data = {
.tpauser = 1,
.hw_swap = 1,
.rmiimode = 1,
+ .magic = 1,
};
#endif /* CONFIG_OF */
--
2.11.0
^ permalink raw reply related
* Re: [PATCH net-next v2] net: dsa: make "label" property optional for dsa2
From: Vivien Didelot @ 2017-01-09 15:45 UTC (permalink / raw)
To: Jiri Pirko
Cc: netdev, linux-kernel, kernel, David S. Miller, Florian Fainelli,
Andrew Lunn, Uwe Kleine-König, Andrey Smirnov
In-Reply-To: <20170109151131.GC1862@nanopsycho>
Hi Jiri,
Jiri Pirko <jiri@resnulli.us> writes:
>>Extra question: shouldn't phys_port_{id,name} be switchdev attributes in
>
> Again, phys_port_id has nothing to do with switches. Should be removed
> from dsa because its use there is incorrect.
Florian, since 3a543ef just got in, can it be reverted?
>>> I guess that it should be enough for you to implement
>>> ndo_get_phys_port_name.
>>
>>Well, if this name must be unique on a system, it's not likely to happen
>>until we agree that we use an ugly tXsYpZ template where X is a tree ID,
>>or we assign system-wide unique IDs to switches, which requires a bit of
>>changes.
>
> No. That should be unique within one switch. In mlxsw we name it "p1",
> "p2", ...
>
> The final netdev names are:
> enp3s0np1, enp3s0np2, ...
OK perfect then, "p%d" sounds good. You seems to avoid "p0" in mlxsw, is
there a reason for that?
>>But again, this is not related to this patch ;-)
>
> It is! You are using phys_port_id, which is completely wrong. You should
> not use it.
I can resend this patch without the udev examples in the commit message
if that can be less confusing.
Thanks,
Vivien
^ permalink raw reply
* Re: [PATCH net-next] bridge: multicast to unicast
From: Johannes Berg @ 2017-01-09 15:47 UTC (permalink / raw)
To: michael-dev
Cc: netdev, bridge, linux-wireless, linux-kernel, David S . Miller,
Felix Fietkau
In-Reply-To: <ebc5050e12c04273fa518e35d7f0b8bc@fami-braun.de>
On Mon, 2017-01-09 at 16:25 +0100, michael-dev wrote:
> Am 09.01.2017 13:15, schrieb Johannes Berg:
> > > That is bridge fdb entries (need to) expire so the bridge might
> > > "forget" a still-connected station not sending but only consuming
> > > broadcast traffic.
> >
> > Ok, that I don't know. Somehow if you address a unicast packet
> > there
> > the bridge has to make a decision - so it really should know?
>
> If the bridge has not learned the unicast destination mac address on
> any port, it will flood the packet on all ports except the port it
> received the packet on.
Ok, so this really needs to be done in mac80211.
I'll send out the pull request soon then.
johannes
^ permalink raw reply
* Re: [for-next V2 06/10] net/mlx5: Add interface to get reference to a UAR
From: David Miller @ 2017-01-09 15:47 UTC (permalink / raw)
To: saeedm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb
Cc: yuval.shaia-QHcLZuEGTsvQT0dZR+AlfA, saeedm-VPRAkNaXOzVWk0Htik3J/w,
dledford-H+wXaHxf7aLQT0dZR+AlfA, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, leonro-VPRAkNaXOzVWk0Htik3J/w,
eli-VPRAkNaXOzVWk0Htik3J/w, matanb-VPRAkNaXOzVWk0Htik3J/w,
leon-DgEjT+Ai2ygdnm+yROfE0A
In-Reply-To: <CALzJLG-yiZfa6J0WciFPowfF2LWrXAqCKc4ETu4x=ZJdn+CfDA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
From: Saeed Mahameed <saeedm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
Date: Mon, 9 Jan 2017 10:31:36 +0200
> We will submit an incremental patch for this, as checkpatch doesn't
> complain about such minor things.
Please fix this and resubmit the series.
Checkpatch not complaining is not an argument for fixing up coding
style issues reported to you in feedback.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [PATCHv2 net-next 0/5] sctp: add support for generating stream reconf chunks
From: David Miller @ 2017-01-09 15:53 UTC (permalink / raw)
To: nhorman; +Cc: lucien.xin, netdev, linux-sctp, marcelo.leitner, vyasevich
In-Reply-To: <20170109124325.GB20973@hmswarspite.think-freely.org>
From: Neil Horman <nhorman@tuxdriver.com>
Date: Mon, 9 Jan 2017 07:43:25 -0500
> These all look reasonably good, but it seems before we accept them,
> there should be an additional patch that actually makes use of the code.
> I presume that is forthcomming?
This all comes from my asking that the original huge set of patches be
split up.
People always just rush this kind of work and never think about laying
out the resubmission properly.
One should always only submit new interfaces along with an actual use
because only with a use can we properly review whether the new
interface is good or not.
^ permalink raw reply
* [PATCH V4 net-next 01/15] net: introduce keepalive function in struct proto
From: Ursula Braun @ 2017-01-09 15:55 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, schwidefsky, heiko.carstens, utz.bacher,
ubraun
In-Reply-To: <20170109155526.10961-1-ubraun@linux.vnet.ibm.com>
Direct call of tcp_set_keepalive() function from protocol-agnostic
sock_setsockopt() function in net/core/sock.c violates network
layering. And newly introduced protocol (SMC-R) will need its own
keepalive function. Therefore, add "keepalive" function pointer
to "struct proto", and call it from sock_setsockopt() via this pointer.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: Utz Bacher <utz.bacher@de.ibm.com>
---
include/net/sock.h | 1 +
net/core/sock.c | 7 ++-----
net/ipv4/tcp_ipv4.c | 1 +
net/ipv4/tcp_timer.c | 1 +
net/ipv6/tcp_ipv6.c | 1 +
5 files changed, 6 insertions(+), 5 deletions(-)
diff --git a/include/net/sock.h b/include/net/sock.h
index f0e867f..99deda6 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -1024,6 +1024,7 @@ struct proto {
int (*getsockopt)(struct sock *sk, int level,
int optname, char __user *optval,
int __user *option);
+ void (*keepalive)(struct sock *sk, int valbool);
#ifdef CONFIG_COMPAT
int (*compat_setsockopt)(struct sock *sk,
int level,
diff --git a/net/core/sock.c b/net/core/sock.c
index f560e08..5018703 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -762,11 +762,8 @@ int sock_setsockopt(struct socket *sock, int level, int optname,
goto set_rcvbuf;
case SO_KEEPALIVE:
-#ifdef CONFIG_INET
- if (sk->sk_protocol == IPPROTO_TCP &&
- sk->sk_type == SOCK_STREAM)
- tcp_set_keepalive(sk, valbool);
-#endif
+ if (sk->sk_prot->keepalive)
+ sk->sk_prot->keepalive(sk, valbool);
sock_valbool_flag(sk, SOCK_KEEPOPEN, valbool);
break;
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 7e4be4f..56d756e 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2376,6 +2376,7 @@ struct proto tcp_prot = {
.shutdown = tcp_shutdown,
.setsockopt = tcp_setsockopt,
.getsockopt = tcp_getsockopt,
+ .keepalive = tcp_set_keepalive,
.recvmsg = tcp_recvmsg,
.sendmsg = tcp_sendmsg,
.sendpage = tcp_sendpage,
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index 3705075..29a9bd5 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -617,6 +617,7 @@ void tcp_set_keepalive(struct sock *sk, int val)
else if (!val)
inet_csk_delete_keepalive_timer(sk);
}
+EXPORT_SYMBOL_GPL(tcp_set_keepalive);
static void tcp_keepalive_timer (unsigned long data)
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index a4cdf6a..228965d 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -1889,6 +1889,7 @@ struct proto tcpv6_prot = {
.shutdown = tcp_shutdown,
.setsockopt = tcp_setsockopt,
.getsockopt = tcp_getsockopt,
+ .keepalive = tcp_set_keepalive,
.recvmsg = tcp_recvmsg,
.sendmsg = tcp_sendmsg,
.sendpage = tcp_sendpage,
--
2.8.4
^ permalink raw reply related
* [PATCH V4 net-next 00/15] net/smc: Shared Memory Communications - RDMA
From: Ursula Braun @ 2017-01-09 15:55 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, schwidefsky, heiko.carstens, utz.bacher,
ubraun
Dave,
here is now V4 of the SMC-R patches having processed your feedback from end
of November. The most important change is the replacement of sysfs by a
generic netlink solution in patch 04. And I tried to get rid of the __packed
attributes. There are still a few usages left due to SMC-R protocol defined
structures.
V4 changes:
The order of patches 03 and 04 for pnet table management and SMC IB-client
establishing has been exchanged, since pnet table management is now built on
top of smc_ib_devices.
Patch 01: Use EXPORT_SYMBOL_GPL().
Patch 02: Define "use_fallback" as bool.
Get rid of useless smc_sock fields clearing in smc_sock_alloc(),
since sk_alloc() clears out the memory.
Patch 03: Postpone smc_ib_remember_port_attr() call till ib_device is
mentioned in the pnet table.
Patch 04: Replace sysfs-usage by a generic netlink approach for pnet table
configuration.
Change layout of pnet table entries to reference net_device and
ib_device instead of dealing with names of net_devices and
ib_devices.
Patch 05: Adapt "use_fallback" usages to new type bool.
Get rid of useless smc_sock fields clearing in smc_sock_alloc()
Avoid __packed where possible.
Check if clc responses are not too big.
Patch 09: Postpone smc_setup_per_ibdev till the first connection with this
ib_device is really created.
Patch 11: Get rid of __packed usage.
V3 changes:
Patch 05: Remove unneeded DEFINE_WAIT
Patch 06: Improve synchronization of link group creation
Patch 07: Rename peer_rmbe_len into peer_rmbe_size to be more consistent
Patch 09: Avoid calls of ib_get_memory_region with IB_ACCESS_LOCAL_WRITE,
use new default local_dma_lkey from protection domain as lkey
instead.
Remove no longer needed function smc_ib_dereg_memory_region().
Patch 14: Switch to state ACTIVE only if still in state INIT.
Return 0 for recvmsg invoked in a socket closing state.
Allow getname call in state APPCLOSEWAIT1
Do not trigger destruction of a socket-in-error queued in accept
queue.
During cleanup of accept queue, make sure sockets are destructed,
and sockets in fallback mode are handled appropriately.
When freeing sndbufs/rmbs, remove them from their list and free
the entry.
Use add_wait_queue() and remove_wait_queue() in close wait
functions.
If actively closing a socket in state for PEERFINCLOSEWAIT, keep
this state.
If passively closing a socket while bytes are to be received, move
to state APPCLOSEWAIT1.
If actively aborting a socket, skip sending the close_abort flag,
since RDMA communication is no longer possible.
When terminating a link group, do not schedule link group freeing a
2nd time, since already done when unregistering the last remaining
connection.
Patch 15: Introduce smc_diag module for monitoring SMC protocol sockets.
This replaces the old patch 0015 dealing with procfs.
V2 changes:
Patch 0002: Add SMC versions for family key strings in net/core/sock.c.
Patch 0006: initialize rb_tree.
Patch 0007: Get rid of unneeded use of xchg() in smc_sndbuf_unuse() and
smc_rmb_unuse().
Patch 0008: Correct error checking logic for ib_function calls.
Define struct smc_link field wr_tx_id as atomic_long_t.
Use "do_div" instead of "%" to be architecture-independent.
Patch 0009: Correct error checking logic for ib_function calls.
Patch 0011: Remove xchg() calls in cursor handling. Use atomic64_t for cursor
overlays on 64-bit architectures. If not available, use plain u64
and add locking for cursor reading and writing.
Implement smc_curs_add() without modulo operator "%".
Patch 0012: Remove xchg() calls in cursor handling.
Implement smc_tx_rdma_writes() without module operator "%".
Patch 0013: Remove xchg() calls in cursor handling.
Patch 0014: Return type bool in smc_wr_tx_has_pending().
Remove unneeded semicolon in smc_close_shutdown_write().
Call smc_close_active() in non-fallback case only.
Get rid of duplicate schedule of sock_put_work().
Take nested sock_lock in smc_listen_work().
Start close stream_wait in case of prepared sends only.
Patch 0015: Remove unneeded socket ref_count in smc_proc_seq_show().
Take lock before list_empty check in smc_proc_sock_list_del().
These patches are the initial part of the implementation of the
"Shared Memory Communications-RDMA" (SMC-R) protocol as defined in
RFC7609 [1]. While SMC-R does not aim to replace TCP,
it taps a wealth of existing data center TCP socket applications
to become more efficient without the need for rewriting them.
SMC-R uses RDMA over Converged Ethernet (RoCE) to save CPU consumption.
For instance, when running 10 parallel connections with uperf, we measured
a decrease of 60% in CPU consumption with SMC-R compared to TCP/IP
(with throughput and latency comparable;
measured on x86_64 with the same RoCE card and port).
SMC-R does not require an RDMA communication manager (RDMA CM).
SMC-R inherits TCP qualities such as reliable connections, host-based
firewall packet filtering (on connection establishment) and unmodified
application of communication encryption such as TLS (transport layer
security) or SSL (secure sockets layer). Since original TCP is used to
establish SMC-R connections, load balancers and packet inspection based
on TCP/IP connection establishment continue to work for SMC-R.
On the other hand, using SMC-R implies:
- either involving a preload library when invoking the unchanged TCP-application
or slightly modifying the source by simply changing the socket family in
the socket() call
- accepting extra overhead and latency in connection establishment due to
SMC Connection Layer Control (CLC) handshake
- explicit coupling of RoCE ports with Ethernet ports
- not routable as currently built on RoCE V1
- bypassing of packet-based networking features
- filtering (netfilter)
- sniffing (libpcap, packet sockets, (E)BPF)
- traffic control (scheduling, shaping)
- bypassing of IP-header based socket options
- bypassing of memory buffer (pressure) management
- unusable together with IPsec
Overview of the SMC-R Protocol described in informational RFC 7609
SMC-R is an open protocol that provides RDMA capabilities over RoCE
transparently for applications exploiting TCP sockets.
A new socket protocol family PF_SMC is introduced.
There are no changes required to applications using the sockets API for TCP
stream sockets other than the specification of the new socket family AF_SMC.
Unmodified applications can be used by means of a dynamic preload shared
library which rewrites the socket API call
socket(AF_INET, SOCK_STREAM, IPPROTO_TCP) into
socket(AF_SMC, SOCK_STREAM, IPPROTO_TCP).
SMC-R re-uses the address family AF_INET for all addressing purposes around
struct sockaddr.
SMC-R system architecture layers:
+=============================================================================+
| | unmodified TCP application |
| native SMC application +--------------------------------------+
| | dynamic preload shared library |
+=============================================================================+
| SMC socket |
+-----------------------------------------------------------------------------+
| | TCP socket (for connection establishment and fallback) |
| IB verbs +--------------------------------------------------------+
| | IP |
+--------------------+--------------------------------------------------------+
| RoCE device driver | some network device driver |
+=============================================================================+
Terms:
A link group is determined by an ordered peer pair of TCP client and TCP server
(IP addresses and subnet). Reversed client server roles cause an own link group.
A link is a logical point-to-point connection based on an
infiniband reliable connected queue pair (RC-QP) between two RoCE ports
(MACs and GIDs) of a peer pair.
A link group can have 1..8 links for failover and load balancing.
This initial Linux implementation always has 1 link per link group.
Each link group on a peer can have 1..255 remote memory buffers (RMBs).
If more RMBs are needed, a peer can open another link group
(this initial Linux implementation) or fall back to TCP.
Each RMB has its own particular size and its own (R)DMA mapping and credentials
(rtoken consisting of rkey and RDMA "virtual address").
This initial Linux implementation uses physically contiguous memory for RMBs
but we are working towards scattered memory because of memory fragmentation.
Each RMB has 1..255 RMB elements (RMBEs) of equal size
to provide multiplexing of connections within an RMB.
An RMBE is the RDMA Write destination organized as wrapping ring buffer
for data transmit of a particular connection in one direction
(duplex by means of mirror symmetry as with TCP).
This initial Linux implementation always has 1 RMBE per RMB
and thus an individual RMB for each connection.
SMC-R connection establishment with subsequent data transfer:
CLIENT SERVER
TCP three-way handshake:
regular TCP SYN
-------------------------------------------------------->
regular TCP SYN ACK
<--------------------------------------------------------
regular TCP ACK
-------------------------------------------------------->
SMC Connection Layer Control (CLC) handshake
exchanges RDMA credentials between peers:
via above TCP connection: SMC CLC Proposal
-------------------------------------------------------->
via above TCP connection: SMC CLC Accept
<--------------------------------------------------------
via above TCP connection: SMC CLC Confirm
-------------------------------------------------------->
SMC Link Layer Control (LLC) (only once per link, i.e. 1st conn. of link group):
RoCE RC-QP: SMC LLC Confirm Link
<========================================================
RoCE RC-QP: SMC LLC Confirm Link response
========================================================>
SMC data transmission (incl. SMC Connection Data Control (CDC) message):
RoCE RC-QP: RDMA Write
========================================================>
RoCE RC-QP: SMC CDC message (flow control)
========================================================>
...
RoCE RC-QP: RDMA Write
<========================================================
RoCE RC-QP: SMC CDC message (flow control)
<========================================================
...
Data flow within an established connection:
+----------------------------------------------------------------------------
| SENDER
| sendmsg()
| |
| | produces into sndbuf [sender's process context]
| v
| +--------+
| | sndbuf | [ring buffer]
| +--------+
| |
| | consumes from sndbuf and produces into receiver's RMBE [any context]
| | by sending RDMA Write followed by SMC CDC message over RoCE RC-QP
| |
+----|-----------------------------------------------------------------------
|
+----|-----------------------------------------------------------------------
| v RECEIVER
| +------+
| | RMBE | [ring buffer, can have size different from sender's sndbuf]
| | | [RMBE represents rcvbuf, no further de-coupling as on sender side]
| +------+
| |
| | consumes from RMBE [receiver's process context]
| v
| recvmsg()
+----------------------------------------------------------------------------
Flow control ("cursor" updates) by means of SMC CDC messages:
SENDER RECEIVER
sends updates via CDC-------------+ sends updates via CDC
on consuming from sndbuf | on consuming from RMBE
and producing into RMBE | by means of recvmsg()
| |
| |
+-----------------------------------|------------+
| |
+--v-------------------------+ +--v-----------------------+
| receiver's consumer cursor | | sender's producer cursor----+
+----------------|-----------+ +--------------------------+ |
| |
| receiver's RMBE |
| +--------------------------+ |
| | | |
+--------------------------------+ | |
| | | |
| v | |
| +------------| |
|-------------+////////////| |
|//RDMA data written by////| |
|////sender that is////////| |
|/available to be consumed/| |
|///////// +---------------| |
|----------+^ | |
| | | |
| +-----------------+
| |
+--------------------------+
Sending updates of the producer cursor is immediate for low latency;
something like Nagle's algorithm (absence of TCP_NODELAY) is optional and
currently not part of this initial Linux implementation.
Sending updates of the consumer cursor is conditional to avoid the
silly window syndrome.
Normal connection termination:
Normal connection termination starts transitioning from socket state
ACTIVE via either "Active Close" or "Passive Close".
shutdown rdwr +-----------------+
or close, +-------------->| INIT / CLOSED |<-------------+
send PeerCon|nClosed +-----------------+ | PeerConnClosed
| | | received
| connection | established |
| V |
+----------------+ +-----------------+ +----------------+
|AppFinCloseWait | | ACTIVE | |PeerFinCloseWait|
+----------------+ +-----------------+ +----------------+
| | | |
| Active Close: | |Passive Close: |
| close or | |PeerConnClosed or |
| shutdown wr or| |PeerDoneWriting |
| shutdown rdwr | |received |
| V V |
PeerConnClo|sed +--------------+ +-------------+ | close or
received +--<----|PeerCloseWait1| |AppCloseWait1|--->----+ shutdown rdwr,
| +--------------+ +-------------+ | send
| PeerDoneWri|ting | shutdown wr, | PeerConnClosed
| received | send Pee|rDoneWriting |
| V V |
| +--------------+ +-------------+ |
+--<----|PeerCloseWait2| |AppCloseWait2|--->----+
+--------------+ +-------------+
In state CLOSED, the socket can be destructed only, once the application has
issued a close().
Abnormal connection termination:
+-----------------+
+-------------->| INIT / CLOSED |<-------------+
| +-----------------+ |
| |
| +-----------------------+ |
| | Any state | |
PeerConnAbo|rt | (before setting | | send
received | | PeerConnClosed | | PeerConnAbort
| | indicator in | |
| | peer's RMBE) | |
| +-----------------------+ |
| | | |
| Active Abort: | | Passive Abort: |
| problem, | | PeerConnAbort |
| send | | received, |
| PeerConnAbort,| | ECONNRESET |
| ECONNABORTED | | |
| V V |
| +--------------+ +--------------+ |
+-------|PeerAbortWait | | ProcessAbort |------+
+--------------+ +--------------+
Implementation notes beyond RFC 7609:
A PNET table in sysfs provides the mapping between network device names and
RoCE Infiniband device names for the transparent switch of data communication.
A PNET table can contain an arbitrary number of PNETIDs.
Each PNETID contains exactly one (Ethernet) network device name
and one or more RoCE Infiniband device names.
Each device name can only exist in at most one PNETID (no overlapping).
This initial Linux implementation allows at most one RoCE Infiniband device
name per PNETID.
After a new TCP connection is established, the network device name
used for egress traffic with the TCP connection's local source IP address
is used as key to lookup the unique PNETID, and the RoCE Infiniband device
of this PNETID is used to switch data communication from TCP to RDMA
during SMC CLC handshake.
Problem determination:
A protocol dissector is available with upstream wireshark for formatting
SMC-R related RoCE LAN traffic.
[https://code.wireshark.org/review/gitweb?p=wireshark.git;a=blob;f=epan/dissectors/packet-smcr.c]
We are working on enhancing the Linux implementation to cover:
- Improve default socket closing asynchronicity
- Address corner cases with many parallel connections
- Tracing
- Integrated load balancing and fail-over within a link group
- Splice and sendpage support
- IPv6 addressing support
- Keepalive, Cork
- Namespaces support
- Urgent data
- More socket options
- Diagnostics
- Statistics support
- SNMP support
References:
[1] SMC-R Informational RFC: http://www.rfc-editor.org/info/rfc7609
Thomas Richter (1):
smc: establish pnet table management
Ursula Braun (14):
net: introduce keepalive function in struct proto
smc: establish new socket family
smc: introduce SMC as an IB-client
smc: CLC handshake (incl. preparation steps)
smc: connection and link group creation
smc: remote memory buffers (RMBs)
smc: work request (WR) base for use by LLC and CDC
smc: initialize IB transport incl. PD, MR, QP, CQ, event, WR
smc: link layer control (LLC)
smc: connection data control (CDC)
smc: send data (through RDMA)
smc: receive data from RMBE
smc: socket closing and linkgroup cleanup
smc: netlink interface for SMC sockets
MAINTAINERS | 7 +
include/linux/socket.h | 7 +-
include/net/smc.h | 20 +
include/net/sock.h | 4 +
include/uapi/linux/netlink.h | 1 +
include/uapi/linux/smc.h | 35 +
include/uapi/linux/smc_diag.h | 85 +++
net/Kconfig | 1 +
net/Makefile | 1 +
net/core/sock.c | 13 +-
net/ipv4/tcp_ipv4.c | 1 +
net/ipv4/tcp_timer.c | 1 +
net/ipv6/tcp_ipv6.c | 1 +
net/smc/Kconfig | 20 +
net/smc/Makefile | 4 +
net/smc/af_smc.c | 1407 +++++++++++++++++++++++++++++++++++++++++
net/smc/smc.h | 274 ++++++++
net/smc/smc_cdc.c | 304 +++++++++
net/smc/smc_cdc.h | 218 +++++++
net/smc/smc_clc.c | 282 +++++++++
net/smc/smc_clc.h | 116 ++++
net/smc/smc_close.c | 442 +++++++++++++
net/smc/smc_close.h | 28 +
net/smc/smc_core.c | 677 ++++++++++++++++++++
net/smc/smc_core.h | 181 ++++++
net/smc/smc_diag.c | 215 +++++++
net/smc/smc_ib.c | 466 ++++++++++++++
net/smc/smc_ib.h | 69 ++
net/smc/smc_llc.c | 158 +++++
net/smc/smc_llc.h | 63 ++
net/smc/smc_pnet.c | 534 ++++++++++++++++
net/smc/smc_pnet.h | 23 +
net/smc/smc_rx.c | 217 +++++++
net/smc/smc_rx.h | 23 +
net/smc/smc_tx.c | 483 ++++++++++++++
net/smc/smc_tx.h | 35 +
net/smc/smc_wr.c | 614 ++++++++++++++++++
net/smc/smc_wr.h | 106 ++++
38 files changed, 7127 insertions(+), 9 deletions(-)
create mode 100644 include/net/smc.h
create mode 100644 include/uapi/linux/smc.h
create mode 100644 include/uapi/linux/smc_diag.h
create mode 100644 net/smc/Kconfig
create mode 100644 net/smc/Makefile
create mode 100644 net/smc/af_smc.c
create mode 100644 net/smc/smc.h
create mode 100644 net/smc/smc_cdc.c
create mode 100644 net/smc/smc_cdc.h
create mode 100644 net/smc/smc_clc.c
create mode 100644 net/smc/smc_clc.h
create mode 100644 net/smc/smc_close.c
create mode 100644 net/smc/smc_close.h
create mode 100644 net/smc/smc_core.c
create mode 100644 net/smc/smc_core.h
create mode 100644 net/smc/smc_diag.c
create mode 100644 net/smc/smc_ib.c
create mode 100644 net/smc/smc_ib.h
create mode 100644 net/smc/smc_llc.c
create mode 100644 net/smc/smc_llc.h
create mode 100644 net/smc/smc_pnet.c
create mode 100644 net/smc/smc_pnet.h
create mode 100644 net/smc/smc_rx.c
create mode 100644 net/smc/smc_rx.h
create mode 100644 net/smc/smc_tx.c
create mode 100644 net/smc/smc_tx.h
create mode 100644 net/smc/smc_wr.c
create mode 100644 net/smc/smc_wr.h
--
2.8.4
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox