tipc: Remove redundant tsk->published flag
Subsystems (3)
PATCH 1/1 tipc: Remove redundant tsk->published flag
It's supposed to always equal !list_empty(tsk->publications), so just replace it with the latter.
I kept the tipc_sk_dump() output unchanged, but perhaps someone who understands how that's used better would like to change it to reflect the simplified structure accurately.
---
I just happened to be looking at the code and noticed that tsk->published could be reduced to a bool. Then I noticed that it could be deleted entirely.
It's a separate patch, but someone might want to move probe_unacked to be with the rest of the bools so struct tipc_sock packs better.
net/tipc/socket.c | 29 +++++++++++------------------ 1 file changed, 11 insertions(+), 18 deletions(-)
net/tipc/socket.c +11 -18
@@ -73,7 +73,6 @@ struct sockaddr_pair {* @sk: socket - interacts with 'port' and with user via the socket API * @conn_type: TIPC type used when connection was established * @conn_instance: TIPC instance used when connection was established -* @published: non-zero if port has one or more associated names * @max_pkt: maximum packet size "hint" used when building messages sent by port * @maxnagle: maximum size of msg which can be subject to nagle * @portid: unique port identity in TIPC socket hash table
@@ -96,7 +95,6 @@ struct tipc_sock {struct sock sk;u32 conn_type;u32 conn_instance; -int published;u32 max_pkt;u32 maxnagle;u32 portid;
@@ -1412,7 +1410,7 @@ static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen)return -EPIPE ;if (sk->sk_state !=TIPC_OPEN )return -EISCONN ; -if (tsk->published) +if (!list_empty (&tsk->publications))return -EOPNOTSUPP ;if (dest->addrtype ==TIPC_ADDR_NAME ) { tsk->conn_type = dest->addr.name.name.type;
@@ -2824,7 +2822,6 @@ static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,list_add (&publ->binding_sock, &tsk->publications); tsk->pub_count++; - tsk->published =1 ;return 0 ; }
@@ -2858,8 +2855,6 @@ static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope, publ->upper, publ->key); rc =0 ; } -if (list_empty (&tsk->publications)) - tsk->published =0 ;return rc; }
@@ -3743,7 +3738,6 @@ int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb)bool tipc_sk_filtering (struct sock *sk) {struct tipc_sock *tsk; -struct publication *p;u32 _port, _sktype, _type, _lower, _upper;u32 type =0 , lower =0 , upper =0 ;
@@ -3767,14 +3761,12 @@ bool tipc_sk_filtering(struct sock *sk)if (_sktype && _sktype != sk->sk_type)return false; -if (tsk->published) { - p =list_first_entry_or_null (&tsk->publications, - struct publication, binding_sock); -if (p) { - type = p->type; - lower = p->lower; - upper = p->upper; - } +if (!list_empty (&tsk->publications)) { +struct publication *p =list_first_entry (&tsk->publications, + struct publication, binding_sock); + type = p->type; + lower = p->lower; + upper = p->upper; }if (!tipc_sk_type_connectionless (sk)) {
@@ -3847,7 +3839,7 @@ int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf)size_t sz = (dqueues) ?SK_LMAX :SK_LMIN ;struct tipc_sock *tsk;struct publication *p; -bool tsk_connected; +bool tsk_connected, tsk_published;if (!sk) { i +=scnprintf (buf, sz,"sk data: (null)\n" );
@@ -3868,8 +3860,9 @@ int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf) i +=scnprintf (buf + i, sz - i," %u" , tsk->conn_type); i +=scnprintf (buf + i, sz - i," %u" , tsk->conn_instance); } - i +=scnprintf (buf + i, sz - i," | %u" , tsk->published); -if (tsk->published) { + tsk_published = !list_empty (&tsk->publications); + i +=scnprintf (buf + i, sz - i," | %u" , tsk_published); +if (tsk_published) { p =list_first_entry_or_null (&tsk->publications, struct publication, binding_sock); i +=scnprintf (buf + i, sz - i," %u" , (p) ? p->type :0 );
Signature · 1 line
2.26.1
David Miller
maintainerFrom: George Spelvin <[email protected]>
Date: Thu, 16 Apr 2020 03:27:35 GMT
net/tipc/socket.c int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf) · line 3843
Quoted 7 lines
net/tipc/socket.c +1
-1@@ -3847,7 +3839,7 @@ int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf)size_t sz = (dqueues) ?SK_LMAX :SK_LMIN ;struct tipc_sock *tsk;struct publication *p; -bool tsk_connected; +bool tsk_connected, tsk_published;
Please preserve the reverse christmas tree ordering of local variables here.
Thank you.
Quoted 13 lines · David Miller
From: George Spelvin <[email protected]>
Date: Thu, 16 Apr 2020 03:27:35 GMTnet/tipc/socket.c int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf) · line 3843
Quoted 7 lines
net/tipc/socket.c +1
-1@@ -3847,7 +3839,7 @@ int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf)size_t sz = (dqueues) ?SK_LMAX :SK_LMIN ;struct tipc_sock *tsk;struct publication *p; -bool tsk_connected; +bool tsk_connected, tsk_published;Please preserve the reverse christmas tree ordering of local variables here.
Happy to, but is that actually defined anywhere? "Preserve" implies that it was present before the patch, and I can't infer a rule which is obeyed by the pre-patch declarations:
int i = 0; size_t sz = (dqueues) ? SK_LMAX : SK_LMIN; struct tipc_sock *tsk; struct publication *p; bool tsk_connected;
One option is to sort by the full line length, including initialization:
size_t sz = (dqueues) ? SK_LMAX : SK_LMIN; struct tipc_sock *tsk; struct publication *p; bool tsk_connected; int i = 0;
The other is to sort by the *declaration* length:
struct tipc_sock *tsk; struct publication *p; bool tsk_connected; size_t sz = (dqueues) ? SK_LMAX : SK_LMIN; int i = 0;
Looking at the local variable declarations in the rest of the file isn't producing any clarity.
Thank you.
Jon Maloy
maintainerQuoted 3 lines · George Spelvin
Quoted 2 lines · David Miller
From: George Spelvin <[email protected]>
Date: Thu, 16 Apr 2020 03:27:35 GMT
[...]
Quoted 5 lines
Quoted 2 lines
Please preserve the reverse christmas tree ordering of local variables here.
Happy to, but is that actually defined anywhere? "Preserve" implies that it was present before the patch, and I can't infer a rule which is obeyed by the pre-patch declarations:
It all depends on the age of the code. In newer code, we always follow the rule, and when refactoring old code at least I try to adapt to this rule even at the cost of reshuffling a few declarations. It is risk free,
- you only have to watch out for dependencies.
Quoted 6 lines
int i = 0; size_t sz = (dqueues) ? SK_LMAX : SK_LMIN; struct tipc_sock *tsk; struct publication *p; bool tsk_connected;
Quoted 6 lines
One option is to sort by the full line length, including initialization:
size_t sz = (dqueues) ? SK_LMAX : SK_LMIN; struct tipc_sock *tsk; struct publication *p; bool tsk_connected; int i = 0;
This is the preferred order, AFAIK.
///jon