* [PATCH] RDS: Simplify code
@ 2016-09-03 5:33 Christophe JAILLET
2016-09-04 12:20 ` Leon Romanovsky
0 siblings, 1 reply; 6+ messages in thread
From: Christophe JAILLET @ 2016-09-03 5:33 UTC (permalink / raw)
To: santosh.shilimkar, davem
Cc: linux-rdma, netdev, linux-kernel, kernel-janitors, Christophe JAILLET
Calling 'list_splice' followed by 'INIT_LIST_HEAD' is equivalent to
'list_splice_init'.
This has been spotted with the following coccinelle script:
/////
@@
expression y,z;
@@
- list_splice(y,z);
- INIT_LIST_HEAD(y);
+ list_splice_init(y,z);
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
net/rds/loop.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/net/rds/loop.c b/net/rds/loop.c
index f2bf78de5688..c3e6da4fdf97 100644
--- a/net/rds/loop.c
+++ b/net/rds/loop.c
@@ -167,8 +167,7 @@ void rds_loop_exit(void)
/* avoid calling conn_destroy with irqs off */
spin_lock_irq(&loop_conns_lock);
- list_splice(&loop_conns, &tmp_list);
- INIT_LIST_HEAD(&loop_conns);
+ list_splice_init(&loop_conns, &tmp_list);
spin_unlock_irq(&loop_conns_lock);
list_for_each_entry_safe(lc, _lc, &tmp_list, loop_node) {
--
2.7.4
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH] RDS: Simplify code
2016-09-03 5:33 [PATCH] RDS: Simplify code Christophe JAILLET
@ 2016-09-04 12:20 ` Leon Romanovsky
2016-09-04 15:57 ` Christophe JAILLET
0 siblings, 1 reply; 6+ messages in thread
From: Leon Romanovsky @ 2016-09-04 12:20 UTC (permalink / raw)
To: Christophe JAILLET
Cc: santosh.shilimkar, davem, linux-rdma, netdev, linux-kernel,
kernel-janitors
[-- Attachment #1: Type: text/plain, Size: 1267 bytes --]
On Sat, Sep 03, 2016 at 07:33:29AM +0200, Christophe JAILLET wrote:
> Calling 'list_splice' followed by 'INIT_LIST_HEAD' is equivalent to
> 'list_splice_init'.
It is not 100% accurate
list_splice(y, z)
INIT_LIST_HEAD(y)
==>
if (!list_empty(y))
__list_splice(y, z, z>next);
INIT_LIST_HEAD(y)
and not
if (!list_empty(y)) {
__list_splice(y, z, z>next);
INIT_LIST_HEAD(y)
}
as list_splice_init will do.
>
> This has been spotted with the following coccinelle script:
> /////
> @@
> expression y,z;
> @@
>
> - list_splice(y,z);
> - INIT_LIST_HEAD(y);
> + list_splice_init(y,z);
>
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
> net/rds/loop.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/net/rds/loop.c b/net/rds/loop.c
> index f2bf78de5688..c3e6da4fdf97 100644
> --- a/net/rds/loop.c
> +++ b/net/rds/loop.c
> @@ -167,8 +167,7 @@ void rds_loop_exit(void)
>
> /* avoid calling conn_destroy with irqs off */
> spin_lock_irq(&loop_conns_lock);
> - list_splice(&loop_conns, &tmp_list);
> - INIT_LIST_HEAD(&loop_conns);
> + list_splice_init(&loop_conns, &tmp_list);
> spin_unlock_irq(&loop_conns_lock);
>
> list_for_each_entry_safe(lc, _lc, &tmp_list, loop_node) {
> --
> 2.7.4
>
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] RDS: Simplify code
2016-09-04 12:20 ` Leon Romanovsky
@ 2016-09-04 15:57 ` Christophe JAILLET
2016-09-04 18:23 ` Leon Romanovsky
0 siblings, 1 reply; 6+ messages in thread
From: Christophe JAILLET @ 2016-09-04 15:57 UTC (permalink / raw)
To: Leon Romanovsky
Cc: santosh.shilimkar, davem, linux-rdma, netdev, linux-kernel,
kernel-janitors
Le 04/09/2016 à 14:20, Leon Romanovsky a écrit :
> On Sat, Sep 03, 2016 at 07:33:29AM +0200, Christophe JAILLET wrote:
>> Calling 'list_splice' followed by 'INIT_LIST_HEAD' is equivalent to
>> 'list_splice_init'.
> It is not 100% accurate
>
> list_splice(y, z)
> INIT_LIST_HEAD(y)
>
> ==>
>
> if (!list_empty(y))
> __list_splice(y, z, z>next);
> INIT_LIST_HEAD(y)
>
> and not
>
> if (!list_empty(y)) {
> __list_splice(y, z, z>next);
> INIT_LIST_HEAD(y)
> }
>
> as list_splice_init will do.
>
You are right but if you dig further you will see that calling
INIT_LIST_HEAD on an empty list is a no-op (AFAIK).
And if this list was not already correctly initialized, then you would
have some other troubles.
CJ
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] RDS: Simplify code
2016-09-04 15:57 ` Christophe JAILLET
@ 2016-09-04 18:23 ` Leon Romanovsky
[not found] ` <73fc4be8-2530-b9ab-945b-887db6dca2a6@wanadoo.fr>
2016-09-06 0:22 ` santosh.shilimkar
0 siblings, 2 replies; 6+ messages in thread
From: Leon Romanovsky @ 2016-09-04 18:23 UTC (permalink / raw)
To: Christophe JAILLET
Cc: santosh.shilimkar, davem, linux-rdma, netdev, linux-kernel,
kernel-janitors
[-- Attachment #1: Type: text/plain, Size: 1214 bytes --]
On Sun, Sep 04, 2016 at 05:57:20PM +0200, Christophe JAILLET wrote:
> Le 04/09/2016 à 14:20, Leon Romanovsky a écrit :
> >On Sat, Sep 03, 2016 at 07:33:29AM +0200, Christophe JAILLET wrote:
> >>Calling 'list_splice' followed by 'INIT_LIST_HEAD' is equivalent to
> >>'list_splice_init'.
> >It is not 100% accurate
> >
> >list_splice(y, z)
> >INIT_LIST_HEAD(y)
> >
> >==>
> >
> >if (!list_empty(y))
> > __list_splice(y, z, z>next);
> >INIT_LIST_HEAD(y)
> >
> >and not
> >
> >if (!list_empty(y)) {
> > __list_splice(y, z, z>next);
> > INIT_LIST_HEAD(y)
> >}
> >
> >as list_splice_init will do.
> >
> You are right but if you dig further you will see that calling
> INIT_LIST_HEAD on an empty list is a no-op (AFAIK).
> And if this list was not already correctly initialized, then you would have
> some other troubles.
Thank you for the suggestion,
It looks like the code after that can be skipped in case of loop_conns
list is empty, the tmp_list will be empty too.
174 list_for_each_entry_safe(lc, _lc, &tmp_list, loop_node) {
175 WARN_ON(lc->conn->c_passive);
176 rds_conn_destroy(lc->conn);
177 }
>
> CJ
>
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] RDS: Simplify code
[not found] ` <73fc4be8-2530-b9ab-945b-887db6dca2a6@wanadoo.fr>
@ 2016-09-05 5:14 ` Leon Romanovsky
0 siblings, 0 replies; 6+ messages in thread
From: Leon Romanovsky @ 2016-09-05 5:14 UTC (permalink / raw)
To: Christophe JAILLET; +Cc: linux-rdma, netdev, linux-kernel, kernel-janitors
[-- Attachment #1: Type: text/plain, Size: 2234 bytes --]
On Mon, Sep 05, 2016 at 06:38:21AM +0200, Christophe JAILLET wrote:
> Le 04/09/2016 à 20:23, Leon Romanovsky a écrit :
> >On Sun, Sep 04, 2016 at 05:57:20PM +0200, Christophe JAILLET wrote:
> >>Le 04/09/2016 à 14:20, Leon Romanovsky a écrit :
> >>>On Sat, Sep 03, 2016 at 07:33:29AM +0200, Christophe JAILLET wrote:
> >>>>Calling 'list_splice' followed by 'INIT_LIST_HEAD' is equivalent to
> >>>>'list_splice_init'.
> >>>It is not 100% accurate
> >>>
> >>>list_splice(y, z)
> >>>INIT_LIST_HEAD(y)
> >>>
> >>>==>
> >>>
> >>>if (!list_empty(y))
> >>> __list_splice(y, z, z>next);
> >>>INIT_LIST_HEAD(y)
> >>>
> >>>and not
> >>>
> >>>if (!list_empty(y)) {
> >>> __list_splice(y, z, z>next);
> >>> INIT_LIST_HEAD(y)
> >>>}
> >>>
> >>>as list_splice_init will do.
> >>>
> >>You are right but if you dig further you will see that calling
> >>INIT_LIST_HEAD on an empty list is a no-op (AFAIK).
> >>And if this list was not already correctly initialized, then you would have
> >>some other troubles.
> >Thank you for the suggestion,
> >It looks like the code after that can be skipped in case of loop_conns
> >list is empty, the tmp_list will be empty too.
> >
> >174 list_for_each_entry_safe(lc, _lc, &tmp_list, loop_node) {
> >175 WARN_ON(lc->conn->c_passive);
> >176 rds_conn_destroy(lc->conn);
> >177 }
> Yes, but this would require some more code and test. This function doesn't
> seem to be in a hot path, so I'm not sure that the added complexity would
> worth it.
> It would require a new 'list_empty()' test and some code rearrangement.
>
> I suppose that testing for emptiness at the beginning or going through a
> list_for_each_entry_safe on a empty list (which will exit immediately and do
> nothing) is more or less the same in term of speed. So keep the code simple
> and readable.
I would expect one list_empty check at the beginning and return
immediately, but anyway it doesn't matter.
>
> CJ
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] RDS: Simplify code
2016-09-04 18:23 ` Leon Romanovsky
[not found] ` <73fc4be8-2530-b9ab-945b-887db6dca2a6@wanadoo.fr>
@ 2016-09-06 0:22 ` santosh.shilimkar
1 sibling, 0 replies; 6+ messages in thread
From: santosh.shilimkar @ 2016-09-06 0:22 UTC (permalink / raw)
To: Leon Romanovsky, Christophe JAILLET
Cc: davem, linux-rdma, netdev, linux-kernel, kernel-janitors
On 9/4/16 11:23 AM, Leon Romanovsky wrote:
> On Sun, Sep 04, 2016 at 05:57:20PM +0200, Christophe JAILLET wrote:
>> Le 04/09/2016 à 14:20, Leon Romanovsky a écrit :
>>> On Sat, Sep 03, 2016 at 07:33:29AM +0200, Christophe JAILLET wrote:
>>>> Calling 'list_splice' followed by 'INIT_LIST_HEAD' is equivalent to
>>>> 'list_splice_init'.
>>> It is not 100% accurate
>>>
>>> list_splice(y, z)
>>> INIT_LIST_HEAD(y)
>>>
>>> ==>
>>>
>>> if (!list_empty(y))
>>> __list_splice(y, z, z>next);
>>> INIT_LIST_HEAD(y)
>>>
>>> and not
>>>
>>> if (!list_empty(y)) {
>>> __list_splice(y, z, z>next);
>>> INIT_LIST_HEAD(y)
>>> }
>>>
>>> as list_splice_init will do.
>>>
>> You are right but if you dig further you will see that calling
>> INIT_LIST_HEAD on an empty list is a no-op (AFAIK).
>> And if this list was not already correctly initialized, then you would have
>> some other troubles.
>
> Thank you for the suggestion,
> It looks like the code after that can be skipped in case of loop_conns
> list is empty, the tmp_list will be empty too.
>
> 174 list_for_each_entry_safe(lc, _lc, &tmp_list, loop_node) {
> 175 WARN_ON(lc->conn->c_passive);
> 176 rds_conn_destroy(lc->conn);
> 177 }
>
Thanks for trying. As already pointed, your change doesn't simplify
much rather change the behavior. The loop cursor already takes care
of list empty case. I don't see any reason to change that code.
Regards,
Santosh
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2016-09-06 0:22 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-09-03 5:33 [PATCH] RDS: Simplify code Christophe JAILLET
2016-09-04 12:20 ` Leon Romanovsky
2016-09-04 15:57 ` Christophe JAILLET
2016-09-04 18:23 ` Leon Romanovsky
[not found] ` <73fc4be8-2530-b9ab-945b-887db6dca2a6@wanadoo.fr>
2016-09-05 5:14 ` Leon Romanovsky
2016-09-06 0:22 ` santosh.shilimkar
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).