Re: [PATCH 5/5] staging: lustre: osc_page.c: Use list_for_each_entry_safe

From: Dilger, Andreas
Date: Tue Mar 07 2017 - 07:02:43 EST


On Mar 6, 2017, at 08:20, James Simmons <jsimmons@xxxxxxxxxxxxx> wrote:
>
>>
>> Doubly linked lists which are iterated using list_empty
>> and list_entry macros have been replaced with list_for_each_entry_safe
>> macro.
>> This makes the iteration simpler and more readable.
>>
>> This patch replaces the while loop containing list_empty and list_entry
>> with list_for_each_entry_safe.
>>
>> This was done with Coccinelle.
>>
>> @@
>> expression E1;
>> identifier I1, I2;
>> type T;
>> iterator name list_for_each_entry_safe;
>> @@
>>
>> T *I1;
>> + T *tmp;
>> ...
>> - while (list_empty(&E1) == 0)
>> + list_for_each_entry_safe (I1, tmp, &E1, I2)
>> {
>> ...when != T *I1;
>> - I1 = list_entry(E1.next, T, I2);
>> ...
>> }
>>
>> Signed-off-by: simran singhal <singhalsimran0@xxxxxxxxx>
>
> NAK!!!!!!
>
> This change was reverted in commit
>
> cd15dd6ef4ea11df87f717b8b1b83aaa738ec8af
>
> Doing these while (list_empty(..)) to list_for_entry...
> are not simple changes and have broken things in lustre
> before. Unless you really understand the state machine of
> the lustre code I don't recommend these kinds of change
> for lustre.

It may be useful to add a comment to these cases where the while() loop cannot be
replaced by list_for_each_entry_safe() (with details of why that is the case) to
avoid such optimization attempts again in the future.

Cheers, Andreas

>
>> ---
>> drivers/staging/lustre/lustre/osc/osc_page.c | 11 ++++-------
>> 1 file changed, 4 insertions(+), 7 deletions(-)
>>
>> diff --git a/drivers/staging/lustre/lustre/osc/osc_page.c b/drivers/staging/lustre/lustre/osc/osc_page.c
>> index ed8a0dc..e8b974f 100644
>> --- a/drivers/staging/lustre/lustre/osc/osc_page.c
>> +++ b/drivers/staging/lustre/lustre/osc/osc_page.c
>> @@ -542,6 +542,7 @@ long osc_lru_shrink(const struct lu_env *env, struct client_obd *cli,
>> struct cl_object *clobj = NULL;
>> struct cl_page **pvec;
>> struct osc_page *opg;
>> + struct osc_page *tmp;
>> int maxscan = 0;
>> long count = 0;
>> int index = 0;
>> @@ -572,7 +573,7 @@ long osc_lru_shrink(const struct lu_env *env, struct client_obd *cli,
>> if (force)
>> cli->cl_lru_reclaim++;
>> maxscan = min(target << 1, atomic_long_read(&cli->cl_lru_in_list));
>> - while (!list_empty(&cli->cl_lru_list)) {
>> + list_for_each_entry_safe(opg, tmp, &cli->cl_lru_list, ops_lru) {
>> struct cl_page *page;
>> bool will_free = false;
>>
>> @@ -582,8 +583,6 @@ long osc_lru_shrink(const struct lu_env *env, struct client_obd *cli,
>> if (--maxscan < 0)
>> break;
>>
>> - opg = list_entry(cli->cl_lru_list.next, struct osc_page,
>> - ops_lru);
>> page = opg->ops_cl.cpl_page;
>> if (lru_page_busy(cli, page)) {
>> list_move_tail(&opg->ops_lru, &cli->cl_lru_list);
>> @@ -1043,6 +1042,7 @@ unsigned long osc_cache_shrink_scan(struct shrinker *sk,
>> {
>> struct client_obd *stop_anchor = NULL;
>> struct client_obd *cli;
>> + struct client_obd *tmp;
>> struct lu_env *env;
>> long shrank = 0;
>> u16 refcheck;
>> @@ -1059,10 +1059,7 @@ unsigned long osc_cache_shrink_scan(struct shrinker *sk,
>> return SHRINK_STOP;
>>
>> spin_lock(&osc_shrink_lock);
>> - while (!list_empty(&osc_shrink_list)) {
>> - cli = list_entry(osc_shrink_list.next, struct client_obd,
>> - cl_shrink_list);
>> -
>> + list_for_each_entry_safe(cli, tmp, &osc_shrink_list, cl_shrink_list) {
>> if (!stop_anchor)
>> stop_anchor = cli;
>> else if (cli == stop_anchor)
>> --
>> 2.7.4

Cheers, Andreas
--
Andreas Dilger
Lustre Principal Architect
Intel Corporation