Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1314479 > unrolled thread

Re: [PATCH] cleancache: constify cleancache_ops structure

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH] cleancache: constify cleancache_ops structure Rasmus Villemoes <linux@rasmusvillemoes.dk> - 2016-01-21 21:10 +0100
    Re: [PATCH] cleancache: constify cleancache_ops structure Julia Lawall <julia.lawall@lip6.fr> - 2016-01-21 21:10 +0100

#1314479 — Re: [PATCH] cleancache: constify cleancache_ops structure

On Wed, Jan 20 2016, Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> wrote:

> On Wed, Dec 23, 2015 at 10:06:24PM +0100, Julia Lawall wrote:
>> The cleancache_ops structure is never modified, so declare it as const.
>> 
>> This also removes the __read_mostly declaration on the cleancache_ops
>> variable declaration, since it seems redundant with const.
>> 
>> Done with the help of Coccinelle.
>> 
>> Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
>> 
>> ---
>> 
>> Not sure that the __read_mostly change is correct.  Does it apply to the
>> variable, or to what the variable points to?
>
> It should just put the structure in the right section (.rodata).
>
> Thanks for the patch!

The __read_mostly marker should probably be left there...

>>   */
>> -static struct cleancache_ops *cleancache_ops __read_mostly;
>> +static const struct cleancache_ops *cleancache_ops;
>>  
>>  /*
>>   * Counters available via /sys/kernel/debug/cleancache (if debugfs is
>> @@ -49,7 +49,7 @@ static void cleancache_register_ops_sb(struct super_block *sb, void *unused)
>>  /*
>>   * Register operations for cleancache. Returns 0 on success.
>>   */
>> -int cleancache_register_ops(struct cleancache_ops *ops)
>> +int cleancache_register_ops(const struct cleancache_ops *ops)
>>  {
>>  	if (cmpxchg(&cleancache_ops, NULL, ops))
>>  		return -EBUSY;
>>

I don't know this code, but I assume that this is mostly a one-time
thing, so once cleancache_ops gets its value assigned, it doesn't
change, and that's what the __read_mostly is about (it applies to the
object declared, not whatever it happens to point to).

(Also, the commit message is slightly inaccurate: it is
tmem_cleancache_ops which is never changed and hence declared const;
changing the various pointers to it to const is just a necessary followup).

Rasmus

[toc] | [next] | [standalone]

#1314485

On Thu, 21 Jan 2016, Rasmus Villemoes wrote:

> On Wed, Jan 20 2016, Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> wrote:
> 
> > On Wed, Dec 23, 2015 at 10:06:24PM +0100, Julia Lawall wrote:
> >> The cleancache_ops structure is never modified, so declare it as const.
> >> 
> >> This also removes the __read_mostly declaration on the cleancache_ops
> >> variable declaration, since it seems redundant with const.
> >> 
> >> Done with the help of Coccinelle.
> >> 
> >> Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
> >> 
> >> ---
> >> 
> >> Not sure that the __read_mostly change is correct.  Does it apply to the
> >> variable, or to what the variable points to?
> >
> > It should just put the structure in the right section (.rodata).
> >
> > Thanks for the patch!
> 
> The __read_mostly marker should probably be left there...

I sent a corrected version this afternoon.

> 
> >>   */
> >> -static struct cleancache_ops *cleancache_ops __read_mostly;
> >> +static const struct cleancache_ops *cleancache_ops;
> >>  
> >>  /*
> >>   * Counters available via /sys/kernel/debug/cleancache (if debugfs is
> >> @@ -49,7 +49,7 @@ static void cleancache_register_ops_sb(struct super_block *sb, void *unused)
> >>  /*
> >>   * Register operations for cleancache. Returns 0 on success.
> >>   */
> >> -int cleancache_register_ops(struct cleancache_ops *ops)
> >> +int cleancache_register_ops(const struct cleancache_ops *ops)
> >>  {
> >>  	if (cmpxchg(&cleancache_ops, NULL, ops))
> >>  		return -EBUSY;
> >>
> 
> I don't know this code, but I assume that this is mostly a one-time
> thing, so once cleancache_ops gets its value assigned, it doesn't
> change, and that's what the __read_mostly is about (it applies to the
> object declared, not whatever it happens to point to).
> 
> (Also, the commit message is slightly inaccurate: it is
> tmem_cleancache_ops which is never changed and hence declared const;
> changing the various pointers to it to const is just a necessary followup).

OK, in general, I have referred to the type rather than the structure name 
in these patches, since there can be more than one structure.

julia

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web