2024-01-14 12:30:39

by Christophe JAILLET

Subject: [PATCH] bcache: Remove usage of the deprecated ida_simple_xx() API

ida_alloc() and ida_free() should be preferred to the deprecated
ida_simple_get() and ida_simple_remove().

Note that the upper limit of ida_simple_get() is exclusive, but the one of
ida_alloc_max() is inclusive. So a -1 has been added when needed.

Signed-off-by: Christophe JAILLET <[email protected]>
---
drivers/md/bcache/super.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
index dc3f50f69714..a2eecd1db126 100644
--- a/drivers/md/bcache/super.c
+++ b/drivers/md/bcache/super.c
@@ -881,8 +881,8 @@ static void bcache_device_free(struct bcache_device *d)
bcache_device_detach(d);

if (disk) {
- ida_simple_remove(&bcache_device_idx,
- first_minor_to_idx(disk->first_minor));
+ ida_free(&bcache_device_idx,
+ first_minor_to_idx(disk->first_minor));
put_disk(disk);
}

@@ -926,8 +926,8 @@ static int bcache_device_init(struct bcache_device *d, unsigned int block_size,
if (!d->full_dirty_stripes)
goto out_free_stripe_sectors_dirty;

- idx = ida_simple_get(&bcache_device_idx, 0,
- BCACHE_DEVICE_IDX_MAX, GFP_KERNEL);
+ idx = ida_alloc_max(&bcache_device_idx, BCACHE_DEVICE_IDX_MAX - 1,
+ GFP_KERNEL);
if (idx < 0)
goto out_free_full_dirty_stripes;

@@ -980,7 +980,7 @@ static int bcache_device_init(struct bcache_device *d, unsigned int block_size,
out_bioset_exit:
bioset_exit(&d->bio_split);
out_ida_remove:
- ida_simple_remove(&bcache_device_idx, idx);
+ ida_free(&bcache_device_idx, idx);
out_free_full_dirty_stripes:
kvfree(d->full_dirty_stripes);
out_free_stripe_sectors_dirty:
--
2.43.0

2024-01-14 15:17:01

by Coly Li

Subject: Re: [PATCH] bcache: Remove usage of the deprecated ida_simple_xx() API

On Sun, Jan 14, 2024 at 01:30:16PM +0100, Christophe JAILLET wrote:
> ida_alloc() and ida_free() should be preferred to the deprecated
> ida_simple_get() and ida_simple_remove().
>
> Note that the upper limit of ida_simple_get() is exclusive, but the one of
> ida_alloc_max() is inclusive. So a -1 has been added when needed.
>
> Signed-off-by: Christophe JAILLET <[email protected]>

It looks good to me. Add this patch into my testing directory.

Thanks.

Coly Li


> ---
> drivers/md/bcache/super.c | 10 +++++-----
> 1 file changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
> index dc3f50f69714..a2eecd1db126 100644
> --- a/drivers/md/bcache/super.c
> +++ b/drivers/md/bcache/super.c
> @@ -881,8 +881,8 @@ static void bcache_device_free(struct bcache_device *d)
> bcache_device_detach(d);
>
> if (disk) {
> - ida_simple_remove(&bcache_device_idx,
> - first_minor_to_idx(disk->first_minor));
> + ida_free(&bcache_device_idx,
> + first_minor_to_idx(disk->first_minor));
> put_disk(disk);
> }
>
> @@ -926,8 +926,8 @@ static int bcache_device_init(struct bcache_device *d, unsigned int block_size,
> if (!d->full_dirty_stripes)
> goto out_free_stripe_sectors_dirty;
>
> - idx = ida_simple_get(&bcache_device_idx, 0,
> - BCACHE_DEVICE_IDX_MAX, GFP_KERNEL);
> + idx = ida_alloc_max(&bcache_device_idx, BCACHE_DEVICE_IDX_MAX - 1,
> + GFP_KERNEL);
> if (idx < 0)
> goto out_free_full_dirty_stripes;
>
> @@ -980,7 +980,7 @@ static int bcache_device_init(struct bcache_device *d, unsigned int block_size,
> out_bioset_exit:
> bioset_exit(&d->bio_split);
> out_ida_remove:
> - ida_simple_remove(&bcache_device_idx, idx);
> + ida_free(&bcache_device_idx, idx);
> out_free_full_dirty_stripes:
> kvfree(d->full_dirty_stripes);
> out_free_stripe_sectors_dirty:
> --
> 2.43.0
>

--
Coly Li

2024-04-14 08:25:05

by Christophe JAILLET

Subject: Re: [PATCH] bcache: Remove usage of the deprecated ida_simple_xx() API

Le 14/01/2024 à 16:16, Coly Li a écrit :
> On Sun, Jan 14, 2024 at 01:30:16PM +0100, Christophe JAILLET wrote:
>> ida_alloc() and ida_free() should be preferred to the deprecated
>> ida_simple_get() and ida_simple_remove().
>>
>> Note that the upper limit of ida_simple_get() is exclusive, but the one of
>> ida_alloc_max() is inclusive. So a -1 has been added when needed.
>>
>> Signed-off-by: Christophe JAILLET <[email protected]>
>
> It looks good to me. Add this patch into my testing directory.
>
> Thanks.
>
> Coly Li

Hi,

polite reminder ;-)

CJ

>
>
>> ---
>> drivers/md/bcache/super.c | 10 +++++-----
>> 1 file changed, 5 insertions(+), 5 deletions(-)
>>
>> diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
>> index dc3f50f69714..a2eecd1db126 100644
>> --- a/drivers/md/bcache/super.c
>> +++ b/drivers/md/bcache/super.c
>> @@ -881,8 +881,8 @@ static void bcache_device_free(struct bcache_device *d)
>> bcache_device_detach(d);
>>
>> if (disk) {
>> - ida_simple_remove(&bcache_device_idx,
>> - first_minor_to_idx(disk->first_minor));
>> + ida_free(&bcache_device_idx,
>> + first_minor_to_idx(disk->first_minor));
>> put_disk(disk);
>> }
>>
>> @@ -926,8 +926,8 @@ static int bcache_device_init(struct bcache_device *d, unsigned int block_size,
>> if (!d->full_dirty_stripes)
>> goto out_free_stripe_sectors_dirty;
>>
>> - idx = ida_simple_get(&bcache_device_idx, 0,
>> - BCACHE_DEVICE_IDX_MAX, GFP_KERNEL);
>> + idx = ida_alloc_max(&bcache_device_idx, BCACHE_DEVICE_IDX_MAX - 1,
>> + GFP_KERNEL);
>> if (idx < 0)
>> goto out_free_full_dirty_stripes;
>>
>> @@ -980,7 +980,7 @@ static int bcache_device_init(struct bcache_device *d, unsigned int block_size,
>> out_bioset_exit:
>> bioset_exit(&d->bio_split);
>> out_ida_remove:
>> - ida_simple_remove(&bcache_device_idx, idx);
>> + ida_free(&bcache_device_idx, idx);
>> out_free_full_dirty_stripes:
>> kvfree(d->full_dirty_stripes);
>> out_free_stripe_sectors_dirty:
>> --
>> 2.43.0
>>
>

2024-04-15 15:00:41

by Coly Li

Subject: Re: [PATCH] bcache: Remove usage of the deprecated ida_simple_xx() API

> 2024年4月14日 16:24，Christophe JAILLET <[email protected]> 写道：
>
> Le 14/01/2024 à 16:16, Coly Li a écrit :
>> On Sun, Jan 14, 2024 at 01:30:16PM +0100, Christophe JAILLET wrote:
>>> ida_alloc() and ida_free() should be preferred to the deprecated
>>> ida_simple_get() and ida_simple_remove().
>>>
>>> Note that the upper limit of ida_simple_get() is exclusive, but the one of
>>> ida_alloc_max() is inclusive. So a -1 has been added when needed.
>>>
>>> Signed-off-by: Christophe JAILLET <[email protected]>
>> It looks good to me. Add this patch into my testing directory.
>> Thanks.
>> Coly Li
>
> Hi,
>
> polite reminder ;-)

Yes, this one is included in my recent performance testing :-)

Thank you.

Coly Li