When a large flex_bg file system is resized, the number of groups to be
added may be small, and a large amount of memory that will not be used will
be allocated. Therefore, resize_bg can be set to the size after the number
of new_group_data to be used is aligned upwards to the power of 2. This
does not affect the disk layout after online resize and saves some memory.

Signed-off-by: Baokun Li <libaokun1@huawei.com>
---
 fs/ext4/resize.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c
index e168a9f59600..4a7430524014 100644
--- a/fs/ext4/resize.c
+++ b/fs/ext4/resize.c
@@ -235,8 +235,10 @@ struct ext4_new_flex_group_data {
  *
  * Returns NULL on failure otherwise address of the allocated structure.
  */
-static struct ext4_new_flex_group_data *alloc_flex_gd(unsigned int flexbg_size)
+static struct ext4_new_flex_group_data *alloc_flex_gd(unsigned int flexbg_size,
+				ext4_group_t o_group, ext4_group_t n_group)
 {
+	ext4_group_t last_group;
 	struct ext4_new_flex_group_data *flex_gd;
 
 	flex_gd = kmalloc(sizeof(*flex_gd), GFP_NOFS);
@@ -248,6 +250,14 @@ static struct ext4_new_flex_group_data *alloc_flex_gd(unsigned int flexbg_size)
 	else
 		flex_gd->resize_bg = flexbg_size;
 
+	/* Avoid allocating new groups that will not be used. */
+	last_group = o_group | (flex_gd->resize_bg - 1);
+	if (n_group <= last_group)
+		flex_gd->resize_bg = 1 << fls(n_group - o_group + 1);
+	else if (n_group - last_group < flex_gd->resize_bg)
+		flex_gd->resize_bg = 1 << max(fls(last_group - o_group + 1),
+					      fls(n_group - last_group));
+
 	flex_gd->groups = kmalloc_array(flex_gd->resize_bg,
 					sizeof(struct ext4_new_group_data),
 					GFP_NOFS);
@@ -2131,7 +2141,7 @@ int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count)
 	if (err)
 		goto out;
 
-	flex_gd = alloc_flex_gd(flexbg_size);
+	flex_gd = alloc_flex_gd(flexbg_size, o_group, n_group);
 	if (flex_gd == NULL) {
 		err = -ENOMEM;
 		goto out;
-- 
2.31.1

On Wed 18-10-23 19:42:21, Baokun Li wrote:
> When a large flex_bg file system is resized, the number of groups to be
> added may be small, and a large amount of memory that will not be used will
> be allocated. Therefore, resize_bg can be set to the size after the number
> of new_group_data to be used is aligned upwards to the power of 2. This
> does not affect the disk layout after online resize and saves some memory.
> 
> Signed-off-by: Baokun Li <libaokun1@huawei.com>

Looks good, just one small comment below. Feel free to add:

Reviewed-by: Jan Kara <jack@suse.cz>


> @@ -248,6 +250,14 @@ static struct ext4_new_flex_group_data *alloc_flex_gd(unsigned int flexbg_size)
>  	else
>  		flex_gd->resize_bg = flexbg_size;
>  
> +	/* Avoid allocating new groups that will not be used. */

Perhaps make the comment more understandable like:
	/* Avoid allocating large 'groups' array if not needed */

									Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

On 2023/10/19 19:57, Jan Kara wrote:
> On Wed 18-10-23 19:42:21, Baokun Li wrote:
>> When a large flex_bg file system is resized, the number of groups to be
>> added may be small, and a large amount of memory that will not be used will
>> be allocated. Therefore, resize_bg can be set to the size after the number
>> of new_group_data to be used is aligned upwards to the power of 2. This
>> does not affect the disk layout after online resize and saves some memory.
>>
>> Signed-off-by: Baokun Li <libaokun1@huawei.com>
> Looks good, just one small comment below. Feel free to add:
>
> Reviewed-by: Jan Kara <jack@suse.cz>
>
>
>> @@ -248,6 +250,14 @@ static struct ext4_new_flex_group_data *alloc_flex_gd(unsigned int flexbg_size)
>>   	else
>>   		flex_gd->resize_bg = flexbg_size;
>>   
>> +	/* Avoid allocating new groups that will not be used. */
> Perhaps make the comment more understandable like:
> 	/* Avoid allocating large 'groups' array if not needed */
>
> 									Honza
Thank you very much for your careful review!
I'll change the comment to this more understandable one in the next version.

Cheers!
-- 
With Best Regards,
Baokun Li
.