Re: [PATCH v4 06/11] md: support to align bio to limits

在 2026/1/12 19:24, Li Nan 写道:
> 
> 
> 在 2026/1/12 12:28, Yu Kuai 写道:
>> For personalities that report optimal IO size, it indicates that users
>> can get the best IO bandwidth if they issue IO with this size. However
>> there is also an implicit condition that IO should also be aligned to the
>> optimal IO size.
>>
>> Currently, bio will only be split by limits, if bio offset is not aligned
>> to limits, then all split bio will not be aligned. This patch add a new
>> feature to align bio to limits first, and following patches will support
>> this for each personality if necessary.
>>
>> Link: 
>> https://lore.kernel.org/linux-raid/[email protected]
>> Signed-off-by: Yu Kuai <[email protected]>
>> Reviewed-by: Li Nan <[email protected]>
>> ---
>>   drivers/md/md.c | 54 +++++++++++++++++++++++++++++++++++++++++++++++++
>>   drivers/md/md.h |  2 ++
>>   2 files changed, 56 insertions(+)
>>
>> diff --git a/drivers/md/md.c b/drivers/md/md.c
>> index 21b0bc3088d2..731ec800f5cb 100644
>> --- a/drivers/md/md.c
>> +++ b/drivers/md/md.c
>> @@ -428,6 +428,56 @@ bool md_handle_request(struct mddev *mddev, struct 
>> bio *bio)
>>   }
>>   EXPORT_SYMBOL(md_handle_request);
>> +static struct bio *__md_bio_align_to_limits(struct mddev *mddev,
>> +                         struct bio *bio)
>> +{
>> +    unsigned int max_sectors = mddev->gendisk->queue->limits.max_sectors;
>> +    sector_t start = bio->bi_iter.bi_sector;
>> +    sector_t end = start + bio_sectors(bio);
>> +    sector_t align_start;
>> +    sector_t align_end;
>> +    u32 rem;
>> +
>> +    /* calculate align_start = roundup(start, max_sectors) */
> 
> Can we use roundup_u64() here?
> 
>> +    align_start = start;
>> +    rem = sector_div(align_start, max_sectors);
>> +    /* already aligned */
>> +    if (!rem)
>> +        return bio;
>> +
>> +    align_start = start + max_sectors - rem;
>> +
>> +    /* calculate align_end = rounddown(end, max_sectors) */
> 
> Use div64_u64_rem() here seems better.

div64_u64_rem is same as sector_div. Please ignore it.

> 
>> +    align_end = end;
>> +    rem = sector_div(align_end, max_sectors);
>> +    align_end = end - rem;
>> +
>> +    /* bio is too small to split */
>> +    if (align_end <= align_start)
>> +        return bio;
>> +
>> +    return bio_submit_split_bioset(bio, align_start - start,
>> +                       &mddev->gendisk->bio_split);
>> +}
>> +

-- 
Thanks,
Nan