Hi,
在 2025/09/01 14:30, Damien Le Moal 写道:
On 9/1/25 12:32 PM, Yu Kuai wrote:
From: Yu Kuai <yukuai3@xxxxxxxxxx>
If bio is split by internal chunksize of badblocks, the corresponding
badblocks ? Unclear.
This is due to raid1/10/5 internal processing, if read/write range
contain badblocks that are recorded in rdev, this bio will be split to
bypass the badblocks range, an example from raid1 is choose_bb_rdev()
will update max_sectors from read_balance(), and caller will split bio
by max_sectors.
trace_block_split() is missing, causing blktrace can't catch the split
events and make it hader to analyze IO behavior.
maybe:
trace_block_split() is missing, resulting in blktrace inability to catch BIO
split events and making it harder to analyze the BIO sequence.
would be better.
OK,
Fixes: 4b1faf931650 ("block: Kill bio_pair_split()")
Missing Cc: stable@xxxxxxxxxxxxxxx
Signed-off-by: Yu Kuai <yukuai3@xxxxxxxxxx>
With that,
Reviewed-by: Damien Le Moal <dlemoal@xxxxxxxxxx>
(maybe drop the RFC on this patch series ? Sending a review tag for RFC patches
is odd...)
Yes, I'll send the next version without RFC now. :)
Thanks,
Kuai
