[PATCH] f2fs: restrict write IO alignment condition

[PATCH] f2fs: restrict write IO alignment condition

[Chao Yu]

We should only align start offset of bio with defined IO_SIZE for below
conditions:
a. write IOs
b. Out-place-update
c. non-meta page

Signed-off-by: Chao Yu <yuchao0@huawei.com>
---
 fs/f2fs/data.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 1375fef11146..3e6b0cc5e36a 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -395,7 +395,9 @@ int f2fs_submit_page_mbio(struct f2fs_io_info *fio)
 		__submit_merged_bio(io);
 alloc_new:
 	if (io->bio == NULL) {
-		if ((fio->type == DATA || fio->type == NODE) &&
+		if (!is_read && (fio->type == DATA || fio->type == NODE) &&
+				bio_page->mapping != META_MAPPING(sbi) &&
+				fio->new_blkaddr != fio->old_blkaddr &&
 				fio->new_blkaddr & F2FS_IO_SIZE_MASK(sbi)) {
 			err = -EAGAIN;
 			if (!is_read)
-- 
2.8.2.295.g3f1c1d0

Re: [PATCH] f2fs: restrict write IO alignment condition

[Jaegeuk Kim]

On 03/10, Chao Yu wrote:
> We should only align start offset of bio with defined IO_SIZE for below
> conditions:
> a. write IOs
> b. Out-place-update
> c. non-meta page
> 
> Signed-off-by: Chao Yu <yuchao0@huawei.com>
> ---
>  fs/f2fs/data.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
> index 1375fef11146..3e6b0cc5e36a 100644
> --- a/fs/f2fs/data.c
> +++ b/fs/f2fs/data.c
> @@ -395,7 +395,9 @@ int f2fs_submit_page_mbio(struct f2fs_io_info *fio)
>  		__submit_merged_bio(io);
>  alloc_new:
>  	if (io->bio == NULL) {
> -		if ((fio->type == DATA || fio->type == NODE) &&
> +		if (!is_read && (fio->type == DATA || fio->type == NODE) &&
> +				bio_page->mapping != META_MAPPING(sbi) &&

BUG_ON(), since it's covered by fio->type == DATA || fio->type == NODE.

> +				fio->new_blkaddr != fio->old_blkaddr &&

BUG_ON(), since LFS mount option doesn't allow any in-place updates.

>  				fio->new_blkaddr & F2FS_IO_SIZE_MASK(sbi)) {

So, is_read is worth to check?

Thanks,

>  			err = -EAGAIN;
>  			if (!is_read)
> -- 
> 2.8.2.295.g3f1c1d0

Re: [PATCH] f2fs: restrict write IO alignment condition

[Chao Yu]

On 2017/3/11 2:50, Jaegeuk Kim wrote:
> On 03/10, Chao Yu wrote:
>> We should only align start offset of bio with defined IO_SIZE for below
>> conditions:
>> a. write IOs
>> b. Out-place-update
>> c. non-meta page
>>
>> Signed-off-by: Chao Yu <yuchao0@huawei.com>
>> ---
>>  fs/f2fs/data.c | 4 +++-
>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
>> index 1375fef11146..3e6b0cc5e36a 100644
>> --- a/fs/f2fs/data.c
>> +++ b/fs/f2fs/data.c
>> @@ -395,7 +395,9 @@ int f2fs_submit_page_mbio(struct f2fs_io_info *fio)
>>  		__submit_merged_bio(io);
>>  alloc_new:
>>  	if (io->bio == NULL) {
>> -		if ((fio->type == DATA || fio->type == NODE) &&
>> +		if (!is_read && (fio->type == DATA || fio->type == NODE) &&
>> +				bio_page->mapping != META_MAPPING(sbi) &&
> 
> BUG_ON(), since it's covered by fio->type == DATA || fio->type == NODE.

Oh, the condition I added here is wrong, since move_encrypted_block will add a
meta page into DATA type bio cache, also we can not add a BUG_ON here.

> 
>> +				fio->new_blkaddr != fio->old_blkaddr &&
> 
> BUG_ON(), since LFS mount option doesn't allow any in-place updates.

You're right. :)

> 
>>  				fio->new_blkaddr & F2FS_IO_SIZE_MASK(sbi)) {
> 
> So, is_read is worth to check?

Yes.

I will fix them and resend v2.

Thanks,

> 
> Thanks,
> 
>>  			err = -EAGAIN;
>>  			if (!is_read)
>> -- 
>> 2.8.2.295.g3f1c1d0
> 
> .
>