Re: [PATCH 1/2] f2fs: swap: remove dead codes

[Chao Yu <chao@kernel.org>]

On 2021/6/5 7:51, Jaegeuk Kim wrote:
> On 06/04, Chao Yu wrote:
>> On 2021/5/27 10:12, Chao Yu wrote:
>>> On 2021/5/27 9:40, Jaegeuk Kim wrote:
>>>> On 05/26, Chao Yu wrote:
>>>>> On 2021/5/26 21:30, Jaegeuk Kim wrote:
>>>>>> On 05/26, Chao Yu wrote:
>>>>>>> After commit af4b6b8edf6a ("f2fs: introduce check_swap_activate_fast()"),
>>>>>>> we will never run into original logic of check_swap_activate() before
>>>>>>> f2fs supports non 4k-sized page, so let's delete those dead codes.
>>>>>>
>>>>>> Why not keeping this for large page support in future maybe?
>>>>>
>>>>> Well, if so, at that time, it would be better to refactor
>>>>> check_swap_activate_fast() implementation based on f2fs_map_block() rather
>>>>> than refactoring check_swap_activate() implementation based on low efficient
>>>>> bmap()?
>>>>
>>>> Let's first check whether we support large page. Have you quickly tested it?
>>>> If we support it now and current flow is just inefficient, I'd say keeping but
>>>> refactoring it later.
>>>
>>> Let me check this, but I guess it missed to handle large page in lots of flows,
>>> not sure where we may panic at... :(
>>
>> I've checked f2fs behavior on 64kb page size in arm64 vm,
>> - 'ls -ls' will cause f2fs printing "invalid namelen(0), ino:%u, run fsck
>> to fix." each time.
>> - after touch a file and umount, fsck reports that image is corrupted.
>>
>> F2FS uses PAGE_SIZE and F2FS_BLKSIZE directly in a lot of places, I doubt
>> most of places may not consider large page size condition.
>>
>> So, thoughts? I guess we have to admit we do not support 64KB page right
>> now... :P
> 
> Yeah, we need to block the large page support unfortunately. :(

Could you please queue this series in dev-test?

Thanks,

> 
>>
>> Thanks,
>>
>>>
>>> Thanks,
>>>
>>>>
>>>>>
>>>>> Thanks,
>>>>>
>>>>>>
>>>>>>>
>>>>>>> Signed-off-by: Chao Yu <yuchao0@huawei.com>
>>>>>>> ---
>>>>>>>     fs/f2fs/data.c | 171 +------------------------------------------------
>>>>>>>     1 file changed, 3 insertions(+), 168 deletions(-)
>>>>>>>
>>>>>>> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
>>>>>>> index 3058c7e28b11..9c23fde93b76 100644
>>>>>>> --- a/fs/f2fs/data.c
>>>>>>> +++ b/fs/f2fs/data.c
>>>>>>> @@ -3830,67 +3830,7 @@ int f2fs_migrate_page(struct address_space *mapping,
>>>>>>>     #endif
>>>>>>>     #ifdef CONFIG_SWAP
>>>>>>> -static int f2fs_is_file_aligned(struct inode *inode)
>>>>>>> -{
>>>>>>> -    struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
>>>>>>> -    block_t main_blkaddr = SM_I(sbi)->main_blkaddr;
>>>>>>> -    block_t cur_lblock;
>>>>>>> -    block_t last_lblock;
>>>>>>> -    block_t pblock;
>>>>>>> -    unsigned long nr_pblocks;
>>>>>>> -    unsigned int blocks_per_sec = BLKS_PER_SEC(sbi);
>>>>>>> -    unsigned int not_aligned = 0;
>>>>>>> -    int ret = 0;
>>>>>>> -
>>>>>>> -    cur_lblock = 0;
>>>>>>> -    last_lblock = bytes_to_blks(inode, i_size_read(inode));
>>>>>>> -
>>>>>>> -    while (cur_lblock < last_lblock) {
>>>>>>> -        struct f2fs_map_blocks map;
>>>>>>> -
>>>>>>> -        memset(&map, 0, sizeof(map));
>>>>>>> -        map.m_lblk = cur_lblock;
>>>>>>> -        map.m_len = last_lblock - cur_lblock;
>>>>>>> -        map.m_next_pgofs = NULL;
>>>>>>> -        map.m_next_extent = NULL;
>>>>>>> -        map.m_seg_type = NO_CHECK_TYPE;
>>>>>>> -        map.m_may_create = false;
>>>>>>> -
>>>>>>> -        ret = f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_FIEMAP);
>>>>>>> -        if (ret)
>>>>>>> -            goto out;
>>>>>>> -
>>>>>>> -        /* hole */
>>>>>>> -        if (!(map.m_flags & F2FS_MAP_FLAGS)) {
>>>>>>> -            f2fs_err(sbi, "Swapfile has holes\n");
>>>>>>> -            ret = -ENOENT;
>>>>>>> -            goto out;
>>>>>>> -        }
>>>>>>> -
>>>>>>> -        pblock = map.m_pblk;
>>>>>>> -        nr_pblocks = map.m_len;
>>>>>>> -
>>>>>>> -        if ((pblock - main_blkaddr) & (blocks_per_sec - 1) ||
>>>>>>> -            nr_pblocks & (blocks_per_sec - 1)) {
>>>>>>> -            if (f2fs_is_pinned_file(inode)) {
>>>>>>> -                f2fs_err(sbi, "Swapfile does not align to section");
>>>>>>> -                ret = -EINVAL;
>>>>>>> -                goto out;
>>>>>>> -            }
>>>>>>> -            not_aligned++;
>>>>>>> -        }
>>>>>>> -
>>>>>>> -        cur_lblock += nr_pblocks;
>>>>>>> -    }
>>>>>>> -    if (not_aligned)
>>>>>>> -        f2fs_warn(sbi, "Swapfile (%u) is not align to section: \n"
>>>>>>> -            "\t1) creat(), 2) ioctl(F2FS_IOC_SET_PIN_FILE), 3) fallocate()",
>>>>>>> -            not_aligned);
>>>>>>> -out:
>>>>>>> -    return ret;
>>>>>>> -}
>>>>>>> -
>>>>>>> -static int check_swap_activate_fast(struct swap_info_struct *sis,
>>>>>>> +static int check_swap_activate(struct swap_info_struct *sis,
>>>>>>>                     struct file *swap_file, sector_t *span)
>>>>>>>     {
>>>>>>>         struct address_space *mapping = swap_file->f_mapping;
>>>>>>> @@ -3907,6 +3847,8 @@ static int check_swap_activate_fast(struct swap_info_struct *sis,
>>>>>>>         unsigned int not_aligned = 0;
>>>>>>>         int ret = 0;
>>>>>>> +    f2fs_bug_on(sbi, PAGE_SIZE != F2FS_BLKSIZE);
>>>>>>> +
>>>>>>>         /*
>>>>>>>          * Map all the blocks into the extent list.  This code doesn't try
>>>>>>>          * to be very smart.
>>>>>>> @@ -3986,113 +3928,6 @@ static int check_swap_activate_fast(struct swap_info_struct *sis,
>>>>>>>         return ret;
>>>>>>>     }
>>>>>>> -/* Copied from generic_swapfile_activate() to check any holes */
>>>>>>> -static int check_swap_activate(struct swap_info_struct *sis,
>>>>>>> -                struct file *swap_file, sector_t *span)
>>>>>>> -{
>>>>>>> -    struct address_space *mapping = swap_file->f_mapping;
>>>>>>> -    struct inode *inode = mapping->host;
>>>>>>> -    struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
>>>>>>> -    unsigned blocks_per_page;
>>>>>>> -    unsigned long page_no;
>>>>>>> -    sector_t probe_block;
>>>>>>> -    sector_t last_block;
>>>>>>> -    sector_t lowest_block = -1;
>>>>>>> -    sector_t highest_block = 0;
>>>>>>> -    int nr_extents = 0;
>>>>>>> -    int ret = 0;
>>>>>>> -
>>>>>>> -    if (PAGE_SIZE == F2FS_BLKSIZE)
>>>>>>> -        return check_swap_activate_fast(sis, swap_file, span);
>>>>>>> -
>>>>>>> -    ret = f2fs_is_file_aligned(inode);
>>>>>>> -    if (ret)
>>>>>>> -        goto out;
>>>>>>> -
>>>>>>> -    blocks_per_page = bytes_to_blks(inode, PAGE_SIZE);
>>>>>>> -
>>>>>>> -    /*
>>>>>>> -     * Map all the blocks into the extent list.  This code doesn't try
>>>>>>> -     * to be very smart.
>>>>>>> -     */
>>>>>>> -    probe_block = 0;
>>>>>>> -    page_no = 0;
>>>>>>> -    last_block = bytes_to_blks(inode, i_size_read(inode));
>>>>>>> -    while ((probe_block + blocks_per_page) <= last_block &&
>>>>>>> -            page_no < sis->max) {
>>>>>>> -        unsigned block_in_page;
>>>>>>> -        sector_t first_block;
>>>>>>> -        sector_t block = 0;
>>>>>>> -
>>>>>>> -        cond_resched();
>>>>>>> -
>>>>>>> -        block = probe_block;
>>>>>>> -        ret = bmap(inode, &block);
>>>>>>> -        if (ret)
>>>>>>> -            goto out;
>>>>>>> -        if (!block)
>>>>>>> -            goto bad_bmap;
>>>>>>> -        first_block = block;
>>>>>>> -
>>>>>>> -        /*
>>>>>>> -         * It must be PAGE_SIZE aligned on-disk
>>>>>>> -         */
>>>>>>> -        if (first_block & (blocks_per_page - 1)) {
>>>>>>> -            probe_block++;
>>>>>>> -            goto reprobe;
>>>>>>> -        }
>>>>>>> -
>>>>>>> -        for (block_in_page = 1; block_in_page < blocks_per_page;
>>>>>>> -                    block_in_page++) {
>>>>>>> -
>>>>>>> -            block = probe_block + block_in_page;
>>>>>>> -            ret = bmap(inode, &block);
>>>>>>> -            if (ret)
>>>>>>> -                goto out;
>>>>>>> -            if (!block)
>>>>>>> -                goto bad_bmap;
>>>>>>> -
>>>>>>> -            if (block != first_block + block_in_page) {
>>>>>>> -                /* Discontiguity */
>>>>>>> -                probe_block++;
>>>>>>> -                goto reprobe;
>>>>>>> -            }
>>>>>>> -        }
>>>>>>> -
>>>>>>> -        first_block >>= (PAGE_SHIFT - inode->i_blkbits);
>>>>>>> -        if (page_no) {    /* exclude the header page */
>>>>>>> -            if (first_block < lowest_block)
>>>>>>> -                lowest_block = first_block;
>>>>>>> -            if (first_block > highest_block)
>>>>>>> -                highest_block = first_block;
>>>>>>> -        }
>>>>>>> -
>>>>>>> -        /*
>>>>>>> -         * We found a PAGE_SIZE-length, PAGE_SIZE-aligned run of blocks
>>>>>>> -         */
>>>>>>> -        ret = add_swap_extent(sis, page_no, 1, first_block);
>>>>>>> -        if (ret < 0)
>>>>>>> -            goto out;
>>>>>>> -        nr_extents += ret;
>>>>>>> -        page_no++;
>>>>>>> -        probe_block += blocks_per_page;
>>>>>>> -reprobe:
>>>>>>> -        continue;
>>>>>>> -    }
>>>>>>> -    ret = nr_extents;
>>>>>>> -    *span = 1 + highest_block - lowest_block;
>>>>>>> -    if (page_no == 0)
>>>>>>> -        page_no = 1;    /* force Empty message */
>>>>>>> -    sis->max = page_no;
>>>>>>> -    sis->pages = page_no - 1;
>>>>>>> -    sis->highest_bit = page_no - 1;
>>>>>>> -out:
>>>>>>> -    return ret;
>>>>>>> -bad_bmap:
>>>>>>> -    f2fs_err(sbi, "Swapfile has holes\n");
>>>>>>> -    return -EINVAL;
>>>>>>> -}
>>>>>>> -
>>>>>>>     static int f2fs_swap_activate(struct swap_info_struct *sis, struct file *file,
>>>>>>>                     sector_t *span)
>>>>>>>     {
>>>>>>> -- 
>>>>>>> 2.29.2
>>>> .
>>>>