On Mon, Nov 11, 2013 at 12:33 PM, Jaegeuk Kim <jaegeuk.kim@xxxxxxxxxxx> wrote:
> Hi Huajun,
>
> 2013-11-05 (화), 21:28 +0800, Huajun Li:
>> Hi Jaegeuk,
>>
>> Got it, and nice to fix it in mkfs.f2fs.
>>
>> Thanks,
>> --Huajun
>> On Tue, Nov 5, 2013 at 12:48 PM, Jaegeuk Kim <jaegeuk.kim@xxxxxxxxxxx> wrote:
>> > Hi Huajun,
>> >
>> > 2013-11-04 (월), 23:40 +0800, Huajun Li:
>> >> Hi Jaegeuk,
>> >>
>> >> On Mon, Nov 4, 2013 at 9:24 AM, Jaegeuk Kim <jaegeuk.kim@xxxxxxxxxxx> wrote:
>> >> > 2013-11-03 (일), 23:08 +0800, Huajun Li:
>> >> >> From: Huajun Li <huajun.li@xxxxxxxxx>
>> >> >>
>> >> >> Normally we expect an empty partition after formatting by
>> >> >> mkfs.f2fs. But in this case, when we format a dirty partition and mount
>> >> >> it again. The former file will be recovered and available again! and
>> >> >> kernel log shows a recovery procedure is evoked.
>> >> >> This patch adds a new flag CP_EXIST_FLAG to indicate whether is a
>> >> >> available CP, and do recovery only when this flag is set.
>> >> >
>> >> > IMO, mkfs.f2fs should do the right thing to avoid this.
>> >> > If storage does not support discard, mkfs.f2fs can simply address the
>> >> > problem by writing one node block with zeros to prevent this.
>> >> >
>
> WRT the below issue, I made a patch for mkfs.f2fs.
> If possible, could you test and write a valid patch?
> Thanks,
>
Hi Jaegeuk,
Test your following fix just now on the same test machine, it can work fine.
This is new implementation to fix the issue, maybe you can
submit/integrate your patch directly. ;)
> ---
> mkfs/f2fs_format.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/mkfs/f2fs_format.c b/mkfs/f2fs_format.c
> index 8234b00..18ded79 100644
> --- a/mkfs/f2fs_format.c
> +++ b/mkfs/f2fs_format.c
> @@ -788,7 +788,12 @@ static int f2fs_write_root_inode(void)
>
> memset(raw_node, 0xff, sizeof(struct f2fs_node));
>
> - main_area_node_seg_blk_offset += F2FS_BLKSIZE;
> + /* avoid power-off-recovery based on roll-forward policy */
> + main_area_node_seg_blk_offset = le32_to_cpu(super_block.main_blkaddr);
> + main_area_node_seg_blk_offset += config.cur_seg[CURSEG_WARM_NODE] *
> + config.blks_per_seg;
> + main_area_node_seg_blk_offset *= blk_size_bytes;
> +
> if (dev_write(raw_node, main_area_node_seg_blk_offset, F2FS_BLKSIZE))
> {
> MSG(1, "\tError: While writing the raw_node to disk!!!\n");
> return -1;
> --
> 1.8.4.474.g128a96c
>
>
> --
> Jaegeuk Kim
> Samsung
>
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
