[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <5c359041-27e8-12d2-5a85-559ab56b835e@huawei.com>
Date: Wed, 12 May 2021 18:13:30 +0800
From: Chao Yu <yuchao0@...wei.com>
To: Jaegeuk Kim <jaegeuk@...nel.org>, <linux-kernel@...r.kernel.org>,
<linux-f2fs-devel@...ts.sourceforge.net>
CC: kernel test robot <oliver.sang@...el.com>
Subject: Re: [f2fs-dev] [PATCH v2] f2fs: avoid swapon failure by giving a
warning first
Jaegeuk,
On 2021/5/12 12:00, Jaegeuk Kim wrote:
> The final solution can be migrating blocks to form a section-aligned file
> internally. Meanwhile, let's ask users to do that when preparing the swap
> file initially like:
> 1) create()
> 2) ioctl(F2FS_IOC_SET_PIN_FILE)
> 3) fallocate()
>
> Reported-by: kernel test robot <oliver.sang@...el.com>
> Fixes: 36e4d95891ed ("f2fs: check if swapfile is section-alligned")
> Signed-off-by: Jaegeuk Kim <jaegeuk@...nel.org>
Reviewed-by: Chao Yu <yuchao0@...wei.com>
I've not tested yet though, any comments on below patch?
[PATCH RFC] f2fs: support migrating swapfile in aligned write mode
Thanks,
> ---
> v2:
> - fix to warn out once
> - cover check_swap_activate_fast
>
> fs/f2fs/data.c | 29 +++++++++++++++++++++++------
> 1 file changed, 23 insertions(+), 6 deletions(-)
>
> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
> index 33e56ae84e35..41e260680b27 100644
> --- a/fs/f2fs/data.c
> +++ b/fs/f2fs/data.c
> @@ -3801,6 +3801,7 @@ static int f2fs_is_file_aligned(struct inode *inode)
> 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;
> @@ -3833,13 +3834,20 @@ static int f2fs_is_file_aligned(struct inode *inode)
>
> if ((pblock - main_blkaddr) & (blocks_per_sec - 1) ||
> nr_pblocks & (blocks_per_sec - 1)) {
> - f2fs_err(sbi, "Swapfile does not align to section");
> - ret = -EINVAL;
> - goto out;
> + 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;
> }
> @@ -3858,6 +3866,7 @@ static int check_swap_activate_fast(struct swap_info_struct *sis,
> int nr_extents = 0;
> unsigned long nr_pblocks;
> unsigned int blocks_per_sec = BLKS_PER_SEC(sbi);
> + unsigned int not_aligned = 0;
> int ret = 0;
>
> /*
> @@ -3896,9 +3905,12 @@ static int check_swap_activate_fast(struct swap_info_struct *sis,
>
> if ((pblock - SM_I(sbi)->main_blkaddr) & (blocks_per_sec - 1) ||
> nr_pblocks & (blocks_per_sec - 1)) {
> - f2fs_err(sbi, "Swapfile does not align to section");
> - ret = -EINVAL;
> - goto out;
> + if (f2fs_is_pinned_file(inode)) {
> + f2fs_err(sbi, "Swapfile does not align to section");
> + ret = -EINVAL;
> + goto out;
> + }
> + not_aligned++;
> }
>
> if (cur_lblock + nr_pblocks >= sis->max)
> @@ -3927,6 +3939,11 @@ static int check_swap_activate_fast(struct swap_info_struct *sis,
> sis->max = cur_lblock;
> sis->pages = cur_lblock - 1;
> sis->highest_bit = cur_lblock - 1;
> +
> + 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;
> }
>
Powered by blists - more mailing lists