[PATCH] xfs,fstrim: fix to return correct minlen

[PATCH] xfs,fstrim: fix to return correct minlen

[Wang Shilong]

From: Wang Shilong <wshilong@ddn.com>

This patch tries to address two problems:

1) return @minlen we used to trim to
user space.

2) return EINVAL if granularity is larger than
avg size, even most of cases, granularity is small(4K),
but if devices return a lager granularity for some reaons
(testing, bugs etc), fstrim should return failure directly.

Signed-off-by: Wang Shilong <wshilong@ddn.com>
---
 fs/xfs/xfs_discard.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_discard.c b/fs/xfs/xfs_discard.c
index 93f07edafd81..66702b03587a 100644
--- a/fs/xfs/xfs_discard.c
+++ b/fs/xfs/xfs_discard.c
@@ -164,6 +164,8 @@ xfs_ioc_trim(
 	if (copy_from_user(&range, urange, sizeof(range)))
 		return -EFAULT;
 
+	range.minlen = max_t(u64, granularity, range.minlen);
+	minlen = BTOBB(range.minlen);
 	/*
 	 * Truncating down the len isn't actually quite correct, but using
 	 * BBTOB would mean we trivially get overflows for values
@@ -178,7 +180,6 @@ xfs_ioc_trim(
 
 	start = BTOBB(range.start);
 	end = start + BTOBBT(range.len) - 1;
-	minlen = BTOBB(max_t(u64, granularity, range.minlen));
 
 	if (end > XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks) - 1)
 		end = XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)- 1;
-- 
2.20.1

Re: [PATCH] xfs,fstrim: fix to return correct minlen

[Darrick J. Wong]

On Wed, Apr 10, 2019 at 05:42:28PM +0900, Wang Shilong wrote:
> From: Wang Shilong <wshilong@ddn.com>
> 
> This patch tries to address two problems:
> 
> 1) return @minlen we used to trim to
> user space.
> 
> 2) return EINVAL if granularity is larger than
> avg size, even most of cases, granularity is small(4K),
> but if devices return a lager granularity for some reaons
> (testing, bugs etc), fstrim should return failure directly.

It really is a pity^W^Wannoying that there's no manual page for FITRIM,
because I have no idea if this is the right or desired behavior for this
ioctl.

Do the other filesystems update @minlen so that userspace knows the
granularity?

--D

> Signed-off-by: Wang Shilong <wshilong@ddn.com>
> ---
>  fs/xfs/xfs_discard.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/xfs/xfs_discard.c b/fs/xfs/xfs_discard.c
> index 93f07edafd81..66702b03587a 100644
> --- a/fs/xfs/xfs_discard.c
> +++ b/fs/xfs/xfs_discard.c
> @@ -164,6 +164,8 @@ xfs_ioc_trim(
>  	if (copy_from_user(&range, urange, sizeof(range)))
>  		return -EFAULT;
>  
> +	range.minlen = max_t(u64, granularity, range.minlen);
> +	minlen = BTOBB(range.minlen);
>  	/*
>  	 * Truncating down the len isn't actually quite correct, but using
>  	 * BBTOB would mean we trivially get overflows for values
> @@ -178,7 +180,6 @@ xfs_ioc_trim(
>  
>  	start = BTOBB(range.start);
>  	end = start + BTOBBT(range.len) - 1;
> -	minlen = BTOBB(max_t(u64, granularity, range.minlen));
>  
>  	if (end > XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks) - 1)
>  		end = XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)- 1;
> -- 
> 2.20.1
> 

回复: [PATCH] xfs,fstrim: fix to return correct minlen

[Wang Shilong]

Hi Darrick,


On Wed, Apr 10, 2019 at 05:42:28PM +0900, Wang Shilong wrote:
> From: Wang Shilong <wshilong@ddn.com>
>
> This patch tries to address two problems:
>
> 1) return @minlen we used to trim to
> user space.
>
> 2) return EINVAL if granularity is larger than
> avg size, even most of cases, granularity is small(4K),
> but if devices return a lager granularity for some reaons
> (testing, bugs etc), fstrim should return failure directly.

It really is a pity^W^Wannoying that there's no manual page for FITRIM,
because I have no idea if this is the right or desired behavior for this
ioctl.

Do the other filesystems update @minlen so that userspace knows the
granularity?

--D

--------->
As far as I checked ext4, f2fs, Btrfs did so, I think it makes sense we
return this value to userspace, we'd better to doc it somewhere 'man fstrim'?

thanks,
Shilong

Re: 回复: [PATCH] xfs,fstrim: fix to return correct minlen

[Darrick J. Wong]

On Thu, Apr 11, 2019 at 02:05:08AM +0000, Wang Shilong wrote:
> Hi Darrick,
> 
> 
> On Wed, Apr 10, 2019 at 05:42:28PM +0900, Wang Shilong wrote:
> > From: Wang Shilong <wshilong@ddn.com>
> >
> > This patch tries to address two problems:
> >
> > 1) return @minlen we used to trim to
> > user space.
> >
> > 2) return EINVAL if granularity is larger than
> > avg size, even most of cases, granularity is small(4K),
> > but if devices return a lager granularity for some reaons
> > (testing, bugs etc), fstrim should return failure directly.
> 
> It really is a pity^W^Wannoying that there's no manual page for FITRIM,
> because I have no idea if this is the right or desired behavior for this
> ioctl.
> 
> Do the other filesystems update @minlen so that userspace knows the
> granularity?
> 
> --D
> 
> --------->
> As far as I checked ext4, f2fs, Btrfs did so,

Fair enough.

> I think it makes sense we return this value to userspace, we'd better
> to doc it somewhere 'man fstrim'?

Ok, please cc me when you send the documentation update.

--D

> thanks,
> Shilong

Re: [PATCH] xfs,fstrim: fix to return correct minlen

[Darrick J. Wong]

On Wed, Apr 10, 2019 at 05:42:28PM +0900, Wang Shilong wrote:
> From: Wang Shilong <wshilong@ddn.com>
> 
> This patch tries to address two problems:
> 
> 1) return @minlen we used to trim to
> user space.
> 
> 2) return EINVAL if granularity is larger than
> avg size, even most of cases, granularity is small(4K),
> but if devices return a lager granularity for some reaons
> (testing, bugs etc), fstrim should return failure directly.
> 
> Signed-off-by: Wang Shilong <wshilong@ddn.com>

Ok, well, I guess all the other filesystems do this, so we should too...

Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>

--D

> ---
>  fs/xfs/xfs_discard.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/xfs/xfs_discard.c b/fs/xfs/xfs_discard.c
> index 93f07edafd81..66702b03587a 100644
> --- a/fs/xfs/xfs_discard.c
> +++ b/fs/xfs/xfs_discard.c
> @@ -164,6 +164,8 @@ xfs_ioc_trim(
>  	if (copy_from_user(&range, urange, sizeof(range)))
>  		return -EFAULT;
>  
> +	range.minlen = max_t(u64, granularity, range.minlen);
> +	minlen = BTOBB(range.minlen);
>  	/*
>  	 * Truncating down the len isn't actually quite correct, but using
>  	 * BBTOB would mean we trivially get overflows for values
> @@ -178,7 +180,6 @@ xfs_ioc_trim(
>  
>  	start = BTOBB(range.start);
>  	end = start + BTOBBT(range.len) - 1;
> -	minlen = BTOBB(max_t(u64, granularity, range.minlen));
>  
>  	if (end > XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks) - 1)
>  		end = XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)- 1;
> -- 
> 2.20.1
>