* [PATCH 0/3] Add alignment check for DAX mount
@ 2016-04-29 20:39 Toshi Kani
2016-04-29 20:39 ` [PATCH 1/3] ext4: " Toshi Kani
` (3 more replies)
0 siblings, 4 replies; 9+ messages in thread
From: Toshi Kani @ 2016-04-29 20:39 UTC (permalink / raw)
To: dan.j.williams, david, jack
Cc: tytso, adilger.kernel, ross.zwisler, toshi.kani, linux-nvdimm,
linux-fsdevel, linux-kernel
When a partition is not aligned by 4KB, mount -o dax succeeds,
but any read/write access to the filesystem fails, except for
metadata update.
Add alignment check to ext4, ext2, and xfs.
---
Toshi Kani (3):
1/3 ext4: Add alignment check for DAX mount
2/3 ext2: Add alignment check for DAX mount
3/3 xfs: Add alignment check for DAX mount
---
fs/ext2/super.c | 6 ++++++
fs/ext4/super.c | 6 ++++++
fs/xfs/xfs_super.c | 6 ++++++
3 files changed, 18 insertions(+)
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH 1/3] ext4: Add alignment check for DAX mount
2016-04-29 20:39 [PATCH 0/3] Add alignment check for DAX mount Toshi Kani
@ 2016-04-29 20:39 ` Toshi Kani
2016-05-01 9:35 ` Boaz Harrosh
2016-05-01 17:31 ` Christoph Hellwig
2016-04-29 20:39 ` [PATCH 2/3] ext2: " Toshi Kani
` (2 subsequent siblings)
3 siblings, 2 replies; 9+ messages in thread
From: Toshi Kani @ 2016-04-29 20:39 UTC (permalink / raw)
To: dan.j.williams, david, jack
Cc: tytso, adilger.kernel, ross.zwisler, toshi.kani, linux-nvdimm,
linux-fsdevel, linux-kernel
When a partition is not aligned by 4KB, mount -o dax succeeds,
but any read/write access to the filesystem fails, except for
metadata update.
Add alignment check to ext4_fill_super() when -o dax is specified.
Reported-by: Micah Parrish <micah.parrish@hpe.com>
Signed-off-by: Toshi Kani <toshi.kani@hpe.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Jan Kara <jack@suse.cz>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
---
fs/ext4/super.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 304c712..90a8670 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -3421,6 +3421,12 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
"error: unsupported blocksize for dax");
goto failed_mount;
}
+ if (sb->s_bdev->bd_part->start_sect % (PAGE_SIZE / 512) ||
+ sb->s_bdev->bd_part->nr_sects % (PAGE_SIZE / 512)) {
+ ext4_msg(sb, KERN_ERR,
+ "error: unaligned partition for dax");
+ goto failed_mount;
+ }
if (!sb->s_bdev->bd_disk->fops->direct_access) {
ext4_msg(sb, KERN_ERR,
"error: device does not support dax");
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH 2/3] ext2: Add alignment check for DAX mount
2016-04-29 20:39 [PATCH 0/3] Add alignment check for DAX mount Toshi Kani
2016-04-29 20:39 ` [PATCH 1/3] ext4: " Toshi Kani
@ 2016-04-29 20:39 ` Toshi Kani
2016-04-29 20:39 ` [PATCH 3/3] xfs: " Toshi Kani
2016-04-29 23:12 ` [PATCH 0/3] " Ross Zwisler
3 siblings, 0 replies; 9+ messages in thread
From: Toshi Kani @ 2016-04-29 20:39 UTC (permalink / raw)
To: dan.j.williams, david, jack
Cc: tytso, adilger.kernel, ross.zwisler, toshi.kani, linux-nvdimm,
linux-fsdevel, linux-kernel
When a partition is not aligned by 4KB, mount -o dax succeeds,
but any read/write access to the filesystem fails, except for
metadata update.
Add alignment check to ext2_fill_super() when -o dax is specified.
Signed-off-by: Toshi Kani <toshi.kani@hpe.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
---
fs/ext2/super.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/fs/ext2/super.c b/fs/ext2/super.c
index b78caf2..34e32d9 100644
--- a/fs/ext2/super.c
+++ b/fs/ext2/super.c
@@ -927,6 +927,12 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent)
"error: unsupported blocksize for dax");
goto failed_mount;
}
+ if (sb->s_bdev->bd_part->start_sect % (PAGE_SIZE / 512) ||
+ sb->s_bdev->bd_part->nr_sects % (PAGE_SIZE / 512)) {
+ ext2_msg(sb, KERN_ERR,
+ "error: unaligned partition for dax");
+ goto failed_mount;
+ }
if (!sb->s_bdev->bd_disk->fops->direct_access) {
ext2_msg(sb, KERN_ERR,
"error: device does not support dax");
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH 3/3] xfs: Add alignment check for DAX mount
2016-04-29 20:39 [PATCH 0/3] Add alignment check for DAX mount Toshi Kani
2016-04-29 20:39 ` [PATCH 1/3] ext4: " Toshi Kani
2016-04-29 20:39 ` [PATCH 2/3] ext2: " Toshi Kani
@ 2016-04-29 20:39 ` Toshi Kani
2016-04-29 23:12 ` [PATCH 0/3] " Ross Zwisler
3 siblings, 0 replies; 9+ messages in thread
From: Toshi Kani @ 2016-04-29 20:39 UTC (permalink / raw)
To: dan.j.williams, david, jack
Cc: tytso, adilger.kernel, ross.zwisler, toshi.kani, linux-nvdimm,
linux-fsdevel, linux-kernel
When a partition is not aligned by 4KB, mount -o dax succeeds,
but any read/write access to the filesystem fails, except for
metadata update.
Add alignment check to xfs_fs_fill_super() when -o dax is
specified.
Signed-off-by: Toshi Kani <toshi.kani@hpe.com>
Cc: Dave Chinner <david@fromorbit.com> (supporter:XFS FILESYSTEM)
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
---
fs/xfs/xfs_super.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index 187e14b..a17aebc 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -1563,6 +1563,11 @@ xfs_fs_fill_super(
xfs_alert(mp,
"Filesystem block size invalid for DAX Turning DAX off.");
mp->m_flags &= ~XFS_MOUNT_DAX;
+ } else if (sb->s_bdev->bd_part->start_sect % (PAGE_SIZE / 512)
+ || sb->s_bdev->bd_part->nr_sects % (PAGE_SIZE / 512)) {
+ xfs_alert(mp,
+ "Partition alignment invalid for DAX Turning DAX off.");
+ mp->m_flags &= ~XFS_MOUNT_DAX;
} else if (!sb->s_bdev->bd_disk->fops->direct_access) {
xfs_alert(mp,
"Block device does not support DAX Turning DAX off.");
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH 0/3] Add alignment check for DAX mount
2016-04-29 20:39 [PATCH 0/3] Add alignment check for DAX mount Toshi Kani
` (2 preceding siblings ...)
2016-04-29 20:39 ` [PATCH 3/3] xfs: " Toshi Kani
@ 2016-04-29 23:12 ` Ross Zwisler
3 siblings, 0 replies; 9+ messages in thread
From: Ross Zwisler @ 2016-04-29 23:12 UTC (permalink / raw)
To: Toshi Kani
Cc: dan.j.williams, david, jack, tytso, adilger.kernel, ross.zwisler,
linux-nvdimm, linux-fsdevel, linux-kernel
On Fri, Apr 29, 2016 at 02:39:32PM -0600, Toshi Kani wrote:
> When a partition is not aligned by 4KB, mount -o dax succeeds,
> but any read/write access to the filesystem fails, except for
> metadata update.
>
> Add alignment check to ext4, ext2, and xfs.
>
> ---
> Toshi Kani (3):
> 1/3 ext4: Add alignment check for DAX mount
> 2/3 ext2: Add alignment check for DAX mount
> 3/3 xfs: Add alignment check for DAX mount
>
> ---
> fs/ext2/super.c | 6 ++++++
> fs/ext4/super.c | 6 ++++++
> fs/xfs/xfs_super.c | 6 ++++++
> 3 files changed, 18 insertions(+)
For the series:
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/3] ext4: Add alignment check for DAX mount
2016-04-29 20:39 ` [PATCH 1/3] ext4: " Toshi Kani
@ 2016-05-01 9:35 ` Boaz Harrosh
2016-05-01 17:31 ` Christoph Hellwig
1 sibling, 0 replies; 9+ messages in thread
From: Boaz Harrosh @ 2016-05-01 9:35 UTC (permalink / raw)
To: Toshi Kani, dan.j.williams, david, jack
Cc: tytso, linux-nvdimm, linux-kernel, adilger.kernel, linux-fsdevel
On 04/29/2016 11:39 PM, Toshi Kani wrote:
> When a partition is not aligned by 4KB, mount -o dax succeeds,
> but any read/write access to the filesystem fails, except for
> metadata update.
>
> Add alignment check to ext4_fill_super() when -o dax is specified.
>
> Reported-by: Micah Parrish <micah.parrish@hpe.com>
> Signed-off-by: Toshi Kani <toshi.kani@hpe.com>
> Cc: "Theodore Ts'o" <tytso@mit.edu>
> Cc: Andreas Dilger <adilger.kernel@dilger.ca>
> Cc: Jan Kara <jack@suse.cz>
> Cc: Dan Williams <dan.j.williams@intel.com>
> Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
> ---
> fs/ext4/super.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> index 304c712..90a8670 100644
> --- a/fs/ext4/super.c
> +++ b/fs/ext4/super.c
> @@ -3421,6 +3421,12 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
> "error: unsupported blocksize for dax");
> goto failed_mount;
> }
> + if (sb->s_bdev->bd_part->start_sect % (PAGE_SIZE / 512) ||
> + sb->s_bdev->bd_part->nr_sects % (PAGE_SIZE / 512)) {
Can you please not do this like this? For me is a layering violation only
the device should know what are its limits.
I would prefer if you just try to bdev_direct_access() the 0 sector and if
it fails then fail here. This way you let the device decide what it needs
and if/how to support unaligned partitions.
(For example it could by shrinking its size)
Thanks
Boaz
> + ext4_msg(sb, KERN_ERR,
> + "error: unaligned partition for dax");
> + goto failed_mount;
> + }
> if (!sb->s_bdev->bd_disk->fops->direct_access) {
> ext4_msg(sb, KERN_ERR,
> "error: device does not support dax");
> _______________________________________________
> Linux-nvdimm mailing list
> Linux-nvdimm@lists.01.org
> https://lists.01.org/mailman/listinfo/linux-nvdimm
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/3] ext4: Add alignment check for DAX mount
2016-04-29 20:39 ` [PATCH 1/3] ext4: " Toshi Kani
2016-05-01 9:35 ` Boaz Harrosh
@ 2016-05-01 17:31 ` Christoph Hellwig
2016-05-02 9:49 ` Boaz Harrosh
1 sibling, 1 reply; 9+ messages in thread
From: Christoph Hellwig @ 2016-05-01 17:31 UTC (permalink / raw)
To: Toshi Kani
Cc: dan.j.williams, david, jack, tytso, adilger.kernel, ross.zwisler,
linux-nvdimm, linux-fsdevel, linux-kernel
On Fri, Apr 29, 2016 at 02:39:33PM -0600, Toshi Kani wrote:
> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> index 304c712..90a8670 100644
> --- a/fs/ext4/super.c
> +++ b/fs/ext4/super.c
> @@ -3421,6 +3421,12 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
> "error: unsupported blocksize for dax");
> goto failed_mount;
> }
> + if (sb->s_bdev->bd_part->start_sect % (PAGE_SIZE / 512) ||
> + sb->s_bdev->bd_part->nr_sects % (PAGE_SIZE / 512)) {
> + ext4_msg(sb, KERN_ERR,
> + "error: unaligned partition for dax");
> + goto failed_mount;
> + }
> if (!sb->s_bdev->bd_disk->fops->direct_access) {
> ext4_msg(sb, KERN_ERR,
> "error: device does not support dax");
Factor your new checks and the ->direct_access into a new helper. It
should take the block device as file systems might have multiple
underlying devices.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/3] ext4: Add alignment check for DAX mount
2016-05-01 17:31 ` Christoph Hellwig
@ 2016-05-02 9:49 ` Boaz Harrosh
2016-05-02 15:23 ` Toshi Kani
0 siblings, 1 reply; 9+ messages in thread
From: Boaz Harrosh @ 2016-05-02 9:49 UTC (permalink / raw)
To: Christoph Hellwig, Toshi Kani
Cc: dan.j.williams, david, jack, tytso, adilger.kernel, ross.zwisler,
linux-nvdimm, linux-fsdevel, linux-kernel
On 05/01/2016 08:31 PM, Christoph Hellwig wrote:
> On Fri, Apr 29, 2016 at 02:39:33PM -0600, Toshi Kani wrote:
>> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
>> index 304c712..90a8670 100644
>> --- a/fs/ext4/super.c
>> +++ b/fs/ext4/super.c
>> @@ -3421,6 +3421,12 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
>> "error: unsupported blocksize for dax");
>> goto failed_mount;
>> }
>> + if (sb->s_bdev->bd_part->start_sect % (PAGE_SIZE / 512) ||
>> + sb->s_bdev->bd_part->nr_sects % (PAGE_SIZE / 512)) {
>> + ext4_msg(sb, KERN_ERR,
>> + "error: unaligned partition for dax");
>> + goto failed_mount;
>> + }
>> if (!sb->s_bdev->bd_disk->fops->direct_access) {
>> ext4_msg(sb, KERN_ERR,
>> "error: device does not support dax");
>
> Factor your new checks and the ->direct_access into a new helper. It
> should take the block device as file systems might have multiple
> underlying devices.
This already exists as part of bdev_direct_access()
All the code needs to do is call bdev_direct_access() on sector ZERO and
see if it is successful. If the alignment is broken it will fail.
Infact the FS does not even need the second if (!...bd_disk->fops->direct_access)
check either because it is also done inside bdev_direct_access(). The only check
needed is calling bdev_direct_access() and the generic layer does all the
checks needed already.
Cheers
Boaz
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 1/3] ext4: Add alignment check for DAX mount
2016-05-02 9:49 ` Boaz Harrosh
@ 2016-05-02 15:23 ` Toshi Kani
0 siblings, 0 replies; 9+ messages in thread
From: Toshi Kani @ 2016-05-02 15:23 UTC (permalink / raw)
To: Boaz Harrosh, Christoph Hellwig
Cc: dan.j.williams, david, jack, tytso, adilger.kernel, ross.zwisler,
linux-nvdimm, linux-fsdevel, linux-kernel
On Mon, 2016-05-02 at 12:49 +0300, Boaz Harrosh wrote:
> On 05/01/2016 08:31 PM, Christoph Hellwig wrote:
> >
> > On Fri, Apr 29, 2016 at 02:39:33PM -0600, Toshi Kani wrote:
> > >
> > > diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> > > index 304c712..90a8670 100644
> > > --- a/fs/ext4/super.c
> > > +++ b/fs/ext4/super.c
> > > @@ -3421,6 +3421,12 @@ static int ext4_fill_super(struct super_block
> > > *sb, void *data, int silent)
> > > "error: unsupported
> > > blocksize for dax");
> > > goto failed_mount;
> > > }
> > > + if (sb->s_bdev->bd_part->start_sect % (PAGE_SIZE /
> > > 512) ||
> > > + sb->s_bdev->bd_part->nr_sects % (PAGE_SIZE /
> > > 512)) {
> > > + ext4_msg(sb, KERN_ERR,
> > > + "error: unaligned partition
> > > for dax");
> > > + goto failed_mount;
> > > + }
> > > if (!sb->s_bdev->bd_disk->fops->direct_access) {
> > > ext4_msg(sb, KERN_ERR,
> > > "error: device does not
> > > support dax");
> >
> > Factor your new checks and the ->direct_access into a new helper. It
> > should take the block device as file systems might have multiple
> > underlying devices.
>
> This already exists as part of bdev_direct_access()
>
> All the code needs to do is call bdev_direct_access() on sector ZERO and
> see if it is successful. If the alignment is broken it will fail.
>
> Infact the FS does not even need the second if (!...bd_disk->fops-
> >direct_access) check either because it is also done inside
> bdev_direct_access(). The only check needed is calling
> bdev_direct_access() and the generic layer does all the
> checks needed already.
Sounds good. I will call bdev_direct_access() for this check. I assume we
do not need to check the alignment of 'bd_part->nr_sects' since the rest of
unaligned sectors will be ignored with block size 4KB.
Thanks!
-Toshi
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2016-05-02 15:33 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-04-29 20:39 [PATCH 0/3] Add alignment check for DAX mount Toshi Kani
2016-04-29 20:39 ` [PATCH 1/3] ext4: " Toshi Kani
2016-05-01 9:35 ` Boaz Harrosh
2016-05-01 17:31 ` Christoph Hellwig
2016-05-02 9:49 ` Boaz Harrosh
2016-05-02 15:23 ` Toshi Kani
2016-04-29 20:39 ` [PATCH 2/3] ext2: " Toshi Kani
2016-04-29 20:39 ` [PATCH 3/3] xfs: " Toshi Kani
2016-04-29 23:12 ` [PATCH 0/3] " Ross Zwisler
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).