[PATCH] Btrfs: do not mount when we have a sectorsize unequal to PAGE_SIZE

[PATCH] Btrfs: do not mount when we have a sectorsize unequal to PAGE_SIZE

[Liu Bo]

Our code is not ready to cope with a sectorsize that's not equal to PAGE_SIZE.
It will lead to hanging-on while writing something.

Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
---
 fs/btrfs/disk-io.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 20196f4..b9866f2 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -2254,9 +2254,9 @@ int open_ctree(struct super_block *sb,
 		goto fail_sb_buffer;
 	}
 
-	if (sectorsize < PAGE_SIZE) {
-		printk(KERN_WARNING "btrfs: Incompatible sector size "
-		       "found on %s\n", sb->s_id);
+	if (sectorsize != PAGE_SIZE) {
+		printk(KERN_WARNING "btrfs: Incompatible sector size(%lu) "
+		       "found on %s\n", (unsigned long)sectorsize, sb->s_id);
 		goto fail_sb_buffer;
 	}
 
-- 
1.6.5.2

Re: [PATCH] Btrfs: do not mount when we have a sectorsize unequal to PAGE_SIZE

[Karel Zak]

On Tue, Apr 03, 2012 at 09:56:53AM +0800, Liu Bo wrote:
> Our code is not ready to cope with a sectorsize that's not equal to PAGE_SIZE.
> It will lead to hanging-on while writing something.
> 
> Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
> ---
>  fs/btrfs/disk-io.c |    6 +++---
>  1 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
> index 20196f4..b9866f2 100644
> --- a/fs/btrfs/disk-io.c
> +++ b/fs/btrfs/disk-io.c
> @@ -2254,9 +2254,9 @@ int open_ctree(struct super_block *sb,
>  		goto fail_sb_buffer;
>  	}
>  
> -	if (sectorsize < PAGE_SIZE) {
> -		printk(KERN_WARNING "btrfs: Incompatible sector size "
> -		       "found on %s\n", sb->s_id);
> +	if (sectorsize != PAGE_SIZE) {
> +		printk(KERN_WARNING "btrfs: Incompatible sector size(%lu) "
> +		       "found on %s\n", (unsigned long)sectorsize, sb->s_id);

 That's strange. Does it mean that if I create the filesystem and then
 reboot to another kernel with different PAGE_SIZE then the filesystem
 is unaccessible for me?

    Karel

-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com

Re: [PATCH] Btrfs: do not mount when we have a sectorsize unequal to PAGE_SIZE

[Liu Bo]

On 04/03/2012 04:06 PM, Karel Zak wrote:
> On Tue, Apr 03, 2012 at 09:56:53AM +0800, Liu Bo wrote:
>> Our code is not ready to cope with a sectorsize that's not equal to PAGE_SIZE.
>> It will lead to hanging-on while writing something.
>>
>> Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
>> ---
>>  fs/btrfs/disk-io.c |    6 +++---
>>  1 files changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
>> index 20196f4..b9866f2 100644
>> --- a/fs/btrfs/disk-io.c
>> +++ b/fs/btrfs/disk-io.c
>> @@ -2254,9 +2254,9 @@ int open_ctree(struct super_block *sb,
>>  		goto fail_sb_buffer;
>>  	}
>>  
>> -	if (sectorsize < PAGE_SIZE) {
>> -		printk(KERN_WARNING "btrfs: Incompatible sector size "
>> -		       "found on %s\n", sb->s_id);
>> +	if (sectorsize != PAGE_SIZE) {
>> +		printk(KERN_WARNING "btrfs: Incompatible sector size(%lu) "
>> +		       "found on %s\n", (unsigned long)sectorsize, sb->s_id);
> 
>  That's strange. Does it mean that if I create the filesystem and then
>  reboot to another kernel with different PAGE_SIZE then the filesystem
>  is unaccessible for me?
> 
>     Karel
> 

I'm afraid yes until we can deal with sectorsize and PAGE_SIZE gracefully.

For now, I didn't test if a read will cause oops or not, but if it's ok, an alternative way is to make FS readonly.

thanks,
liubo

Re: [PATCH] Btrfs: do not mount when we have a sectorsize unequal to PAGE_SIZE

[David Sterba]

On Tue, Apr 03, 2012 at 04:35:28PM +0800, Liu Bo wrote:
> >  That's strange. Does it mean that if I create the filesystem and then
> >  reboot to another kernel with different PAGE_SIZE then the filesystem
> >  is unaccessible for me?
> 
> I'm afraid yes until we can deal with sectorsize and PAGE_SIZE
> gracefully.
> 
> For now, I didn't test if a read will cause oops or not, but if it's
> ok, an alternative way is to make FS readonly.

Writes will cause hang. I have tested that read-only mount of a 64k
sectorsize is ok, file reads do not hang, but any file larger than the
inline limit is empty due to failed csums.

Example read of

65483 ./drivers/scsi/qla2xxx/qla_isr.c

leads to

[ 9249.370133] btrfs csum failed ino 13837 off 0 csum 4169121208 private 3722900035

while reading

65471 ./arch/powerpc/platforms/cell/spufs/file.c

is just fine.

So for now even forcing a read-only mount is not good option, though we
may try to enhance the csum verification.


david