[v2] btrfs: fix a bogus warning when converting only data or metadata
diff mbox

Message ID 20170307223444.7526-1-kilobyte@angband.pl
State New
Headers show

Commit Message

Adam Borowski March 7, 2017, 10:34 p.m. UTC
If your filesystem has, eg, data:raid0 metadata:raid1, and you run "btrfs
balance -dconvert=raid1", the meta.target field will be uninitialized.
That's otherwise ok, as it's unused except for this warning.

Thus, let's use the existing set of raid levels for the comparison.

As a side effect, non-convert balances will now nag about data>metadata.

Signed-off-by: Adam Borowski <kilobyte@angband.pl>
---
To reproduce:
dd if=/dev/zero bs=1048576 count=1 seek=4095 of=ra
dd if=/dev/zero bs=1048576 count=1 seek=4095 of=rb
mkfs.btrfs ra rb     # defaults to -draid0 -mraid1
losetup -f ra
losetup -f rb
mount /dev/loop0 /mnt/vol1
btrfs balance start -dconvert=raid1 /mnt/vol1


On Tue, Mar 07, 2017 at 12:34:07PM -0800, Liu Bo wrote:
> This looks good, but this also brings another side effect, @bctl would
> also be kept in balance_item which will be used to resume balance in
> case of crash, so it may see a different bctl->meta.target.
>
> So would you please use local varibles for meta.target and data.target?

Okay.

I'm not sure why storing a bogus value that came from userspace and was
uninitialized there (0 in normal use) would be better, but here we go:
v2 doesn't overwrite what we got anymore.

Unrelated: I wonder if the profiles in the warning message shouldn't be
printk'ed as words (akin to ebce0e01), but we don't have a function to do
that, have we?


Meow!

 fs/btrfs/volumes.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

Comments

Liu Bo March 8, 2017, 1:27 a.m. UTC | #1
On Tue, Mar 07, 2017 at 11:34:44PM +0100, Adam Borowski wrote:
> If your filesystem has, eg, data:raid0 metadata:raid1, and you run "btrfs
> balance -dconvert=raid1", the meta.target field will be uninitialized.
> That's otherwise ok, as it's unused except for this warning.
> 
> Thus, let's use the existing set of raid levels for the comparison.
> 
> As a side effect, non-convert balances will now nag about data>metadata.
> 
> Signed-off-by: Adam Borowski <kilobyte@angband.pl>
> ---
> To reproduce:
> dd if=/dev/zero bs=1048576 count=1 seek=4095 of=ra
> dd if=/dev/zero bs=1048576 count=1 seek=4095 of=rb
> mkfs.btrfs ra rb     # defaults to -draid0 -mraid1
> losetup -f ra
> losetup -f rb
> mount /dev/loop0 /mnt/vol1
> btrfs balance start -dconvert=raid1 /mnt/vol1
> 
> 
> On Tue, Mar 07, 2017 at 12:34:07PM -0800, Liu Bo wrote:
> > This looks good, but this also brings another side effect, @bctl would
> > also be kept in balance_item which will be used to resume balance in
> > case of crash, so it may see a different bctl->meta.target.
> >
> > So would you please use local varibles for meta.target and data.target?
> 
> Okay.
> 
> I'm not sure why storing a bogus value that came from userspace and was
> uninitialized there (0 in normal use) would be better, but here we go:
> v2 doesn't overwrite what we got anymore.
> 
> Unrelated: I wonder if the profiles in the warning message shouldn't be
> printk'ed as words (akin to ebce0e01), but we don't have a function to do
> that, have we?
>

We don't have it in kernel, but in progs I believe.

For this patch,

Reviewed-by: Liu Bo <bo.li.liu@oracle.com>

Thanks,

-liubo


> 
> Meow!
> 
>  fs/btrfs/volumes.c | 12 +++++++++---
>  1 file changed, 9 insertions(+), 3 deletions(-)
> 
> diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
> index 3645af2749f8..987f395ddec5 100644
> --- a/fs/btrfs/volumes.c
> +++ b/fs/btrfs/volumes.c
> @@ -3750,6 +3750,7 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
>  		  struct btrfs_ioctl_balance_args *bargs)
>  {
>  	struct btrfs_fs_info *fs_info = bctl->fs_info;
> +	__u64 meta_target, data_target;
>  	u64 allowed;
>  	int mixed = 0;
>  	int ret;
> @@ -3846,11 +3847,16 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
>  		}
>  	} while (read_seqretry(&fs_info->profiles_lock, seq));
>  
> -	if (btrfs_get_num_tolerated_disk_barrier_failures(bctl->meta.target) <
> -		btrfs_get_num_tolerated_disk_barrier_failures(bctl->data.target)) {
> +	/* if we're not converting, the target field is uninitialized */
> +	meta_target = (bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) ?
> +		bctl->meta.target : fs_info->avail_metadata_alloc_bits;
> +	data_target = (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) ?
> +		bctl->data.target : fs_info->avail_data_alloc_bits;
> +	if (btrfs_get_num_tolerated_disk_barrier_failures(meta_target) <
> +		btrfs_get_num_tolerated_disk_barrier_failures(data_target)) {
>  		btrfs_warn(fs_info,
>  			   "metadata profile 0x%llx has lower redundancy than data profile 0x%llx",
> -			   bctl->meta.target, bctl->data.target);
> +			   meta_target, data_target);
>  	}
>  
>  	ret = insert_balance_item(fs_info, bctl);
> -- 
> 2.11.0
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Sterba March 9, 2017, 1:43 p.m. UTC | #2
On Tue, Mar 07, 2017 at 11:34:44PM +0100, Adam Borowski wrote:
> On Tue, Mar 07, 2017 at 12:34:07PM -0800, Liu Bo wrote:
> > This looks good, but this also brings another side effect, @bctl would
> > also be kept in balance_item which will be used to resume balance in
> > case of crash, so it may see a different bctl->meta.target.
> >
> > So would you please use local varibles for meta.target and data.target?
> 
> Okay.
> 
> I'm not sure why storing a bogus value that came from userspace and was
> uninitialized there (0 in normal use) would be better, but here we go:
> v2 doesn't overwrite what we got anymore.
> 
> Unrelated: I wonder if the profiles in the warning message shouldn't be
> printk'ed as words (akin to ebce0e01), but we don't have a function to do
> that, have we?

Printing human readable strings would be better, you can copy the
missing functions from progs as Lui pointed out.

> @@ -3750,6 +3750,7 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
>  		  struct btrfs_ioctl_balance_args *bargs)
>  {
>  	struct btrfs_fs_info *fs_info = bctl->fs_info;
> +	__u64 meta_target, data_target;

Changed to u64, the __u64 is for kernel/userspace interfaces.

>  	u64 allowed;
>  	int mixed = 0;
>  	int ret;
> @@ -3846,11 +3847,16 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
>  		}
>  	} while (read_seqretry(&fs_info->profiles_lock, seq));
>  
> -	if (btrfs_get_num_tolerated_disk_barrier_failures(bctl->meta.target) <
> -		btrfs_get_num_tolerated_disk_barrier_failures(bctl->data.target)) {
> +	/* if we're not converting, the target field is uninitialized */
> +	meta_target = (bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) ?
> +		bctl->meta.target : fs_info->avail_metadata_alloc_bits;
> +	data_target = (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) ?
> +		bctl->data.target : fs_info->avail_data_alloc_bits;
> +	if (btrfs_get_num_tolerated_disk_barrier_failures(meta_target) <
> +		btrfs_get_num_tolerated_disk_barrier_failures(data_target)) {
>  		btrfs_warn(fs_info,
>  			   "metadata profile 0x%llx has lower redundancy than data profile 0x%llx",
> -			   bctl->meta.target, bctl->data.target);
> +			   meta_target, data_target);
>  	}

The patch does not apply to current master, there's

  3861         if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
  3862                 fs_info->num_tolerated_disk_barrier_failures = min(
  3863                         btrfs_calc_num_tolerated_disk_barrier_failures(fs_info),
  3864                         btrfs_get_num_tolerated_disk_barrier_failures(
  3865                                 bctl->sys.target));
  3866         }
  3867

followed by the line below, I'll merge that manually but slightly wonder
which branch you have used as a base. No big deal though, I'll add the
patch to the queue. Thanks.

>  	ret = insert_balance_item(fs_info, bctl);
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Adam Borowski March 9, 2017, 9:56 p.m. UTC | #3
On Thu, Mar 09, 2017 at 02:43:27PM +0100, David Sterba wrote:
> The patch does not apply to current master, there's
> 
>   3861         if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
>   3862                 fs_info->num_tolerated_disk_barrier_failures = min(
>   3863                         btrfs_calc_num_tolerated_disk_barrier_failures(fs_info),
>   3864                         btrfs_get_num_tolerated_disk_barrier_failures(
>   3865                                 bctl->sys.target));
>   3866         }
>   3867
> 
> followed by the line below, I'll merge that manually but slightly wonder
> which branch you have used as a base. No big deal though, I'll add the
> patch to the queue. Thanks.
>
> >       ret = insert_balance_item(fs_info, bctl);

I had Qu's chunk counting patchset in my tree (and some irrelevant stuff,
like tty patches).  But this means there'll be a minor conflict between the
two once you get around to merging Qu's stuff.

It's obvious how they interact; the bogus warning happens with and without
chunk counting.

Patch
diff mbox

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 3645af2749f8..987f395ddec5 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3750,6 +3750,7 @@  int btrfs_balance(struct btrfs_balance_control *bctl,
 		  struct btrfs_ioctl_balance_args *bargs)
 {
 	struct btrfs_fs_info *fs_info = bctl->fs_info;
+	__u64 meta_target, data_target;
 	u64 allowed;
 	int mixed = 0;
 	int ret;
@@ -3846,11 +3847,16 @@  int btrfs_balance(struct btrfs_balance_control *bctl,
 		}
 	} while (read_seqretry(&fs_info->profiles_lock, seq));
 
-	if (btrfs_get_num_tolerated_disk_barrier_failures(bctl->meta.target) <
-		btrfs_get_num_tolerated_disk_barrier_failures(bctl->data.target)) {
+	/* if we're not converting, the target field is uninitialized */
+	meta_target = (bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) ?
+		bctl->meta.target : fs_info->avail_metadata_alloc_bits;
+	data_target = (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) ?
+		bctl->data.target : fs_info->avail_data_alloc_bits;
+	if (btrfs_get_num_tolerated_disk_barrier_failures(meta_target) <
+		btrfs_get_num_tolerated_disk_barrier_failures(data_target)) {
 		btrfs_warn(fs_info,
 			   "metadata profile 0x%llx has lower redundancy than data profile 0x%llx",
-			   bctl->meta.target, bctl->data.target);
+			   meta_target, data_target);
 	}
 
 	ret = insert_balance_item(fs_info, bctl);