Skip to contentSkip to conversation

R: Re: [PATCH v2 5/5] Btrfs: Add BTRFS_IOC_SUBVOL_GETFLAGS/SETFLAGS ioctl

DISCUSSION Goffredo Baroncelli <[email protected]> 1 participant 1 message Dec 10, 2010 Btrfs file system

lore
Share & tools
lore
#0 lore raw
Quoted 20 lines

----Messaggio originale----
Da: [email protected]
Data: 10/12/2010 8.12
A: <[email protected]>
Cc: <[email protected]>
Ogg: Re: [PATCH v2 5/5] Btrfs: Add BTRFS_IOC_SUBVOL_GETFLAGS/SETFLAGS ioctl

Quoted 10 lines · Goffredo Baroncelli

Hi Li,

Quoted 7 lines · Li Zefan

This allows us to set a snapshot or a subvolume readonly or writable on the fly.

Usage:

Set BTRFS_SUBVOL_RDONLY of btrfs_ioctl_vol_arg_v2->flags, and then call ioctl(BTRFS_IOCTL_SUBVOL_SETFLAGS);

[...]

Quoted 6 lines
Quoted 6 lines

Review

Quoted 3 lines
(quoted diff) +3 -0
+	/* nothing to do */
+	if (!!(flags & BTRFS_SUBVOL_RDONLY) == root->readonly)
+		goto out_unlock;

This is only an aesthetic comment: I prefer a simpler style like

Quoted 6 lines
Quoted 3 lines
	if ((flags & BTRFS_SUBVOL_RDONLY) && root->readonly)
		goto out_unlock;

They are not equivalent. The former checks if the flags and the root both have readonly bit set or cleared.

Right, my fault

Quoted 13 lines
Quoted 13 lines

But I know that every body has its style :-)

Review

Quoted 9 lines
(quoted diff) +9 -0
+
+	root_flags = btrfs_root_flags(&root->root_item);
+	if (flags & BTRFS_SUBVOL_RDONLY)
+		btrfs_set_root_flags(&root->root_item,
+				     root_flags | BTRFS_ROOT_SNAP_RDONLY);
+	else
+		btrfs_set_root_flags(&root->root_item,
+				     root_flags & ~BTRFS_ROOT_SNAP_RDONLY);
+	root->readonly = !root->readonly;

I double checked this line. But if I read the code correctly I think that

the

Quoted 7 lines
Quoted 3 lines

line above is wrong: the field "root->readonly" is flipped regardless the value of the flags passed by the user.

Yep, that's because if we don't need to flip it, we've already exited early.

Note, there's only one flag.

Yes, it is true. However I strongly suggest to avoid that. If someone adds another flag this may miss...
Obviously if we get rid of the root->readonly we solve at the root the problem :)

Quoted 2 lines
Quoted 2 lines

Moreover I have another question: why internally the flags is BTRFS_ROOT_SNAP_RDONLY, instead in user space the flag is

BTRFS_SUBVOL_RDONLY

Quoted 9 lines
Quoted 2 lines

?

That's my carelessness..

Quoted 4 lines

I suggest to
- rename BTRFS_SUBVOL_RDONLY in BTRFS_SUBVOL_CREATE_SNAP_RDONLY (like BTRFS_SUBVOL_CREATE_SNAP_ASYNC)
- rename BTRFS_ROOT_SNAP_RDONLY in BTRFS_ROOT_FLAGS_RDONLY and use both

in

Quoted 1 line
Quoted 1 line

userspace and in kernel space this flag. I suggested to remove SNAP

because

Quoted 9 lines
Quoted 2 lines

the flag make sense also for a "standard" subvolume.

I'd prefer not to mix flags for root_item flags and vol ioctl flags.

And CREATE_RDONLY and CREATE_ASYNC can also be valid for subvolumes too. Support for async subvolume creation is already there, just lack of an user interface. So I've changed _CREATE_SNAP_ASYNC to _CREATE_ASYNC in the patch I sent just now.

I fully gree

Quoted 5 lines

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html