On 6/25/14, 12:14 AM, Eric Sandeen wrote:

On 6/24/14, 9:22 PM, Gui Hecheng wrote:

quoted

quoted

On Tue, 2014-06-24 at 21:17 -0500, Eric Sandeen wrote:

quoted

quoted

On 6/11/14, 9:25 PM, Gui Hecheng wrote:

quoted

quoted

When run chunk-recover on a health btrfs(data profile raid0, with
plenty of data), the program has a chance to abort on the number
of mirrors of an extent.

According to the kernel code, the max mirror number of an extent
is 3 not 2:
	ctree.h: 		BTRFS_MAX_MIRRORS	3
	chunk-recover.c :	BTRFS_NUM_MIRRORS	2
just change BTRFS_NUM_MIRRORS to 3, and everything goes well.

Wouldn't it make a lot more sense, then, to change the userspace
macro to be called BTRFS_MAX_MIRRORS as well?

-Eric

Yes, Eric, unify the names between userspace and kernelspace is really a
good point. Also, I plan to move the macro into ctree.h, what do you
think?

It's only used in chunk-recover.c, so I don't see much point to moving it
to a new file.

Sorry, I take that back.  Actually -

Yes, I think it does make sense, just so that userspace moves slightly closer to
kernelspace.

-Eric (who said long ago that he wanted to try to sync things up, but
found himself daunted by the task, and failed)