Re: [PATCH v3 0/3] __btrfs_drop_extents() BUG_ON reproducer

[David Disseldorp <ddiss@suse.de>]

Ping, any Btrfsers get a chance to look at this patch series?
I'd like to get it into the QA tree.

On Fri,  7 Feb 2014 11:35:38 +0100, David Disseldorp wrote:

> This patch-set provides a reproducer for hitting the 3.14.0-rc1 BUG_ON()
> at:
>  692 int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
> ...
>  839                 /*
>  840                  *  | ---- range to drop ----- |
>  841                  *      | -------- extent -------- |
>  842                  */
>  843                 if (start <= key.offset && end < extent_end) {
>  844                         BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE);
>  845 
>  846                         memcpy(&new_key, &key, sizeof(new_key));
> 
> The first patch adds a small cloner binary which is used by btrfs/035 to
> dispatch BTRFS_IOC_CLONE_RANGE requests.
> 
> This workload resembles that of Samba's vfs_btrfs module, when a Windows
> client restores a file from a shadow-copy (snapshot) using server-side
> copy requests.
> 
> Changes since V2:
> - Remove explicit write error checks
> 
> Changes since V1:
> - Use strtoull instead of atoi
> - Print error conditions in cloner
> - Check for cloner binary before running test
> - Continue test on failure
> - Add cloner to .gitignore
> 
> Feedback appreciated.
> 
> Cheers, David
> 
> 
>  .gitignore          |   1 +
>  configure.ac        |   1 +
>  src/Makefile        |   2 +-
>  src/cloner.c        | 192 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/btrfs/035     |  77 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/btrfs/035.out |   3 +++
>  tests/btrfs/group   |   1 +
>  7 files changed, 276 insertions(+), 1 deletion(-)