Re: [PATCH 4/6] btrfs-progs: mkfs.c overwrites fd without appropriate close

[Josef Bacik <jbacik@fusionio.com>]

On Wed, Aug 14, 2013 at 11:17:05AM +0800, Anand Jain wrote:
> 
> 
> On 08/14/2013 10:04 AM, Anand Jain wrote:
> >
> >
> >On 08/14/2013 03:14 AM, Josef Bacik wrote:
> >>On Fri, Jul 26, 2013 at 01:35:28AM +0800, Anand Jain wrote:
> >>>Signed-off-by: Anand Jain <anand.jain@oracle.com>
> >>>---
> >>>  mkfs.c |    3 ++-
> >>>  1 files changed, 2 insertions(+), 1 deletions(-)
> >>>
> >>>diff --git a/mkfs.c b/mkfs.c
> >>>index 60f906c..66f558a 100644
> >>>--- a/mkfs.c
> >>>+++ b/mkfs.c
> >>>@@ -1570,6 +1570,8 @@ int main(int ac, char **av)
> >>>           * occur by the following processing.
> >>>           * (btrfs_register_one_device() fails if O_EXCL is on)
> >>>           */
> >>>+        if (fd > 0)
> >>>+            close(fd);
> >>>          fd = open(file, O_RDWR);
> >>>          if (fd < 0) {
> >>>              fprintf(stderr, "unable to open %s: %s\n", file,
> >>>@@ -1581,7 +1583,6 @@ int main(int ac, char **av)
> >>>          if (ret) {
> >>>              fprintf(stderr, "skipping duplicate device %s in FS\n",
> >>>                  file);
> >>>-            close(fd);
> >>>              continue;
> >>>          }
> >>>          ret = btrfs_prepare_device(fd, file, zero_end,
> >>>&dev_block_count,
> >>
> >>This breaks mkfs with multiple disks.
> >
> >  I can't believe as I have been playing with multiple disks
> >  quite a lot recently. let me dig more.
> 
>  Sorry my mistake.
> 
>  Indeed further down btrfs_add_to_fsid() stores fd. closing a
>  stored fd is not correct theoretically.
> 
>  Josef, Would be keen to know which xfstest caught this. ?
> 

It was btrfs/265, the one that does raid tests and such.

Josef