[PATCH] mkfs: don't try to detect filesystems on regular files via blkid

[Eric Sandeen]

Jim Meyering wrote:
> Eric Sandeen wrote:
>> https://bugzilla.redhat.com/show_bug.cgi?id=561870
>>
>> # dd if=/dev/zero of=k bs=1MB count=2 seek=20; mkfs.xfs k
>> # mkfs.xfs: probe of k failed, cannot detect existing filesystem.
>> # mkfs.xfs: Use the -f option to force overwrite.
>>
>> blkid fails to do a probe of a regular file.
>>
>> I wish blkid would cope with this, but for now it might
>> be better to just turn it off.
>>
>> Reported-by: Jim Meyering <meyering at redhat.com>
>> Signed-off-by: Eric Sandeen <sandeen at redhat.com>
>> ---
>>
>> diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
>> index 9baf116..de87647 100644
>> --- a/mkfs/xfs_mkfs.c
>> +++ b/mkfs/xfs_mkfs.c
>> @@ -300,10 +300,15 @@ check_overwrite(
>>  	int		fd;
>>  	long long	size;
>>  	int		bsz;
>> +	struct stat	statbuf;
>>
>>  	if (!device || !*device)
>>  		return 0;
>>
>> +	/* blkid can't get info from a regular file */
>> +	if (!stat(device, &statbuf) && S_ISREG(statbuf.st_mode))
>> +		return 0;
>> +
>>  	ret = -1; /* will reset on success of all setup calls */
>>
>>  	fd = open(device, O_RDONLY);
> 
> Hi Eric,
> 
> Did you consider calling fstat after opening,
> rather than "stat" before?
> 
>         /* blkid can't get info from a regular file */
>         if (!fstat(fd, &statbuf) && S_ISREG(statbuf.st_mode)) {
>                 close(fd);
>                 return 0;
>         }
> 
> That's slightly more efficient, and not prone to confusion
> in the unlikely event that "device" changes inode between
> the stat and the open calls.

Ok, yes that's probably better, will resend.  thanks.

-Eric