On 6/19/15 8:20 AM, Brian Foster wrote:
> On Thu, Jun 18, 2015 at 04:29:33PM -0500, Eric Sandeen wrote:
>> If we specify "-f" to xfs_repair, it recognizes that it's working
>> on a file, and if the underlying filesystem sector size differs
>> such that direct IO won't work, it disables direct IO.
>>
>> It's odd, though, that we'd need to specify this, and the failure
>> is non-obvious:
>>
>> # xfs_repair /mnt/test/foo.img
>> Phase 1 - find and verify superblock...
>> xfs_repair: read failed: Invalid argument
>>
>> I see no advantage to requirin the administrator to jump through
>> this hoop; why not just detect that it's a file, and move on?
>>
>> Signed-off-by: Eric Sandeen <sandeen@xxxxxxxxxx>
>> ---
>>
>> diff --git a/repair/xfs_repair.c b/repair/xfs_repair.c
>> index 834697a..2d376be 100644
>> --- a/repair/xfs_repair.c
>> +++ b/repair/xfs_repair.c
>> @@ -573,6 +573,18 @@ main(int argc, char **argv)
>> exit(1);
>> }
>>
>> + /* -f forces this, but let's be nice and autodetect it, as well. */
>> + if (!isa_file) {
>> + int fd = libxfs_device_to_fd(x.ddev);
>> + struct stat64 statbuf;
>> +
>> + if (fstat64(fd, &statbuf) < 0)
>> + do_warn(_("%s: couldn't stat \"%s\"\n"),
>> + progname, fs_name);
>> + if (S_ISREG(statbuf.st_mode))
>> + isa_file = 1;
>
> We probably shouldn't query the statbuf if the stat call failed (who
> knows what's in there). Otherwise this seems like a good change to me.
yeargh, right, last-minute decision to not fail on the failed stat.
Sigh. V2 coming.
(I think Jan is working on a more wholesale cleanup of this crap, but in the
meantime I think maybe a couple targeted fixes to get things going are probably
ok).
-Eric
> Brian
>
>> + }
>> +
>> /*
>> * if the sector size of the filesystem we are trying to repair is
>> * smaller than that of the underlying filesystem (i.e. we are repairing
>>
>> _______________________________________________
>> xfs mailing list
>> xfs@xxxxxxxxxxx
>> http://oss.sgi.com/mailman/listinfo/xfs
|