[Ocfs2-test-devel] [PATCH 4/9] Ocfs2-test: Add utility funcs for both single-node and multi-nodes tests.

[tristan.ye]

On Tue, 2009-03-24 at 10:51 +0800, Tao Ma wrote:
> Hi Tristan,
> 	Just some minor comments.
> 	I see a large part of this patch isn't concerned with reflink 
> operation. Just about read/write and mmap. 

Actually, there were no other generic funcs more than read/write/mmap
operations. Not sure if it deserves our efforts to build them to a lib
or not. maybe we can mainatain a C common lib like shell and python in
ocfs2-tests, but the situation was not like the ocfs2-tools somehow, our
current tests in ocfs2-tests seems a little bit independent and
separate.

> So is there any possibility 
> that we can create a lib for it. So that our future test can use it(just 
> like some general functions we create in libocfs2 of ocfs2-tools).
> 
> Tristan Ye wrote:
> > All common funcs shared by single-node and multi-nodes testing binaries
> > gathered here.
> > 
> > Signed-off-by: Tristan Ye <tristan.ye at oracle.com>
> <snip>
> > +extern int open_ocfs2_volume(char *device_name)
> > +{
> > +	int open_flags = OCFS2_FLAG_HEARTBEAT_DEV_OK | OCFS2_FLAG_RO;
> > +	int ret;
> > +	uint64_t superblock = 0, block_size = 0;
> > +
> > +	ret = ocfs2_open(device_name, open_flags, superblock, block_size,
> > +			 &fs);
> > +	if (ret < 0) {
> > +		fprintf(stderr, "Not a ocfs2 volume!\n");
> ocfs2_open's return value give the right answer. So use it please 
> because it may be caused other issues.

I'll take care of this.

> > +		return ret;
> > +	}
> > +
> > +	ocfs2_sb = OCFS2_RAW_SB(fs->fs_super);
> > +	if (!(ocfs2_sb->s_feature_incompat &
> > +	      OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE)) {
> > +		fprintf(stderr, "refcount is not supported"
> > +			" on this ocfs2 volume\n");
> > +		return -1;
> > +	}
> > +
> > +	blocksize = 1 << ocfs2_sb->s_blocksize_bits;
> > +	clustersize = 1 << ocfs2_sb->s_clustersize_bits;
> > +	max_inline_size = ocfs2_max_inline_data(blocksize);
> I never seen ocfs2_close in your patch. So it is in another file?

Good catch, thanks.

> > +
> > +	return 0;
> > +}
> 
> Regards,
> Tao