Message-ID: <20200220231000.GA11036@suse.com>
Date:   Thu, 20 Feb 2020 23:10:00 +0000
From:   Luis Henriques <lhenriques@...e.com>
To:     Jeff Layton <jlayton@...nel.org>
Cc:     Sage Weil <sage@...hat.com>, Ilya Dryomov <idryomov@...il.com>,
        "Yan, Zheng" <zyan@...hat.com>, ceph-devel@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] ceph: re-org copy_file_range and fix error handling paths

On Thu, Feb 20, 2020 at 05:54:03PM -0500, Jeff Layton wrote:
> On Thu, 2020-02-20 at 22:36 +0000, Luis Henriques wrote:
> > On Thu, Feb 20, 2020 at 03:41:14PM -0500, Jeff Layton wrote:
> > > On Mon, 2020-02-17 at 12:36 +0000, Luis Henriques wrote:
> > > > This patch re-organizes copy_file_range, trying to fix a few issues in
> > > > error handling.  Here's the summary:
> > > > 
> > > > - Abort copy if initial do_splice_direct() returns fewer bytes than
> > > >   requested.
> > > > 
> > > > - Move the 'size' initialization (with i_size_read()) further down in the
> > > >   code, after the initial call to do_splice_direct().  This avoids issues
> > > >   with a possibly stale value if a manual copy is done.
> > > > 
> > > > - Move the object copy loop into a separate function.  This makes it
> > > >   easier to handle errors (e.g, dirtying caps and updating the MDS
> > > >   metadata if only some objects have been copied before an error has
> > > >   occurred).
> > > > 
> > > > - Added calls to ceph_oloc_destroy() to avoid leaking memory with src_oloc
> > > >   and dst_oloc
> > > > 
> > > > - After the object copy loop, the new file size to be reported to the MDS
> > > >   (if there's file size change) is now the actual file size, and not the
> > > >   size after an eventual extra manual copy.
> > > > 
> > > > - Added a few dout() to show the number of bytes copied in the two manual
> > > >   copies and in the object copy loop.
> > > > 
> > > > Signed-off-by: Luis Henriques <lhenriques@...e.com>
> > > > ---
> > > > Hi!
> > > > 
> > > > Initially I was going to have this patch split in a series, but then I
> > > > decided not to do that as this big patch allows (IMO) to better see the
> > > > whole picture.  But please let me know if you think otherwise and I can
> > > > split it in a few smaller patches.
> > > > 
> > > > I tried to cover all the issues that have been pointed out by Ilya, but I
> > > > may have missed something or, more likely, introduced new bugs ;-)
> > > > 
> > > > Cheers,
> > > > --
> > > > Luis
> > > > 
> > > 
> > > Sorry for the delay in review!
> > 
> > No worries, I appreciate the feedback but I obviously don't expect it to
> > happen immediately :-)
> > 
> > > >  fs/ceph/file.c | 169 ++++++++++++++++++++++++++++---------------------
> > > >  1 file changed, 96 insertions(+), 73 deletions(-)
> > > > 
> > > > diff --git a/fs/ceph/file.c b/fs/ceph/file.c
> > > > index c3b8e8e0bf17..4d90a275f9a5 100644
> > > > --- a/fs/ceph/file.c
> > > > +++ b/fs/ceph/file.c
> > > > @@ -1931,6 +1931,71 @@ static int is_file_size_ok(struct inode *src_inode, struct inode *dst_inode,
> > > >  	return 0;
> > > >  }
> > > >  
> > > > +static ssize_t ceph_do_objects_copy(struct ceph_inode_info *src_ci, u64 *src_off,
> > > > +				    struct ceph_inode_info *dst_ci, u64 *dst_off,
> > > > +				    struct ceph_fs_client *fsc,
> > > > +				    size_t len, unsigned int flags)
> > > > +{
> > > > +	struct ceph_object_locator src_oloc, dst_oloc;
> > > > +	struct ceph_object_id src_oid, dst_oid;
> > > > +	size_t bytes = 0;
> > > > +	u64 src_objnum, src_objoff, dst_objnum, dst_objoff;
> > > > +	u32 src_objlen, dst_objlen;
> > > > +	u32 object_size = src_ci->i_layout.object_size;
> > > > +	int ret;
> > > > +
> > > > +	src_oloc.pool = src_ci->i_layout.pool_id;
> > > > +	src_oloc.pool_ns = ceph_try_get_string(src_ci->i_layout.pool_ns);
> > > > +	dst_oloc.pool = dst_ci->i_layout.pool_id;
> > > > +	dst_oloc.pool_ns = ceph_try_get_string(dst_ci->i_layout.pool_ns);
> > > > +
> > > > +	while (len >= object_size) {
> > > > +		ceph_calc_file_object_mapping(&src_ci->i_layout, *src_off,
> > > > +					      object_size, &src_objnum,
> > > > +					      &src_objoff, &src_objlen);
> > > > +		ceph_calc_file_object_mapping(&dst_ci->i_layout, *dst_off,
> > > > +					      object_size, &dst_objnum,
> > > > +					      &dst_objoff, &dst_objlen);
> > > > +		ceph_oid_init(&src_oid);
> > > > +		ceph_oid_printf(&src_oid, "%llx.%08llx",
> > > > +				src_ci->i_vino.ino, src_objnum);
> > > > +		ceph_oid_init(&dst_oid);
> > > > +		ceph_oid_printf(&dst_oid, "%llx.%08llx",
> > > > +				dst_ci->i_vino.ino, dst_objnum);
> > > > +		/* Do an object remote copy */
> > > > +		ret = ceph_osdc_copy_from(&fsc->client->osdc,
> > > > +					  src_ci->i_vino.snap, 0,
> > > > +					  &src_oid, &src_oloc,
> > > > +					  CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |
> > > > +					  CEPH_OSD_OP_FLAG_FADVISE_NOCACHE,
> > > > +					  &dst_oid, &dst_oloc,
> > > > +					  CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |
> > > > +					  CEPH_OSD_OP_FLAG_FADVISE_DONTNEED,
> > > > +					  dst_ci->i_truncate_seq,
> > > > +					  dst_ci->i_truncate_size,
> > > > +					  CEPH_OSD_COPY_FROM_FLAG_TRUNCATE_SEQ);
> > > > +		if (ret) {
> > > > +			if (ret == -EOPNOTSUPP) {
> > > > +				fsc->have_copy_from2 = false;
> > > > +				pr_notice("OSDs don't support copy-from2; disabling copy offload\n");
> > > > +			}
> > > > +			dout("ceph_osdc_copy_from returned %d\n", ret);
> > > > +			if (!bytes)
> > > > +				bytes = ret;
> > > > +			goto out;
> > > > +		}
> > > > +		len -= object_size;
> > > > +		bytes += object_size;
> > > > +		*src_off += object_size;
> > > > +		*dst_off += object_size;
> > > > +	}
> > > > +
> > > > +out:
> > > > +	ceph_oloc_destroy(&src_oloc);
> > > > +	ceph_oloc_destroy(&dst_oloc);
> > > > +	return bytes;
> > > > +}
> > > > +
> > > >  static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> > > >  				      struct file *dst_file, loff_t dst_off,
> > > >  				      size_t len, unsigned int flags)
> > > > @@ -1941,14 +2006,11 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> > > >  	struct ceph_inode_info *dst_ci = ceph_inode(dst_inode);
> > > >  	struct ceph_cap_flush *prealloc_cf;
> > > >  	struct ceph_fs_client *src_fsc = ceph_inode_to_client(src_inode);
> > > > -	struct ceph_object_locator src_oloc, dst_oloc;
> > > > -	struct ceph_object_id src_oid, dst_oid;
> > > > -	loff_t endoff = 0, size;
> > > > -	ssize_t ret = -EIO;
> > > > +	loff_t size;
> > > > +	ssize_t ret = -EIO, bytes;
> > > >  	u64 src_objnum, dst_objnum, src_objoff, dst_objoff;
> > > > -	u32 src_objlen, dst_objlen, object_size;
> > > > +	u32 src_objlen, dst_objlen;
> > > >  	int src_got = 0, dst_got = 0, err, dirty;
> > > > -	bool do_final_copy = false;
> > > >  
> > > >  	if (src_inode->i_sb != dst_inode->i_sb) {
> > > >  		struct ceph_fs_client *dst_fsc = ceph_inode_to_client(dst_inode);
> > > > @@ -2026,22 +2088,14 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> > > >  	if (ret < 0)
> > > >  		goto out_caps;
> > > >  
> > > > -	size = i_size_read(dst_inode);
> > > > -	endoff = dst_off + len;
> > > > -
> > > >  	/* Drop dst file cached pages */
> > > >  	ret = invalidate_inode_pages2_range(dst_inode->i_mapping,
> > > >  					    dst_off >> PAGE_SHIFT,
> > > > -					    endoff >> PAGE_SHIFT);
> > > > +					    (dst_off + len) >> PAGE_SHIFT);
> > > >  	if (ret < 0) {
> > > >  		dout("Failed to invalidate inode pages (%zd)\n", ret);
> > > >  		ret = 0; /* XXX */
> > > >  	}
> > > > -	src_oloc.pool = src_ci->i_layout.pool_id;
> > > > -	src_oloc.pool_ns = ceph_try_get_string(src_ci->i_layout.pool_ns);
> > > > -	dst_oloc.pool = dst_ci->i_layout.pool_id;
> > > > -	dst_oloc.pool_ns = ceph_try_get_string(dst_ci->i_layout.pool_ns);
> > > > -
> > > >  	ceph_calc_file_object_mapping(&src_ci->i_layout, src_off,
> > > >  				      src_ci->i_layout.object_size,
> > > >  				      &src_objnum, &src_objoff, &src_objlen);
> > > > @@ -2060,6 +2114,8 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> > > >  	 * starting at the src_off
> > > >  	 */
> > > >  	if (src_objoff) {
> > > > +		dout("Initial partial copy of %u bytes\n", src_objlen);
> > > > +
> > > >  		/*
> > > >  		 * we need to temporarily drop all caps as we'll be calling
> > > >  		 * {read,write}_iter, which will get caps again.
> > > > @@ -2067,8 +2123,9 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> > > >  		put_rd_wr_caps(src_ci, src_got, dst_ci, dst_got);
> > > >  		ret = do_splice_direct(src_file, &src_off, dst_file,
> > > >  				       &dst_off, src_objlen, flags);
> > > > -		if (ret < 0) {
> > > > -			dout("do_splice_direct returned %d\n", err);
> > > > +		/* Abort on short copies or on error */
> > > > +		if (ret < src_objlen) {
> > > > +			dout("Failed partial copy (%zd)\n", ret);
> > > >  			goto out;
> > > >  		}
> > > >  		len -= ret;
> > > > @@ -2081,62 +2138,29 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> > > >  		if (err < 0)
> > > >  			goto out_caps;
> > > >  	}
> > > > -	object_size = src_ci->i_layout.object_size;
> > > > -	while (len >= object_size) {
> > > > -		ceph_calc_file_object_mapping(&src_ci->i_layout, src_off,
> > > > -					      object_size, &src_objnum,
> > > > -					      &src_objoff, &src_objlen);
> > > > -		ceph_calc_file_object_mapping(&dst_ci->i_layout, dst_off,
> > > > -					      object_size, &dst_objnum,
> > > > -					      &dst_objoff, &dst_objlen);
> > > > -		ceph_oid_init(&src_oid);
> > > > -		ceph_oid_printf(&src_oid, "%llx.%08llx",
> > > > -				src_ci->i_vino.ino, src_objnum);
> > > > -		ceph_oid_init(&dst_oid);
> > > > -		ceph_oid_printf(&dst_oid, "%llx.%08llx",
> > > > -				dst_ci->i_vino.ino, dst_objnum);
> > > > -		/* Do an object remote copy */
> > > > -		err = ceph_osdc_copy_from(
> > > > -			&src_fsc->client->osdc,
> > > > -			src_ci->i_vino.snap, 0,
> > > > -			&src_oid, &src_oloc,
> > > > -			CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |
> > > > -			CEPH_OSD_OP_FLAG_FADVISE_NOCACHE,
> > > > -			&dst_oid, &dst_oloc,
> > > > -			CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |
> > > > -			CEPH_OSD_OP_FLAG_FADVISE_DONTNEED,
> > > > -			dst_ci->i_truncate_seq, dst_ci->i_truncate_size,
> > > > -			CEPH_OSD_COPY_FROM_FLAG_TRUNCATE_SEQ);
> > > > -		if (err) {
> > > > -			if (err == -EOPNOTSUPP) {
> > > > -				src_fsc->have_copy_from2 = false;
> > > > -				pr_notice("OSDs don't support copy-from2; disabling copy offload\n");
> > > > -			}
> > > > -			dout("ceph_osdc_copy_from returned %d\n", err);
> > > > -			if (!ret)
> > > > -				ret = err;
> > > > -			goto out_caps;
> > > > -		}
> > > > -		len -= object_size;
> > > > -		src_off += object_size;
> > > > -		dst_off += object_size;
> > > > -		ret += object_size;
> > > > -	}
> > > >  
> > > > -	if (len)
> > > > -		/* We still need one final local copy */
> > > > -		do_final_copy = true;
> > > > +	size = i_size_read(dst_inode);
> > > > +	bytes = ceph_do_objects_copy(src_ci, &src_off, dst_ci, &dst_off,
> > > > +				     src_fsc, len, flags);
> > > > +	if (bytes <= 0) {
> > > > +		if (!ret)
> > > > +			ret = bytes;
> > > > +		goto out_caps;
> > > > +	}
> > > 
> > > Suppose we did the front part with do_splice_direct (ret > 0), but then
> > > ceph_do_objects_copy fails (bytes < 0). We "goto out_caps" but then...
> > > 
> > > [...]
> > > 
> > > > @@ -2152,15 +2176,14 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> > > >  out_caps:
> > > >  	put_rd_wr_caps(src_ci, src_got, dst_ci, dst_got);
> > > >  
> > > > -	if (do_final_copy) {
> > > > -		err = do_splice_direct(src_file, &src_off, dst_file,
> > > > -				       &dst_off, len, flags);
> > > > -		if (err < 0) {
> > > > -			dout("do_splice_direct returned %d\n", err);
> > > > -			goto out;
> > > > -		}
> > > > -		len -= err;
> > > > -		ret += err;
> > > > +	if (len && (ret >= 0)) {
> > > 
> > > ...len is still positive and we do_splice_direct again (probably at the
> > > wrong offset?), instead of just returning a short copy. I think we
> > > probably want to just stop any copying if it fails at any point along
> > > the way, right?
> > 
> > Well... To be honest I deliberately wanted to try a do_splice_direct in
> > case the remote object copies fail (basically, reverting to something
> > similar to the generic_copy_file_range).  Now, I've been staring at this
> > code for some time and I may be missing the obvious (again!).  But:
> > 
> >  - the offsets should be OK because ceph_do_objects_copy() only updates
> >    them after each successful object copy
> > 
> >  - len should also be consistent:
> >    * If 'bytes' was <= 0, it should contain what was written by
> >      do_splice_direct
> >    * if 'bytes' was > 0, but possibly < expected (e.g. an OSD returned an
> >      error after a few object copies), len should still be consistent
> > 
> > Anyway, I'm not too attached to this approach, and if you rather have this
> > function to return in the scenario you've described (and eventually have
> > the user to retry the operation) I'm OK with that.
> > 
> 
> Yes, sorry. I wasn't disputing whether this would fall over, but whether
> it was intended behavior.
> 
> I'm not sure we gain anything by doing a second splice once we've had a
> failure though. I think if this isn't going to (largely) use copy
> offloading then we probably ought to stop and just return to userland as
> quickly as we can.

Sure, makes sense.  I'll work on v2 to change that behaviour.  (Probably I
won't be able to send it tomorrow, as I'll be traveling and won't be able
to test it.)

Cheers,
--
Luís


> 
> > > > +		dout("Final partial copy of %zu bytes\n", len);
> > > > +		bytes = do_splice_direct(src_file, &src_off, dst_file,
> > > > +					 &dst_off, len, flags);
> > > > +		if (bytes > 0)
> > > > +			ret += bytes;
> > > > +		else
> > > > +			dout("Failed partial copy (%zd)\n", bytes);
> > > >  	}
> > > >  
> > > >  out:
> > > 
> > > -- 
> > > Jeff Layton <jlayton@...nel.org>
> > > 
> 
> -- 
> Jeff Layton <jlayton@...nel.org>
> 

Powered by blists - more mailing lists