From mboxrd@z Thu Jan 1 00:00:00 1970 From: Josh Durgin Subject: Re: [PATCH] rbd: support reading parent page data for writes Date: Sat, 11 May 2013 12:27:00 -0700 Message-ID: <518E9B84.20100@inktank.com> References: <518E81E1.2000806@inktank.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from mail-pa0-f42.google.com ([209.85.220.42]:38906 "EHLO mail-pa0-f42.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753276Ab3EKT05 (ORCPT ); Sat, 11 May 2013 15:26:57 -0400 Received: by mail-pa0-f42.google.com with SMTP id bj3so3675209pad.29 for ; Sat, 11 May 2013 12:26:57 -0700 (PDT) In-Reply-To: <518E81E1.2000806@inktank.com> Sender: ceph-devel-owner@vger.kernel.org List-ID: To: Alex Elder Cc: ceph-devel@vger.kernel.org On 05/11/2013 10:37 AM, Alex Elder wrote: > Currently, rbd_img_obj_parent_read_full() assumes the incoming > object request contains bio data. But if a layered image is part of > a multi-layer stack of images it will result in read requests of > page data to parent images. > > This is handling the same kind of issue as was resolved by this > commit: > 5b2ab72d rbd: support reading parent page data > > This resolves: > http://tracker.ceph.com/issues/5027 > > Signed-off-by: Alex Elder > --- > drivers/block/rbd.c | 21 ++++++++++++++------- > 1 file changed, 14 insertions(+), 7 deletions(-) > > diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c > index 51c45e7..b9221a0 100644 > --- a/drivers/block/rbd.c > +++ b/drivers/block/rbd.c > @@ -2160,7 +2160,8 @@ rbd_img_obj_parent_read_full_callback(struct > rbd_img_request *img_request) > struct page **pages; > int result; > u64 obj_size; > - u64 xferred; > + u64 offset; > + u64 length; > > rbd_assert(img_request_child_test(img_request)); > > @@ -2172,10 +2173,10 @@ rbd_img_obj_parent_read_full_callback(struct > rbd_img_request *img_request) > > orig_request = img_request->obj_request; > rbd_assert(orig_request != NULL); > - rbd_assert(orig_request->type == OBJ_REQUEST_BIO); > + rbd_assert(obj_request_type_valid(orig_request->type)); > result = img_request->result; > obj_size = img_request->length; Naming this obj_size is a bit confusing now, since a parent read that would cross the overlap point will be shortened to less than an object size. The rest looks good. Reviewed-by: Josh Durgin > - xferred = img_request->xferred; > + rbd_assert(obj_size == img_request->xferred); > rbd_img_request_put(img_request); > > rbd_assert(orig_request->img_request); > @@ -2204,11 +2205,17 @@ rbd_img_obj_parent_read_full_callback(struct > rbd_img_request *img_request) > > /* Then the original write request op */ > > + offset = orig_request->offset; > + length = orig_request->length; > osd_req_op_extent_init(osd_req, 1, CEPH_OSD_OP_WRITE, > - orig_request->offset, > - orig_request->length, 0, 0); > - osd_req_op_extent_osd_data_bio(osd_req, 1, orig_request->bio_list, > - orig_request->length); > + offset, length, 0, 0); > + if (orig_request->type == OBJ_REQUEST_BIO) > + osd_req_op_extent_osd_data_bio(osd_req, 1, > + orig_request->bio_list, length); > + else > + osd_req_op_extent_osd_data_pages(osd_req, 1, > + orig_request->pages, length, > + offset & ~PAGE_MASK, false, false); > > rbd_osd_req_format_write(orig_request); >