Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   =?UTF-8?q?Lu=C3=ADs=20Henriques?= <lhenriques@suse.de>
To:     Jeff Layton <jlayton@kernel.org>,
        Ilya Dryomov <idryomov@gmail.com>, Xiubo Li <xiubli@redhat.com>
Cc:     Patrick Donnelly <pdonnell@redhat.com>, ceph-devel@vger.kernel.org,
        linux-kernel@vger.kernel.org,
        =?UTF-8?q?Lu=C3=ADs=20Henriques?= <lhenriques@suse.de>
Subject: [PATCH 0/2] ceph: metrics for remote object copies
Date:   Wed,  3 Nov 2021 11:24:03 +0000
Message-Id: <20211103112405.8733-1-lhenriques@suse.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

Following this email, I'm sending two patches that add support for a an
extra metric in the cephfs metrics infrastructure.

The 1st patch modifies libceph so that ceph_osdc_copy_from() returns an
OSD request and makes it responsibility of the caller to do the wait (and
release the request).  This is required so that the callers (currently
only the copy_file_range() syscall code) can access the request latency
timestamps.

The 2nd patch effectively adds support for the 'copyfrom' metrics.

Luís Henriques (2):
  libceph: have ceph_osdc_copy_from() return the osd request
  ceph: add a new metric to keep track of remote object copies

 fs/ceph/debugfs.c               |  3 ++-
 fs/ceph/file.c                  | 13 ++++++++++++-
 fs/ceph/metric.h                |  8 ++++++++
 include/linux/ceph/osd_client.h | 21 +++++++++++----------
 net/ceph/osd_client.c           | 27 ++++++++++++++-------------
 5 files changed, 47 insertions(+), 25 deletions(-)