On Sat, Nov 16, 2019 at 05:23:28PM +0100, Ilya Dryomov wrote:
On Sat, Nov 16, 2019 at 4:43 PM Sasha Levin <sashal@xxxxxxxxxx> wrote:From: Ilya Dryomov <idryomov@xxxxxxxxx> [ Upstream commit 894868330a1e038ea4a65dbb81741eef70ad71b1 ] Because send_mds_reconnect() wants to send a message with a pagelist and pass the ownership to the messenger, ceph_msg_data_add_pagelist() consumes a ref which is then put in ceph_msg_data_destroy(). This makes managing pagelists in the OSD client (where they are wrapped in ceph_osd_data) unnecessarily hard because the handoff only happens in ceph_osdc_start_request() instead of when the pagelist is passed to ceph_osd_data_pagelist_init(). I counted several memory leaks on various error paths. Fix up ceph_msg_data_add_pagelist() and carry a pagelist ref in ceph_osd_data. Signed-off-by: Ilya Dryomov <idryomov@xxxxxxxxx> Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx> --- fs/ceph/mds_client.c | 2 +- net/ceph/messenger.c | 1 + net/ceph/osd_client.c | 8 ++++++++ 3 files changed, 10 insertions(+), 1 deletion(-) diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 09db6d08614d2..94494d05a94cb 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -2184,7 +2184,6 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc, if (req->r_pagelist) { struct ceph_pagelist *pagelist = req->r_pagelist; - refcount_inc(&pagelist->refcnt); ceph_msg_data_add_pagelist(msg, pagelist); msg->hdr.data_len = cpu_to_le32(pagelist->length); } else { @@ -3289,6 +3288,7 @@ static void send_mds_reconnect(struct ceph_mds_client *mdsc, mutex_unlock(&mdsc->mutex); up_read(&mdsc->snap_rwsem); + ceph_pagelist_release(pagelist); return; fail: diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c index f7d7f32ac673c..2c8cd339d59ea 100644 --- a/net/ceph/messenger.c +++ b/net/ceph/messenger.c @@ -3323,6 +3323,7 @@ void ceph_msg_data_add_pagelist(struct ceph_msg *msg, data = ceph_msg_data_create(CEPH_MSG_DATA_PAGELIST); BUG_ON(!data); + refcount_inc(&pagelist->refcnt); data->pagelist = pagelist; list_add_tail(&data->links, &msg->data); diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c index 76c41a84550e7..c3494c1fb3a9a 100644 --- a/net/ceph/osd_client.c +++ b/net/ceph/osd_client.c @@ -126,6 +126,9 @@ static void ceph_osd_data_init(struct ceph_osd_data *osd_data) osd_data->type = CEPH_OSD_DATA_TYPE_NONE; } +/* + * Consumes @pages if @own_pages is true. + */ static void ceph_osd_data_pages_init(struct ceph_osd_data *osd_data, struct page **pages, u64 length, u32 alignment, bool pages_from_pool, bool own_pages) @@ -138,6 +141,9 @@ static void ceph_osd_data_pages_init(struct ceph_osd_data *osd_data, osd_data->own_pages = own_pages; } +/* + * Consumes a ref on @pagelist. + */ static void ceph_osd_data_pagelist_init(struct ceph_osd_data *osd_data, struct ceph_pagelist *pagelist) { @@ -362,6 +368,8 @@ static void ceph_osd_data_release(struct ceph_osd_data *osd_data) num_pages = calc_pages_for((u64)osd_data->alignment, (u64)osd_data->length); ceph_release_page_vector(osd_data->pages, num_pages); + } else if (osd_data->type == CEPH_OSD_DATA_TYPE_PAGELIST) { + ceph_pagelist_release(osd_data->pagelist); } ceph_osd_data_init(osd_data); }Hi Sasha, This commit was part of a larger series and shouldn't be backported on its own. Please drop it.
Dropped, thanks! -- Thanks, Sasha
