From b372e96bd0a32729d55d27f613c8bc80708a82e1 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 25 Mar 2024 09:21:20 +1100 Subject: [PATCH 1/3] ceph: redirty page before returning AOP_WRITEPAGE_ACTIVATE The page has been marked clean before writepage is called. If we don't redirty it before postponing the write, it might never get written. Cc: stable@vger.kernel.org Fixes: 503d4fa6ee28 ("ceph: remove reliance on bdi congestion") Signed-off-by: NeilBrown Reviewed-by: Jeff Layton Reviewed-by: Xiubo Li Signed-off-by: Ilya Dryomov --- fs/ceph/addr.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index 1340d77124ae..ee9caf7916fb 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -795,8 +795,10 @@ static int ceph_writepage(struct page *page, struct writeback_control *wbc) ihold(inode); if (wbc->sync_mode == WB_SYNC_NONE && - ceph_inode_to_fs_client(inode)->write_congested) + ceph_inode_to_fs_client(inode)->write_congested) { + redirty_page_for_writepage(wbc, page); return AOP_WRITEPAGE_ACTIVATE; + } wait_on_page_fscache(page); From 17f8dc2db52185460f212052f3a692c1fdc167ba Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Tue, 9 Apr 2024 08:56:03 +0800 Subject: [PATCH 2/3] ceph: switch to use cap_delay_lock for the unlink delay list The same list item will be used in both cap_delay_list and cap_unlink_delay_list, so it's buggy to use two different locks to protect them. Cc: stable@vger.kernel.org Fixes: dbc347ef7f0c ("ceph: add ceph_cap_unlink_work to fire check_caps() immediately") Link: https://lists.ceph.io/hyperkitty/list/ceph-users@ceph.io/thread/AODC76VXRAMXKLFDCTK4TKFDDPWUSCN5 Reported-by: Marc Ruhmann Signed-off-by: Xiubo Li Reviewed-by: Ilya Dryomov Tested-by: Marc Ruhmann Signed-off-by: Ilya Dryomov --- fs/ceph/caps.c | 4 ++-- fs/ceph/mds_client.c | 9 ++++----- fs/ceph/mds_client.h | 3 +-- 3 files changed, 7 insertions(+), 9 deletions(-) diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 55051ad09c19..c4941ba245ac 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -4783,13 +4783,13 @@ int ceph_drop_caps_for_unlink(struct inode *inode) doutc(mdsc->fsc->client, "%p %llx.%llx\n", inode, ceph_vinop(inode)); - spin_lock(&mdsc->cap_unlink_delay_lock); + spin_lock(&mdsc->cap_delay_lock); ci->i_ceph_flags |= CEPH_I_FLUSH; if (!list_empty(&ci->i_cap_delay_list)) list_del_init(&ci->i_cap_delay_list); list_add_tail(&ci->i_cap_delay_list, &mdsc->cap_unlink_delay_list); - spin_unlock(&mdsc->cap_unlink_delay_lock); + spin_unlock(&mdsc->cap_delay_lock); /* * Fire the work immediately, because the MDS maybe diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 3ab9c268a8bb..360b686c3c67 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -2504,7 +2504,7 @@ static void ceph_cap_unlink_work(struct work_struct *work) struct ceph_client *cl = mdsc->fsc->client; doutc(cl, "begin\n"); - spin_lock(&mdsc->cap_unlink_delay_lock); + spin_lock(&mdsc->cap_delay_lock); while (!list_empty(&mdsc->cap_unlink_delay_list)) { struct ceph_inode_info *ci; struct inode *inode; @@ -2516,15 +2516,15 @@ static void ceph_cap_unlink_work(struct work_struct *work) inode = igrab(&ci->netfs.inode); if (inode) { - spin_unlock(&mdsc->cap_unlink_delay_lock); + spin_unlock(&mdsc->cap_delay_lock); doutc(cl, "on %p %llx.%llx\n", inode, ceph_vinop(inode)); ceph_check_caps(ci, CHECK_CAPS_FLUSH); iput(inode); - spin_lock(&mdsc->cap_unlink_delay_lock); + spin_lock(&mdsc->cap_delay_lock); } } - spin_unlock(&mdsc->cap_unlink_delay_lock); + spin_unlock(&mdsc->cap_delay_lock); doutc(cl, "done\n"); } @@ -5404,7 +5404,6 @@ int ceph_mdsc_init(struct ceph_fs_client *fsc) INIT_LIST_HEAD(&mdsc->cap_wait_list); spin_lock_init(&mdsc->cap_delay_lock); INIT_LIST_HEAD(&mdsc->cap_unlink_delay_list); - spin_lock_init(&mdsc->cap_unlink_delay_lock); INIT_LIST_HEAD(&mdsc->snap_flush_list); spin_lock_init(&mdsc->snap_flush_lock); mdsc->last_cap_flush_tid = 1; diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h index 03f8ff00874f..b88e80415224 100644 --- a/fs/ceph/mds_client.h +++ b/fs/ceph/mds_client.h @@ -461,9 +461,8 @@ struct ceph_mds_client { struct delayed_work delayed_work; /* delayed work */ unsigned long last_renew_caps; /* last time we renewed our caps */ struct list_head cap_delay_list; /* caps with delayed release */ - spinlock_t cap_delay_lock; /* protects cap_delay_list */ struct list_head cap_unlink_delay_list; /* caps with delayed release for unlink */ - spinlock_t cap_unlink_delay_lock; /* protects cap_unlink_delay_list */ + spinlock_t cap_delay_lock; /* protects cap_delay_list and cap_unlink_delay_list */ struct list_head snap_flush_list; /* cap_snaps ready to flush */ spinlock_t snap_flush_lock; From d3e0469306793972fd2b1ea016fa7ab0658c9849 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Tue, 9 Apr 2024 07:01:57 -0400 Subject: [PATCH 3/3] MAINTAINERS: remove myself as a Reviewer for Ceph It has been a couple of years since I stepped down as CephFS maintainer. I'm not involved in any meaningful way with the project these days, so while I'm happy to help review the occasional patch, I don't need to be cc'ed on all of them. Signed-off-by: Jeff Layton Signed-off-by: Ilya Dryomov --- MAINTAINERS | 2 -- 1 file changed, 2 deletions(-) diff --git a/MAINTAINERS b/MAINTAINERS index aea47e04c3a5..c45e2c3f6ff9 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -4869,7 +4869,6 @@ F: drivers/power/supply/cw2015_battery.c CEPH COMMON CODE (LIBCEPH) M: Ilya Dryomov M: Xiubo Li -R: Jeff Layton L: ceph-devel@vger.kernel.org S: Supported W: http://ceph.com/ @@ -4881,7 +4880,6 @@ F: net/ceph/ CEPH DISTRIBUTED FILE SYSTEM CLIENT (CEPH) M: Xiubo Li M: Ilya Dryomov -R: Jeff Layton L: ceph-devel@vger.kernel.org S: Supported W: http://ceph.com/