From cbcb358b744bf8d8c52b35d5efb1a960438c31cd Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Thu, 24 Aug 2023 17:55:51 +0800 Subject: [PATCH 01/12] ceph: skip reconnecting if MDS is not ready When MDS closed the session the kclient will send to reconnect to it immediately, but if the MDS just restarted and still not ready yet, such as still in the up:replay state and the sessionmap journal logs hasn't be replayed, the MDS will close the session. And then the kclient could remove the session and later when the mdsmap is in RECONNECT phrase it will skip reconnecting. But the MDS will wait until timeout and then evict the kclient. Just skip sending the reconnection request until the MDS is ready. Link: https://tracker.ceph.com/issues/62489 Signed-off-by: Xiubo Li Reviewed-by: Ilya Dryomov Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index d95eb525519a..be00c189ed46 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -5870,7 +5870,8 @@ static void mds_peer_reset(struct ceph_connection *con) pr_warn_client(mdsc->fsc->client, "mds%d closed our session\n", s->s_mds); - if (READ_ONCE(mdsc->fsc->mount_state) != CEPH_MOUNT_FENCE_IO) + if (READ_ONCE(mdsc->fsc->mount_state) != CEPH_MOUNT_FENCE_IO && + ceph_mdsmap_get_state(mdsc->mdsmap, s->s_mds) >= CEPH_MDS_STATE_RECONNECT) send_mds_reconnect(mdsc, s); } From f48e0342a74d7770cdf1d11894bdc3b6d989b29e Mon Sep 17 00:00:00 2001 From: Venky Shankar Date: Mon, 6 Nov 2023 10:02:32 +0530 Subject: [PATCH 02/12] ceph: reinitialize mds feature bit even when session in open Following along the same lines as per the user-space fix. Right now this isn't really an issue with the ceph kernel driver because of the feature bit laginess, however, that can change over time (when the new snaprealm info type is ported to the kernel driver) and depending on the MDS version that's being upgraded can cause message decoding issues - so, fix that early on. Link: http://tracker.ceph.com/issues/63188 Signed-off-by: Venky Shankar Reviewed-by: Xiubo Li Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index be00c189ed46..6781438f8782 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -4128,12 +4128,12 @@ static void handle_session(struct ceph_mds_session *session, pr_info_client(cl, "mds%d reconnect success\n", session->s_mds); + session->s_features = features; if (session->s_state == CEPH_MDS_SESSION_OPEN) { pr_notice_client(cl, "mds%d is already opened\n", session->s_mds); } else { session->s_state = CEPH_MDS_SESSION_OPEN; - session->s_features = features; renewed_caps(mdsc, session, 0); if (test_bit(CEPHFS_FEATURE_METRIC_COLLECT, &session->s_features)) From b79e4a0aa902322756ced7361a2c637d462c3c1c Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Tue, 7 Nov 2023 09:37:47 +0800 Subject: [PATCH 03/12] libceph: remove MAX_EXTENTS check for sparse reads There is no any limit for the extent array size and it's possible that when reading with a large size contents the total number of extents will exceed 4096. Then the messager will fail by reseting the connection and keeps resending the inflight IOs infinitely. [ idryomov: adjust error message ] Link: https://tracker.ceph.com/issues/62081 Signed-off-by: Xiubo Li Reviewed-by: Ilya Dryomov Signed-off-by: Ilya Dryomov --- net/ceph/osd_client.c | 17 ++++------------- 1 file changed, 4 insertions(+), 13 deletions(-) diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c index d3a759e052c8..625622016f57 100644 --- a/net/ceph/osd_client.c +++ b/net/ceph/osd_client.c @@ -5850,8 +5850,6 @@ static inline void convert_extent_map(struct ceph_sparse_read *sr) } #endif -#define MAX_EXTENTS 4096 - static int osd_sparse_read(struct ceph_connection *con, struct ceph_msg_data_cursor *cursor, char **pbuf) @@ -5882,23 +5880,16 @@ static int osd_sparse_read(struct ceph_connection *con, if (count > 0) { if (!sr->sr_extent || count > sr->sr_ext_len) { - /* - * Apply a hard cap to the number of extents. - * If we have more, assume something is wrong. - */ - if (count > MAX_EXTENTS) { - dout("%s: OSD returned 0x%x extents in a single reply!\n", - __func__, count); - return -EREMOTEIO; - } - /* no extent array provided, or too short */ kfree(sr->sr_extent); sr->sr_extent = kmalloc_array(count, sizeof(*sr->sr_extent), GFP_NOIO); - if (!sr->sr_extent) + if (!sr->sr_extent) { + pr_err("%s: failed to allocate %u extents\n", + __func__, count); return -ENOMEM; + } sr->sr_ext_len = count; } ret = count * sizeof(*sr->sr_extent); From aaefabc4a5f7ae48682c4d2d5d10faaf95c08eb9 Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Tue, 7 Nov 2023 10:44:41 +0800 Subject: [PATCH 04/12] ceph: try to allocate a smaller extent map for sparse read In fscrypt case and for a smaller read length we can predict the max count of the extent map. And for small read length use cases this could save some memories. [ idryomov: squash into a single patch to avoid build break, drop redundant variable in ceph_alloc_sparse_ext_map() ] Signed-off-by: Xiubo Li Reviewed-by: Ilya Dryomov Signed-off-by: Ilya Dryomov --- fs/ceph/addr.c | 4 +++- fs/ceph/file.c | 8 ++++++-- fs/ceph/super.h | 14 ++++++++++++++ include/linux/ceph/osd_client.h | 7 +++++-- 4 files changed, 28 insertions(+), 5 deletions(-) diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index 85be3bf18cdf..a5caafc1859e 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -357,6 +357,7 @@ static void ceph_netfs_issue_read(struct netfs_io_subrequest *subreq) u64 len = subreq->len; bool sparse = IS_ENCRYPTED(inode) || ceph_test_mount_opt(fsc, SPARSEREAD); u64 off = subreq->start; + int extent_cnt; if (ceph_inode_is_shutdown(inode)) { err = -EIO; @@ -379,7 +380,8 @@ static void ceph_netfs_issue_read(struct netfs_io_subrequest *subreq) } if (sparse) { - err = ceph_alloc_sparse_ext_map(&req->r_ops[0]); + extent_cnt = __ceph_sparse_read_ext_count(inode, len); + err = ceph_alloc_sparse_ext_map(&req->r_ops[0], extent_cnt); if (err) goto out; } diff --git a/fs/ceph/file.c b/fs/ceph/file.c index 3b5aae29e944..4dde0da10079 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -1028,6 +1028,7 @@ ssize_t __ceph_sync_read(struct inode *inode, loff_t *ki_pos, struct ceph_osd_req_op *op; u64 read_off = off; u64 read_len = len; + int extent_cnt; /* determine new offset/length if encrypted */ ceph_fscrypt_adjust_off_and_len(inode, &read_off, &read_len); @@ -1067,7 +1068,8 @@ ssize_t __ceph_sync_read(struct inode *inode, loff_t *ki_pos, op = &req->r_ops[0]; if (sparse) { - ret = ceph_alloc_sparse_ext_map(op); + extent_cnt = __ceph_sparse_read_ext_count(inode, read_len); + ret = ceph_alloc_sparse_ext_map(op, extent_cnt); if (ret) { ceph_osdc_put_request(req); break; @@ -1464,6 +1466,7 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, ssize_t len; struct ceph_osd_req_op *op; int readop = sparse ? CEPH_OSD_OP_SPARSE_READ : CEPH_OSD_OP_READ; + int extent_cnt; if (write) size = min_t(u64, size, fsc->mount_options->wsize); @@ -1527,7 +1530,8 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, osd_req_op_extent_osd_data_bvecs(req, 0, bvecs, num_pages, len); op = &req->r_ops[0]; if (sparse) { - ret = ceph_alloc_sparse_ext_map(op); + extent_cnt = __ceph_sparse_read_ext_count(inode, size); + ret = ceph_alloc_sparse_ext_map(op, extent_cnt); if (ret) { ceph_osdc_put_request(req); break; diff --git a/fs/ceph/super.h b/fs/ceph/super.h index fe0f64a0acb2..b06e2bc86221 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -3,6 +3,7 @@ #define _FS_CEPH_SUPER_H #include +#include #include #include @@ -1407,6 +1408,19 @@ static inline void __ceph_update_quota(struct ceph_inode_info *ci, ceph_adjust_quota_realms_count(&ci->netfs.inode, has_quota); } +static inline int __ceph_sparse_read_ext_count(struct inode *inode, u64 len) +{ + int cnt = 0; + + if (IS_ENCRYPTED(inode)) { + cnt = len >> CEPH_FSCRYPT_BLOCK_SHIFT; + if (cnt > CEPH_SPARSE_EXT_ARRAY_INITIAL) + cnt = 0; + } + + return cnt; +} + extern void ceph_handle_quota(struct ceph_mds_client *mdsc, struct ceph_mds_session *session, struct ceph_msg *msg); diff --git a/include/linux/ceph/osd_client.h b/include/linux/ceph/osd_client.h index b8610e9d2471..fa018d5864e7 100644 --- a/include/linux/ceph/osd_client.h +++ b/include/linux/ceph/osd_client.h @@ -572,9 +572,12 @@ int __ceph_alloc_sparse_ext_map(struct ceph_osd_req_op *op, int cnt); */ #define CEPH_SPARSE_EXT_ARRAY_INITIAL 16 -static inline int ceph_alloc_sparse_ext_map(struct ceph_osd_req_op *op) +static inline int ceph_alloc_sparse_ext_map(struct ceph_osd_req_op *op, int cnt) { - return __ceph_alloc_sparse_ext_map(op, CEPH_SPARSE_EXT_ARRAY_INITIAL); + if (!cnt) + cnt = CEPH_SPARSE_EXT_ARRAY_INITIAL; + + return __ceph_alloc_sparse_ext_map(op, cnt); } extern void ceph_osdc_get_request(struct ceph_osd_request *req); From b493ad718b1f0357394d2cdecbf00a44a36fa085 Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Fri, 17 Nov 2023 13:26:18 +0800 Subject: [PATCH 05/12] ceph: fix deadlock or deadcode of misusing dget() The lock order is incorrect between denty and its parent, we should always make sure that the parent get the lock first. But since this deadcode is never used and the parent dir will always be set from the callers, let's just remove it. Link: https://lore.kernel.org/r/20231116081919.GZ1957730@ZenIV Reported-by: Al Viro Signed-off-by: Xiubo Li Reviewed-by: Jeff Layton Signed-off-by: Ilya Dryomov --- fs/ceph/caps.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 2c0b8dc3dd0d..9c02f328c966 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -4887,13 +4887,15 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry, struct inode *dir, int mds, int drop, int unless) { - struct dentry *parent = NULL; struct ceph_mds_request_release *rel = *p; struct ceph_dentry_info *di = ceph_dentry(dentry); struct ceph_client *cl; int force = 0; int ret; + /* This shouldn't happen */ + BUG_ON(!dir); + /* * force an record for the directory caps if we have a dentry lease. * this is racy (can't take i_ceph_lock and d_lock together), but it @@ -4903,14 +4905,9 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry, spin_lock(&dentry->d_lock); if (di->lease_session && di->lease_session->s_mds == mds) force = 1; - if (!dir) { - parent = dget(dentry->d_parent); - dir = d_inode(parent); - } spin_unlock(&dentry->d_lock); ret = ceph_encode_inode_release(p, dir, mds, drop, unless, force); - dput(parent); cl = ceph_inode_to_client(dir); spin_lock(&dentry->d_lock); From 9c896d6bc3dfef86659a6a1fb25ccdea5dbef6a3 Mon Sep 17 00:00:00 2001 From: Eric Biggers Date: Wed, 22 Nov 2023 19:08:38 -0800 Subject: [PATCH 06/12] ceph: select FS_ENCRYPTION_ALGS if FS_ENCRYPTION The kconfig options for filesystems that support FS_ENCRYPTION are supposed to select FS_ENCRYPTION_ALGS. This is needed to ensure that required crypto algorithms get enabled as loadable modules or builtin as is appropriate for the set of enabled filesystems. Do this for CEPH_FS so that there aren't any missing algorithms if someone happens to have CEPH_FS as their only enabled filesystem that supports encryption. Cc: stable@vger.kernel.org Fixes: f061feda6c54 ("ceph: add fscrypt ioctls and ceph.fscrypt.auth vxattr") Signed-off-by: Eric Biggers Reviewed-by: Xiubo Li Signed-off-by: Ilya Dryomov --- fs/ceph/Kconfig | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/ceph/Kconfig b/fs/ceph/Kconfig index 94df854147d3..7249d70e1a43 100644 --- a/fs/ceph/Kconfig +++ b/fs/ceph/Kconfig @@ -7,6 +7,7 @@ config CEPH_FS select CRYPTO_AES select CRYPTO select NETFS_SUPPORT + select FS_ENCRYPTION_ALGS if FS_ENCRYPTION default n help Choose Y or M here to include support for mounting the From 66207de308df82242da0bf88035b24bcd4377562 Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Thu, 16 Nov 2023 09:46:19 +0800 Subject: [PATCH 07/12] ceph: rename create_session_open_msg() to create_session_full_msg() Makes the create session msg helper to be more general and could be used by other ops. Signed-off-by: Xiubo Li Reviewed-by: Ilya Dryomov Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 6781438f8782..a7eb722c1b48 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -1534,7 +1534,8 @@ static int encode_metric_spec(void **p, void *end) * session message, specialization for CEPH_SESSION_REQUEST_OPEN * to include additional client metadata fields. */ -static struct ceph_msg *create_session_open_msg(struct ceph_mds_client *mdsc, u64 seq) +static struct ceph_msg * +create_session_full_msg(struct ceph_mds_client *mdsc, int op, u64 seq) { struct ceph_msg *msg; struct ceph_mds_session_head *h; @@ -1589,7 +1590,7 @@ static struct ceph_msg *create_session_open_msg(struct ceph_mds_client *mdsc, u6 end = p + msg->front.iov_len; h = p; - h->op = cpu_to_le32(CEPH_SESSION_REQUEST_OPEN); + h->op = cpu_to_le32(op); h->seq = cpu_to_le64(seq); /* @@ -1663,7 +1664,8 @@ static int __open_session(struct ceph_mds_client *mdsc, session->s_renew_requested = jiffies; /* send connect message */ - msg = create_session_open_msg(mdsc, session->s_seq); + msg = create_session_full_msg(mdsc, CEPH_SESSION_REQUEST_OPEN, + session->s_seq); if (IS_ERR(msg)) return PTR_ERR(msg); ceph_con_send(&session->s_con, msg); From 6df89bf220fdac9f40b0d35cd132eef54cf99d4b Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Thu, 16 Nov 2023 10:56:24 +0800 Subject: [PATCH 08/12] ceph: send oldest_client_tid when renewing caps Update the oldest_client_tid via the session renew caps msg to make sure that the MDSs won't pile up the completed request list in a very large size. [ idryomov: drop inapplicable comment ] Link: https://tracker.ceph.com/issues/63364 Signed-off-by: Xiubo Li Reviewed-by: Ilya Dryomov Signed-off-by: Ilya Dryomov --- fs/ceph/mds_client.c | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-) diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index a7eb722c1b48..6cfeba08a360 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -1579,6 +1579,9 @@ create_session_full_msg(struct ceph_mds_client *mdsc, int op, u64 seq) size = METRIC_BYTES(count); extra_bytes += 2 + 4 + 4 + size; + /* flags, mds auth caps and oldest_client_tid */ + extra_bytes += 4 + 4 + 8; + /* Allocate the message */ msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h) + extra_bytes, GFP_NOFS, false); @@ -1597,9 +1600,9 @@ create_session_full_msg(struct ceph_mds_client *mdsc, int op, u64 seq) * Serialize client metadata into waiting buffer space, using * the format that userspace expects for map * - * ClientSession messages with metadata are v4 + * ClientSession messages with metadata are v7 */ - msg->hdr.version = cpu_to_le16(4); + msg->hdr.version = cpu_to_le16(7); msg->hdr.compat_version = cpu_to_le16(1); /* The write pointer, following the session_head structure */ @@ -1635,6 +1638,15 @@ create_session_full_msg(struct ceph_mds_client *mdsc, int op, u64 seq) return ERR_PTR(ret); } + /* version == 5, flags */ + ceph_encode_32(&p, 0); + + /* version == 6, mds auth caps */ + ceph_encode_32(&p, 0); + + /* version == 7, oldest_client_tid */ + ceph_encode_64(&p, mdsc->oldest_tid); + msg->front.iov_len = p - msg->front.iov_base; msg->hdr.front_len = cpu_to_le32(msg->front.iov_len); @@ -2030,10 +2042,10 @@ static int send_renew_caps(struct ceph_mds_client *mdsc, doutc(cl, "to mds%d (%s)\n", session->s_mds, ceph_mds_state_name(state)); - msg = ceph_create_session_msg(CEPH_SESSION_REQUEST_RENEWCAPS, + msg = create_session_full_msg(mdsc, CEPH_SESSION_REQUEST_RENEWCAPS, ++session->s_renew_seq); - if (!msg) - return -ENOMEM; + if (IS_ERR(msg)) + return PTR_ERR(msg); ceph_con_send(&session->s_con, msg); return 0; } From b36b03344f5fccb81e5cf3b3ede68b7e7a7e930a Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Wed, 22 Nov 2023 15:55:14 +0800 Subject: [PATCH 09/12] ceph: remove duplicated code in ceph_netfs_issue_read() When allocating an osd request the libceph.ko will add the 'read_from_replica' flag by default. Signed-off-by: Xiubo Li Reviewed-by: Ilya Dryomov Signed-off-by: Ilya Dryomov --- fs/ceph/addr.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index a5caafc1859e..792bbbf8de64 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -371,8 +371,8 @@ static void ceph_netfs_issue_read(struct netfs_io_subrequest *subreq) req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout, vino, off, &len, 0, 1, sparse ? CEPH_OSD_OP_SPARSE_READ : CEPH_OSD_OP_READ, - CEPH_OSD_FLAG_READ | fsc->client->osdc.client->options->read_from_replica, - NULL, ci->i_truncate_seq, ci->i_truncate_size, false); + CEPH_OSD_FLAG_READ, NULL, ci->i_truncate_seq, + ci->i_truncate_size, false); if (IS_ERR(req)) { err = PTR_ERR(req); req = NULL; From 0f4cf64eabc6e16cfc2704f1960e82dc79d91c8d Mon Sep 17 00:00:00 2001 From: Wenchao Hao Date: Thu, 23 Nov 2023 09:53:40 +0800 Subject: [PATCH 10/12] ceph: fix invalid pointer access if get_quota_realm return ERR_PTR This issue is reported by smatch that get_quota_realm() might return ERR_PTR but we did not handle it. It's not a immediate bug, while we still should address it to avoid potential bugs if get_quota_realm() is changed to return other ERR_PTR in future. Set ceph_snap_realm's pointer in get_quota_realm()'s to address this issue, the pointer would be set to NULL if get_quota_realm() failed to get struct ceph_snap_realm, so no ERR_PTR would happen any more. [ xiubli: minor code style clean up ] Signed-off-by: Wenchao Hao Reviewed-by: Xiubo Li Signed-off-by: Ilya Dryomov --- fs/ceph/quota.c | 39 ++++++++++++++++++++++----------------- 1 file changed, 22 insertions(+), 17 deletions(-) diff --git a/fs/ceph/quota.c b/fs/ceph/quota.c index 9d36c3532de1..06ee397e0c3a 100644 --- a/fs/ceph/quota.c +++ b/fs/ceph/quota.c @@ -197,10 +197,10 @@ void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc) } /* - * This function walks through the snaprealm for an inode and returns the - * ceph_snap_realm for the first snaprealm that has quotas set (max_files, + * This function walks through the snaprealm for an inode and set the + * realmp with the first snaprealm that has quotas set (max_files, * max_bytes, or any, depending on the 'which_quota' argument). If the root is - * reached, return the root ceph_snap_realm instead. + * reached, set the realmp with the root ceph_snap_realm instead. * * Note that the caller is responsible for calling ceph_put_snap_realm() on the * returned realm. @@ -211,10 +211,9 @@ void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc) * this function will return -EAGAIN; otherwise, the snaprealms walk-through * will be restarted. */ -static struct ceph_snap_realm *get_quota_realm(struct ceph_mds_client *mdsc, - struct inode *inode, - enum quota_get_realm which_quota, - bool retry) +static int get_quota_realm(struct ceph_mds_client *mdsc, struct inode *inode, + enum quota_get_realm which_quota, + struct ceph_snap_realm **realmp, bool retry) { struct ceph_client *cl = mdsc->fsc->client; struct ceph_inode_info *ci = NULL; @@ -222,8 +221,10 @@ static struct ceph_snap_realm *get_quota_realm(struct ceph_mds_client *mdsc, struct inode *in; bool has_quota; + if (realmp) + *realmp = NULL; if (ceph_snap(inode) != CEPH_NOSNAP) - return NULL; + return 0; restart: realm = ceph_inode(inode)->i_snap_realm; @@ -250,7 +251,7 @@ static struct ceph_snap_realm *get_quota_realm(struct ceph_mds_client *mdsc, break; ceph_put_snap_realm(mdsc, realm); if (!retry) - return ERR_PTR(-EAGAIN); + return -EAGAIN; goto restart; } @@ -259,8 +260,11 @@ static struct ceph_snap_realm *get_quota_realm(struct ceph_mds_client *mdsc, iput(in); next = realm->parent; - if (has_quota || !next) - return realm; + if (has_quota || !next) { + if (realmp) + *realmp = realm; + return 0; + } ceph_get_snap_realm(mdsc, next); ceph_put_snap_realm(mdsc, realm); @@ -269,7 +273,7 @@ static struct ceph_snap_realm *get_quota_realm(struct ceph_mds_client *mdsc, if (realm) ceph_put_snap_realm(mdsc, realm); - return NULL; + return 0; } bool ceph_quota_is_same_realm(struct inode *old, struct inode *new) @@ -277,6 +281,7 @@ bool ceph_quota_is_same_realm(struct inode *old, struct inode *new) struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(old->i_sb); struct ceph_snap_realm *old_realm, *new_realm; bool is_same; + int ret; restart: /* @@ -286,9 +291,9 @@ bool ceph_quota_is_same_realm(struct inode *old, struct inode *new) * dropped and we can then restart the whole operation. */ down_read(&mdsc->snap_rwsem); - old_realm = get_quota_realm(mdsc, old, QUOTA_GET_ANY, true); - new_realm = get_quota_realm(mdsc, new, QUOTA_GET_ANY, false); - if (PTR_ERR(new_realm) == -EAGAIN) { + get_quota_realm(mdsc, old, QUOTA_GET_ANY, &old_realm, true); + ret = get_quota_realm(mdsc, new, QUOTA_GET_ANY, &new_realm, false); + if (ret == -EAGAIN) { up_read(&mdsc->snap_rwsem); if (old_realm) ceph_put_snap_realm(mdsc, old_realm); @@ -492,8 +497,8 @@ bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, struct kstatfs *buf) bool is_updated = false; down_read(&mdsc->snap_rwsem); - realm = get_quota_realm(mdsc, d_inode(fsc->sb->s_root), - QUOTA_GET_MAX_BYTES, true); + get_quota_realm(mdsc, d_inode(fsc->sb->s_root), QUOTA_GET_MAX_BYTES, + &realm, true); up_read(&mdsc->snap_rwsem); if (!realm) return false; From f6fb21b22fbe443f92b0d580391a7fb46d1840df Mon Sep 17 00:00:00 2001 From: Al Viro Date: Wed, 20 Dec 2023 05:20:54 +0000 Subject: [PATCH 11/12] ceph: d_obtain_{alias,root}(ERR_PTR(...)) will do the right thing Clean up the code. Signed-off-by: Al Viro Reviewed-by: Xiubo Li Signed-off-by: Ilya Dryomov --- fs/ceph/export.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/fs/ceph/export.c b/fs/ceph/export.c index 726af69d4d62..a79f163ae4ed 100644 --- a/fs/ceph/export.c +++ b/fs/ceph/export.c @@ -286,8 +286,6 @@ static struct dentry *__snapfh_to_dentry(struct super_block *sb, doutc(cl, "%llx.%llx parent %llx hash %x err=%d", vino.ino, vino.snap, sfh->parent_ino, sfh->hash, err); } - if (IS_ERR(inode)) - return ERR_CAST(inode); /* see comments in ceph_get_parent() */ return unlinked ? d_obtain_root(inode) : d_obtain_alias(inode); } From 2a965d1b15d28065b35ab4ebd1e51558fcd91aa5 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Wed, 20 Dec 2023 05:29:25 +0000 Subject: [PATCH 12/12] ceph: get rid of passing callbacks in __dentry_leases_walk() __dentry_leases_walk() gets a callback and calls it for a bunch of denties; there are exactly two callers and we already have a flag telling them apart - lwc->dir_lease. Seeing that indirect calls are costly these days, let's get rid of the callback and just call the right function directly. Has a side benefit of saner signatures... [ xiubli: a minor fix in the commit title ] Signed-off-by: Al Viro Reviewed-by: Xiubo Li Signed-off-by: Ilya Dryomov --- fs/ceph/dir.c | 21 +++++++++++++-------- 1 file changed, 13 insertions(+), 8 deletions(-) diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c index 91709934c8b1..768158743750 100644 --- a/fs/ceph/dir.c +++ b/fs/ceph/dir.c @@ -1593,10 +1593,12 @@ struct ceph_lease_walk_control { unsigned long dir_lease_ttl; }; +static int __dir_lease_check(const struct dentry *, struct ceph_lease_walk_control *); +static int __dentry_lease_check(const struct dentry *); + static unsigned long __dentry_leases_walk(struct ceph_mds_client *mdsc, - struct ceph_lease_walk_control *lwc, - int (*check)(struct dentry*, void*)) + struct ceph_lease_walk_control *lwc) { struct ceph_dentry_info *di, *tmp; struct dentry *dentry, *last = NULL; @@ -1624,7 +1626,10 @@ __dentry_leases_walk(struct ceph_mds_client *mdsc, goto next; } - ret = check(dentry, lwc); + if (lwc->dir_lease) + ret = __dir_lease_check(dentry, lwc); + else + ret = __dentry_lease_check(dentry); if (ret & TOUCH) { /* move it into tail of dir lease list */ __dentry_dir_lease_touch(mdsc, di); @@ -1681,7 +1686,7 @@ __dentry_leases_walk(struct ceph_mds_client *mdsc, return freed; } -static int __dentry_lease_check(struct dentry *dentry, void *arg) +static int __dentry_lease_check(const struct dentry *dentry) { struct ceph_dentry_info *di = ceph_dentry(dentry); int ret; @@ -1696,9 +1701,9 @@ static int __dentry_lease_check(struct dentry *dentry, void *arg) return DELETE; } -static int __dir_lease_check(struct dentry *dentry, void *arg) +static int __dir_lease_check(const struct dentry *dentry, + struct ceph_lease_walk_control *lwc) { - struct ceph_lease_walk_control *lwc = arg; struct ceph_dentry_info *di = ceph_dentry(dentry); int ret = __dir_lease_try_check(dentry); @@ -1737,7 +1742,7 @@ int ceph_trim_dentries(struct ceph_mds_client *mdsc) lwc.dir_lease = false; lwc.nr_to_scan = CEPH_CAPS_PER_RELEASE * 2; - freed = __dentry_leases_walk(mdsc, &lwc, __dentry_lease_check); + freed = __dentry_leases_walk(mdsc, &lwc); if (!lwc.nr_to_scan) /* more invalid leases */ return -EAGAIN; @@ -1747,7 +1752,7 @@ int ceph_trim_dentries(struct ceph_mds_client *mdsc) lwc.dir_lease = true; lwc.expire_dir_lease = freed < count; lwc.dir_lease_ttl = mdsc->fsc->mount_options->caps_wanted_delay_max * HZ; - freed +=__dentry_leases_walk(mdsc, &lwc, __dir_lease_check); + freed +=__dentry_leases_walk(mdsc, &lwc); if (!lwc.nr_to_scan) /* more to check */ return -EAGAIN;