diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2020-04-08 21:44:05 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2020-04-08 21:44:05 -0700 |
commit | fcc95f06403c956e3f50ca4a82db12b66a3078e0 (patch) | |
tree | 74bb3e387b635601ba351f621c7eea575e185aac /fs | |
parent | c6b80eb89b55590b12db11103913088735205b5c (diff) | |
parent | ef9157259fb7bb3bc2c61df227e36f1b861a4753 (diff) |
Merge tag 'ceph-for-5.7-rc1' of git://github.com/ceph/ceph-client
Pull ceph updates from Ilya Dryomov:
"The main items are:
- support for asynchronous create and unlink (Jeff Layton).
Creates and unlinks are satisfied locally, without waiting for a
reply from the MDS, provided the client has been granted
appropriate caps (new in v15.y.z ("Octopus") release). This can be
a big help for metadata heavy workloads such as tar and rsync.
Opt-in with the new nowsync mount option.
- multiple blk-mq queues for rbd (Hannes Reinecke and myself).
When the driver was converted to blk-mq, we settled on a single
blk-mq queue because of a global lock in libceph and some other
technical debt. These have since been addressed, so allocate a
queue per CPU to enhance parallelism.
- don't hold onto caps that aren't actually needed (Zheng Yan).
This has been our long-standing behavior, but it causes issues with
some active/standby applications (synchronous I/O, stalls if the
standby goes down, etc).
- .snap directory timestamps consistent with ceph-fuse (Luis
Henriques)"
* tag 'ceph-for-5.7-rc1' of git://github.com/ceph/ceph-client: (49 commits)
ceph: fix snapshot directory timestamps
ceph: wait for async creating inode before requesting new max size
ceph: don't skip updating wanted caps when cap is stale
ceph: request new max size only when there is auth cap
ceph: cleanup return error of try_get_cap_refs()
ceph: return ceph_mdsc_do_request() errors from __get_parent()
ceph: check all mds' caps after page writeback
ceph: update i_requested_max_size only when sending cap msg to auth mds
ceph: simplify calling of ceph_get_fmode()
ceph: remove delay check logic from ceph_check_caps()
ceph: consider inode's last read/write when calculating wanted caps
ceph: always renew caps if mds_wanted is insufficient
ceph: update dentry lease for async create
ceph: attempt to do async create when possible
ceph: cache layout in parent dir on first sync create
ceph: add new MDS req field to hold delegated inode number
ceph: decode interval_sets for delegated inos
ceph: make ceph_fill_inode non-static
ceph: perform asynchronous unlink if we have sufficient caps
ceph: don't take refs to want mask unless we have all bits
...
Diffstat (limited to 'fs')
-rw-r--r-- | fs/ceph/addr.c | 90 | ||||
-rw-r--r-- | fs/ceph/cache.c | 2 | ||||
-rw-r--r-- | fs/ceph/caps.c | 536 | ||||
-rw-r--r-- | fs/ceph/debugfs.c | 16 | ||||
-rw-r--r-- | fs/ceph/dir.c | 132 | ||||
-rw-r--r-- | fs/ceph/export.c | 5 | ||||
-rw-r--r-- | fs/ceph/file.c | 486 | ||||
-rw-r--r-- | fs/ceph/inode.c | 84 | ||||
-rw-r--r-- | fs/ceph/ioctl.c | 2 | ||||
-rw-r--r-- | fs/ceph/locks.c | 31 | ||||
-rw-r--r-- | fs/ceph/mds_client.c | 240 | ||||
-rw-r--r-- | fs/ceph/mds_client.h | 30 | ||||
-rw-r--r-- | fs/ceph/super.c | 28 | ||||
-rw-r--r-- | fs/ceph/super.h | 70 |
14 files changed, 1267 insertions, 485 deletions
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c index 7ab616601141..6f4678d98df7 100644 --- a/fs/ceph/addr.c +++ b/fs/ceph/addr.c @@ -159,8 +159,6 @@ static void ceph_invalidatepage(struct page *page, unsigned int offset, if (!PagePrivate(page)) return; - ClearPageChecked(page); - dout("%p invalidatepage %p idx %lu full dirty page\n", inode, page, page->index); @@ -183,6 +181,47 @@ static int ceph_releasepage(struct page *page, gfp_t g) } /* + * Read some contiguous pages. If we cross a stripe boundary, shorten + * *plen. Return number of bytes read, or error. + */ +static int ceph_sync_readpages(struct ceph_fs_client *fsc, + struct ceph_vino vino, + struct ceph_file_layout *layout, + u64 off, u64 *plen, + u32 truncate_seq, u64 truncate_size, + struct page **pages, int num_pages, + int page_align) +{ + struct ceph_osd_client *osdc = &fsc->client->osdc; + struct ceph_osd_request *req; + int rc = 0; + + dout("readpages on ino %llx.%llx on %llu~%llu\n", vino.ino, + vino.snap, off, *plen); + req = ceph_osdc_new_request(osdc, layout, vino, off, plen, 0, 1, + CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ, + NULL, truncate_seq, truncate_size, + false); + if (IS_ERR(req)) + return PTR_ERR(req); + + /* it may be a short read due to an object boundary */ + osd_req_op_extent_osd_data_pages(req, 0, + pages, *plen, page_align, false, false); + + dout("readpages final extent is %llu~%llu (%llu bytes align %d)\n", + off, *plen, *plen, page_align); + + rc = ceph_osdc_start_request(osdc, req, false); + if (!rc) + rc = ceph_osdc_wait_request(osdc, req); + + ceph_osdc_put_request(req); + dout("readpages result %d\n", rc); + return rc; +} + +/* * read a single page, without unlocking it. */ static int ceph_do_readpage(struct file *filp, struct page *page) @@ -218,7 +257,7 @@ static int ceph_do_readpage(struct file *filp, struct page *page) dout("readpage inode %p file %p page %p index %lu\n", inode, filp, page, page->index); - err = ceph_osdc_readpages(&fsc->client->osdc, ceph_vino(inode), + err = ceph_sync_readpages(fsc, ceph_vino(inode), &ci->i_layout, off, &len, ci->i_truncate_seq, ci->i_truncate_size, &page, 1, 0); @@ -571,6 +610,47 @@ static u64 get_writepages_data_length(struct inode *inode, } /* + * do a synchronous write on N pages + */ +static int ceph_sync_writepages(struct ceph_fs_client *fsc, + struct ceph_vino vino, + struct ceph_file_layout *layout, + struct ceph_snap_context *snapc, + u64 off, u64 len, + u32 truncate_seq, u64 truncate_size, + struct timespec64 *mtime, + struct page **pages, int num_pages) +{ + struct ceph_osd_client *osdc = &fsc->client->osdc; + struct ceph_osd_request *req; + int rc = 0; + int page_align = off & ~PAGE_MASK; + + req = ceph_osdc_new_request(osdc, layout, vino, off, &len, 0, 1, + CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE, + snapc, truncate_seq, truncate_size, + true); + if (IS_ERR(req)) + return PTR_ERR(req); + + /* it may be a short write due to an object boundary */ + osd_req_op_extent_osd_data_pages(req, 0, pages, len, page_align, + false, false); + dout("writepages %llu~%llu (%llu bytes)\n", off, len, len); + + req->r_mtime = *mtime; + rc = ceph_osdc_start_request(osdc, req, true); + if (!rc) + rc = ceph_osdc_wait_request(osdc, req); + + ceph_osdc_put_request(req); + if (rc == 0) + rc = len; + dout("writepages result %d\n", rc); + return rc; +} + +/* * Write a single page, but leave the page locked. * * If we get a write error, mark the mapping for error, but still adjust the @@ -628,7 +708,7 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc) set_bdi_congested(inode_to_bdi(inode), BLK_RW_ASYNC); set_page_writeback(page); - err = ceph_osdc_writepages(&fsc->client->osdc, ceph_vino(inode), + err = ceph_sync_writepages(fsc, ceph_vino(inode), &ci->i_layout, snapc, page_off, len, ceph_wbc.truncate_seq, ceph_wbc.truncate_size, @@ -1575,7 +1655,7 @@ static vm_fault_t ceph_page_mkwrite(struct vm_fault *vmf) do { lock_page(page); - if ((off > size) || (page->mapping != inode->i_mapping)) { + if (page_mkwrite_check_truncate(page, inode) < 0) { unlock_page(page); ret = VM_FAULT_NOPAGE; break; diff --git a/fs/ceph/cache.c b/fs/ceph/cache.c index 270b769607a2..2f5cb6bc78e1 100644 --- a/fs/ceph/cache.c +++ b/fs/ceph/cache.c @@ -32,7 +32,7 @@ struct ceph_fscache_entry { size_t uniq_len; /* The following members must be last */ struct ceph_fsid fsid; - char uniquifier[0]; + char uniquifier[]; }; static const struct fscache_cookie_def ceph_fscache_fsid_object_def = { diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 28ae0c134700..185db76300b3 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -490,13 +490,10 @@ static void __cap_set_timeouts(struct ceph_mds_client *mdsc, struct ceph_inode_info *ci) { struct ceph_mount_options *opt = mdsc->fsc->mount_options; - - ci->i_hold_caps_min = round_jiffies(jiffies + - opt->caps_wanted_delay_min * HZ); ci->i_hold_caps_max = round_jiffies(jiffies + opt->caps_wanted_delay_max * HZ); - dout("__cap_set_timeouts %p min %lu max %lu\n", &ci->vfs_inode, - ci->i_hold_caps_min - jiffies, ci->i_hold_caps_max - jiffies); + dout("__cap_set_timeouts %p %lu\n", &ci->vfs_inode, + ci->i_hold_caps_max - jiffies); } /* @@ -508,10 +505,9 @@ static void __cap_set_timeouts(struct ceph_mds_client *mdsc, * -> we take mdsc->cap_delay_lock */ static void __cap_delay_requeue(struct ceph_mds_client *mdsc, - struct ceph_inode_info *ci, - bool set_timeout) + struct ceph_inode_info *ci) { - dout("__cap_delay_requeue %p flags %d at %lu\n", &ci->vfs_inode, + dout("__cap_delay_requeue %p flags 0x%lx at %lu\n", &ci->vfs_inode, ci->i_ceph_flags, ci->i_hold_caps_max); if (!mdsc->stopping) { spin_lock(&mdsc->cap_delay_lock); @@ -520,8 +516,7 @@ static void __cap_delay_requeue(struct ceph_mds_client *mdsc, goto no_change; list_del_init(&ci->i_cap_delay_list); } - if (set_timeout) - __cap_set_timeouts(mdsc, ci); + __cap_set_timeouts(mdsc, ci); list_add_tail(&ci->i_cap_delay_list, &mdsc->cap_delay_list); no_change: spin_unlock(&mdsc->cap_delay_lock); @@ -561,19 +556,20 @@ static void __cap_delay_cancel(struct ceph_mds_client *mdsc, spin_unlock(&mdsc->cap_delay_lock); } -/* - * Common issue checks for add_cap, handle_cap_grant. - */ +/* Common issue checks for add_cap, handle_cap_grant. */ static void __check_cap_issue(struct ceph_inode_info *ci, struct ceph_cap *cap, unsigned issued) { unsigned had = __ceph_caps_issued(ci, NULL); + lockdep_assert_held(&ci->i_ceph_lock); + /* * Each time we receive FILE_CACHE anew, we increment * i_rdcache_gen. */ - if ((issued & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) && + if (S_ISREG(ci->vfs_inode.i_mode) && + (issued & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) && (had & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0) { ci->i_rdcache_gen++; } @@ -592,6 +588,13 @@ static void __check_cap_issue(struct ceph_inode_info *ci, struct ceph_cap *cap, __ceph_dir_clear_complete(ci); } } + + /* Wipe saved layout if we're losing DIR_CREATE caps */ + if (S_ISDIR(ci->vfs_inode.i_mode) && (had & CEPH_CAP_DIR_CREATE) && + !(issued & CEPH_CAP_DIR_CREATE)) { + ceph_put_string(rcu_dereference_raw(ci->i_cached_layout.pool_ns)); + memset(&ci->i_cached_layout, 0, sizeof(ci->i_cached_layout)); + } } /* @@ -605,7 +608,7 @@ static void __check_cap_issue(struct ceph_inode_info *ci, struct ceph_cap *cap, */ void ceph_add_cap(struct inode *inode, struct ceph_mds_session *session, u64 cap_id, - int fmode, unsigned issued, unsigned wanted, + unsigned issued, unsigned wanted, unsigned seq, unsigned mseq, u64 realmino, int flags, struct ceph_cap **new_cap) { @@ -621,13 +624,6 @@ void ceph_add_cap(struct inode *inode, dout("add_cap %p mds%d cap %llx %s seq %d\n", inode, session->s_mds, cap_id, ceph_cap_string(issued), seq); - /* - * If we are opening the file, include file mode wanted bits - * in wanted. - */ - if (fmode >= 0) - wanted |= ceph_caps_for_mode(fmode); - spin_lock(&session->s_gen_ttl_lock); gen = session->s_cap_gen; spin_unlock(&session->s_gen_ttl_lock); @@ -725,7 +721,7 @@ void ceph_add_cap(struct inode *inode, dout(" issued %s, mds wanted %s, actual %s, queueing\n", ceph_cap_string(issued), ceph_cap_string(wanted), ceph_cap_string(actual_wanted)); - __cap_delay_requeue(mdsc, ci, true); + __cap_delay_requeue(mdsc, ci); } if (flags & CEPH_CAP_FLAG_AUTH) { @@ -752,9 +748,6 @@ void ceph_add_cap(struct inode *inode, cap->issue_seq = seq; cap->mseq = mseq; cap->cap_gen = gen; - - if (fmode >= 0) - __ceph_get_fmode(ci, fmode); } /* @@ -958,29 +951,97 @@ int __ceph_caps_used(struct ceph_inode_info *ci) if (ci->i_rd_ref) used |= CEPH_CAP_FILE_RD; if (ci->i_rdcache_ref || - (!S_ISDIR(ci->vfs_inode.i_mode) && /* ignore readdir cache */ + (S_ISREG(ci->vfs_inode.i_mode) && ci->vfs_inode.i_data.nrpages)) used |= CEPH_CAP_FILE_CACHE; if (ci->i_wr_ref) used |= CEPH_CAP_FILE_WR; if (ci->i_wb_ref || ci->i_wrbuffer_ref) used |= CEPH_CAP_FILE_BUFFER; + if (ci->i_fx_ref) + used |= CEPH_CAP_FILE_EXCL; return used; } +#define FMODE_WAIT_BIAS 1000 + /* * wanted, by virtue of open file modes */ int __ceph_caps_file_wanted(struct ceph_inode_info *ci) { - int i, bits = 0; - for (i = 0; i < CEPH_FILE_MODE_BITS; i++) { - if (ci->i_nr_by_mode[i]) - bits |= 1 << i; + const int PIN_SHIFT = ffs(CEPH_FILE_MODE_PIN); + const int RD_SHIFT = ffs(CEPH_FILE_MODE_RD); + const int WR_SHIFT = ffs(CEPH_FILE_MODE_WR); + const int LAZY_SHIFT = ffs(CEPH_FILE_MODE_LAZY); + struct ceph_mount_options *opt = + ceph_inode_to_client(&ci->vfs_inode)->mount_options; + unsigned long used_cutoff = jiffies - opt->caps_wanted_delay_max * HZ; + unsigned long idle_cutoff = jiffies - opt->caps_wanted_delay_min * HZ; + + if (S_ISDIR(ci->vfs_inode.i_mode)) { + int want = 0; + + /* use used_cutoff here, to keep dir's wanted caps longer */ + if (ci->i_nr_by_mode[RD_SHIFT] > 0 || + time_after(ci->i_last_rd, used_cutoff)) + want |= CEPH_CAP_ANY_SHARED; + + if (ci->i_nr_by_mode[WR_SHIFT] > 0 || + time_after(ci->i_last_wr, used_cutoff)) { + want |= CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_EXCL; + if (opt->flags & CEPH_MOUNT_OPT_ASYNC_DIROPS) + want |= CEPH_CAP_ANY_DIR_OPS; + } + + if (want || ci->i_nr_by_mode[PIN_SHIFT] > 0) + want |= CEPH_CAP_PIN; + + return want; + } else { + int bits = 0; + + if (ci->i_nr_by_mode[RD_SHIFT] > 0) { + if (ci->i_nr_by_mode[RD_SHIFT] >= FMODE_WAIT_BIAS || + time_after(ci->i_last_rd, used_cutoff)) + bits |= 1 << RD_SHIFT; + } else if (time_after(ci->i_last_rd, idle_cutoff)) { + bits |= 1 << RD_SHIFT; + } + + if (ci->i_nr_by_mode[WR_SHIFT] > 0) { + if (ci->i_nr_by_mode[WR_SHIFT] >= FMODE_WAIT_BIAS || + time_after(ci->i_last_wr, used_cutoff)) + bits |= 1 << WR_SHIFT; + } else if (time_after(ci->i_last_wr, idle_cutoff)) { + bits |= 1 << WR_SHIFT; + } + + /* check lazyio only when read/write is wanted */ + if ((bits & (CEPH_FILE_MODE_RDWR << 1)) && + ci->i_nr_by_mode[LAZY_SHIFT] > 0) + bits |= 1 << LAZY_SHIFT; + + return bits ? ceph_caps_for_mode(bits >> 1) : 0; } - if (bits == 0) - return 0; - return ceph_caps_for_mode(bits >> 1); +} + +/* + * wanted, by virtue of open file modes AND cap refs (buffered/cached data) + */ +int __ceph_caps_wanted(struct ceph_inode_info *ci) +{ + int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); + if (S_ISDIR(ci->vfs_inode.i_mode)) { + /* we want EXCL if holding caps of dir ops */ + if (w & CEPH_CAP_ANY_DIR_OPS) + w |= CEPH_CAP_FILE_EXCL; + } else { + /* we want EXCL if dirty data */ + if (w & CEPH_CAP_FILE_BUFFER) + w |= CEPH_CAP_FILE_EXCL; + } + return w; } /* @@ -1004,14 +1065,6 @@ int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check) return mds_wanted; } -/* - * called under i_ceph_lock - */ -static int __ceph_is_single_caps(struct ceph_inode_info *ci) -{ - return rb_first(&ci->i_caps) == rb_last(&ci->i_caps); -} - int ceph_is_any_caps(struct inode *inode) { struct ceph_inode_info *ci = ceph_inode(inode); @@ -1274,9 +1327,15 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, struct cap_msg_args arg; int held, revoking; int wake = 0; - int delayed = 0; int ret; + /* Don't send anything if it's still being created. Return delayed */ + if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) { + spin_unlock(&ci->i_ceph_lock); + dout("%s async create in flight for %p\n", __func__, inode); + return 1; + } + held = cap->issued | cap->implemented; revoking = cap->implemented & ~cap->issued; retain &= ~revoking; @@ -1287,28 +1346,7 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, ceph_cap_string(revoking)); BUG_ON((retain & CEPH_CAP_PIN) == 0); - arg.session = cap->session; - - /* don't release wanted unless we've waited a bit. */ - if ((ci->i_ceph_flags & CEPH_I_NODELAY) == 0 && - time_before(jiffies, ci->i_hold_caps_min)) { - dout(" delaying issued %s -> %s, wanted %s -> %s on send\n", - ceph_cap_string(cap->issued), - ceph_cap_string(cap->issued & retain), - ceph_cap_string(cap->mds_wanted), - ceph_cap_string(want)); - want |= cap->mds_wanted; - retain |= cap->issued; - delayed = 1; - } - ci->i_ceph_flags &= ~(CEPH_I_NODELAY | CEPH_I_FLUSH); - if (want & ~cap->mds_wanted) { - /* user space may open/close single file frequently. - * This avoids droping mds_wanted immediately after - * requesting new mds_wanted. - */ - __cap_set_timeouts(mdsc, ci); - } + ci->i_ceph_flags &= ~CEPH_I_FLUSH; cap->issued &= retain; /* drop bits we don't want */ if (cap->implemented & ~cap->issued) { @@ -1323,6 +1361,7 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, cap->implemented &= cap->issued | used; cap->mds_wanted = want; + arg.session = cap->session; arg.ino = ceph_vino(inode).ino; arg.cid = cap->cap_id; arg.follows = flushing ? ci->i_head_snapc->seq : 0; @@ -1332,7 +1371,8 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, arg.size = inode->i_size; ci->i_reported_size = arg.size; arg.max_size = ci->i_wanted_max_size; - ci->i_requested_max_size = arg.max_size; + if (cap == ci->i_auth_cap) + ci->i_requested_max_size = arg.max_size; if (flushing & CEPH_CAP_XATTR_EXCL) { old_blob = __ceph_build_xattrs_blob(ci); @@ -1383,14 +1423,19 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, ret = send_cap_msg(&arg); if (ret < 0) { - dout("error sending cap msg, must requeue %p\n", inode); - delayed = 1; + pr_err("error sending cap msg, ino (%llx.%llx) " + "flushing %s tid %llu, requeue\n", + ceph_vinop(inode), ceph_cap_string(flushing), + flush_tid); + spin_lock(&ci->i_ceph_lock); + __cap_delay_requeue(mdsc, ci); + spin_unlock(&ci->i_ceph_lock); } if (wake) wake_up_all(&ci->i_cap_wq); - return delayed; + return ret; } static inline int __send_flush_snap(struct inode *inode, @@ -1617,6 +1662,8 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, int was = ci->i_dirty_caps; int dirty = 0; + lockdep_assert_held(&ci->i_ceph_lock); + if (!ci->i_auth_cap) { pr_warn("__mark_dirty_caps %p %llx mask %s, " "but no auth cap (session was closed?)\n", @@ -1654,7 +1701,7 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, if (((was | ci->i_flushing_caps) & CEPH_CAP_FILE_BUFFER) && (mask & CEPH_CAP_FILE_BUFFER)) dirty |= I_DIRTY_DATASYNC; - __cap_delay_requeue(mdsc, ci, true); + __cap_delay_requeue(mdsc, ci); return dirty; } @@ -1726,6 +1773,7 @@ static u64 __mark_caps_flushing(struct inode *inode, struct ceph_cap_flush *cf = NULL; int flushing; + lockdep_assert_held(&ci->i_ceph_lock); BUG_ON(ci->i_dirty_caps == 0); BUG_ON(list_empty(&ci->i_dirty_item)); BUG_ON(!ci->i_prealloc_cap_flush); @@ -1805,8 +1853,6 @@ bool __ceph_should_report_size(struct ceph_inode_info *ci) * versus held caps. Release, flush, ack revoked caps to mds as * appropriate. * - * CHECK_CAPS_NODELAY - caller is delayed work and we should not delay - * cap release further. * CHECK_CAPS_AUTHONLY - we should only check the auth cap * CHECK_CAPS_FLUSH - we should flush any dirty caps immediately, without * further delay. @@ -1825,24 +1871,13 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags, int mds = -1; /* keep track of how far we've gone through i_caps list to avoid an infinite loop on retry */ struct rb_node *p; - int delayed = 0, sent = 0; - bool no_delay = flags & CHECK_CAPS_NODELAY; bool queue_invalidate = false; bool tried_invalidate = false; - /* if we are unmounting, flush any unused caps immediately. */ - if (mdsc->stopping) - no_delay = true; - spin_lock(&ci->i_ceph_lock); - if (ci->i_ceph_flags & CEPH_I_FLUSH) flags |= CHECK_CAPS_FLUSH; - if (!(flags & CHECK_CAPS_AUTHONLY) || - (ci->i_auth_cap && __ceph_is_single_caps(ci))) - __cap_delay_cancel(mdsc, ci); - goto retry_locked; retry: spin_lock(&ci->i_ceph_lock); @@ -1866,10 +1901,11 @@ retry_locked: * revoking the shared cap on every create/unlink * operation. */ - if (IS_RDONLY(inode)) + if (IS_RDONLY(inode)) { want = CEPH_CAP_ANY_SHARED; - else - want = CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_EXCL; + } else { + want |= CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_EXCL; + } retain |= want; } else { @@ -1885,14 +1921,13 @@ retry_locked: } dout("check_caps %p file_want %s used %s dirty %s flushing %s" - " issued %s revoking %s retain %s %s%s%s\n", inode, + " issued %s revoking %s retain %s %s%s\n", inode, ceph_cap_string(file_wanted), ceph_cap_string(used), ceph_cap_string(ci->i_dirty_caps), ceph_cap_string(ci->i_flushing_caps), ceph_cap_string(issued), ceph_cap_string(revoking), ceph_cap_string(retain), (flags & CHECK_CAPS_AUTHONLY) ? " AUTHONLY" : "", - (flags & CHECK_CAPS_NODELAY) ? " NODELAY" : "", (flags & CHECK_CAPS_FLUSH) ? " FLUSH" : ""); /* @@ -1900,8 +1935,8 @@ retry_locked: * have cached pages, but don't want them, then try to invalidate. * If we fail, it's because pages are locked.... try again later. */ - if ((!no_delay || mdsc->stopping) && - !S_ISDIR(inode->i_mode) && /* ignore readdir cache */ + if ((!(flags & CHECK_CAPS_NOINVAL) || mdsc->stopping) && + S_ISREG(inode->i_mode) && !(ci->i_wb_ref || ci->i_wrbuffer_ref) && /* no dirty pages... */ inode->i_data.nrpages && /* have cached pages */ (revoking & (CEPH_CAP_FILE_CACHE| @@ -1973,28 +2008,17 @@ retry_locked: } /* want more caps from mds? */ - if (want & ~(cap->mds_wanted | cap->issued)) - goto ack; + if (want & ~cap->mds_wanted) { + if (want & ~(cap->mds_wanted | cap->issued)) + goto ack; + if (!__cap_is_valid(cap)) + goto ack; + } /* things we might delay */ if ((cap->issued & ~retain) == 0) continue; /* nope, all good */ - if (no_delay) - goto ack; - - /* delay? */ - if ((ci->i_ceph_flags & CEPH_I_NODELAY) == 0 && - time_before(jiffies, ci->i_hold_caps_max)) { - dout(" delaying issued %s -> %s, wanted %s -> %s\n", - ceph_cap_string(cap->issued), - ceph_cap_string(cap->issued & retain), - ceph_cap_string(cap->mds_wanted), - ceph_cap_string(want)); - delayed++; - continue; - } - ack: if (session && session != cap->session) { dout("oops, wrong session %p mutex\n", session); @@ -2055,18 +2079,20 @@ ack: } mds = cap->mds; /* remember mds, so we don't repeat */ - sent++; /* __send_cap drops i_ceph_lock */ - delayed += __send_cap(mdsc, cap, CEPH_CAP_OP_UPDATE, 0, - cap_used, want, retain, flushing, - flush_tid, oldest_flush_tid); + __send_cap(mdsc, cap, CEPH_CAP_OP_UPDATE, 0, cap_used, want, + retain, flushing, flush_tid, oldest_flush_tid); goto retry; /* retake i_ceph_lock and restart our cap scan. */ } - /* Reschedule delayed caps release if we delayed anything */ - if (delayed) - __cap_delay_requeue(mdsc, ci, false); + /* periodically re-calculate caps wanted by open files */ + if (__ceph_is_any_real_caps(ci) && + list_empty(&ci->i_cap_delay_list) && + (file_wanted & ~CEPH_CAP_PIN) && + !(used & (CEPH_CAP_FILE_RD | CEPH_CAP_ANY_FILE_WR))) { + __cap_delay_requeue(mdsc, ci); + } spin_unlock(&ci->i_ceph_lock); @@ -2095,7 +2121,6 @@ retry: retry_locked: if (ci->i_dirty_caps && ci->i_auth_cap) { struct ceph_cap *cap = ci->i_auth_cap; - int delayed; if (session != cap->session) { spin_unlock(&ci->i_ceph_lock); @@ -2124,18 +2149,10 @@ retry_locked: &oldest_flush_tid); /* __send_cap drops i_ceph_lock */ - delayed = __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH, - CEPH_CLIENT_CAPS_SYNC, - __ceph_caps_used(ci), - __ceph_caps_wanted(ci), - (cap->issued | cap->implemented), - flushing, flush_tid, oldest_flush_tid); - - if (delayed) { - spin_lock(&ci->i_ceph_lock); - __cap_delay_requeue(mdsc, ci, true); - spin_unlock(&ci->i_ceph_lock); - } + __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH, CEPH_CLIENT_CAPS_SYNC, + __ceph_caps_used(ci), __ceph_caps_wanted(ci), + (cap->issued | cap->implemented), + flushing, flush_tid, oldest_flush_tid); } else { if (!list_empty(&ci->i_cap_flush_list)) { struct ceph_cap_flush *cf = @@ -2233,6 +2250,10 @@ int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync) if (datasync) goto out; + ret = ceph_wait_on_async_create(inode); + if (ret) + goto out; + dirty = try_flush_caps(inode, &flush_tid); dout("fsync dirty caps are %s\n", ceph_cap_string(dirty)); @@ -2335,22 +2356,13 @@ static void __kick_flushing_caps(struct ceph_mds_client *mdsc, if (cf->caps) { dout("kick_flushing_caps %p cap %p tid %llu %s\n", inode, cap, cf->tid, ceph_cap_string(cf->caps)); - ci->i_ceph_flags |= CEPH_I_NODELAY; - - ret = __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH, + __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH, (cf->tid < last_snap_flush ? CEPH_CLIENT_CAPS_PENDING_CAPSNAP : 0), __ceph_caps_used(ci), __ceph_caps_wanted(ci), (cap->issued | cap->implemented), cf->caps, cf->tid, oldest_flush_tid); - if (ret) { - pr_err("kick_flushing_caps: error sending " - "cap flush, ino (%llx.%llx) " - "tid %llu flushing %s\n", - ceph_vinop(inode), cf->tid, - ceph_cap_string(cf->caps)); - } } else { struct ceph_cap_snap *capsnap = container_of(cf, struct ceph_cap_snap, @@ -2457,16 +2469,15 @@ void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, } } -static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc, - struct ceph_mds_session *session, - struct inode *inode) - __releases(ci->i_ceph_lock) +void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, + struct ceph_inode_info *ci) { - struct ceph_inode_info *ci = ceph_inode(inode); - struct ceph_cap *cap; + struct ceph_mds_client *mdsc = session->s_mdsc; + struct ceph_cap *cap = ci->i_auth_cap; + + lockdep_assert_held(&ci->i_ceph_lock); - cap = ci->i_auth_cap; - dout("kick_flushing_inode_caps %p flushing %s\n", inode, + dout("%s %p flushing %s\n", __func__, &ci->vfs_inode, ceph_cap_string(ci->i_flushing_caps)); if (!list_empty(&ci->i_cap_flush_list)) { @@ -2478,9 +2489,6 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc, spin_unlock(&mdsc->cap_dirty_lock); __kick_flushing_caps(mdsc, session, ci, oldest_flush_tid); - spin_unlock(&ci->i_ceph_lock); - } else { - spin_unlock(&ci->i_ceph_lock); } } @@ -2488,18 +2496,20 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc, /* * Take references to capabilities we hold, so that we don't release * them to the MDS prematurely. - * - * Protected by i_ceph_lock. */ -static void __take_cap_refs(struct ceph_inode_info *ci, int got, +void ceph_take_cap_refs(struct ceph_inode_info *ci, int got, bool snap_rwsem_locked) { + lockdep_assert_held(&ci->i_ceph_lock); + if (got & CEPH_CAP_PIN) ci->i_pin_ref++; if (got & CEPH_CAP_FILE_RD) ci->i_rd_ref++; if (got & CEPH_CAP_FILE_CACHE) ci->i_rdcache_ref++; + if (got & CEPH_CAP_FILE_EXCL) + ci->i_fx_ref++; if (got & CEPH_CAP_FILE_WR) { if (ci->i_wr_ref == 0 && !ci->i_head_snapc) { BUG_ON(!snap_rwsem_locked); @@ -2512,7 +2522,7 @@ static void __take_cap_refs(struct ceph_inode_info *ci, int got, if (ci->i_wb_ref == 0) ihold(&ci->vfs_inode); ci->i_wb_ref++; - dout("__take_cap_refs %p wb %d -> %d (?)\n", + dout("%s %p wb %d -> %d (?)\n", __func__, &ci->vfs_inode, ci->i_wb_ref-1, ci->i_wb_ref); } } @@ -2524,14 +2534,16 @@ static void __take_cap_refs(struct ceph_inode_info *ci, int got, * Note that caller is responsible for ensuring max_size increases are * requested from the MDS. * - * Returns 0 if caps were not able to be acquired (yet), a 1 if they were, - * or a negative error code. - * - * FIXME: how does a 0 return differ from -EAGAIN? + * Returns 0 if caps were not able to be acquired (yet), 1 if succeed, + * or a negative error code. There are 3 speical error codes: + * -EAGAIN: need to sleep but non-blocking is specified + * -EFBIG: ask caller to call check_max_size() and try again. + * -ESTALE: ask caller to call ceph_renew_caps() and try again. */ enum { - NON_BLOCKING = 1, - CHECK_FILELOCK = 2, + /* first 8 bits are reserved for CEPH_FILE_MODE_FOO */ + NON_BLOCKING = (1 << 8), + CHECK_FILELOCK = (1 << 9), }; static int try_get_cap_refs(struct inode *inode, int need, int want, @@ -2541,7 +2553,6 @@ static int try_get_cap_refs(struct inode *inode, int need, int want, struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc; int ret = 0; int have, implemented; - int file_wanted; bool snap_rwsem_locked = false; dout("get_cap_refs %p need %s want %s\n", inode, @@ -2557,15 +2568,6 @@ again: goto out_unlock; } - /* make sure file is actually open */ - file_wanted = __ceph_caps_file_wanted(ci); - if ((file_wanted & need) != need) { - dout("try_get_cap_refs need %s file_wanted %s, EBADF\n", - ceph_cap_string(need), ceph_cap_string(file_wanted)); - ret = -EBADF; - goto out_unlock; - } - /* finish pending truncate */ while (ci->i_truncate_pending) { spin_unlock(&ci->i_ceph_lock); @@ -2584,7 +2586,7 @@ again: dout("get_cap_refs %p endoff %llu > maxsize %llu\n", inode, endoff, ci->i_max_size); if (endoff > ci->i_requested_max_size) - ret = -EAGAIN; + ret = ci->i_auth_cap ? -EFBIG : -ESTALE; goto out_unlock; } /* @@ -2630,51 +2632,55 @@ again: } snap_rwsem_locked = true; } - *got = need | (have & want); - if ((need & CEPH_CAP_FILE_RD) && + if ((have & want) == want) + *got = need | want; + else + *got = need; + if (S_ISREG(inode->i_mode) && + (need & CEPH_CAP_FILE_RD) && !(*got & CEPH_CAP_FILE_CACHE)) ceph_disable_fscache_readpage(ci); - __take_cap_refs(ci, *got, true); + ceph_take_cap_refs(ci, *got, true); ret = 1; } } else { int session_readonly = false; - if ((need & CEPH_CAP_FILE_WR) && ci->i_auth_cap) { + int mds_wanted; + if (ci->i_auth_cap && + (need & (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_EXCL))) { struct ceph_mds_session *s = ci->i_auth_cap->session; spin_lock(&s->s_cap_lock); session_readonly = s->s_readonly; spin_unlock(&s->s_cap_lock); } if (session_readonly) { - dout("get_cap_refs %p needed %s but mds%d readonly\n", + dout("get_cap_refs %p need %s but mds%d readonly\n", inode, ceph_cap_string(need), ci->i_auth_cap->mds); ret = -EROFS; goto out_unlock; } - if (ci->i_ceph_flags & CEPH_I_CAP_DROPPED) { - int mds_wanted; - if (READ_ONCE(mdsc->fsc->mount_state) == - CEPH_MOUNT_SHUTDOWN) { - dout("get_cap_refs %p forced umount\n", inode); - ret = -EIO; - goto out_unlock; - } - mds_wanted = __ceph_caps_mds_wanted(ci, false); - if (need & ~(mds_wanted & need)) { - dout("get_cap_refs %p caps were dropped" - " (session killed?)\n", inode); - ret = -ESTALE; - goto out_unlock; - } - if (!(file_wanted & ~mds_wanted)) - ci->i_ceph_flags &= ~CEPH_I_CAP_DROPPED; + if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) { + dout("get_cap_refs %p forced umount\n", inode); + ret = -EIO; + goto out_unlock; + } + mds_wanted = __ceph_caps_mds_wanted(ci, false); + if (need & ~mds_wanted) { + dout("get_cap_refs %p need %s > mds_wanted %s\n", + inode, ceph_cap_string(need), + ceph_cap_ |