summaryrefslogtreecommitdiff
path: root/fs/ceph
diff options
context:
space:
mode:
Diffstat (limited to 'fs/ceph')
-rw-r--r--fs/ceph/addr.c193
-rw-r--r--fs/ceph/caps.c12
-rw-r--r--fs/ceph/dir.c87
-rw-r--r--fs/ceph/inode.c56
-rw-r--r--fs/ceph/ioctl.c34
-rw-r--r--fs/ceph/ioctl.h55
-rw-r--r--fs/ceph/mds_client.c21
-rw-r--r--fs/ceph/super.c65
-rw-r--r--fs/ceph/super.h42
9 files changed, 375 insertions, 190 deletions
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index 5a3953db8118..4144caf2f9d3 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -228,102 +228,155 @@ static int ceph_readpage(struct file *filp, struct page *page)
}
/*
- * Build a vector of contiguous pages from the provided page list.
+ * Finish an async read(ahead) op.
*/
-static struct page **page_vector_from_list(struct list_head *page_list,
- unsigned *nr_pages)
+static void finish_read(struct ceph_osd_request *req, struct ceph_msg *msg)
{
- struct page **pages;
- struct page *page;
- int next_index, contig_pages = 0;
+ struct inode *inode = req->r_inode;
+ struct ceph_osd_reply_head *replyhead;
+ int rc, bytes;
+ int i;
- /* build page vector */
- pages = kmalloc(sizeof(*pages) * *nr_pages, GFP_NOFS);
- if (!pages)
- return ERR_PTR(-ENOMEM);
+ /* parse reply */
+ replyhead = msg->front.iov_base;
+ WARN_ON(le32_to_cpu(replyhead->num_ops) == 0);
+ rc = le32_to_cpu(replyhead->result);
+ bytes = le32_to_cpu(msg->hdr.data_len);
- BUG_ON(list_empty(page_list));
- next_index = list_entry(page_list->prev, struct page, lru)->index;
- list_for_each_entry_reverse(page, page_list, lru) {
- if (page->index == next_index) {
- dout("readpages page %d %p\n", contig_pages, page);
- pages[contig_pages] = page;
- contig_pages++;
- next_index++;
- } else {
- break;
+ dout("finish_read %p req %p rc %d bytes %d\n", inode, req, rc, bytes);
+
+ /* unlock all pages, zeroing any data we didn't read */
+ for (i = 0; i < req->r_num_pages; i++, bytes -= PAGE_CACHE_SIZE) {
+ struct page *page = req->r_pages[i];
+
+ if (bytes < (int)PAGE_CACHE_SIZE) {
+ /* zero (remainder of) page */
+ int s = bytes < 0 ? 0 : bytes;
+ zero_user_segment(page, s, PAGE_CACHE_SIZE);
}
+ dout("finish_read %p uptodate %p idx %lu\n", inode, page,
+ page->index);
+ flush_dcache_page(page);
+ SetPageUptodate(page);
+ unlock_page(page);
+ page_cache_release(page);
}
- *nr_pages = contig_pages;
- return pages;
+ kfree(req->r_pages);
}
/*
- * Read multiple pages. Leave pages we don't read + unlock in page_list;
- * the caller (VM) cleans them up.
+ * start an async read(ahead) operation. return nr_pages we submitted
+ * a read for on success, or negative error code.
*/
-static int ceph_readpages(struct file *file, struct address_space *mapping,
- struct list_head *page_list, unsigned nr_pages)
+static int start_read(struct inode *inode, struct list_head *page_list, int max)
{
- struct inode *inode = file->f_dentry->d_inode;
- struct ceph_inode_info *ci = ceph_inode(inode);
struct ceph_osd_client *osdc =
&ceph_inode_to_client(inode)->client->osdc;
- int rc = 0;
- struct page **pages;
- loff_t offset;
+ struct ceph_inode_info *ci = ceph_inode(inode);
+ struct page *page = list_entry(page_list->prev, struct page, lru);
+ struct ceph_osd_request *req;
+ u64 off;
u64 len;
+ int i;
+ struct page **pages;
+ pgoff_t next_index;
+ int nr_pages = 0;
+ int ret;
- dout("readpages %p file %p nr_pages %d\n",
- inode, file, nr_pages);
-
- pages = page_vector_from_list(page_list, &nr_pages);
- if (IS_ERR(pages))
- return PTR_ERR(pages);
+ off = page->index << PAGE_CACHE_SHIFT;
- /* guess read extent */
- offset = pages[0]->index << PAGE_CACHE_SHIFT;
+ /* count pages */
+ next_index = page->index;
+ list_for_each_entry_reverse(page, page_list, lru) {
+ if (page->index != next_index)
+ break;
+ nr_pages++;
+ next_index++;
+ if (max && nr_pages == max)
+ break;
+ }
len = nr_pages << PAGE_CACHE_SHIFT;
- rc = ceph_osdc_readpages(osdc, ceph_vino(inode), &ci->i_layout,
- offset, &len,
- ci->i_truncate_seq, ci->i_truncate_size,
- pages, nr_pages, 0);
- if (rc == -ENOENT)
- rc = 0;
- if (rc < 0)
- goto out;
-
- for (; !list_empty(page_list) && len > 0;
- rc -= PAGE_CACHE_SIZE, len -= PAGE_CACHE_SIZE) {
- struct page *page =
- list_entry(page_list->prev, struct page, lru);
+ dout("start_read %p nr_pages %d is %lld~%lld\n", inode, nr_pages,
+ off, len);
+
+ req = ceph_osdc_new_request(osdc, &ci->i_layout, ceph_vino(inode),
+ off, &len,
+ CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ,
+ NULL, 0,
+ ci->i_truncate_seq, ci->i_truncate_size,
+ NULL, false, 1, 0);
+ if (!req)
+ return -ENOMEM;
+ /* build page vector */
+ nr_pages = len >> PAGE_CACHE_SHIFT;
+ pages = kmalloc(sizeof(*pages) * nr_pages, GFP_NOFS);
+ ret = -ENOMEM;
+ if (!pages)
+ goto out;
+ for (i = 0; i < nr_pages; ++i) {
+ page = list_entry(page_list->prev, struct page, lru);
+ BUG_ON(PageLocked(page));
list_del(&page->lru);
-
- if (rc < (int)PAGE_CACHE_SIZE) {
- /* zero (remainder of) page */
- int s = rc < 0 ? 0 : rc;
- zero_user_segment(page, s, PAGE_CACHE_SIZE);
- }
-
- if (add_to_page_cache_lru(page, mapping, page->index,
+
+ dout("start_read %p adding %p idx %lu\n", inode, page,
+ page->index);
+ if (add_to_page_cache_lru(page, &inode->i_data, page->index,
GFP_NOFS)) {
page_cache_release(page);
- dout("readpages %p add_to_page_cache failed %p\n",
+ dout("start_read %p add_to_page_cache failed %p\n",
inode, page);
- continue;
+ nr_pages = i;
+ goto out_pages;
}
- dout("readpages %p adding %p idx %lu\n", inode, page,
- page->index);
- flush_dcache_page(page);
- SetPageUptodate(page);
- unlock_page(page);
- page_cache_release(page);
+ pages[i] = page;
}
- rc = 0;
+ req->r_pages = pages;
+ req->r_num_pages = nr_pages;
+ req->r_callback = finish_read;
+ req->r_inode = inode;
+
+ dout("start_read %p starting %p %lld~%lld\n", inode, req, off, len);
+ ret = ceph_osdc_start_request(osdc, req, false);
+ if (ret < 0)
+ goto out_pages;
+ ceph_osdc_put_request(req);
+ return nr_pages;
+out_pages:
+ ceph_release_page_vector(pages, nr_pages);
+out:
+ ceph_osdc_put_request(req);
+ return ret;
+}
+
+
+/*
+ * Read multiple pages. Leave pages we don't read + unlock in page_list;
+ * the caller (VM) cleans them up.
+ */
+static int ceph_readpages(struct file *file, struct address_space *mapping,
+ struct list_head *page_list, unsigned nr_pages)
+{
+ struct inode *inode = file->f_dentry->d_inode;
+ struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+ int rc = 0;
+ int max = 0;
+
+ if (fsc->mount_options->rsize >= PAGE_CACHE_SIZE)
+ max = (fsc->mount_options->rsize + PAGE_CACHE_SIZE - 1)
+ >> PAGE_SHIFT;
+
+ dout("readpages %p file %p nr_pages %d max %d\n", inode, file, nr_pages,
+ max);
+ while (!list_empty(page_list)) {
+ rc = start_read(inode, page_list, max);
+ if (rc < 0)
+ goto out;
+ BUG_ON(rc == 0);
+ }
out:
- kfree(pages);
+ dout("readpages %p file %p ret %d\n", inode, file, rc);
return rc;
}
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index 8d74ad7ba556..0f327c6c9679 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -487,17 +487,15 @@ static void __check_cap_issue(struct ceph_inode_info *ci, struct ceph_cap *cap,
ci->i_rdcache_gen++;
/*
- * if we are newly issued FILE_SHARED, clear I_COMPLETE; we
+ * if we are newly issued FILE_SHARED, clear D_COMPLETE; we
* don't know what happened to this directory while we didn't
* have the cap.
*/
if ((issued & CEPH_CAP_FILE_SHARED) &&
(had & CEPH_CAP_FILE_SHARED) == 0) {
ci->i_shared_gen++;
- if (S_ISDIR(ci->vfs_inode.i_mode)) {
- dout(" marking %p NOT complete\n", &ci->vfs_inode);
- ci->i_ceph_flags &= ~CEPH_I_COMPLETE;
- }
+ if (S_ISDIR(ci->vfs_inode.i_mode))
+ ceph_dir_clear_complete(&ci->vfs_inode);
}
}
@@ -945,7 +943,7 @@ static int send_cap_msg(struct ceph_mds_session *session,
seq, issue_seq, mseq, follows, size, max_size,
xattr_version, xattrs_buf ? (int)xattrs_buf->vec.iov_len : 0);
- msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, sizeof(*fc), GFP_NOFS);
+ msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, sizeof(*fc), GFP_NOFS, false);
if (!msg)
return -ENOMEM;
@@ -2363,7 +2361,7 @@ static void handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant,
}
if ((issued & CEPH_CAP_LINK_EXCL) == 0)
- inode->i_nlink = le32_to_cpu(grant->nlink);
+ set_nlink(inode, le32_to_cpu(grant->nlink));
if ((issued & CEPH_CAP_XATTR_EXCL) == 0 && grant->xattr_len) {
int len = le32_to_cpu(grant->xattr_len);
diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 382abc9a6a54..2abd0dfad7f8 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -108,7 +108,7 @@ static unsigned fpos_off(loff_t p)
* falling back to a "normal" sync readdir if any dentries in the dir
* are dropped.
*
- * I_COMPLETE tells indicates we have all dentries in the dir. It is
+ * D_COMPLETE tells indicates we have all dentries in the dir. It is
* defined IFF we hold CEPH_CAP_FILE_SHARED (which will be revoked by
* the MDS if/when the directory is modified).
*/
@@ -199,8 +199,8 @@ more:
filp->f_pos++;
/* make sure a dentry wasn't dropped while we didn't have parent lock */
- if (!ceph_i_test(dir, CEPH_I_COMPLETE)) {
- dout(" lost I_COMPLETE on %p; falling back to mds\n", dir);
+ if (!ceph_dir_test_complete(dir)) {
+ dout(" lost D_COMPLETE on %p; falling back to mds\n", dir);
err = -EAGAIN;
goto out;
}
@@ -285,7 +285,7 @@ static int ceph_readdir(struct file *filp, void *dirent, filldir_t filldir)
if ((filp->f_pos == 2 || fi->dentry) &&
!ceph_test_mount_opt(fsc, NOASYNCREADDIR) &&
ceph_snap(inode) != CEPH_SNAPDIR &&
- (ci->i_ceph_flags & CEPH_I_COMPLETE) &&
+ ceph_dir_test_complete(inode) &&
__ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1)) {
spin_unlock(&inode->i_lock);
err = __dcache_readdir(filp, dirent, filldir);
@@ -351,7 +351,7 @@ more:
if (!req->r_did_prepopulate) {
dout("readdir !did_prepopulate");
- fi->dir_release_count--; /* preclude I_COMPLETE */
+ fi->dir_release_count--; /* preclude D_COMPLETE */
}
/* note next offset and last dentry name */
@@ -430,8 +430,7 @@ more:
*/
spin_lock(&inode->i_lock);
if (ci->i_release_count == fi->dir_release_count) {
- dout(" marking %p complete\n", inode);
- /* ci->i_ceph_flags |= CEPH_I_COMPLETE; */
+ ceph_dir_set_complete(inode);
ci->i_max_offset = filp->f_pos;
}
spin_unlock(&inode->i_lock);
@@ -614,7 +613,7 @@ static struct dentry *ceph_lookup(struct inode *dir, struct dentry *dentry,
fsc->mount_options->snapdir_name,
dentry->d_name.len) &&
!is_root_ceph_dentry(dir, dentry) &&
- (ci->i_ceph_flags & CEPH_I_COMPLETE) &&
+ ceph_dir_test_complete(dir) &&
(__ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1))) {
spin_unlock(&dir->i_lock);
dout(" dir %p complete, -ENOENT\n", dir);
@@ -934,7 +933,7 @@ static int ceph_rename(struct inode *old_dir, struct dentry *old_dentry,
*/
/* d_move screws up d_subdirs order */
- ceph_i_clear(new_dir, CEPH_I_COMPLETE);
+ ceph_dir_clear_complete(new_dir);
d_move(old_dentry, new_dentry);
@@ -1092,7 +1091,75 @@ static int ceph_snapdir_d_revalidate(struct dentry *dentry,
return 1;
}
+/*
+ * Set/clear/test dir complete flag on the dir's dentry.
+ */
+static struct dentry * __d_find_any_alias(struct inode *inode)
+{
+ struct dentry *alias;
+
+ if (list_empty(&inode->i_dentry))
+ return NULL;
+ alias = list_first_entry(&inode->i_dentry, struct dentry, d_alias);
+ return alias;
+}
+
+void ceph_dir_set_complete(struct inode *inode)
+{
+ struct dentry *dentry = __d_find_any_alias(inode);
+
+ if (dentry && ceph_dentry(dentry)) {
+ dout(" marking %p (%p) complete\n", inode, dentry);
+ set_bit(CEPH_D_COMPLETE, &ceph_dentry(dentry)->flags);
+ }
+}
+
+void ceph_dir_clear_complete(struct inode *inode)
+{
+ struct dentry *dentry = __d_find_any_alias(inode);
+
+ if (dentry && ceph_dentry(dentry)) {
+ dout(" marking %p (%p) NOT complete\n", inode, dentry);
+ clear_bit(CEPH_D_COMPLETE, &ceph_dentry(dentry)->flags);
+ }
+}
+
+bool ceph_dir_test_complete(struct inode *inode)
+{
+ struct dentry *dentry = __d_find_any_alias(inode);
+
+ if (dentry && ceph_dentry(dentry))
+ return test_bit(CEPH_D_COMPLETE, &ceph_dentry(dentry)->flags);
+ return false;
+}
+
+/*
+ * When the VFS prunes a dentry from the cache, we need to clear the
+ * complete flag on the parent directory.
+ *
+ * Called under dentry->d_lock.
+ */
+static void ceph_d_prune(struct dentry *dentry)
+{
+ struct ceph_dentry_info *di;
+
+ dout("d_release %p\n", dentry);
+
+ /* do we have a valid parent? */
+ if (!dentry->d_parent || IS_ROOT(dentry))
+ return;
+ /* if we are not hashed, we don't affect D_COMPLETE */
+ if (d_unhashed(dentry))
+ return;
+
+ /*
+ * we hold d_lock, so d_parent is stable, and d_fsdata is never
+ * cleared until d_release
+ */
+ di = ceph_dentry(dentry->d_parent);
+ clear_bit(CEPH_D_COMPLETE, &di->flags);
+}
/*
* read() on a dir. This weird interface hack only works if mounted
@@ -1306,6 +1373,7 @@ const struct inode_operations ceph_dir_iops = {
const struct dentry_operations ceph_dentry_ops = {
.d_revalidate = ceph_d_revalidate,
.d_release = ceph_d_release,
+ .d_prune = ceph_d_prune,
};
const struct dentry_operations ceph_snapdir_dentry_ops = {
@@ -1315,4 +1383,5 @@ const struct dentry_operations ceph_snapdir_dentry_ops = {
const struct dentry_operations ceph_snap_dentry_ops = {
.d_release = ceph_d_release,
+ .d_prune = ceph_d_prune,
};
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 095799ba9dd1..e392bfce84a3 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -9,7 +9,6 @@
#include <linux/namei.h>
#include <linux/writeback.h>
#include <linux/vmalloc.h>
-#include <linux/pagevec.h>
#include "super.h"
#include "mds_client.h"
@@ -619,7 +618,7 @@ static int fill_inode(struct inode *inode,
}
if ((issued & CEPH_CAP_LINK_EXCL) == 0)
- inode->i_nlink = le32_to_cpu(info->nlink);
+ set_nlink(inode, le32_to_cpu(info->nlink));
/* be careful with mtime, atime, size */
ceph_decode_timespec(&atime, &info->atime);
@@ -772,9 +771,9 @@ no_change:
ceph_snap(inode) == CEPH_NOSNAP &&
(le32_to_cpu(info->cap.caps) & CEPH_CAP_FILE_SHARED) &&
(issued & CEPH_CAP_FILE_EXCL) == 0 &&
- (ci->i_ceph_flags & CEPH_I_COMPLETE) == 0) {
+ !ceph_dir_test_complete(inode)) {
dout(" marking %p complete (empty)\n", inode);
- /* ci->i_ceph_flags |= CEPH_I_COMPLETE; */
+ ceph_dir_set_complete(inode);
ci->i_max_offset = 2;
}
@@ -857,7 +856,7 @@ static void ceph_set_dentry_offset(struct dentry *dn)
di = ceph_dentry(dn);
spin_lock(&inode->i_lock);
- if ((ceph_inode(inode)->i_ceph_flags & CEPH_I_COMPLETE) == 0) {
+ if (!ceph_dir_test_complete(inode)) {
spin_unlock(&inode->i_lock);
return;
}
@@ -1057,7 +1056,7 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req,
* d_move() puts the renamed dentry at the end of
* d_subdirs. We need to assign it an appropriate
* directory offset so we can behave when holding
- * I_COMPLETE.
+ * D_COMPLETE.
*/
ceph_set_dentry_offset(req->r_old_dentry);
dout("dn %p gets new offset %lld\n", req->r_old_dentry,
@@ -1364,49 +1363,6 @@ void ceph_queue_invalidate(struct inode *inode)
}
/*
- * invalidate any pages that are not dirty or under writeback. this
- * includes pages that are clean and mapped.
- */
-static void ceph_invalidate_nondirty_pages(struct address_space *mapping)
-{
- struct pagevec pvec;
- pgoff_t next = 0;
- int i;
-
- pagevec_init(&pvec, 0);
- while (pagevec_lookup(&pvec, mapping, next, PAGEVEC_SIZE)) {
- for (i = 0; i < pagevec_count(&pvec); i++) {
- struct page *page = pvec.pages[i];
- pgoff_t index;
- int skip_page =
- (PageDirty(page) || PageWriteback(page));
-
- if (!skip_page)
- skip_page = !trylock_page(page);
-
- /*
- * We really shouldn't be looking at the ->index of an
- * unlocked page. But we're not allowed to lock these
- * pages. So we rely upon nobody altering the ->index
- * of this (pinned-by-us) page.
- */
- index = page->index;
- if (index > next)
- next = index;
- next++;
-
- if (skip_page)
- continue;
-
- generic_error_remove_page(mapping, page);
- unlock_page(page);
- }
- pagevec_release(&pvec);
- cond_resched();
- }
-}
-
-/*
* Invalidate inode pages in a worker thread. (This can't be done
* in the message handler context.)
*/
@@ -1429,7 +1385,7 @@ static void ceph_invalidate_work(struct work_struct *work)
orig_gen = ci->i_rdcache_gen;
spin_unlock(&inode->i_lock);
- ceph_invalidate_nondirty_pages(inode->i_mapping);
+ truncate_inode_pages(&inode->i_data, 0);
spin_lock(&inode->i_lock);
if (orig_gen == ci->i_rdcache_gen &&
diff --git a/fs/ceph/ioctl.c b/fs/ceph/ioctl.c
index 3b256b50f7d8..5a14c29cbba6 100644
--- a/fs/ceph/ioctl.c
+++ b/fs/ceph/ioctl.c
@@ -42,17 +42,39 @@ static long ceph_ioctl_set_layout(struct file *file, void __user *arg)
struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
struct ceph_mds_request *req;
struct ceph_ioctl_layout l;
+ struct ceph_inode_info *ci = ceph_inode(file->f_dentry->d_inode);
+ struct ceph_ioctl_layout nl;
int err, i;
- /* copy and validate */
if (copy_from_user(&l, arg, sizeof(l)))
return -EFAULT;
- if ((l.object_size & ~PAGE_MASK) ||
- (l.stripe_unit & ~PAGE_MASK) ||
- !l.stripe_unit ||
- (l.object_size &&
- (unsigned)l.object_size % (unsigned)l.stripe_unit))
+ /* validate changed params against current layout */
+ err = ceph_do_getattr(file->f_dentry->d_inode, CEPH_STAT_CAP_LAYOUT);
+ if (!err) {
+ nl.stripe_unit = ceph_file_layout_su(ci->i_layout);
+ nl.stripe_count = ceph_file_layout_stripe_count(ci->i_layout);
+ nl.object_size = ceph_file_layout_object_size(ci->i_layout);
+ nl.data_pool = le32_to_cpu(ci->i_layout.fl_pg_pool);
+ nl.preferred_osd =
+ (s32)le32_to_cpu(ci->i_layout.fl_pg_preferred);
+ } else
+ return err;
+
+ if (l.stripe_count)
+ nl.stripe_count = l.stripe_count;
+ if (l.stripe_unit)
+ nl.stripe_unit = l.stripe_unit;
+ if (l.object_size)
+ nl.object_size = l.object_size;
+ if (l.data_pool)
+ nl.data_pool = l.data_pool;
+ if (l.preferred_osd)
+ nl.preferred_osd = l.preferred_osd;
+
+ if ((nl.object_size & ~PAGE_MASK) ||
+ (nl.stripe_unit & ~PAGE_MASK) ||
+ ((unsigned)nl.object_size % (unsigned)nl.stripe_unit))
return -EINVAL;
/* make sure it's a valid data pool */
diff --git a/fs/ceph/ioctl.h b/fs/ceph/ioctl.h
index 0c5167e43180..be4a60487333 100644
--- a/fs/ceph/ioctl.h
+++ b/fs/ceph/ioctl.h
@@ -6,7 +6,31 @@
#define CEPH_IOCTL_MAGIC 0x97
-/* just use u64 to align sanely on all archs */
+/*
+ * CEPH_IOC_GET_LAYOUT - get file layout or dir layout policy
+ * CEPH_IOC_SET_LAYOUT - set file layout
+ * CEPH_IOC_SET_LAYOUT_POLICY - set dir layout policy
+ *
+ * The file layout specifies how file data is striped over objects in
+ * the distributed object store, which object pool they belong to (if
+ * it differs from the default), and an optional 'preferred osd' to
+ * store them on.
+ *
+ * Files get a new layout based on the policy set on the containing
+ * directory or one of its ancestors. The GET_LAYOUT ioctl will let
+ * you examine the layout for a file or the policy on a directory.
+ *
+ * SET_LAYOUT will let you set a layout on a newly created file. This
+ * only works immediately after the file is created and before any
+ * data is written to it.
+ *
+ * SET_LAYOUT_POLICY will let you set a layout policy (default layout)
+ * on a directory that will apply to any new files created in that
+ * directory (or any child directory that doesn't specify a layout of
+ * its own).
+ */
+
+/* use u64 to align sanely on all archs */
struct ceph_ioctl_layout {
__u64 stripe_unit, stripe_count, object_size;
__u64 data_pool;
@@ -21,6 +45,8 @@ struct ceph_ioctl_layout {
struct ceph_ioctl_layout)
/*
+ * CEPH_IOC_GET_DATALOC - get location of file data in the cluster
+ *
* Extract identity, address of the OSD and object storing a given
* file offset.
*/
@@ -39,7 +65,34 @@ struct ceph_ioctl_dataloc {
#define CEPH_IOC_GET_DATALOC _IOWR(CEPH_IOCTL_MAGIC, 3, \
struct ceph_ioctl_dataloc)
+/*
+ * CEPH_IOC_LAZYIO - relax consistency
+ *
+ * Normally Ceph switches to synchronous IO when multiple clients have
+ * the file open (and or more for write). Reads and writes bypass the
+ * page cache and go directly to the OSD. Setting this flag on a file
+ * descriptor will allow buffered IO for this file in cases where the
+ * application knows it won't interfere with other nodes (or doesn't
+ * care).
+ */
#define CEPH_IOC_LAZYIO _IO(CEPH_IOCTL_MAGIC, 4)
+
+/*
+ * CEPH_IOC_SYNCIO - force synchronous IO
+ *
+ * This ioctl sets a file flag that forces the synchronous IO that
+ * bypasses the page cache, even if it is not necessary. This is
+ * essentially the opposite behavior of IOC_LAZYIO. This forces the
+ * same read/write path as a file opened by multiple clients when one
+ * or more of those clients is opened for write.
+ *
+ * Note that this type of sync IO takes a different path than a file
+ * opened with O_SYNC/D_SYNC (writes hit the page cache and are
+ * immediately flushed on page boundaries). It is very similar to
+ * O_DIRECT (writes bypass the page cache) excep that O_DIRECT writes
+ * are not copied (user page must remain stable) and O_DIRECT writes
+ * have alignment restrictions (on the buffer and file offset).
+ */
#define CEPH_IOC_SYNCIO _IO(CEPH_IOCTL_MAGIC, 5)
#endif
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 86c59e16ba74..264ab701154f 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -619,7 +619,7 @@ static void __unregister_request(struct ceph_mds_client *mdsc,
*
* Called under mdsc->mutex.
*/
-struct dentry *get_nonsnap_parent(struct dentry *dentry)
+static struct dentry *get_nonsnap_parent(struct dentry *dentry)
{
/*
* we don't need to worry about protecting the d_parent access
@@ -764,7 +764,8 @@ static struct ceph_msg *create_session_msg(u32 op, u64 seq)
struct ceph_msg *msg;
struct ceph_mds_session_head *h;
- msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h), GFP_NOFS);
+ msg = ceph_msg_new(CEPH_MSG_CLIENT_SESSION, sizeof(*h), GFP_NOFS,
+ false);
if (!msg) {
pr_err("create_session_msg ENOMEM creating msg\n");
return NULL;
@@ -1240,7 +1241,7 @@ int ceph_add_cap_releases(struct ceph_mds_client *mdsc,
while (session->s_num_cap_releases < session->s_nr_caps + extra) {
spin_unlock(&session->s_cap_lock);
msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPRELEASE, PAGE_CACHE_SIZE,
- GFP_NOFS);
+ GFP_NOFS, false);
if (!msg)
goto out_unlocked;
dout("add_cap_releases %p msg %p now %d\n", session, msg,
@@ -1652,7 +1653,7 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc,
if (req->r_old_dentry_drop)
len += req->r_old_dentry->d_name.len;
- msg = ceph_msg_new(CEPH_MSG_CLIENT_REQUEST, len, GFP_NOFS);
+ msg = ceph_msg_new(CEPH_MSG_CLIENT_REQUEST, len, GFP_NOFS, false);
if (!msg) {
msg = ERR_PTR(-ENOMEM);
goto out_free2;
@@ -2001,7 +2002,7 @@ out:
}
/*
- * Invalidate dir I_COMPLETE, dentry lease state on an aborted MDS
+ * Invalidate dir D_COMPLETE, dentry lease state on an aborted MDS
* namespace request.
*/
void ceph_invalidate_dir_request(struct ceph_mds_request *req)
@@ -2009,9 +2010,9 @@ void ceph_invalidate_dir_request(struct ceph_mds_request *req)
struct inode *inode = req->r_locked_dir;
struct ceph_inode_info *ci = ceph_inode(inode);
- dout("invalidate_dir_request %p (I_COMPLETE, lease(s))\n", inode);
+ dout("invalidate_dir_request %p (D_COMPLETE, lease(s))\n", inode);
spin_lock(&inode->i_lock);
- ci->i_ceph_flags &= ~CEPH_I_COMPLETE;
+ ceph_dir_clear_complete(inode);
ci->i_release_count++;
spin_unlock(&inode->i_lock);
@@ -2518,7 +2519,7 @@ static void send_mds_reconnect(struct ceph_mds_client *mdsc,
goto fail_nopagelist;
ceph_pagelist_init(pagelist);
- reply = ceph_msg_new(CEPH_MSG_CLIENT_RECONNECT, 0, GFP_NOFS);
+ reply = ceph_msg_new(CEPH_MSG_CLIENT_RECONNECT, 0, GFP_NOFS, false);
if (!reply)
goto fail_nomsg;
@@ -2831,7 +2832,7 @@ void ceph_mdsc_lease_send_msg(struct ceph_mds_session *session,
dnamelen = dentry->d_name.len;
len += dnamelen;
- msg = ceph_msg_new(CEPH_MSG_CLIENT_LEASE, len, GFP_NOFS);
+ msg = ceph_msg_new(CEPH_MSG_CLIENT_LEASE, len, GFP_NOFS, false);
if (!msg)
return;
lease = msg->front.iov_base;
@@ -3153,7 +3154,7 @@ void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
/*
* true if all sessions are closed, or we force unmount
*/
-bool done_closing_sessions(struct ceph_mds_client *mdsc)
+static bool done_closing_sessions(struct ceph_mds_client *mdsc)
{
int i, n = 0;
diff --git a/fs/ceph/super.c b/fs/ceph/super.c
index 88bacaf385d9..a90846fac759 100644
--- a/fs/ceph/super.c
+++ b/fs/ceph/super.c
@@ -114,6 +114,7 @@ static int ceph_sync_fs(struct super_block *sb, int wait)
enum {
Opt_wsize,
Opt_rsize,
+ Opt_rasize,
Opt_caps_wanted_delay_min,
Opt_caps_wanted_delay_max,
Opt_cap_release_safety,
@@ -136,6 +137,7 @@ enum {
static match_table_t fsopt_tokens = {
{Opt_wsize, "wsize=%d"},
{Opt_rsize, "rsize=%d"},
+ {Opt_rasize, "rasize=%d"},
{Opt_caps_wanted_delay_min, "caps_wanted_delay_min=%d"},
{Opt_caps_wanted_delay_max, "caps_wanted_delay_max=%d"},
{Opt_cap_release_safety, "cap_release_safety=%d"},
@@ -196,6 +198,9 @@ static int parse_fsopt_token(char *c, void *private)
case Opt_rsize:
fsopt->rsize = intval;
break;
+ case Opt_rasize:
+ fsopt->rasize = intval;
+ break;
case Opt_caps_wanted_delay_min:
fsopt->caps_wanted_delay_min = intval;
break;
@@ -289,28 +294,29 @@ static int parse_mount_options(struct ceph_mount_options **pfsopt,
dout("parse_mount_options %p, dev_name '%s'\n", fsopt, dev_name);
- fsopt->sb_flags = flags;
- fsopt->flags = CEPH_MOUNT_OPT_DEFAULT;
+ fsopt->sb_flags = flags;
+ fsopt->flags = CEPH_MOUNT_OPT_DEFAULT;
- fsopt->rsize = CEPH_RSIZE_DEFAULT;
- fsopt->snapdir_name = kstrdup(CEPH_SNAPDIRNAME_DEFAULT, GFP_KERNEL);
+ fsopt->rsize = CEPH_RSIZE_DEFAULT;
+ fsopt->rasize = CEPH_RASIZE_DEFAULT;
+ fsopt->snapdir_name = kstrdup(CEPH_SNAPDIRNAME_DEFAULT, GFP_KERNEL);
fsopt->caps_wanted_delay_min = CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT;
fsopt->caps_wanted_delay_max = CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT;
- fsopt->cap_release_safety = CEPH_CAP_RELEASE_SAFETY_DEFAULT;
- fsopt->max_readdir = CEPH_MAX_READDIR_DEFAULT;
- fsopt->max_readdir_bytes = CEPH_MAX_READDIR_BYTES_DEFAULT;
- fsopt->congestion_kb = default_congestion_kb();
-
- /* ip1[:port1][,ip2[:port2]...]:/subdir/in/fs */
- err = -EINVAL;
- if (!dev_name)
- goto out;
- *path = strstr(dev_name, ":/");
- if (*path == NULL) {
- pr_err("device name is missing path (no :/ in %s)\n",
- dev_name);
- goto out;
- }
+ fsopt->cap_release_safety = CEPH_CAP_RELEASE_SAFETY_DEFAULT;
+ fsopt->max_readdir = CEPH_MAX_READDIR_DEFAULT;
+ fsopt->max_readdir_bytes = CEPH_MAX_READDIR_BYTES_DEFAULT;
+ fsopt->congestion_kb = default_congestion_kb();
+
+ /* ip1[:port1][,ip2[:port2]...]:/subdir/in/fs */
+ err = -EINVAL;
+ if (!dev_name)
+ goto out;
+ *path = strstr(dev_name, ":/");
+ if (*path == NULL) {
+ pr_err("device name is missing path (no :/ in %s)\n",
+ dev_name);
+ goto out;
+ }
dev_name_end = *path;
dout("device name '%.*s'\n", (int)(dev_name_end - dev_name), dev_name);
@@ -376,6 +382,8 @@ static int ceph_show_options(struct seq_file *m, struct vfsmount *mnt)
seq_printf(m, ",wsize=%d", fsopt->wsize);
if (fsopt->rsize != CEPH_RSIZE_DEFAULT)
seq_printf(m, ",rsize=%d", fsopt->rsize);
+ if (fsopt->rasize != CEPH_RASIZE_DEFAULT)
+ seq_printf(m, ",rasize=%d", fsopt->rsize);
if (fsopt->congestion_kb != default_congestion_kb())
seq_printf(m, ",write_congestion_kb=%d", fsopt->congestion_kb);
if (fsopt->caps_wanted_delay_min != CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT)
@@ -418,24 +426,27 @@ static int extra_mon_dispatch(struct ceph_client *client, struct ceph_msg *msg)
/*
* create a new fs client
*/
-struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt,
+static struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt,
struct ceph_options *opt)
{
struct ceph_fs_client *fsc;
+ const unsigned supported_features =
+ CEPH_FEATURE_FLOCK |
+ CEPH_FEATURE_DIRLAYOUTHASH;
+ const unsigned required_features = 0;
int err = -ENOMEM;
fsc = kzalloc(sizeof(*fsc), GFP_KERNEL);
if (!fsc)
return ERR_PTR(-ENOMEM);
- fsc->client = ceph_create_client(opt, fsc);
+ fsc->client = ceph_create_client(opt, fsc, supported_features,
+ required_features);
if (IS_ERR(fsc->client)) {
err = PTR_ERR(fsc->client);
goto fail;
}
fsc->client->extra_mon_dispatch = extra_mon_dispatch;
- fsc->client->supported_features |= CEPH_FEATURE_FLOCK |
- CEPH_FEATURE_DIRLAYOUTHASH;
fsc->client->monc.want_mdsmap = 1;
fsc->mount_options = fsopt;
@@ -491,7 +502,7 @@ fail:
return ERR_PTR(err);
}
-void destroy_fs_client(struct ceph_fs_client *fsc)
+static void destroy_fs_client(struct ceph_fs_client *fsc)
{
dout("destroy_fs_client %p\n", fsc);
@@ -774,10 +785,10 @@ static int ceph_register_bdi(struct super_block *sb,
{
int err;
- /* set ra_pages based on rsize mount option? */
- if (fsc->mount_options->rsize >= PAGE_CACHE_SIZE)
+ /* set ra_pages based on rasize mount option? */
+ if (fsc->mount_options->rasize >= PAGE_CACHE_SIZE)
fsc->backing_dev_info.ra_pages =
- (fsc->mount_options->rsize + PAGE_CACHE_SIZE - 1)
+ (fsc->mount_options->rasize + PAGE_CACHE_SIZE - 1)
>> PAGE_SHIFT;
else
fsc->backing_dev_info.ra_pages =
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index a23eed526f05..01bf189e08a9 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -36,7 +36,8 @@
#define ceph_test_mount_opt(fsc, opt) \
(!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
-#define CEPH_RSIZE_DEFAULT (512*1024) /* readahead */
+#define CEPH_RSIZE_DEFAULT 0 /* max read size */
+#define CEPH_RASIZE_DEFAULT (8192*1024) /* readahead */
#define CEPH_MAX_READDIR_DEFAULT 1024
#define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024)
#define CEPH_SNAPDIRNAME_DEFAULT ".snap"
@@ -45,8 +46,9 @@ struct ceph_mount_options {
int flags;
int sb_flags;
- int wsize;
- int rsize; /* max readahead */
+ int wsize; /* max write size */
+ int rsize; /* max read size */
+ int rasize; /* max readahead */
int congestion_kb; /* max writeback in flight */
int caps_wanted_delay_min, caps_wanted_delay_max;
int cap_release_safety;
@@ -201,6 +203,7 @@ struct ceph_inode_xattr {
* Ceph dentry state
*/
struct ceph_dentry_info {
+ unsigned long flags;
struct ceph_mds_session *lease_session;
u32 lease_gen, lease_shared_gen;
u32 lease_seq;
@@ -211,6 +214,18 @@ struct ceph_dentry_info {
u64 offset;
};
+/*
+ * dentry flags
+ *
+ * The locking for D_COMPLETE is a bit odd:
+ * - we can clear it at almost any time (see ceph_d_prune)
+ * - it is only meaningful if:
+ * - we hold dir inode i_lock
+ * - we hold dir FILE_SHARED caps
+ * - the dentry D_COMPLETE is set
+ */
+#define CEPH_D_COMPLETE 1 /* if set, d_u.d_subdirs is complete directory */
+
struct ceph_inode_xattrs_info {
/*
* (still encoded) xattr blob. we avoid the overhead of parsing
@@ -249,7 +264,7 @@ struct ceph_inode_info {
struct timespec i_rctime;
u64 i_rbytes, i_rfiles, i_rsubdirs;
u64 i_files, i_subdirs;
- u64 i_max_offset; /* largest readdir offset, set with I_COMPLETE */
+ u64 i_max_offset; /* largest readdir offset, set with D_COMPLETE */
struct rb_root i_fragtree;
struct mutex i_fragtree_mutex;
@@ -344,9 +359,10 @@ static inline struct ceph_vino ceph_vino(struct inode *inode)
* x86_64+ino32 64 32
* x86_64 64 64
*/
-static inline u32 ceph_ino_to_ino32(ino_t ino)
+static inline u32 ceph_ino_to_ino32(__u64 vino)
{
- ino ^= ino >> (sizeof(ino) * 8 - 32);
+ u32 ino = vino & 0xffffffff;
+ ino ^= vino >> 32;
if (!ino)
ino = 1;
return ino;
@@ -357,11 +373,11 @@ static inline u32 ceph_ino_to_ino32(ino_t ino)
*/
static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
{
- ino_t ino = (ino_t)vino.ino; /* ^ (vino.snap << 20); */
#if BITS_PER_LONG == 32
- ino = ceph_ino_to_ino32(ino);
+ return ceph_ino_to_ino32(vino.ino);
+#else
+ return (ino_t)vino.ino;
#endif
- return ino;
}
/*
@@ -413,7 +429,6 @@ static inline struct inode *ceph_find_inode(struct super_block *sb,
/*
* Ceph inode.
*/
-#define CEPH_I_COMPLETE 1 /* we have complete directory cached */
#define CEPH_I_NODELAY 4 /* do not delay cap release */
#define CEPH_I_FLUSH 8 /* do not delay flush of dirty metadata */
#define CEPH_I_NOFLUSH 16 /* do not flush dirty caps */
@@ -471,6 +486,13 @@ static inline loff_t ceph_make_fpos(unsigned frag, unsigned off)
}
/*
+ * set/clear directory D_COMPLETE flag
+ */
+void ceph_dir_set_complete(struct inode *inode);
+void ceph_dir_clear_complete(struct inode *inode);
+bool ceph_dir_test_complete(struct inode *inode);
+
+/*
* caps helpers
*/
static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
generated by cgit v1.2.3 (git 2.39.0) at 2025-02-02 13:09:58 +0300