Message-Id: <1226624178-3761-8-git-send-email-sage@newdream.net>
Date:	Thu, 13 Nov 2008 16:56:06 -0800
From:	Sage Weil <sage@...dream.net>
To:	linux-fsdevel@...r.kernel.org
Cc:	linux-kernel@...r.kernel.org, Sage Weil <sage@...dream.net>
Subject: [PATCH 07/19] ceph: file operations

File open and close operations, and read and write methods that ensure
we have obtained the proper capabilities from the MDS cluster before
performing IO on a file.  We take references on held capabilities for
the duration of the read/write to avoid prematurely releasing them
back to the MDS.

Signed-off-by: Sage Weil <sage@...dream.net>
---
 fs/ceph/file.c |  446 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 446 insertions(+), 0 deletions(-)
 create mode 100644 fs/ceph/file.c

diff --git a/fs/ceph/file.c b/fs/ceph/file.c
new file mode 100644
index 0000000..19e93f2
--- /dev/null
+++ b/fs/ceph/file.c
@@ -0,0 +1,446 @@
+
+#include <linux/sched.h>
+#include <linux/file.h>
+
+#include "ceph_debug.h"
+
+int ceph_debug_file = -1;
+#define DOUT_MASK DOUT_MASK_FILE
+#define DOUT_VAR ceph_debug_file
+#define DOUT_PREFIX "file: "
+#include "super.h"
+
+#include "mds_client.h"
+
+#include <linux/namei.h>
+
+
+/*
+ * Prepare an open request.  Preallocate ceph_cap to avoid an
+ * inopportune ENOMEM later.
+ */
+static struct ceph_mds_request *
+prepare_open_request(struct super_block *sb, struct dentry *dentry,
+		     int flags, int create_mode)
+{
+	struct ceph_client *client = ceph_sb_to_client(sb);
+	struct ceph_mds_client *mdsc = &client->mdsc;
+	u64 pathbase;
+	char *path;
+	int pathlen;
+	struct ceph_mds_request *req;
+	struct ceph_mds_request_head *rhead;
+	int want_auth = USE_ANY_MDS;
+
+	if (flags & (O_WRONLY|O_RDWR|O_CREAT|O_TRUNC))
+		want_auth = USE_AUTH_MDS;
+
+	dout(5, "prepare_open_request dentry %p name '%s' flags %d\n", dentry,
+	     dentry->d_name.name, flags);
+	path = ceph_build_path(dentry, &pathlen, &pathbase, 0);
+	if (IS_ERR(path))
+		return ERR_PTR(PTR_ERR(path));
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_OPEN, pathbase, path,
+				       0, NULL,
+				       dentry, want_auth);
+	kfree(path);
+	if (IS_ERR(req))
+		goto out;
+	req->r_expected_cap = kmalloc(sizeof(struct ceph_cap), GFP_NOFS);
+	if (!req->r_expected_cap) {
+		ceph_mdsc_put_request(req);
+		return ERR_PTR(-ENOMEM);
+	}
+	req->r_fmode = ceph_flags_to_mode(flags);
+
+	rhead = req->r_request->front.iov_base;
+	rhead->args.open.flags = cpu_to_le32(flags);
+	rhead->args.open.mode = cpu_to_le32(create_mode);
+out:
+	return req;
+}
+
+/*
+ * initialize private struct file data.
+ * if we fail, clean up by dropping fmode reference on the ceph_inode
+ */
+static int ceph_init_file(struct inode *inode, struct file *file, int fmode)
+{
+	struct ceph_file_info *cf;
+
+	cf = kzalloc(sizeof(*cf), GFP_NOFS);
+	if (cf == NULL) {
+		ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
+		return -ENOMEM;
+	}
+	cf->mode = fmode;
+	file->private_data = cf;
+	return 0;
+}
+
+/*
+ * If the filp already has private_data, that means the file was
+ * already opened by intent during lookup, and we do nothing.
+ *
+ * If we already have the requisite capabilities, we can satisfy
+ * the open request locally (no need to request new caps from the
+ * MDS).
+ */
+int ceph_open(struct inode *inode, struct file *file)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_client *client = ceph_sb_to_client(inode->i_sb);
+	struct ceph_mds_client *mdsc = &client->mdsc;
+	struct dentry *dentry;
+	struct ceph_mds_request *req;
+	struct ceph_file_info *cf = file->private_data;
+	int err;
+	int flags, fmode, wantcaps;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP && (file->f_mode & FMODE_WRITE))
+		return -EROFS;
+
+	if (cf) {
+		dout(5, "open file %p is already opened\n", file);
+		return 0;
+	}
+
+	/* filter out O_CREAT|O_EXCL; vfs did that already.  yuck. */
+	flags = file->f_flags & ~(O_CREAT|O_EXCL);
+	if (S_ISDIR(inode->i_mode))
+		flags = O_DIRECTORY;  /* mds likes to know */
+
+	dout(5, "open inode %p ino %llx.%llx file %p flags %d (%d)\n", inode,
+	     ceph_vinop(inode), file, flags, file->f_flags);
+	fmode = ceph_flags_to_mode(flags);
+	wantcaps = ceph_caps_for_mode(fmode);
+
+	/* can we re-use existing caps? */
+	spin_lock(&inode->i_lock);
+	if ((__ceph_caps_issued(ci, NULL) & wantcaps) == wantcaps) {
+		dout(10, "open fmode %d caps %d using existing on %p\n",
+		     fmode, wantcaps, inode);
+		__ceph_get_fmode(ci, fmode);
+		spin_unlock(&inode->i_lock);
+		return ceph_init_file(inode, file, fmode);
+	}
+	spin_unlock(&inode->i_lock);
+	dout(10, "open fmode %d, don't have caps %d\n", fmode, wantcaps);
+
+	dentry = d_find_alias(inode);
+	if (!dentry)
+		return -ESTALE;  /* yuck */
+	ceph_mdsc_lease_release(mdsc, inode, NULL, CEPH_LOCK_ICONTENT);
+	req = prepare_open_request(inode->i_sb, dentry, flags, 0);
+	if (IS_ERR(req)) {
+		err = PTR_ERR(req);
+		goto out;
+	}
+	err = ceph_mdsc_do_request(mdsc, req);
+	if (!err)
+		err = ceph_init_file(inode, file, req->r_fmode);
+	ceph_mdsc_put_request(req);
+	dout(5, "open result=%d on %llx.%llx\n", err, ceph_vinop(inode));
+out:
+	dput(dentry);
+	return err;
+}
+
+
+/*
+ * Do a lookup + open with a single request.
+ *
+ * If this succeeds, but some subsequent check in the vfs
+ * may_open() fails, the struct *file gets cleaned up (i.e.
+ * ceph_release gets called).  So fear not!
+ */
+/*
+ * flags
+ *  path_lookup_open   -> LOOKUP_OPEN
+ *  path_lookup_create -> LOOKUP_OPEN|LOOKUP_CREATE
+ */
+struct dentry *ceph_lookup_open(struct inode *dir, struct dentry *dentry,
+				struct nameidata *nd, int mode,
+				int locked_dir)
+{
+	struct ceph_client *client = ceph_sb_to_client(dir->i_sb);
+	struct ceph_mds_client *mdsc = &client->mdsc;
+	struct file *file = nd->intent.open.file;
+	struct ceph_mds_request *req;
+	int err;
+	int flags = nd->intent.open.flags - 1;  /* silly vfs! */
+
+	dout(5, "ceph_lookup_open dentry %p '%.*s' flags %d mode 0%o\n",
+	     dentry, dentry->d_name.len, dentry->d_name.name, flags, mode);
+
+	/* do the open */
+	req = prepare_open_request(dir->i_sb, dentry, flags, mode);
+	if (IS_ERR(req))
+		return ERR_PTR(PTR_ERR(req));
+	if (flags & O_CREAT)
+		ceph_mdsc_lease_release(mdsc, dir, NULL, CEPH_LOCK_ICONTENT);
+	req->r_last_dentry = dget(dentry); /* use this dentry in fill_trace */
+	req->r_locked_dir = dir;           /* caller holds dir->i_mutex */
+	err = ceph_mdsc_do_request(mdsc, req);
+	dentry = ceph_finish_lookup(req, dentry, err);
+	if (!err)
+		err = ceph_init_file(req->r_last_inode, file, req->r_fmode);
+	ceph_mdsc_put_request(req);
+	dout(5, "ceph_lookup_open result=%p\n", dentry);
+	return dentry;
+}
+
+int ceph_release(struct inode *inode, struct file *file)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_file_info *cf = file->private_data;
+
+	dout(5, "release inode %p file %p\n", inode, file);
+
+	/*
+	 * MYSTERY: why is file->f_flags now different than
+	 * file->f_flags (actually, nd->intent.open.flags) on
+	 * open?  e.g., on ceph_lookup_open,
+	 *   ceph_file: opened 000000006fa3ebd0 flags 0101102 mode 2 nr now 1.\
+	 *  wanted 0 -> 30
+	 * and on release,
+	 *   ceph_file: released 000000006fa3ebd0 flags 0100001 mode 3 nr now \
+	 * -1.  wanted 30 was 30
+	 *
+	 * Huh.  Well, for now, store the open mode in ceph_file_info.
+	 */
+
+	ceph_put_fmode(ci, cf->mode);
+	if (cf->last_readdir)
+		ceph_mdsc_put_request(cf->last_readdir);
+	kfree(cf->dir_info);
+	kfree(cf);
+	return 0;
+}
+
+/*
+ * Completely synchronous read and write methods.  Direct from __user
+ * buffer to osd.
+ */
+static ssize_t ceph_sync_read(struct file *file, char __user *data,
+			       size_t count, loff_t *offset)
+{
+	struct inode *inode = file->f_dentry->d_inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_client *client = ceph_inode_to_client(inode);
+	int ret = 0;
+	off_t pos = *offset;
+
+	dout(10, "sync_read on file %p %lld~%u\n", file, *offset,
+	     (unsigned)count);
+
+	ret = ceph_osdc_sync_read(&client->osdc, ceph_vino(inode),
+				  &ci->i_layout,
+				  pos, count, data);
+	if (ret > 0)
+		*offset = pos + ret;
+	return ret;
+}
+
+static ssize_t ceph_sync_write(struct file *file, const char __user *data,
+			       size_t count, loff_t *offset)
+{
+	struct inode *inode = file->f_dentry->d_inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_client *client = ceph_inode_to_client(inode);
+	int ret = 0;
+	off_t pos = *offset;
+
+	if (ceph_snap(file->f_dentry->d_inode) != CEPH_NOSNAP)
+		return -EROFS;
+
+	dout(10, "sync_write on file %p %lld~%u\n", file, *offset,
+	     (unsigned)count);
+
+	if (file->f_flags & O_APPEND)
+		pos = i_size_read(inode);
+
+	ret = ceph_osdc_sync_write(&client->osdc, ceph_vino(inode),
+				   &ci->i_layout,
+				   ci->i_snap_realm->cached_context,
+				   pos, count, data);
+	if (ret > 0) {
+		pos += ret;
+		*offset = pos;
+		if (pos > i_size_read(inode))
+			ceph_inode_set_size(inode, pos);
+	}
+
+	return ret;
+}
+
+/*
+ * Wrap generic_file_aio_read with checks for cap bits on the inode.
+ * Atomically grab references, so that those bits are not released
+ * back to the MDS mid-read.
+ *
+ * Hmm, the sync reach case isn't actually async... should it be?
+ */
+static ssize_t ceph_aio_read(struct kiocb *iocb, const struct iovec *iov,
+		      unsigned long nr_segs, loff_t pos)
+{
+	struct file *filp = iocb->ki_filp;
+	loff_t *ppos = &iocb->ki_pos;
+	size_t len = iov->iov_len;
+	struct inode *inode = filp->f_dentry->d_inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	ssize_t ret;
+	int got = 0;
+
+	__ceph_do_pending_vmtruncate(inode);
+
+	dout(10, "aio_read %llx.%llx %llu~%u trying to get caps on %p\n",
+	     ceph_vinop(inode), pos, (unsigned)len, inode);
+	ret = wait_event_interruptible(ci->i_cap_wq,
+				       ceph_get_cap_refs(ci, CEPH_CAP_RD,
+							 CEPH_CAP_RDCACHE,
+							 &got, -1));
+	if (ret < 0)
+		goto out;
+	dout(10, "aio_read %llx.%llx %llu~%u got cap refs %d\n",
+	     ceph_vinop(inode), pos, (unsigned)len, got);
+
+	if ((got & CEPH_CAP_RDCACHE) == 0 ||
+	    (inode->i_sb->s_flags & MS_SYNCHRONOUS))
+		/* hmm, this isn't really async... */
+		ret = ceph_sync_read(filp, iov->iov_base, len, ppos);
+	else
+		ret = generic_file_aio_read(iocb, iov, nr_segs, pos);
+
+out:
+	dout(10, "aio_read %llx.%llx dropping cap refs on %d\n",
+	     ceph_vinop(inode), got);
+	ceph_put_cap_refs(ci, got);
+	return ret;
+}
+
+/*
+ * Check the offset we are writing up to against our current
+ * max_size.  If necessary, tell the MDS we want to write to
+ * a larger offset.
+ */
+static void check_max_size(struct inode *inode, loff_t endoff)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	int check = 0;
+
+	/* do we need to explicitly request a larger max_size? */
+	spin_lock(&inode->i_lock);
+	if ((endoff >= ci->i_max_size ||
+	     endoff > (inode->i_size << 1)) &&
+	    endoff > ci->i_wanted_max_size) {
+		dout(10, "write %p at large endoff %llu, req max_size\n",
+		     inode, endoff);
+		ci->i_wanted_max_size = endoff;
+		check = 1;
+	}
+	spin_unlock(&inode->i_lock);
+	if (check)
+		ceph_check_caps(ci, 0);
+}
+
+/*
+ * Take cap references to avoid releasing caps to MDS mid-write.
+ *
+ * If we are synchronous, and write with an old snap context, the OSD
+ * may return EOLDSNAPC.  In that case, retry the write.. _after_
+ * dropping our cap refs and allowing the pending snap to logically
+ * complete _before_ this write occurs.
+ *
+ * If we are near ENOSPC, write synchronously.
+ */
+static ssize_t ceph_aio_write(struct kiocb *iocb, const struct iovec *iov,
+		       unsigned long nr_segs, loff_t pos)
+{
+	struct file *file = iocb->ki_filp;
+	struct inode *inode = file->f_dentry->d_inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_osd_client *osdc = &ceph_client(inode->i_sb)->osdc;
+	loff_t endoff = pos + iov->iov_len;
+	int got = 0;
+	int ret;
+
+	if (ceph_snap(inode) != CEPH_NOSNAP)
+		return -EROFS;
+
+retry_snap:
+	if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL))
+		return -ENOSPC;
+	__ceph_do_pending_vmtruncate(inode);
+	check_max_size(inode, endoff);
+	dout(10, "aio_write %p %llu~%u getting caps. i_size %llu\n",
+	     inode, pos, (unsigned)iov->iov_len, inode->i_size);
+	ret = wait_event_interruptible(ci->i_cap_wq,
+				       ceph_get_cap_refs(ci, CEPH_CAP_WR,
+							 CEPH_CAP_WRBUFFER,
+							 &got, endoff));
+	if (ret < 0)
+		goto out;
+
+	dout(10, "aio_write %p %llu~%u  got cap refs on %d\n",
+	     inode, pos, (unsigned)iov->iov_len, got);
+
+	if ((got & CEPH_CAP_WRBUFFER) == 0 ||
+	    ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_NEARFULL) ||
+	    (inode->i_sb->s_flags & MS_SYNCHRONOUS))
+		/* fixme, this isn't actually async! */
+		ret = ceph_sync_write(file, iov->iov_base, iov->iov_len,
+				      &iocb->ki_pos);
+	else
+		ret = generic_file_aio_write(iocb, iov, nr_segs, pos);
+
+out:
+	dout(10, "aio_write %p %llu~%u  dropping cap refs on %d\n",
+	     inode, pos, (unsigned)iov->iov_len, got);
+	ceph_put_cap_refs(ci, got);
+
+	if (ret == -EOLDSNAPC) {
+		dout(10, "aio_write %p %llu~%u got EOLDSNAPC, retrying\n",
+		     inode, pos, (unsigned)iov->iov_len);
+		goto retry_snap;
+	}
+
+	return ret;
+}
+
+static int ceph_fsync(struct file *file, struct dentry *dentry, int datasync)
+{
+	struct inode *inode = dentry->d_inode;
+	int ret;
+
+	dout(10, "fsync on inode %p\n", inode);
+	ret = write_inode_now(inode, 1);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * HMM: should we also ensure that caps are flushed to mds?
+	 * It's not strictly necessary, since with the data on the
+	 * osds the mds can/will always reconstruct the file size.
+	 * Not mtime, though.
+	 */
+
+	return 0;
+}
+
+const struct file_operations ceph_file_fops = {
+	.open = ceph_open,
+	.release = ceph_release,
+	.llseek = generic_file_llseek,
+	.read = do_sync_read,
+	.write = do_sync_write,
+	.aio_read = ceph_aio_read,
+	.aio_write = ceph_aio_write,
+	.mmap = ceph_mmap,
+	.fsync = ceph_fsync,
+	.splice_read = generic_file_splice_read,
+	.splice_write = generic_file_splice_write,
+	.unlocked_ioctl = ceph_ioctl,
+};
+
+
-- 
1.5.6.5

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists