ceph: use i_ceph_lock instead of i_lock
We have been using i_lock to protect all kinds of data structures in the
ceph_inode_info struct, including lists of inodes that we need to iterate
over while avoiding races with inode destruction. That requires grabbing
a reference to the inode with the list lock protected, but igrab() now
takes i_lock to check the inode flags.
Changing the list lock ordering would be a painful process.
However, using a ceph-specific i_ceph_lock in the ceph inode instead of
i_lock is a simple mechanical change and avoids the ordering constraints
imposed by igrab().
Reported-by: Amon Ott <a.ott@m-privacy.de>
Signed-off-by: Sage Weil <sage@newdream.net>
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 116f365..87fb132 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -297,6 +297,8 @@
dout("alloc_inode %p\n", &ci->vfs_inode);
+ spin_lock_init(&ci->i_ceph_lock);
+
ci->i_version = 0;
ci->i_time_warp_seq = 0;
ci->i_ceph_flags = 0;
@@ -583,7 +585,7 @@
iinfo->xattr_len);
}
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
/*
* provided version will be odd if inode value is projected,
@@ -680,7 +682,7 @@
char *sym;
BUG_ON(symlen != inode->i_size);
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
err = -ENOMEM;
sym = kmalloc(symlen+1, GFP_NOFS);
@@ -689,7 +691,7 @@
memcpy(sym, iinfo->symlink, symlen);
sym[symlen] = 0;
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
if (!ci->i_symlink)
ci->i_symlink = sym;
else
@@ -715,7 +717,7 @@
}
no_change:
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
/* queue truncate if we saw i_size decrease */
if (queue_trunc)
@@ -750,13 +752,13 @@
info->cap.flags,
caps_reservation);
} else {
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
dout(" %p got snap_caps %s\n", inode,
ceph_cap_string(le32_to_cpu(info->cap.caps)));
ci->i_snap_caps |= le32_to_cpu(info->cap.caps);
if (cap_fmode >= 0)
__ceph_get_fmode(ci, cap_fmode);
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
}
} else if (cap_fmode >= 0) {
pr_warning("mds issued no caps on %llx.%llx\n",
@@ -849,19 +851,20 @@
{
struct dentry *dir = dn->d_parent;
struct inode *inode = dir->d_inode;
+ struct ceph_inode_info *ci = ceph_inode(inode);
struct ceph_dentry_info *di;
BUG_ON(!inode);
di = ceph_dentry(dn);
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
if (!ceph_dir_test_complete(inode)) {
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
return;
}
di->offset = ceph_inode(inode)->i_max_offset++;
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
spin_lock(&dir->d_lock);
spin_lock_nested(&dn->d_lock, DENTRY_D_LOCK_NESTED);
@@ -1308,7 +1311,7 @@
struct ceph_inode_info *ci = ceph_inode(inode);
int ret = 0;
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
dout("set_size %p %llu -> %llu\n", inode, inode->i_size, size);
inode->i_size = size;
inode->i_blocks = (size + (1 << 9) - 1) >> 9;
@@ -1318,7 +1321,7 @@
(ci->i_reported_size << 1) < ci->i_max_size)
ret = 1;
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
return ret;
}
@@ -1376,20 +1379,20 @@
u32 orig_gen;
int check = 0;
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
dout("invalidate_pages %p gen %d revoking %d\n", inode,
ci->i_rdcache_gen, ci->i_rdcache_revoking);
if (ci->i_rdcache_revoking != ci->i_rdcache_gen) {
/* nevermind! */
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
goto out;
}
orig_gen = ci->i_rdcache_gen;
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
truncate_inode_pages(&inode->i_data, 0);
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
if (orig_gen == ci->i_rdcache_gen &&
orig_gen == ci->i_rdcache_revoking) {
dout("invalidate_pages %p gen %d successful\n", inode,
@@ -1401,7 +1404,7 @@
inode, orig_gen, ci->i_rdcache_gen,
ci->i_rdcache_revoking);
}
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
if (check)
ceph_check_caps(ci, 0, NULL);
@@ -1460,10 +1463,10 @@
int wrbuffer_refs, wake = 0;
retry:
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
if (ci->i_truncate_pending == 0) {
dout("__do_pending_vmtruncate %p none pending\n", inode);
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
return;
}
@@ -1474,7 +1477,7 @@
if (ci->i_wrbuffer_ref_head < ci->i_wrbuffer_ref) {
dout("__do_pending_vmtruncate %p flushing snaps first\n",
inode);
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
filemap_write_and_wait_range(&inode->i_data, 0,
inode->i_sb->s_maxbytes);
goto retry;
@@ -1484,15 +1487,15 @@
wrbuffer_refs = ci->i_wrbuffer_ref;
dout("__do_pending_vmtruncate %p (%d) to %lld\n", inode,
ci->i_truncate_pending, to);
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
truncate_inode_pages(inode->i_mapping, to);
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
ci->i_truncate_pending--;
if (ci->i_truncate_pending == 0)
wake = 1;
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
if (wrbuffer_refs == 0)
ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
@@ -1547,7 +1550,7 @@
if (IS_ERR(req))
return PTR_ERR(req);
- spin_lock(&inode->i_lock);
+ spin_lock(&ci->i_ceph_lock);
issued = __ceph_caps_issued(ci, NULL);
dout("setattr %p issued %s\n", inode, ceph_cap_string(issued));
@@ -1695,7 +1698,7 @@
}
release &= issued;
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
if (inode_dirty_flags)
__mark_inode_dirty(inode, inode_dirty_flags);
@@ -1717,7 +1720,7 @@
__ceph_do_pending_vmtruncate(inode);
return err;
out:
- spin_unlock(&inode->i_lock);
+ spin_unlock(&ci->i_ceph_lock);
ceph_mdsc_put_request(req);
return err;
}