[JFFS2] Tidy up fix for ACL/permissions problem.

[In commit 9ed437c50d we fixed a problem 
with standard permissions on newly-created inodes, when POSIX ACLs are 
enabled. This cleans it up...]

The attached patch separate jffs2_init_acl() into two parts.

The one is jffs2_init_acl_pre() called from jffs2_new_inode().
It compute ACL oriented inode->i_mode bits, and allocate in-memory ACL
objects associated with the new inode just before when inode meta
infomation is written to the medium.

The other is jffs2_init_acl_post() called from jffs2_symlink(),
jffs2_mkdir(), jffs2_mknod() and jffs2_do_create().
It actually writes in-memory ACL objects into the medium next to
the success of writing meta-information.

In the current implementation, we have to write a same inode meta
infomation twice when inode->i_mode is updated by the default ACL.
However, we can avoid the behavior by putting an updated i_mode
before it is written at first, as jffs2_init_acl_pre() doing.

Signed-off-by: KaiGai Kohei <kaigai@ak.jp.nec.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
This commit is contained in:
KaiGai Kohei
2007-09-14 15:16:35 +09:00
committed by David Woodhouse
parent e8b8c97773
commit cfc8dc6f6f
6 changed files with 94 additions and 87 deletions

View File

@@ -465,6 +465,14 @@ int jffs2_do_create(struct jffs2_sb_info *c, struct jffs2_inode_info *dir_f, str
up(&f->sem);
jffs2_complete_reservation(c);
ret = jffs2_init_security(&f->vfs_inode, &dir_f->vfs_inode);
if (ret)
return ret;
ret = jffs2_init_acl_post(&f->vfs_inode);
if (ret)
return ret;
ret = jffs2_reserve_space(c, sizeof(*rd)+namelen, &alloclen,
ALLOC_NORMAL, JFFS2_SUMMARY_DIRENT_SIZE(namelen));