Skip to content

Commit

Permalink
xfs: allow 2048 acl entries per inode (XFS_ACL_MAX_ENTRIES)
Browse files Browse the repository at this point in the history 
We need to keep in mind the following:
- In V5-superblock, XFS allows to have up to 64KB of xattr data, which is about 5400 ACLs,
  so our value is still less than that.
  See:
    commit 0a8aa1939777dd114479677f0044652c1fd72398 xfs: increase number of ACL entries for V5 superblocks
    http://oss.sgi.com/archives/xfs/2013-06/msg00020.html
- commit 2dc164f2965b92a6efd2edb9e2813271741e96db (xfs: fix memory allocation failures with ACLs)
  uses vmalloc to allocate large buffers for ACLs. In our commit, we also use vzalloc instead of kzalloc.
  However, note the comments in kmem_zalloc_large() of the mainline code.
- We change XFS_ACL_MAX_ENTRIES without changing the superblock version.
  This is alright for us, as long as we never downgrade VPSA version (which
  we should never ever do).
- struct xfs_acl_entry in our code does not have the "ae_pad" field (which it has upstream),
  but the size of the structure is (somehow) still the same (12 bytes)

DDF4624D7A2FF62052FC99E
  • Loading branch information
@alex-zadara
alex-zadara committed Jan 21, 2015
1 parent 21dc2e0 commit 094df94
Show file tree
Hide file tree
Showing 5 changed files with 169 additions and 1 deletion.
133 changes: 133 additions & 0 deletions fs/xfs/xfs_acl.c
View file
Expand Up @@ -16,6 +16,10 @@
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "xfs.h"
#ifdef CONFIG_XFS_ZADARA
#include "xfs_sb.h"
#include "xfs_mount.h"
#endif /*CONFIG_XFS_ZADARA*/
#include "xfs_acl.h"
#include "xfs_attr.h"
#include "xfs_bmap_btree.h"
Expand Down Expand Up @@ -134,9 +138,17 @@ xfs_get_acl(struct inode *inode, int type)
* go out to the disk.
*/

#ifdef CONFIG_XFS_ZADARA
xfs_acl = vzalloc(sizeof(struct xfs_acl));
if (!xfs_acl) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu vzalloc(sizeof(struct xfs_acl)(%lu)) failed", ip->i_ino, sizeof(struct xfs_acl));
return ERR_PTR(-ENOMEM);
}
#else /*CONFIG_XFS_ZADARA*/
xfs_acl = kzalloc(sizeof(struct xfs_acl), GFP_KERNEL);
if (!xfs_acl)
return ERR_PTR(-ENOMEM);
#endif /*CONFIG_XFS_ZADARA*/

error = -xfs_attr_get(ip, ea_name, (unsigned char *)xfs_acl,
&len, ATTR_ROOT);
Expand All @@ -160,7 +172,11 @@ xfs_get_acl(struct inode *inode, int type)
out_update_cache:
set_cached_acl(inode, type, acl);
out:
#ifdef CONFIG_XFS_ZADARA
vfree(xfs_acl);
#else /*CONFIG_XFS_ZADARA*/
kfree(xfs_acl);
#endif /*CONFIG_XFS_ZADARA*/
return acl;
}

Expand All @@ -171,29 +187,55 @@ xfs_set_acl(struct inode *inode, int type, struct posix_acl *acl)
unsigned char *ea_name;
int error;

#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, Z_KDEB1, ZKLOG_TAG_XATTR, "ino=%llu type=0x%x acl=%p", ip->i_ino, type, acl);
if (S_ISLNK(inode->i_mode)) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu S_ISLNK => EOPNOTSUPP", ip->i_ino);
return -EOPNOTSUPP;
}
#else /*CONFIG_XFS_ZADARA*/
if (S_ISLNK(inode->i_mode))
return -EOPNOTSUPP;
#endif /*CONFIG_XFS_ZADARA*/

switch (type) {
case ACL_TYPE_ACCESS:
ea_name = SGI_ACL_FILE;
break;
case ACL_TYPE_DEFAULT:
#ifdef CONFIG_XFS_ZADARA
if (!S_ISDIR(inode->i_mode)) {
ZXFSLOG_TAG(ip->i_mount, Z_KWARN, ZKLOG_TAG_XATTR,"ino=%llu ACL_TYPE_DEFAULT !S_ISDIR ret=%d", ip->i_ino, acl ? -EACCES : 0);
return acl ? -EACCES : 0;
}
#else /*CONFIG_XFS_ZADARA*/
if (!S_ISDIR(inode->i_mode))
return acl ? -EACCES : 0;
#endif /*CONFIG_XFS_ZADARA*/
ea_name = SGI_ACL_DEFAULT;
break;
default:
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu unknown type=0x%x => EINVAL", ip->i_ino, type);
#endif /*CONFIG_XFS_ZADARA*/
return -EINVAL;
}

if (acl) {
struct xfs_acl *xfs_acl;
int len;

#ifdef CONFIG_XFS_ZADARA
xfs_acl = vzalloc(sizeof(struct xfs_acl));
if (!xfs_acl) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu vzalloc(struct xfs_acl(%lu)) ENOMEM", ip->i_ino, sizeof(struct xfs_acl));
return -ENOMEM;
}
#else /*CONFIG_XFS_ZADARA*/
xfs_acl = kzalloc(sizeof(struct xfs_acl), GFP_KERNEL);
if (!xfs_acl)
return -ENOMEM;
#endif /*CONFIG_XFS_ZADARA*/

xfs_acl_to_disk(xfs_acl, acl);
len = sizeof(struct xfs_acl) -
Expand All @@ -203,12 +245,20 @@ xfs_set_acl(struct inode *inode, int type, struct posix_acl *acl)
error = -xfs_attr_set(ip, ea_name, (unsigned char *)xfs_acl,
len, ATTR_ROOT);

#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_attr_set(ea_name=%s, len=%d) ret=%d", ip->i_ino, ea_name, len, error);
vfree(xfs_acl);
#else /*CONFIG_XFS_ZADARA*/
kfree(xfs_acl);
#endif /*CONFIG_XFS_ZADARA*/
} else {
/*
* A NULL ACL argument means we want to remove the ACL.
*/
error = -xfs_attr_remove(ip, ea_name, ATTR_ROOT);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_attr_remove(ea_name=%s) ret=%d", ip->i_ino, ea_name, error);
#endif /*CONFIG_XFS_ZADARA*/

/*
* If the attribute didn't exist to start with that's fine.
Expand All @@ -235,6 +285,9 @@ xfs_set_mode(struct inode *inode, umode_t mode)
iattr.ia_ctime = current_fs_time(inode->i_sb);

error = -xfs_setattr_nonsize(XFS_I(inode), &iattr, XFS_ATTR_NOACL);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(XFS_I(inode)->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_setattr_nonsize(mode=0%o) ret=%d", XFS_I(inode)->i_ino, mode, error);
#endif /*CONFIG_XFS_ZADARA*/
}

return error;
Expand Down Expand Up @@ -271,14 +324,24 @@ xfs_inherit_acl(struct inode *inode, struct posix_acl *acl)
{
umode_t mode = inode->i_mode;
int error = 0, inherit = 0;
#ifdef CONFIG_XFS_ZADARA
struct xfs_inode *ip = XFS_I(inode);
ZXFSLOG_TAG(ip->i_mount, Z_KDEB1, ZKLOG_TAG_XATTR, "ino=%llu mode=0%o", ip->i_ino, inode->i_mode);
#endif /*CONFIG_XFS_ZADARA*/

if (S_ISDIR(inode->i_mode)) {
error = xfs_set_acl(inode, ACL_TYPE_DEFAULT, acl);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_set_acl(ACL_TYPE_DEFAULT) ret=%d", ip->i_ino, error);
#endif /*CONFIG_XFS_ZADARA*/
if (error)
goto out;
}

error = posix_acl_create(&acl, GFP_KERNEL, &mode);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error >= 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu posix_acl_create() mode=0%o ret=%d", ip->i_ino, mode, error);
#endif /*CONFIG_XFS_ZADARA*/
if (error < 0)
return error;

Expand All @@ -291,11 +354,18 @@ xfs_inherit_acl(struct inode *inode, struct posix_acl *acl)
inherit = 1;

error = xfs_set_mode(inode, mode);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_set_mode(mode=0%o) ret=%d", ip->i_ino, mode, error);
#endif /*CONFIG_XFS_ZADARA*/
if (error)
goto out;

if (inherit)
error = xfs_set_acl(inode, ACL_TYPE_ACCESS, acl);
#ifdef CONFIG_XFS_ZADARA
if (inherit)
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_set_acl(ACL_TYPE_ACCESS) ret=%d", ip->i_ino, error);
#endif /*CONFIG_XFS_ZADARA*/

out:
posix_acl_release(acl);
Expand All @@ -308,18 +378,35 @@ xfs_acl_chmod(struct inode *inode)
struct posix_acl *acl;
int error;

#ifdef CONFIG_XFS_ZADARA
struct xfs_inode *ip = XFS_I(inode);
if (S_ISLNK(inode->i_mode)) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu S_ISLNK(inode->i_mode=0%o) EOPNOTSUPP", ip->i_ino, inode->i_mode);
return -EOPNOTSUPP;
}
#else /*CONFIG_XFS_ZADARA*/
if (S_ISLNK(inode->i_mode))
return -EOPNOTSUPP;
#endif /*CONFIG_XFS_ZADARA*/

acl = xfs_get_acl(inode, ACL_TYPE_ACCESS);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, IS_ERR(acl) || !acl ? Z_KWARN : Z_KDEB1, ZKLOG_TAG_XATTR, "ino=%llu xfs_get_acl=%p/%d", ip->i_ino, acl, (int)(IS_ERR(acl) ? PTR_ERR(acl) : 0));
#endif /*CONFIG_XFS_ZADARA*/
if (IS_ERR(acl) || !acl)
return PTR_ERR(acl);

error = posix_acl_chmod(&acl, GFP_KERNEL, inode->i_mode);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu posix_acl_chmod(inode->i_mode=0%o) ret=%d", ip->i_ino, inode->i_mode, error);
#endif /*CONFIG_XFS_ZADARA*/
if (error)
return error;

error = xfs_set_acl(inode, ACL_TYPE_ACCESS, acl);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_set_acl(ACL_TYPE_ACCESS) ret=%d", ip->i_ino, error);
#endif /*CONFIG_XFS_ZADARA*/
posix_acl_release(acl);
return error;
}
Expand Down Expand Up @@ -351,12 +438,29 @@ xfs_xattr_acl_set(struct dentry *dentry, const char *name,
struct posix_acl *acl = NULL;
int error = 0;

#ifdef CONFIG_XFS_ZADARA
struct xfs_inode *ip = XFS_I(inode);
ZXFSLOG_TAG(ip->i_mount, Z_KDEB1, ZKLOG_TAG_XATTR, "ino=%llu name=[%s] flags=0x%x type=0x%x val=[%.*s]", ip->i_ino, name, flags, type, min_t(int, size, 16), (const char*)(value ? value : "NULL"));
if (flags & XATTR_CREATE) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu name=[%s] flags & XATTR_CREATE => -EINVAL", ip->i_ino, name);
return -EINVAL;
}
if (type == ACL_TYPE_DEFAULT && !S_ISDIR(inode->i_mode)) {
ZXFSLOG_TAG(ip->i_mount, Z_KWARN, ZKLOG_TAG_XATTR, "ino=%llu ACL_TYPE_DEFAULT !S_ISDIR ret=%d", ip->i_ino, value ? -EACCES : 0);
return value ? -EACCES : 0;
}
if ((current_fsuid() != inode->i_uid) && !capable(CAP_FOWNER)) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu current_fsuid(%u)!=inode->i_uid(%u) && !capable(CAP_FOWNER) => -EPERM", ip->i_ino, current_fsuid(), inode->i_uid);
return -EPERM;
}
#else /*CONFIG_XFS_ZADARA*/
if (flags & XATTR_CREATE)
return -EINVAL;
if (type == ACL_TYPE_DEFAULT && !S_ISDIR(inode->i_mode))
return value ? -EACCES : 0;
if ((current_fsuid() != inode->i_uid) && !capable(CAP_FOWNER))
return -EPERM;
#endif /*CONFIG_XFS_ZADARA*/

if (!value)
goto set_acl;
Expand All @@ -367,40 +471,69 @@ xfs_xattr_acl_set(struct dentry *dentry, const char *name,
* acl_set_file(3) may request that we set default ACLs with
* zero length -- defend (gracefully) against that here.
*/
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, Z_KWARN, ZKLOG_TAG_XATTR, "ino=%llu posix_acl_from_xattr() ret=NULL", ip->i_ino);
#endif /*CONFIG_XFS_ZADARA*/
goto out;
}
if (IS_ERR(acl)) {
error = PTR_ERR(acl);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu posix_acl_from_xattr() ret=%d", ip->i_ino, error);
#endif /*CONFIG_XFS_ZADARA*/
goto out;
}

error = posix_acl_valid(acl);
#ifdef CONFIG_XFS_ZADARA
if (error) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu posix_acl_valid() ret=%d", ip->i_ino, error);
goto out_release;
}
#else /*CONFIG_XFS_ZADARA*/
if (error)
goto out_release;
#endif /*CONFIG_XFS_ZADARA*/

error = -EINVAL;
#ifdef CONFIG_XFS_ZADARA
if (acl->a_count > XFS_ACL_MAX_ENTRIES) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu acl->a_count(%u) > XFS_ACL_MAX_ENTRIES(%u)", ip->i_ino, acl->a_count, XFS_ACL_MAX_ENTRIES);
goto out_release;
}
#else /*CONFIG_XFS_ZADARA*/
if (acl->a_count > XFS_ACL_MAX_ENTRIES)
goto out_release;
#endif /*CONFIG_XFS_ZADARA*/

if (type == ACL_TYPE_ACCESS) {
umode_t mode = inode->i_mode;
error = posix_acl_equiv_mode(acl, &mode);

if (error <= 0) {
posix_acl_release(acl);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu posix_acl_equiv_mode() mode=0%o ret=%d", ip->i_ino, mode, error);
#endif /*CONFIG_XFS_ZADARA*/
acl = NULL;

if (error < 0)
return error;
}

error = xfs_set_mode(inode, mode);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_set_mode(mode=0%o) ret=%d", ip->i_ino, mode, error);
#endif /*CONFIG_XFS_ZADARA*/
if (error)
goto out_release;
}

set_acl:
error = xfs_set_acl(inode, type, acl);
#ifdef CONFIG_XFS_ZADARA
ZXFSLOG_TAG(ip->i_mount, error == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_set_acl(type=0x%x) ret=%d", ip->i_ino, type, error);
#endif /*CONFIG_XFS_ZADARA*/
out_release:
posix_acl_release(acl);
out:
Expand Down
4 changes: 4 additions & 0 deletions fs/xfs/xfs_acl.h
View file
Expand Up @@ -22,7 +22,11 @@ struct inode;
struct posix_acl;
struct xfs_inode;

#ifdef CONFIG_XFS_ZADARA
#define XFS_ACL_MAX_ENTRIES 2048
#else /*CONFIG_XFS_ZADARA*/
#define XFS_ACL_MAX_ENTRIES 25
#endif /*CONFIG_XFS_ZADARA*/
#define XFS_ACL_NOT_PRESENT (-1)

/* On-disk XFS access control list structure */
Expand Down
26 changes: 25 additions & 1 deletion fs/xfs/xfs_xattr.c
View file
Expand Up @@ -24,6 +24,10 @@
#include "xfs_attr_leaf.h"
#include "xfs_acl.h"
#include "xfs_vnodeops.h"
#ifdef CONFIG_XFS_ZADARA
#include "xfs_sb.h"
#include "xfs_mount.h"
#endif /*CONFIG_XFS_ZADARA*/

#include <linux/posix_acl_xattr.h>
#include <linux/xattr.h>
Expand Down Expand Up @@ -56,20 +60,40 @@ xfs_xattr_set(struct dentry *dentry, const char *name, const void *value,
size_t size, int flags, int xflags)
{
struct xfs_inode *ip = XFS_I(dentry->d_inode);

#ifdef CONFIG_XFS_ZADARA
int ret = 0;
ZXFSLOG_TAG(ip->i_mount, Z_KDEB1, ZKLOG_TAG_XATTR, "ino=%llu name=[%s] flags=0x%x xflags=0x%x val=[%.*s]", ip->i_ino, name, flags, xflags, min_t(int, size, 16), (const char*)(value ? value : "NULL"));
if (strcmp(name, "") == 0) {
ZXFSLOG_TAG(ip->i_mount, Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu name=empty => -EINVAL", ip->i_ino);
return -EINVAL;
}
#else /*CONFIG_XFS_ZADARA*/
if (strcmp(name, "") == 0)
return -EINVAL;
#endif /*CONFIG_XFS_ZADARA*/

/* Convert Linux syscall to XFS internal ATTR flags */
if (flags & XATTR_CREATE)
xflags |= ATTR_CREATE;
if (flags & XATTR_REPLACE)
xflags |= ATTR_REPLACE;

#ifdef CONFIG_XFS_ZADARA
if (!value) {
ret = -xfs_attr_remove(ip, (unsigned char *)name, xflags);
ZXFSLOG_TAG(ip->i_mount, ret == 0 || ret == -ENODATA ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_attr_remove(%s) flags=0x%x xflags=0x%x ret=%d", ip->i_ino, name, flags, xflags, ret);
} else {
ret = -xfs_attr_set(ip, (unsigned char *)name,
(void *)value, size, xflags);
ZXFSLOG_TAG(ip->i_mount, ret == 0 ? Z_KDEB1 : Z_KERR, ZKLOG_TAG_XATTR, "ino=%llu xfs_attr_set(%s) flags=0x%x xflags=0x%x ret=%d", ip->i_ino, name, flags, xflags, ret);
}
return ret;
#else /*CONFIG_XFS_ZADARA*/
if (!value)
return -xfs_attr_remove(ip, (unsigned char *)name, xflags);
return -xfs_attr_set(ip, (unsigned char *)name,
(void *)value, size, xflags);
#endif /*CONFIG_XFS_ZADARA*/
}

static const struct xattr_handler xfs_xattr_user_handler = {
Expand Down
6 changes: 6 additions & 0 deletions fs/xfs/zxfs.c
View file
Expand Up @@ -12,6 +12,7 @@ zklog_tag_t ZKLOG_TAG_AGF = 0;
zklog_tag_t ZKLOG_TAG_BUSY_EXT = 0;
zklog_tag_t ZKLOG_TAG_DISCARD = 0;
zklog_tag_t ZKLOG_TAG_RESIZE = 0;
zklog_tag_t ZKLOG_TAG_XATTR = 0;

/****** MISC stuff ****************************************/

Expand Down Expand Up @@ -262,6 +263,11 @@ STATIC int zinit_xfs_klog(void)
zklog(Z_KERR, "zklog_add_tag('sz') failed, ret=%d", error);
goto out;
}
error = zklog_add_tag("xa", "XATTR", Z_KINFO, &ZKLOG_TAG_XATTR);
if (error != 0) {
zklog(Z_KERR, "zklog_add_tag('xa') failed, ret=%d", error);
goto out;
}

error = 0;

Expand Down

0 comments on commit 094df94

Please sign in to comment.