Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / fs / cifs / cifsacl.c
diff --git a/kernel/fs/cifs/cifsacl.c b/kernel/fs/cifs/cifsacl.c
new file mode 100644 (file)
index 0000000..1ea780b
--- /dev/null
@@ -0,0 +1,1119 @@
+/*
+ *   fs/cifs/cifsacl.c
+ *
+ *   Copyright (C) International Business Machines  Corp., 2007,2008
+ *   Author(s): Steve French (sfrench@us.ibm.com)
+ *
+ *   Contains the routines for mapping CIFS/NTFS ACLs
+ *
+ *   This library is free software; you can redistribute it and/or modify
+ *   it under the terms of the GNU Lesser General Public License as published
+ *   by the Free Software Foundation; either version 2.1 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This library is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU Lesser General Public License for more details.
+ *
+ *   You should have received a copy of the GNU Lesser General Public License
+ *   along with this library; if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/slab.h>
+#include <linux/string.h>
+#include <linux/keyctl.h>
+#include <linux/key-type.h>
+#include <keys/user-type.h>
+#include "cifspdu.h"
+#include "cifsglob.h"
+#include "cifsacl.h"
+#include "cifsproto.h"
+#include "cifs_debug.h"
+
+/* security id for everyone/world system group */
+static const struct cifs_sid sid_everyone = {
+       1, 1, {0, 0, 0, 0, 0, 1}, {0} };
+/* security id for Authenticated Users system group */
+static const struct cifs_sid sid_authusers = {
+       1, 1, {0, 0, 0, 0, 0, 5}, {cpu_to_le32(11)} };
+/* group users */
+static const struct cifs_sid sid_user = {1, 2 , {0, 0, 0, 0, 0, 5}, {} };
+
+static const struct cred *root_cred;
+
+static int
+cifs_idmap_key_instantiate(struct key *key, struct key_preparsed_payload *prep)
+{
+       char *payload;
+
+       /*
+        * If the payload is less than or equal to the size of a pointer, then
+        * an allocation here is wasteful. Just copy the data directly to the
+        * payload.value union member instead.
+        *
+        * With this however, you must check the datalen before trying to
+        * dereference payload.data!
+        */
+       if (prep->datalen <= sizeof(key->payload)) {
+               key->payload.value = 0;
+               memcpy(&key->payload.value, prep->data, prep->datalen);
+               key->datalen = prep->datalen;
+               return 0;
+       }
+       payload = kmemdup(prep->data, prep->datalen, GFP_KERNEL);
+       if (!payload)
+               return -ENOMEM;
+
+       key->payload.data = payload;
+       key->datalen = prep->datalen;
+       return 0;
+}
+
+static inline void
+cifs_idmap_key_destroy(struct key *key)
+{
+       if (key->datalen > sizeof(key->payload))
+               kfree(key->payload.data);
+}
+
+static struct key_type cifs_idmap_key_type = {
+       .name        = "cifs.idmap",
+       .instantiate = cifs_idmap_key_instantiate,
+       .destroy     = cifs_idmap_key_destroy,
+       .describe    = user_describe,
+};
+
+static char *
+sid_to_key_str(struct cifs_sid *sidptr, unsigned int type)
+{
+       int i, len;
+       unsigned int saval;
+       char *sidstr, *strptr;
+       unsigned long long id_auth_val;
+
+       /* 3 bytes for prefix */
+       sidstr = kmalloc(3 + SID_STRING_BASE_SIZE +
+                        (SID_STRING_SUBAUTH_SIZE * sidptr->num_subauth),
+                        GFP_KERNEL);
+       if (!sidstr)
+               return sidstr;
+
+       strptr = sidstr;
+       len = sprintf(strptr, "%cs:S-%hhu", type == SIDOWNER ? 'o' : 'g',
+                       sidptr->revision);
+       strptr += len;
+
+       /* The authority field is a single 48-bit number */
+       id_auth_val = (unsigned long long)sidptr->authority[5];
+       id_auth_val |= (unsigned long long)sidptr->authority[4] << 8;
+       id_auth_val |= (unsigned long long)sidptr->authority[3] << 16;
+       id_auth_val |= (unsigned long long)sidptr->authority[2] << 24;
+       id_auth_val |= (unsigned long long)sidptr->authority[1] << 32;
+       id_auth_val |= (unsigned long long)sidptr->authority[0] << 48;
+
+       /*
+        * MS-DTYP states that if the authority is >= 2^32, then it should be
+        * expressed as a hex value.
+        */
+       if (id_auth_val <= UINT_MAX)
+               len = sprintf(strptr, "-%llu", id_auth_val);
+       else
+               len = sprintf(strptr, "-0x%llx", id_auth_val);
+
+       strptr += len;
+
+       for (i = 0; i < sidptr->num_subauth; ++i) {
+               saval = le32_to_cpu(sidptr->sub_auth[i]);
+               len = sprintf(strptr, "-%u", saval);
+               strptr += len;
+       }
+
+       return sidstr;
+}
+
+/*
+ * if the two SIDs (roughly equivalent to a UUID for a user or group) are
+ * the same returns zero, if they do not match returns non-zero.
+ */
+static int
+compare_sids(const struct cifs_sid *ctsid, const struct cifs_sid *cwsid)
+{
+       int i;
+       int num_subauth, num_sat, num_saw;
+
+       if ((!ctsid) || (!cwsid))
+               return 1;
+
+       /* compare the revision */
+       if (ctsid->revision != cwsid->revision) {
+               if (ctsid->revision > cwsid->revision)
+                       return 1;
+               else
+                       return -1;
+       }
+
+       /* compare all of the six auth values */
+       for (i = 0; i < NUM_AUTHS; ++i) {
+               if (ctsid->authority[i] != cwsid->authority[i]) {
+                       if (ctsid->authority[i] > cwsid->authority[i])
+                               return 1;
+                       else
+                               return -1;
+               }
+       }
+
+       /* compare all of the subauth values if any */
+       num_sat = ctsid->num_subauth;
+       num_saw = cwsid->num_subauth;
+       num_subauth = num_sat < num_saw ? num_sat : num_saw;
+       if (num_subauth) {
+               for (i = 0; i < num_subauth; ++i) {
+                       if (ctsid->sub_auth[i] != cwsid->sub_auth[i]) {
+                               if (le32_to_cpu(ctsid->sub_auth[i]) >
+                                       le32_to_cpu(cwsid->sub_auth[i]))
+                                       return 1;
+                               else
+                                       return -1;
+                       }
+               }
+       }
+
+       return 0; /* sids compare/match */
+}
+
+static void
+cifs_copy_sid(struct cifs_sid *dst, const struct cifs_sid *src)
+{
+       int i;
+
+       dst->revision = src->revision;
+       dst->num_subauth = min_t(u8, src->num_subauth, SID_MAX_SUB_AUTHORITIES);
+       for (i = 0; i < NUM_AUTHS; ++i)
+               dst->authority[i] = src->authority[i];
+       for (i = 0; i < dst->num_subauth; ++i)
+               dst->sub_auth[i] = src->sub_auth[i];
+}
+
+static int
+id_to_sid(unsigned int cid, uint sidtype, struct cifs_sid *ssid)
+{
+       int rc;
+       struct key *sidkey;
+       struct cifs_sid *ksid;
+       unsigned int ksid_size;
+       char desc[3 + 10 + 1]; /* 3 byte prefix + 10 bytes for value + NULL */
+       const struct cred *saved_cred;
+
+       rc = snprintf(desc, sizeof(desc), "%ci:%u",
+                       sidtype == SIDOWNER ? 'o' : 'g', cid);
+       if (rc >= sizeof(desc))
+               return -EINVAL;
+
+       rc = 0;
+       saved_cred = override_creds(root_cred);
+       sidkey = request_key(&cifs_idmap_key_type, desc, "");
+       if (IS_ERR(sidkey)) {
+               rc = -EINVAL;
+               cifs_dbg(FYI, "%s: Can't map %cid %u to a SID\n",
+                        __func__, sidtype == SIDOWNER ? 'u' : 'g', cid);
+               goto out_revert_creds;
+       } else if (sidkey->datalen < CIFS_SID_BASE_SIZE) {
+               rc = -EIO;
+               cifs_dbg(FYI, "%s: Downcall contained malformed key (datalen=%hu)\n",
+                        __func__, sidkey->datalen);
+               goto invalidate_key;
+       }
+
+       /*
+        * A sid is usually too large to be embedded in payload.value, but if
+        * there are no subauthorities and the host has 8-byte pointers, then
+        * it could be.
+        */
+       ksid = sidkey->datalen <= sizeof(sidkey->payload) ?
+               (struct cifs_sid *)&sidkey->payload.value :
+               (struct cifs_sid *)sidkey->payload.data;
+
+       ksid_size = CIFS_SID_BASE_SIZE + (ksid->num_subauth * sizeof(__le32));
+       if (ksid_size > sidkey->datalen) {
+               rc = -EIO;
+               cifs_dbg(FYI, "%s: Downcall contained malformed key (datalen=%hu, ksid_size=%u)\n",
+                        __func__, sidkey->datalen, ksid_size);
+               goto invalidate_key;
+       }
+
+       cifs_copy_sid(ssid, ksid);
+out_key_put:
+       key_put(sidkey);
+out_revert_creds:
+       revert_creds(saved_cred);
+       return rc;
+
+invalidate_key:
+       key_invalidate(sidkey);
+       goto out_key_put;
+}
+
+static int
+sid_to_id(struct cifs_sb_info *cifs_sb, struct cifs_sid *psid,
+               struct cifs_fattr *fattr, uint sidtype)
+{
+       int rc;
+       struct key *sidkey;
+       char *sidstr;
+       const struct cred *saved_cred;
+       kuid_t fuid = cifs_sb->mnt_uid;
+       kgid_t fgid = cifs_sb->mnt_gid;
+
+       /*
+        * If we have too many subauthorities, then something is really wrong.
+        * Just return an error.
+        */
+       if (unlikely(psid->num_subauth > SID_MAX_SUB_AUTHORITIES)) {
+               cifs_dbg(FYI, "%s: %u subauthorities is too many!\n",
+                        __func__, psid->num_subauth);
+               return -EIO;
+       }
+
+       sidstr = sid_to_key_str(psid, sidtype);
+       if (!sidstr)
+               return -ENOMEM;
+
+       saved_cred = override_creds(root_cred);
+       sidkey = request_key(&cifs_idmap_key_type, sidstr, "");
+       if (IS_ERR(sidkey)) {
+               rc = -EINVAL;
+               cifs_dbg(FYI, "%s: Can't map SID %s to a %cid\n",
+                        __func__, sidstr, sidtype == SIDOWNER ? 'u' : 'g');
+               goto out_revert_creds;
+       }
+
+       /*
+        * FIXME: Here we assume that uid_t and gid_t are same size. It's
+        * probably a safe assumption but might be better to check based on
+        * sidtype.
+        */
+       BUILD_BUG_ON(sizeof(uid_t) != sizeof(gid_t));
+       if (sidkey->datalen != sizeof(uid_t)) {
+               rc = -EIO;
+               cifs_dbg(FYI, "%s: Downcall contained malformed key (datalen=%hu)\n",
+                        __func__, sidkey->datalen);
+               key_invalidate(sidkey);
+               goto out_key_put;
+       }
+
+       if (sidtype == SIDOWNER) {
+               kuid_t uid;
+               uid_t id;
+               memcpy(&id, &sidkey->payload.value, sizeof(uid_t));
+               uid = make_kuid(&init_user_ns, id);
+               if (uid_valid(uid))
+                       fuid = uid;
+       } else {
+               kgid_t gid;
+               gid_t id;
+               memcpy(&id, &sidkey->payload.value, sizeof(gid_t));
+               gid = make_kgid(&init_user_ns, id);
+               if (gid_valid(gid))
+                       fgid = gid;
+       }
+
+out_key_put:
+       key_put(sidkey);
+out_revert_creds:
+       revert_creds(saved_cred);
+       kfree(sidstr);
+
+       /*
+        * Note that we return 0 here unconditionally. If the mapping
+        * fails then we just fall back to using the mnt_uid/mnt_gid.
+        */
+       if (sidtype == SIDOWNER)
+               fattr->cf_uid = fuid;
+       else
+               fattr->cf_gid = fgid;
+       return 0;
+}
+
+int
+init_cifs_idmap(void)
+{
+       struct cred *cred;
+       struct key *keyring;
+       int ret;
+
+       cifs_dbg(FYI, "Registering the %s key type\n",
+                cifs_idmap_key_type.name);
+
+       /* create an override credential set with a special thread keyring in
+        * which requests are cached
+        *
+        * this is used to prevent malicious redirections from being installed
+        * with add_key().
+        */
+       cred = prepare_kernel_cred(NULL);
+       if (!cred)
+               return -ENOMEM;
+
+       keyring = keyring_alloc(".cifs_idmap",
+                               GLOBAL_ROOT_UID, GLOBAL_ROOT_GID, cred,
+                               (KEY_POS_ALL & ~KEY_POS_SETATTR) |
+                               KEY_USR_VIEW | KEY_USR_READ,
+                               KEY_ALLOC_NOT_IN_QUOTA, NULL);
+       if (IS_ERR(keyring)) {
+               ret = PTR_ERR(keyring);
+               goto failed_put_cred;
+       }
+
+       ret = register_key_type(&cifs_idmap_key_type);
+       if (ret < 0)
+               goto failed_put_key;
+
+       /* instruct request_key() to use this special keyring as a cache for
+        * the results it looks up */
+       set_bit(KEY_FLAG_ROOT_CAN_CLEAR, &keyring->flags);
+       cred->thread_keyring = keyring;
+       cred->jit_keyring = KEY_REQKEY_DEFL_THREAD_KEYRING;
+       root_cred = cred;
+
+       cifs_dbg(FYI, "cifs idmap keyring: %d\n", key_serial(keyring));
+       return 0;
+
+failed_put_key:
+       key_put(keyring);
+failed_put_cred:
+       put_cred(cred);
+       return ret;
+}
+
+void
+exit_cifs_idmap(void)
+{
+       key_revoke(root_cred->thread_keyring);
+       unregister_key_type(&cifs_idmap_key_type);
+       put_cred(root_cred);
+       cifs_dbg(FYI, "Unregistered %s key type\n", cifs_idmap_key_type.name);
+}
+
+/* copy ntsd, owner sid, and group sid from a security descriptor to another */
+static void copy_sec_desc(const struct cifs_ntsd *pntsd,
+                               struct cifs_ntsd *pnntsd, __u32 sidsoffset)
+{
+       struct cifs_sid *owner_sid_ptr, *group_sid_ptr;
+       struct cifs_sid *nowner_sid_ptr, *ngroup_sid_ptr;
+
+       /* copy security descriptor control portion */
+       pnntsd->revision = pntsd->revision;
+       pnntsd->type = pntsd->type;
+       pnntsd->dacloffset = cpu_to_le32(sizeof(struct cifs_ntsd));
+       pnntsd->sacloffset = 0;
+       pnntsd->osidoffset = cpu_to_le32(sidsoffset);
+       pnntsd->gsidoffset = cpu_to_le32(sidsoffset + sizeof(struct cifs_sid));
+
+       /* copy owner sid */
+       owner_sid_ptr = (struct cifs_sid *)((char *)pntsd +
+                               le32_to_cpu(pntsd->osidoffset));
+       nowner_sid_ptr = (struct cifs_sid *)((char *)pnntsd + sidsoffset);
+       cifs_copy_sid(nowner_sid_ptr, owner_sid_ptr);
+
+       /* copy group sid */
+       group_sid_ptr = (struct cifs_sid *)((char *)pntsd +
+                               le32_to_cpu(pntsd->gsidoffset));
+       ngroup_sid_ptr = (struct cifs_sid *)((char *)pnntsd + sidsoffset +
+                                       sizeof(struct cifs_sid));
+       cifs_copy_sid(ngroup_sid_ptr, group_sid_ptr);
+
+       return;
+}
+
+
+/*
+   change posix mode to reflect permissions
+   pmode is the existing mode (we only want to overwrite part of this
+   bits to set can be: S_IRWXU, S_IRWXG or S_IRWXO ie 00700 or 00070 or 00007
+*/
+static void access_flags_to_mode(__le32 ace_flags, int type, umode_t *pmode,
+                                umode_t *pbits_to_set)
+{
+       __u32 flags = le32_to_cpu(ace_flags);
+       /* the order of ACEs is important.  The canonical order is to begin with
+          DENY entries followed by ALLOW, otherwise an allow entry could be
+          encountered first, making the subsequent deny entry like "dead code"
+          which would be superflous since Windows stops when a match is made
+          for the operation you are trying to perform for your user */
+
+       /* For deny ACEs we change the mask so that subsequent allow access
+          control entries do not turn on the bits we are denying */
+       if (type == ACCESS_DENIED) {
+               if (flags & GENERIC_ALL)
+                       *pbits_to_set &= ~S_IRWXUGO;
+
+               if ((flags & GENERIC_WRITE) ||
+                       ((flags & FILE_WRITE_RIGHTS) == FILE_WRITE_RIGHTS))
+                       *pbits_to_set &= ~S_IWUGO;
+               if ((flags & GENERIC_READ) ||
+                       ((flags & FILE_READ_RIGHTS) == FILE_READ_RIGHTS))
+                       *pbits_to_set &= ~S_IRUGO;
+               if ((flags & GENERIC_EXECUTE) ||
+                       ((flags & FILE_EXEC_RIGHTS) == FILE_EXEC_RIGHTS))
+                       *pbits_to_set &= ~S_IXUGO;
+               return;
+       } else if (type != ACCESS_ALLOWED) {
+               cifs_dbg(VFS, "unknown access control type %d\n", type);
+               return;
+       }
+       /* else ACCESS_ALLOWED type */
+
+       if (flags & GENERIC_ALL) {
+               *pmode |= (S_IRWXUGO & (*pbits_to_set));
+               cifs_dbg(NOISY, "all perms\n");
+               return;
+       }
+       if ((flags & GENERIC_WRITE) ||
+                       ((flags & FILE_WRITE_RIGHTS) == FILE_WRITE_RIGHTS))
+               *pmode |= (S_IWUGO & (*pbits_to_set));
+       if ((flags & GENERIC_READ) ||
+                       ((flags & FILE_READ_RIGHTS) == FILE_READ_RIGHTS))
+               *pmode |= (S_IRUGO & (*pbits_to_set));
+       if ((flags & GENERIC_EXECUTE) ||
+                       ((flags & FILE_EXEC_RIGHTS) == FILE_EXEC_RIGHTS))
+               *pmode |= (S_IXUGO & (*pbits_to_set));
+
+       cifs_dbg(NOISY, "access flags 0x%x mode now 0x%x\n", flags, *pmode);
+       return;
+}
+
+/*
+   Generate access flags to reflect permissions mode is the existing mode.
+   This function is called for every ACE in the DACL whose SID matches
+   with either owner or group or everyone.
+*/
+
+static void mode_to_access_flags(umode_t mode, umode_t bits_to_use,
+                               __u32 *pace_flags)
+{
+       /* reset access mask */
+       *pace_flags = 0x0;
+
+       /* bits to use are either S_IRWXU or S_IRWXG or S_IRWXO */
+       mode &= bits_to_use;
+
+       /* check for R/W/X UGO since we do not know whose flags
+          is this but we have cleared all the bits sans RWX for
+          either user or group or other as per bits_to_use */
+       if (mode & S_IRUGO)
+               *pace_flags |= SET_FILE_READ_RIGHTS;
+       if (mode & S_IWUGO)
+               *pace_flags |= SET_FILE_WRITE_RIGHTS;
+       if (mode & S_IXUGO)
+               *pace_flags |= SET_FILE_EXEC_RIGHTS;
+
+       cifs_dbg(NOISY, "mode: 0x%x, access flags now 0x%x\n",
+                mode, *pace_flags);
+       return;
+}
+
+static __u16 fill_ace_for_sid(struct cifs_ace *pntace,
+                       const struct cifs_sid *psid, __u64 nmode, umode_t bits)
+{
+       int i;
+       __u16 size = 0;
+       __u32 access_req = 0;
+
+       pntace->type = ACCESS_ALLOWED;
+       pntace->flags = 0x0;
+       mode_to_access_flags(nmode, bits, &access_req);
+       if (!access_req)
+               access_req = SET_MINIMUM_RIGHTS;
+       pntace->access_req = cpu_to_le32(access_req);
+
+       pntace->sid.revision = psid->revision;
+       pntace->sid.num_subauth = psid->num_subauth;
+       for (i = 0; i < NUM_AUTHS; i++)
+               pntace->sid.authority[i] = psid->authority[i];
+       for (i = 0; i < psid->num_subauth; i++)
+               pntace->sid.sub_auth[i] = psid->sub_auth[i];
+
+       size = 1 + 1 + 2 + 4 + 1 + 1 + 6 + (psid->num_subauth * 4);
+       pntace->size = cpu_to_le16(size);
+
+       return size;
+}
+
+
+#ifdef CONFIG_CIFS_DEBUG2
+static void dump_ace(struct cifs_ace *pace, char *end_of_acl)
+{
+       int num_subauth;
+
+       /* validate that we do not go past end of acl */
+
+       if (le16_to_cpu(pace->size) < 16) {
+               cifs_dbg(VFS, "ACE too small %d\n", le16_to_cpu(pace->size));
+               return;
+       }
+
+       if (end_of_acl < (char *)pace + le16_to_cpu(pace->size)) {
+               cifs_dbg(VFS, "ACL too small to parse ACE\n");
+               return;
+       }
+
+       num_subauth = pace->sid.num_subauth;
+       if (num_subauth) {
+               int i;
+               cifs_dbg(FYI, "ACE revision %d num_auth %d type %d flags %d size %d\n",
+                        pace->sid.revision, pace->sid.num_subauth, pace->type,
+                        pace->flags, le16_to_cpu(pace->size));
+               for (i = 0; i < num_subauth; ++i) {
+                       cifs_dbg(FYI, "ACE sub_auth[%d]: 0x%x\n",
+                                i, le32_to_cpu(pace->sid.sub_auth[i]));
+               }
+
+               /* BB add length check to make sure that we do not have huge
+                       num auths and therefore go off the end */
+       }
+
+       return;
+}
+#endif
+
+
+static void parse_dacl(struct cifs_acl *pdacl, char *end_of_acl,
+                      struct cifs_sid *pownersid, struct cifs_sid *pgrpsid,
+                      struct cifs_fattr *fattr)
+{
+       int i;
+       int num_aces = 0;
+       int acl_size;
+       char *acl_base;
+       struct cifs_ace **ppace;
+
+       /* BB need to add parm so we can store the SID BB */
+
+       if (!pdacl) {
+               /* no DACL in the security descriptor, set
+                  all the permissions for user/group/other */
+               fattr->cf_mode |= S_IRWXUGO;
+               return;
+       }
+
+       /* validate that we do not go past end of acl */
+       if (end_of_acl < (char *)pdacl + le16_to_cpu(pdacl->size)) {
+               cifs_dbg(VFS, "ACL too small to parse DACL\n");
+               return;
+       }
+
+       cifs_dbg(NOISY, "DACL revision %d size %d num aces %d\n",
+                le16_to_cpu(pdacl->revision), le16_to_cpu(pdacl->size),
+                le32_to_cpu(pdacl->num_aces));
+
+       /* reset rwx permissions for user/group/other.
+          Also, if num_aces is 0 i.e. DACL has no ACEs,
+          user/group/other have no permissions */
+       fattr->cf_mode &= ~(S_IRWXUGO);
+
+       acl_base = (char *)pdacl;
+       acl_size = sizeof(struct cifs_acl);
+
+       num_aces = le32_to_cpu(pdacl->num_aces);
+       if (num_aces > 0) {
+               umode_t user_mask = S_IRWXU;
+               umode_t group_mask = S_IRWXG;
+               umode_t other_mask = S_IRWXU | S_IRWXG | S_IRWXO;
+
+               if (num_aces > ULONG_MAX / sizeof(struct cifs_ace *))
+                       return;
+               ppace = kmalloc(num_aces * sizeof(struct cifs_ace *),
+                               GFP_KERNEL);
+               if (!ppace)
+                       return;
+
+               for (i = 0; i < num_aces; ++i) {
+                       ppace[i] = (struct cifs_ace *) (acl_base + acl_size);
+#ifdef CONFIG_CIFS_DEBUG2
+                       dump_ace(ppace[i], end_of_acl);
+#endif
+                       if (compare_sids(&(ppace[i]->sid), pownersid) == 0)
+                               access_flags_to_mode(ppace[i]->access_req,
+                                                    ppace[i]->type,
+                                                    &fattr->cf_mode,
+                                                    &user_mask);
+                       if (compare_sids(&(ppace[i]->sid), pgrpsid) == 0)
+                               access_flags_to_mode(ppace[i]->access_req,
+                                                    ppace[i]->type,
+                                                    &fattr->cf_mode,
+                                                    &group_mask);
+                       if (compare_sids(&(ppace[i]->sid), &sid_everyone) == 0)
+                               access_flags_to_mode(ppace[i]->access_req,
+                                                    ppace[i]->type,
+                                                    &fattr->cf_mode,
+                                                    &other_mask);
+                       if (compare_sids(&(ppace[i]->sid), &sid_authusers) == 0)
+                               access_flags_to_mode(ppace[i]->access_req,
+                                                    ppace[i]->type,
+                                                    &fattr->cf_mode,
+                                                    &other_mask);
+
+
+/*                     memcpy((void *)(&(cifscred->aces[i])),
+                               (void *)ppace[i],
+                               sizeof(struct cifs_ace)); */
+
+                       acl_base = (char *)ppace[i];
+                       acl_size = le16_to_cpu(ppace[i]->size);
+               }
+
+               kfree(ppace);
+       }
+
+       return;
+}
+
+
+static int set_chmod_dacl(struct cifs_acl *pndacl, struct cifs_sid *pownersid,
+                       struct cifs_sid *pgrpsid, __u64 nmode)
+{
+       u16 size = 0;
+       struct cifs_acl *pnndacl;
+
+       pnndacl = (struct cifs_acl *)((char *)pndacl + sizeof(struct cifs_acl));
+
+       size += fill_ace_for_sid((struct cifs_ace *) ((char *)pnndacl + size),
+                                       pownersid, nmode, S_IRWXU);
+       size += fill_ace_for_sid((struct cifs_ace *)((char *)pnndacl + size),
+                                       pgrpsid, nmode, S_IRWXG);
+       size += fill_ace_for_sid((struct cifs_ace *)((char *)pnndacl + size),
+                                        &sid_everyone, nmode, S_IRWXO);
+
+       pndacl->size = cpu_to_le16(size + sizeof(struct cifs_acl));
+       pndacl->num_aces = cpu_to_le32(3);
+
+       return 0;
+}
+
+
+static int parse_sid(struct cifs_sid *psid, char *end_of_acl)
+{
+       /* BB need to add parm so we can store the SID BB */
+
+       /* validate that we do not go past end of ACL - sid must be at least 8
+          bytes long (assuming no sub-auths - e.g. the null SID */
+       if (end_of_acl < (char *)psid + 8) {
+               cifs_dbg(VFS, "ACL too small to parse SID %p\n", psid);
+               return -EINVAL;
+       }
+
+#ifdef CONFIG_CIFS_DEBUG2
+       if (psid->num_subauth) {
+               int i;
+               cifs_dbg(FYI, "SID revision %d num_auth %d\n",
+                        psid->revision, psid->num_subauth);
+
+               for (i = 0; i < psid->num_subauth; i++) {
+                       cifs_dbg(FYI, "SID sub_auth[%d]: 0x%x\n",
+                                i, le32_to_cpu(psid->sub_auth[i]));
+               }
+
+               /* BB add length check to make sure that we do not have huge
+                       num auths and therefore go off the end */
+               cifs_dbg(FYI, "RID 0x%x\n",
+                        le32_to_cpu(psid->sub_auth[psid->num_subauth-1]));
+       }
+#endif
+
+       return 0;
+}
+
+
+/* Convert CIFS ACL to POSIX form */
+static int parse_sec_desc(struct cifs_sb_info *cifs_sb,
+               struct cifs_ntsd *pntsd, int acl_len, struct cifs_fattr *fattr)
+{
+       int rc = 0;
+       struct cifs_sid *owner_sid_ptr, *group_sid_ptr;
+       struct cifs_acl *dacl_ptr; /* no need for SACL ptr */
+       char *end_of_acl = ((char *)pntsd) + acl_len;
+       __u32 dacloffset;
+
+       if (pntsd == NULL)
+               return -EIO;
+
+       owner_sid_ptr = (struct cifs_sid *)((char *)pntsd +
+                               le32_to_cpu(pntsd->osidoffset));
+       group_sid_ptr = (struct cifs_sid *)((char *)pntsd +
+                               le32_to_cpu(pntsd->gsidoffset));
+       dacloffset = le32_to_cpu(pntsd->dacloffset);
+       dacl_ptr = (struct cifs_acl *)((char *)pntsd + dacloffset);
+       cifs_dbg(NOISY, "revision %d type 0x%x ooffset 0x%x goffset 0x%x sacloffset 0x%x dacloffset 0x%x\n",
+                pntsd->revision, pntsd->type, le32_to_cpu(pntsd->osidoffset),
+                le32_to_cpu(pntsd->gsidoffset),
+                le32_to_cpu(pntsd->sacloffset), dacloffset);
+/*     cifs_dump_mem("owner_sid: ", owner_sid_ptr, 64); */
+       rc = parse_sid(owner_sid_ptr, end_of_acl);
+       if (rc) {
+               cifs_dbg(FYI, "%s: Error %d parsing Owner SID\n", __func__, rc);
+               return rc;
+       }
+       rc = sid_to_id(cifs_sb, owner_sid_ptr, fattr, SIDOWNER);
+       if (rc) {
+               cifs_dbg(FYI, "%s: Error %d mapping Owner SID to uid\n",
+                        __func__, rc);
+               return rc;
+       }
+
+       rc = parse_sid(group_sid_ptr, end_of_acl);
+       if (rc) {
+               cifs_dbg(FYI, "%s: Error %d mapping Owner SID to gid\n",
+                        __func__, rc);
+               return rc;
+       }
+       rc = sid_to_id(cifs_sb, group_sid_ptr, fattr, SIDGROUP);
+       if (rc) {
+               cifs_dbg(FYI, "%s: Error %d mapping Group SID to gid\n",
+                        __func__, rc);
+               return rc;
+       }
+
+       if (dacloffset)
+               parse_dacl(dacl_ptr, end_of_acl, owner_sid_ptr,
+                          group_sid_ptr, fattr);
+       else
+               cifs_dbg(FYI, "no ACL\n"); /* BB grant all or default perms? */
+
+       return rc;
+}
+
+/* Convert permission bits from mode to equivalent CIFS ACL */
+static int build_sec_desc(struct cifs_ntsd *pntsd, struct cifs_ntsd *pnntsd,
+       __u32 secdesclen, __u64 nmode, kuid_t uid, kgid_t gid, int *aclflag)
+{
+       int rc = 0;
+       __u32 dacloffset;
+       __u32 ndacloffset;
+       __u32 sidsoffset;
+       struct cifs_sid *owner_sid_ptr, *group_sid_ptr;
+       struct cifs_sid *nowner_sid_ptr, *ngroup_sid_ptr;
+       struct cifs_acl *dacl_ptr = NULL;  /* no need for SACL ptr */
+       struct cifs_acl *ndacl_ptr = NULL; /* no need for SACL ptr */
+
+       if (nmode != NO_CHANGE_64) { /* chmod */
+               owner_sid_ptr = (struct cifs_sid *)((char *)pntsd +
+                               le32_to_cpu(pntsd->osidoffset));
+               group_sid_ptr = (struct cifs_sid *)((char *)pntsd +
+                               le32_to_cpu(pntsd->gsidoffset));
+               dacloffset = le32_to_cpu(pntsd->dacloffset);
+               dacl_ptr = (struct cifs_acl *)((char *)pntsd + dacloffset);
+               ndacloffset = sizeof(struct cifs_ntsd);
+               ndacl_ptr = (struct cifs_acl *)((char *)pnntsd + ndacloffset);
+               ndacl_ptr->revision = dacl_ptr->revision;
+               ndacl_ptr->size = 0;
+               ndacl_ptr->num_aces = 0;
+
+               rc = set_chmod_dacl(ndacl_ptr, owner_sid_ptr, group_sid_ptr,
+                                       nmode);
+               sidsoffset = ndacloffset + le16_to_cpu(ndacl_ptr->size);
+               /* copy sec desc control portion & owner and group sids */
+               copy_sec_desc(pntsd, pnntsd, sidsoffset);
+               *aclflag = CIFS_ACL_DACL;
+       } else {
+               memcpy(pnntsd, pntsd, secdesclen);
+               if (uid_valid(uid)) { /* chown */
+                       uid_t id;
+                       owner_sid_ptr = (struct cifs_sid *)((char *)pnntsd +
+                                       le32_to_cpu(pnntsd->osidoffset));
+                       nowner_sid_ptr = kmalloc(sizeof(struct cifs_sid),
+                                                               GFP_KERNEL);
+                       if (!nowner_sid_ptr)
+                               return -ENOMEM;
+                       id = from_kuid(&init_user_ns, uid);
+                       rc = id_to_sid(id, SIDOWNER, nowner_sid_ptr);
+                       if (rc) {
+                               cifs_dbg(FYI, "%s: Mapping error %d for owner id %d\n",
+                                        __func__, rc, id);
+                               kfree(nowner_sid_ptr);
+                               return rc;
+                       }
+                       cifs_copy_sid(owner_sid_ptr, nowner_sid_ptr);
+                       kfree(nowner_sid_ptr);
+                       *aclflag = CIFS_ACL_OWNER;
+               }
+               if (gid_valid(gid)) { /* chgrp */
+                       gid_t id;
+                       group_sid_ptr = (struct cifs_sid *)((char *)pnntsd +
+                                       le32_to_cpu(pnntsd->gsidoffset));
+                       ngroup_sid_ptr = kmalloc(sizeof(struct cifs_sid),
+                                                               GFP_KERNEL);
+                       if (!ngroup_sid_ptr)
+                               return -ENOMEM;
+                       id = from_kgid(&init_user_ns, gid);
+                       rc = id_to_sid(id, SIDGROUP, ngroup_sid_ptr);
+                       if (rc) {
+                               cifs_dbg(FYI, "%s: Mapping error %d for group id %d\n",
+                                        __func__, rc, id);
+                               kfree(ngroup_sid_ptr);
+                               return rc;
+                       }
+                       cifs_copy_sid(group_sid_ptr, ngroup_sid_ptr);
+                       kfree(ngroup_sid_ptr);
+                       *aclflag = CIFS_ACL_GROUP;
+               }
+       }
+
+       return rc;
+}
+
+struct cifs_ntsd *get_cifs_acl_by_fid(struct cifs_sb_info *cifs_sb,
+               const struct cifs_fid *cifsfid, u32 *pacllen)
+{
+       struct cifs_ntsd *pntsd = NULL;
+       unsigned int xid;
+       int rc;
+       struct tcon_link *tlink = cifs_sb_tlink(cifs_sb);
+
+       if (IS_ERR(tlink))
+               return ERR_CAST(tlink);
+
+       xid = get_xid();
+       rc = CIFSSMBGetCIFSACL(xid, tlink_tcon(tlink), cifsfid->netfid, &pntsd,
+                               pacllen);
+       free_xid(xid);
+
+       cifs_put_tlink(tlink);
+
+       cifs_dbg(FYI, "%s: rc = %d ACL len %d\n", __func__, rc, *pacllen);
+       if (rc)
+               return ERR_PTR(rc);
+       return pntsd;
+}
+
+static struct cifs_ntsd *get_cifs_acl_by_path(struct cifs_sb_info *cifs_sb,
+               const char *path, u32 *pacllen)
+{
+       struct cifs_ntsd *pntsd = NULL;
+       int oplock = 0;
+       unsigned int xid;
+       int rc, create_options = 0;
+       struct cifs_tcon *tcon;
+       struct tcon_link *tlink = cifs_sb_tlink(cifs_sb);
+       struct cifs_fid fid;
+       struct cifs_open_parms oparms;
+
+       if (IS_ERR(tlink))
+               return ERR_CAST(tlink);
+
+       tcon = tlink_tcon(tlink);
+       xid = get_xid();
+
+       if (backup_cred(cifs_sb))
+               create_options |= CREATE_OPEN_BACKUP_INTENT;
+
+       oparms.tcon = tcon;
+       oparms.cifs_sb = cifs_sb;
+       oparms.desired_access = READ_CONTROL;
+       oparms.create_options = create_options;
+       oparms.disposition = FILE_OPEN;
+       oparms.path = path;
+       oparms.fid = &fid;
+       oparms.reconnect = false;
+
+       rc = CIFS_open(xid, &oparms, &oplock, NULL);
+       if (!rc) {
+               rc = CIFSSMBGetCIFSACL(xid, tcon, fid.netfid, &pntsd, pacllen);
+               CIFSSMBClose(xid, tcon, fid.netfid);
+       }
+
+       cifs_put_tlink(tlink);
+       free_xid(xid);
+
+       cifs_dbg(FYI, "%s: rc = %d ACL len %d\n", __func__, rc, *pacllen);
+       if (rc)
+               return ERR_PTR(rc);
+       return pntsd;
+}
+
+/* Retrieve an ACL from the server */
+struct cifs_ntsd *get_cifs_acl(struct cifs_sb_info *cifs_sb,
+                                     struct inode *inode, const char *path,
+                                     u32 *pacllen)
+{
+       struct cifs_ntsd *pntsd = NULL;
+       struct cifsFileInfo *open_file = NULL;
+
+       if (inode)
+               open_file = find_readable_file(CIFS_I(inode), true);
+       if (!open_file)
+               return get_cifs_acl_by_path(cifs_sb, path, pacllen);
+
+       pntsd = get_cifs_acl_by_fid(cifs_sb, &open_file->fid, pacllen);
+       cifsFileInfo_put(open_file);
+       return pntsd;
+}
+
+ /* Set an ACL on the server */
+int set_cifs_acl(struct cifs_ntsd *pnntsd, __u32 acllen,
+                       struct inode *inode, const char *path, int aclflag)
+{
+       int oplock = 0;
+       unsigned int xid;
+       int rc, access_flags, create_options = 0;
+       struct cifs_tcon *tcon;
+       struct cifs_sb_info *cifs_sb = CIFS_SB(inode->i_sb);
+       struct tcon_link *tlink = cifs_sb_tlink(cifs_sb);
+       struct cifs_fid fid;
+       struct cifs_open_parms oparms;
+
+       if (IS_ERR(tlink))
+               return PTR_ERR(tlink);
+
+       tcon = tlink_tcon(tlink);
+       xid = get_xid();
+
+       if (backup_cred(cifs_sb))
+               create_options |= CREATE_OPEN_BACKUP_INTENT;
+
+       if (aclflag == CIFS_ACL_OWNER || aclflag == CIFS_ACL_GROUP)
+               access_flags = WRITE_OWNER;
+       else
+               access_flags = WRITE_DAC;
+
+       oparms.tcon = tcon;
+       oparms.cifs_sb = cifs_sb;
+       oparms.desired_access = access_flags;
+       oparms.create_options = create_options;
+       oparms.disposition = FILE_OPEN;
+       oparms.path = path;
+       oparms.fid = &fid;
+       oparms.reconnect = false;
+
+       rc = CIFS_open(xid, &oparms, &oplock, NULL);
+       if (rc) {
+               cifs_dbg(VFS, "Unable to open file to set ACL\n");
+               goto out;
+       }
+
+       rc = CIFSSMBSetCIFSACL(xid, tcon, fid.netfid, pnntsd, acllen, aclflag);
+       cifs_dbg(NOISY, "SetCIFSACL rc = %d\n", rc);
+
+       CIFSSMBClose(xid, tcon, fid.netfid);
+out:
+       free_xid(xid);
+       cifs_put_tlink(tlink);
+       return rc;
+}
+
+/* Translate the CIFS ACL (simlar to NTFS ACL) for a file into mode bits */
+int
+cifs_acl_to_fattr(struct cifs_sb_info *cifs_sb, struct cifs_fattr *fattr,
+                 struct inode *inode, const char *path,
+                 const struct cifs_fid *pfid)
+{
+       struct cifs_ntsd *pntsd = NULL;
+       u32 acllen = 0;
+       int rc = 0;
+       struct tcon_link *tlink = cifs_sb_tlink(cifs_sb);
+       struct cifs_tcon *tcon;
+
+       cifs_dbg(NOISY, "converting ACL to mode for %s\n", path);
+
+       if (IS_ERR(tlink))
+               return PTR_ERR(tlink);
+       tcon = tlink_tcon(tlink);
+
+       if (pfid && (tcon->ses->server->ops->get_acl_by_fid))
+               pntsd = tcon->ses->server->ops->get_acl_by_fid(cifs_sb, pfid,
+                                                         &acllen);
+       else if (tcon->ses->server->ops->get_acl)
+               pntsd = tcon->ses->server->ops->get_acl(cifs_sb, inode, path,
+                                                       &acllen);
+       else {
+               cifs_put_tlink(tlink);
+               return -EOPNOTSUPP;
+       }
+       /* if we can retrieve the ACL, now parse Access Control Entries, ACEs */
+       if (IS_ERR(pntsd)) {
+               rc = PTR_ERR(pntsd);
+               cifs_dbg(VFS, "%s: error %d getting sec desc\n", __func__, rc);
+       } else {
+               rc = parse_sec_desc(cifs_sb, pntsd, acllen, fattr);
+               kfree(pntsd);
+               if (rc)
+                       cifs_dbg(VFS, "parse sec desc failed rc = %d\n", rc);
+       }
+
+       cifs_put_tlink(tlink);
+
+       return rc;
+}
+
+/* Convert mode bits to an ACL so we can update the ACL on the server */
+int
+id_mode_to_cifs_acl(struct inode *inode, const char *path, __u64 nmode,
+                       kuid_t uid, kgid_t gid)
+{
+       int rc = 0;
+       int aclflag = CIFS_ACL_DACL; /* default flag to set */
+       __u32 secdesclen = 0;
+       struct cifs_ntsd *pntsd = NULL; /* acl obtained from server */
+       struct cifs_ntsd *pnntsd = NULL; /* modified acl to be sent to server */
+       struct cifs_sb_info *cifs_sb = CIFS_SB(inode->i_sb);
+       struct tcon_link *tlink = cifs_sb_tlink(cifs_sb);
+       struct cifs_tcon *tcon;
+
+       if (IS_ERR(tlink))
+               return PTR_ERR(tlink);
+       tcon = tlink_tcon(tlink);
+
+       cifs_dbg(NOISY, "set ACL from mode for %s\n", path);
+
+       /* Get the security descriptor */
+
+       if (tcon->ses->server->ops->get_acl == NULL) {
+               cifs_put_tlink(tlink);
+               return -EOPNOTSUPP;
+       }
+
+       pntsd = tcon->ses->server->ops->get_acl(cifs_sb, inode, path,
+                                               &secdesclen);
+       if (IS_ERR(pntsd)) {
+               rc = PTR_ERR(pntsd);
+               cifs_dbg(VFS, "%s: error %d getting sec desc\n", __func__, rc);
+               cifs_put_tlink(tlink);
+               return rc;
+       }
+
+       /*
+        * Add three ACEs for owner, group, everyone getting rid of other ACEs
+        * as chmod disables ACEs and set the security descriptor. Allocate
+        * memory for the smb header, set security descriptor request security
+        * descriptor parameters, and secuirty descriptor itself
+        */
+       secdesclen = max_t(u32, secdesclen, DEFAULT_SEC_DESC_LEN);
+       pnntsd = kmalloc(secdesclen, GFP_KERNEL);
+       if (!pnntsd) {
+               kfree(pntsd);
+               cifs_put_tlink(tlink);
+               return -ENOMEM;
+       }
+
+       rc = build_sec_desc(pntsd, pnntsd, secdesclen, nmode, uid, gid,
+                               &aclflag);
+
+       cifs_dbg(NOISY, "build_sec_desc rc: %d\n", rc);
+
+       if (tcon->ses->server->ops->set_acl == NULL)
+               rc = -EOPNOTSUPP;
+
+       if (!rc) {
+               /* Set the security descriptor */
+               rc = tcon->ses->server->ops->set_acl(pnntsd, secdesclen, inode,
+                                                    path, aclflag);
+               cifs_dbg(NOISY, "set_cifs_acl rc: %d\n", rc);
+       }
+       cifs_put_tlink(tlink);
+
+       kfree(pnntsd);
+       kfree(pntsd);
+       return rc;
+}