Merge tag 'rds-odp-for-5.5' into rdma.git for-next

From https://git.kernel.org/pub/scm/linux/kernel/git/leon/linux-rdma

Leon Romanovsky says:

====================
Use ODP MRs for kernel ULPs

The following series extends MR creation routines to allow creation of
user MRs through kernel ULPs as a proxy. The immediate use case is to
allow RDS to work over FS-DAX, which requires ODP (on-demand-paging)
MRs to be created and such MRs were not possible to create prior this
series.

The first part of this patchset extends RDMA to have special verb
ib_reg_user_mr(). The common use case that uses this function is a
userspace application that allocates memory for HCA access but the
responsibility to register the memory at the HCA is on an kernel ULP.
This ULP acts as an agent for the userspace application.

The second part provides advise MR functionality for ULPs. This is
integral part of ODP flows and used to trigger pagefaults in advance
to prepare memory before running working set.

The third part is actual user of those in-kernel APIs.
====================

* tag 'rds-odp-for-5.5':
  net/rds: Use prefetch for On-Demand-Paging MR
  net/rds: Handle ODP mr registration/unregistration
  net/rds: Detect need of On-Demand-Paging memory registration
  RDMA/mlx5: Fix handling of IOVA != user_va in ODP paths
  IB/mlx5: Mask out unsupported ODP capabilities for kernel QPs
  RDMA/mlx5: Don't fake udata for kernel path
  IB/mlx5: Add ODP WQE handlers for kernel QPs
  IB/core: Add interface to advise_mr for kernel users
  IB/core: Introduce ib_reg_user_mr
  IB: Allow calls to ib_umem_get from kernel ULPs

Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

10 files changed, 82 insertions, 91 deletions

diff --git a/security/apparmor/apparmorfs.c b/security/apparmor/apparmorfs.c
index 09996f2552ee..47aff8700547 100644
--- a/security/apparmor/apparmorfs.c
+++ b/security/apparmor/apparmorfs.c
@@ -623,7 +623,7 @@ static __poll_t ns_revision_poll(struct file *file, poll_table *pt)
 
 void __aa_bump_ns_revision(struct aa_ns *ns)
 {
-	ns->revision++;
+	WRITE_ONCE(ns->revision, ns->revision + 1);
 	wake_up_interruptible(&ns->wait);
 }
 
diff --git a/security/apparmor/domain.c b/security/apparmor/domain.c
index 9be7ccb8379e..6ceb74e0f789 100644
--- a/security/apparmor/domain.c
+++ b/security/apparmor/domain.c
@@ -317,6 +317,7 @@ static int aa_xattrs_match(const struct linux_binprm *bprm,
 
 	if (!bprm || !profile->xattr_count)
 		return 0;
+	might_sleep();
 
 	/* transition from exec match to xattr set */
 	state = aa_dfa_null_transition(profile->xmatch, state);
@@ -361,10 +362,11 @@ out:
 }
 
 /**
- * __attach_match_ - find an attachment match
+ * find_attach - do attachment search for unconfined processes
  * @bprm - binprm structure of transitioning task
- * @name - to match against  (NOT NULL)
+ * @ns: the current namespace  (NOT NULL)
  * @head - profile list to walk  (NOT NULL)
+ * @name - to match against  (NOT NULL)
  * @info - info message if there was an error (NOT NULL)
  *
  * Do a linear search on the profiles in the list.  There is a matching
@@ -374,12 +376,11 @@ out:
  *
  * Requires: @head not be shared or have appropriate locks held
  *
- * Returns: profile or NULL if no match found
+ * Returns: label or NULL if no match found
  */
-static struct aa_profile *__attach_match(const struct linux_binprm *bprm,
-					 const char *name,
-					 struct list_head *head,
-					 const char **info)
+static struct aa_label *find_attach(const struct linux_binprm *bprm,
+				    struct aa_ns *ns, struct list_head *head,
+				    const char *name, const char **info)
 {
 	int candidate_len = 0, candidate_xattrs = 0;
 	bool conflict = false;
@@ -388,6 +389,8 @@ static struct aa_profile *__attach_match(const struct linux_binprm *bprm,
 	AA_BUG(!name);
 	AA_BUG(!head);
 
+	rcu_read_lock();
+restart:
 	list_for_each_entry_rcu(profile, head, base.list) {
 		if (profile->label.flags & FLAG_NULL &&
 		    &profile->label == ns_unconfined(profile->ns))
@@ -413,16 +416,32 @@ static struct aa_profile *__attach_match(const struct linux_binprm *bprm,
 			perm = dfa_user_allow(profile->xmatch, state);
 			/* any accepting state means a valid match. */
 			if (perm & MAY_EXEC) {
-				int ret;
+				int ret = 0;
 
 				if (count < candidate_len)
 					continue;
 
-				ret = aa_xattrs_match(bprm, profile, state);
-				/* Fail matching if the xattrs don't match */
-				if (ret < 0)
-					continue;
-
+				if (bprm && profile->xattr_count) {
+					long rev = READ_ONCE(ns->revision);
+
+					if (!aa_get_profile_not0(profile))
+						goto restart;
+					rcu_read_unlock();
+					ret = aa_xattrs_match(bprm, profile,
+							      state);
+					rcu_read_lock();
+					aa_put_profile(profile);
+					if (rev !=
+					    READ_ONCE(ns->revision))
+						/* policy changed */
+						goto restart;
+					/*
+					 * Fail matching if the xattrs don't
+					 * match
+					 */
+					if (ret < 0)
+						continue;
+				}
 				/*
 				 * TODO: allow for more flexible best match
 				 *
@@ -445,43 +464,28 @@ static struct aa_profile *__attach_match(const struct linux_binprm *bprm,
 				candidate_xattrs = ret;
 				conflict = false;
 			}
-		} else if (!strcmp(profile->base.name, name))
+		} else if (!strcmp(profile->base.name, name)) {
 			/*
 			 * old exact non-re match, without conditionals such
 			 * as xattrs. no more searching required
 			 */
-			return profile;
+			candidate = profile;
+			goto out;
+		}
 	}
 
-	if (conflict) {
-		*info = "conflicting profile attachments";
+	if (!candidate || conflict) {
+		if (conflict)
+			*info = "conflicting profile attachments";
+		rcu_read_unlock();
 		return NULL;
 	}
 
-	return candidate;
-}
-
-/**
- * find_attach - do attachment search for unconfined processes
- * @bprm - binprm structure of transitioning task
- * @ns: the current namespace  (NOT NULL)
- * @list: list to search  (NOT NULL)
- * @name: the executable name to match against  (NOT NULL)
- * @info: info message if there was an error
- *
- * Returns: label or NULL if no match found
- */
-static struct aa_label *find_attach(const struct linux_binprm *bprm,
-				    struct aa_ns *ns, struct list_head *list,
-				    const char *name, const char **info)
-{
-	struct aa_profile *profile;
-
-	rcu_read_lock();
-	profile = aa_get_profile(__attach_match(bprm, name, list, info));
+out:
+	candidate = aa_get_newest_profile(candidate);
 	rcu_read_unlock();
 
-	return profile ? &profile->label : NULL;
+	return &candidate->label;
 }
 
 static const char *next_name(int xtype, const char *name)
diff --git a/security/apparmor/file.c b/security/apparmor/file.c
index fe2ebe5e865e..f1caf3674e1c 100644
--- a/security/apparmor/file.c
+++ b/security/apparmor/file.c
@@ -618,8 +618,7 @@ int aa_file_perm(const char *op, struct aa_label *label, struct file *file,
 	fctx = file_ctx(file);
 
 	rcu_read_lock();
-	flabel  = aa_get_newest_label(rcu_dereference(fctx->label));
-	rcu_read_unlock();
+	flabel  = rcu_dereference(fctx->label);
 	AA_BUG(!flabel);
 
 	/* revalidate access, if task is unconfined, or the cached cred
@@ -631,9 +630,13 @@ int aa_file_perm(const char *op, struct aa_label *label, struct file *file,
 	 */
 	denied = request & ~fctx->allow;
 	if (unconfined(label) || unconfined(flabel) ||
-	    (!denied && aa_label_is_subset(flabel, label)))
+	    (!denied && aa_label_is_subset(flabel, label))) {
+		rcu_read_unlock();
 		goto done;
+	}
 
+	flabel  = aa_get_newest_label(flabel);
+	rcu_read_unlock();
 	/* TODO: label cross check */
 
 	if (file->f_path.mnt && path_mediated_fs(file->f_path.dentry))
@@ -643,8 +646,9 @@ int aa_file_perm(const char *op, struct aa_label *label, struct file *file,
 	else if (S_ISSOCK(file_inode(file)->i_mode))
 		error = __file_sock_perm(op, label, flabel, file, request,
 					 denied);
-done:
 	aa_put_label(flabel);
+
+done:
 	return error;
 }
 
diff --git a/security/apparmor/mount.c b/security/apparmor/mount.c
index 4ed6688f9d40..e0828ee7a345 100644
--- a/security/apparmor/mount.c
+++ b/security/apparmor/mount.c
@@ -442,7 +442,7 @@ int aa_bind_mount(struct aa_label *label, const struct path *path,
 	buffer = aa_get_buffer(false);
 	old_buffer = aa_get_buffer(false);
 	error = -ENOMEM;
-	if (!buffer || old_buffer)
+	if (!buffer || !old_buffer)
 		goto out;
 
 	error = fn_for_each_confined(label, profile,
diff --git a/security/apparmor/policy.c b/security/apparmor/policy.c
index 03104830c913..269f2f53c0b1 100644
--- a/security/apparmor/policy.c
+++ b/security/apparmor/policy.c
@@ -1125,8 +1125,8 @@ ssize_t aa_remove_profiles(struct aa_ns *policy_ns, struct aa_label *subj,
 	if (!name) {
 		/* remove namespace - can only happen if fqname[0] == ':' */
 		mutex_lock_nested(&ns->parent->lock, ns->level);
-		__aa_remove_ns(ns);
 		__aa_bump_ns_revision(ns);
+		__aa_remove_ns(ns);
 		mutex_unlock(&ns->parent->lock);
 	} else {
 		/* remove profile */
@@ -1138,9 +1138,9 @@ ssize_t aa_remove_profiles(struct aa_ns *policy_ns, struct aa_label *subj,
 			goto fail_ns_lock;
 		}
 		name = profile->base.hname;
+		__aa_bump_ns_revision(ns);
 		__remove_profile(profile);
 		__aa_labelset_update_subtree(ns);
-		__aa_bump_ns_revision(ns);
 		mutex_unlock(&ns->lock);
 	}
 
diff --git a/security/tomoyo/common.c b/security/tomoyo/common.c
index dd3d5942e669..c36bafbcd77e 100644
--- a/security/tomoyo/common.c
+++ b/security/tomoyo/common.c
@@ -951,7 +951,8 @@ static bool tomoyo_manager(void)
 	exe = tomoyo_get_exe();
 	if (!exe)
 		return false;
-	list_for_each_entry_rcu(ptr, &tomoyo_kernel_namespace.policy_list[TOMOYO_ID_MANAGER], head.list) {
+	list_for_each_entry_rcu(ptr, &tomoyo_kernel_namespace.policy_list[TOMOYO_ID_MANAGER], head.list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (!ptr->head.is_deleted &&
 		    (!tomoyo_pathcmp(domainname, ptr->manager) ||
 		     !strcmp(exe, ptr->manager->name))) {
@@ -1095,7 +1096,8 @@ static int tomoyo_delete_domain(char *domainname)
 	if (mutex_lock_interruptible(&tomoyo_policy_lock))
 		return -EINTR;
 	/* Is there an active domain? */
-	list_for_each_entry_rcu(domain, &tomoyo_domain_list, list) {
+	list_for_each_entry_rcu(domain, &tomoyo_domain_list, list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		/* Never delete tomoyo_kernel_domain */
 		if (domain == &tomoyo_kernel_domain)
 			continue;
@@ -2778,7 +2780,8 @@ void tomoyo_check_profile(void)
 
 	tomoyo_policy_loaded = true;
 	pr_info("TOMOYO: 2.6.0\n");
-	list_for_each_entry_rcu(domain, &tomoyo_domain_list, list) {
+	list_for_each_entry_rcu(domain, &tomoyo_domain_list, list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		const u8 profile = domain->profile;
 		struct tomoyo_policy_namespace *ns = domain->ns;
 
diff --git a/security/tomoyo/domain.c b/security/tomoyo/domain.c
index 8526a0a74023..7869d6a9980b 100644
--- a/security/tomoyo/domain.c
+++ b/security/tomoyo/domain.c
@@ -41,7 +41,8 @@ int tomoyo_update_policy(struct tomoyo_acl_head *new_entry, const int size,
 
 	if (mutex_lock_interruptible(&tomoyo_policy_lock))
 		return -ENOMEM;
-	list_for_each_entry_rcu(entry, list, list) {
+	list_for_each_entry_rcu(entry, list, list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (entry->is_deleted == TOMOYO_GC_IN_PROGRESS)
 			continue;
 		if (!check_duplicate(entry, new_entry))
@@ -119,7 +120,8 @@ int tomoyo_update_domain(struct tomoyo_acl_info *new_entry, const int size,
 	}
 	if (mutex_lock_interruptible(&tomoyo_policy_lock))
 		goto out;
-	list_for_each_entry_rcu(entry, list, list) {
+	list_for_each_entry_rcu(entry, list, list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (entry->is_deleted == TOMOYO_GC_IN_PROGRESS)
 			continue;
 		if (!tomoyo_same_acl_head(entry, new_entry) ||
@@ -166,7 +168,8 @@ void tomoyo_check_acl(struct tomoyo_request_info *r,
 	u16 i = 0;
 
 retry:
-	list_for_each_entry_rcu(ptr, list, list) {
+	list_for_each_entry_rcu(ptr, list, list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (ptr->is_deleted || ptr->type != r->param_type)
 			continue;
 		if (!check_entry(r, ptr))
@@ -298,7 +301,8 @@ static inline bool tomoyo_scan_transition
 {
 	const struct tomoyo_transition_control *ptr;
 
-	list_for_each_entry_rcu(ptr, list, head.list) {
+	list_for_each_entry_rcu(ptr, list, head.list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (ptr->head.is_deleted || ptr->type != type)
 			continue;
 		if (ptr->domainname) {
@@ -735,7 +739,8 @@ retry:
 
 		/* Check 'aggregator' directive. */
 		candidate = &exename;
-		list_for_each_entry_rcu(ptr, list, head.list) {
+		list_for_each_entry_rcu(ptr, list, head.list,
+					srcu_read_lock_held(&tomoyo_ss)) {
 			if (ptr->head.is_deleted ||
 			    !tomoyo_path_matches_pattern(&exename,
 							 ptr->original_name))
diff --git a/security/tomoyo/group.c b/security/tomoyo/group.c
index a37c7dc66e44..1cecdd797597 100644
--- a/security/tomoyo/group.c
+++ b/security/tomoyo/group.c
@@ -133,7 +133,8 @@ tomoyo_path_matches_group(const struct tomoyo_path_info *pathname,
 {
 	struct tomoyo_path_group *member;
 
-	list_for_each_entry_rcu(member, &group->member_list, head.list) {
+	list_for_each_entry_rcu(member, &group->member_list, head.list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (member->head.is_deleted)
 			continue;
 		if (!tomoyo_path_matches_pattern(pathname, member->member_name))
@@ -161,7 +162,8 @@ bool tomoyo_number_matches_group(const unsigned long min,
 	struct tomoyo_number_group *member;
 	bool matched = false;
 
-	list_for_each_entry_rcu(member, &group->member_list, head.list) {
+	list_for_each_entry_rcu(member, &group->member_list, head.list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (member->head.is_deleted)
 			continue;
 		if (min > member->number.values[1] ||
@@ -191,7 +193,8 @@ bool tomoyo_address_matches_group(const bool is_ipv6, const __be32 *address,
 	bool matched = false;
 	const u8 size = is_ipv6 ? 16 : 4;
 
-	list_for_each_entry_rcu(member, &group->member_list, head.list) {
+	list_for_each_entry_rcu(member, &group->member_list, head.list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (member->head.is_deleted)
 			continue;
 		if (member->address.is_ipv6 != is_ipv6)
diff --git a/security/tomoyo/realpath.c b/security/tomoyo/realpath.c
index e7832448d721..bf38fc1b59b2 100644
--- a/security/tomoyo/realpath.c
+++ b/security/tomoyo/realpath.c
@@ -218,31 +218,6 @@ out:
 }
 
 /**
- * tomoyo_get_socket_name - Get the name of a socket.
- *
- * @path:   Pointer to "struct path".
- * @buffer: Pointer to buffer to return value in.
- * @buflen: Sizeof @buffer.
- *
- * Returns the buffer.
- */
-static char *tomoyo_get_socket_name(const struct path *path, char * const buffer,
-				    const int buflen)
-{
-	struct inode *inode = d_backing_inode(path->dentry);
-	struct socket *sock = inode ? SOCKET_I(inode) : NULL;
-	struct sock *sk = sock ? sock->sk : NULL;
-
-	if (sk) {
-		snprintf(buffer, buflen, "socket:[family=%u:type=%u:protocol=%u]",
-			 sk->sk_family, sk->sk_type, sk->sk_protocol);
-	} else {
-		snprintf(buffer, buflen, "socket:[unknown]");
-	}
-	return buffer;
-}
-
-/**
  * tomoyo_realpath_from_path - Returns realpath(3) of the given pathname but ignores chroot'ed root.
  *
  * @path: Pointer to "struct path".
@@ -279,12 +254,7 @@ char *tomoyo_realpath_from_path(const struct path *path)
 			break;
 		/* To make sure that pos is '\0' terminated. */
 		buf[buf_len - 1] = '\0';
-		/* Get better name for socket. */
-		if (sb->s_magic == SOCKFS_MAGIC) {
-			pos = tomoyo_get_socket_name(path, buf, buf_len - 1);
-			goto encode;
-		}
-		/* For "pipe:[\$]". */
+		/* For "pipe:[\$]" and "socket:[\$]". */
 		if (dentry->d_op && dentry->d_op->d_dname) {
 			pos = dentry->d_op->d_dname(dentry, buf, buf_len - 1);
 			goto encode;
diff --git a/security/tomoyo/util.c b/security/tomoyo/util.c
index 52752e1a84ed..eba0b3395851 100644
--- a/security/tomoyo/util.c
+++ b/security/tomoyo/util.c
@@ -594,7 +594,8 @@ struct tomoyo_domain_info *tomoyo_find_domain(const char *domainname)
 
 	name.name = domainname;
 	tomoyo_fill_path_info(&name);
-	list_for_each_entry_rcu(domain, &tomoyo_domain_list, list) {
+	list_for_each_entry_rcu(domain, &tomoyo_domain_list, list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		if (!domain->is_deleted &&
 		    !tomoyo_pathcmp(&name, domain->domainname))
 			return domain;
@@ -1028,7 +1029,8 @@ bool tomoyo_domain_quota_is_ok(struct tomoyo_request_info *r)
 		return false;
 	if (!domain)
 		return true;
-	list_for_each_entry_rcu(ptr, &domain->acl_info_list, list) {
+	list_for_each_entry_rcu(ptr, &domain->acl_info_list, list,
+				srcu_read_lock_held(&tomoyo_ss)) {
 		u16 perm;
 		u8 i;