Merge tag 'slab-for-6.5' of git://git.kernel.org/pub/scm/linux/kernel/git/vbabka/slab

Pull slab updates from Vlastimil Babka:

 - SLAB deprecation:

   Following the discussion at LSF/MM 2023 [1] and no objections, the
   SLAB allocator is deprecated by renaming the config option (to make
   its users notice) to CONFIG_SLAB_DEPRECATED with updated help text.
   SLUB should be used instead. Existing defconfigs with CONFIG_SLAB are
   also updated.

 - SLAB_NO_MERGE kmem_cache flag (Jesper Dangaard Brouer):

   There are (very limited) cases where kmem_cache merging is
   undesirable, and existing ways to prevent it are hacky. Introduce a
   new flag to do that cleanly and convert the existing hacky users.
   Btrfs plans to use this for debug kernel builds (that use case is
   always fine), networking for performance reasons (that should be very
   rare).

 - Replace the usage of weak PRNGs (David Keisar Schmidt):

   In addition to using stronger RNGs for the security related features,
   the code is a bit cleaner.

 - Misc code cleanups (SeongJae Parki, Xiongwei Song, Zhen Lei, and
   zhaoxinchao)

Link: https://lwn.net/Articles/932201/ [1]

* tag 'slab-for-6.5' of git://git.kernel.org/pub/scm/linux/kernel/git/vbabka/slab:
  mm/slab_common: use SLAB_NO_MERGE instead of negative refcount
  mm/slab: break up RCU readers on SLAB_TYPESAFE_BY_RCU example code
  mm/slab: add a missing semicolon on SLAB_TYPESAFE_BY_RCU example code
  mm/slab_common: reduce an if statement in create_cache()
  mm/slab: introduce kmem_cache flag SLAB_NO_MERGE
  mm/slab: rename CONFIG_SLAB to CONFIG_SLAB_DEPRECATED
  mm/slab: remove HAVE_HARDENED_USERCOPY_ALLOCATOR
  mm/slab_common: Replace invocation of weak PRNG
  mm/slab: Replace invocation of weak PRNG
  slub: Don't read nr_slabs and total_objects directly
  slub: Remove slabs_node() function
  slub: Remove CONFIG_SMP defined check
  slub: Put objects_show() into CONFIG_SLUB_DEBUG enabled block
  slub: Correct the error code when slab_kset is NULL
  mm/slab: correct return values in comment for _kmem_cache_create()

1 files changed, 18 insertions, 4 deletions

diff --git a/include/linux/slab.h b/include/linux/slab.h
index ca53425e9b32..791f7453a04f 100644
--- a/include/linux/slab.h
+++ b/include/linux/slab.h
@@ -54,16 +54,18 @@
  * stays valid, the trick to using this is relying on an independent
  * object validation pass. Something like:
  *
- *  rcu_read_lock()
- * again:
+ * begin:
+ *  rcu_read_lock();
  *  obj = lockless_lookup(key);
  *  if (obj) {
  *    if (!try_get_ref(obj)) // might fail for free objects
- *      goto again;
+ *      rcu_read_unlock();
+ *      goto begin;
  *
  *    if (obj->key != key) { // not the object we expected
  *      put_ref(obj);
- *      goto again;
+ *      rcu_read_unlock();
+ *      goto begin;
  *    }
  *  }
  *  rcu_read_unlock();
@@ -107,6 +109,18 @@
 /* Avoid kmemleak tracing */
 #define SLAB_NOLEAKTRACE	((slab_flags_t __force)0x00800000U)
 
+/*
+ * Prevent merging with compatible kmem caches. This flag should be used
+ * cautiously. Valid use cases:
+ *
+ * - caches created for self-tests (e.g. kunit)
+ * - general caches created and used by a subsystem, only when a
+ *   (subsystem-specific) debug option is enabled
+ * - performance critical caches, should be very rare and consulted with slab
+ *   maintainers, and not used together with CONFIG_SLUB_TINY
+ */
+#define SLAB_NO_MERGE		((slab_flags_t __force)0x01000000U)
+
 /* Fault injection mark */
 #ifdef CONFIG_FAILSLAB
 # define SLAB_FAILSLAB		((slab_flags_t __force)0x02000000U)