[PATCH bpf] bpf: do not use kmalloc_nolock when !HAVE_CMPXCHG_DOUBLE

[PATCH bpf] bpf: do not use kmalloc_nolock when !HAVE_CMPXCHG_DOUBLE

[Levi Zim via B4 Relay]

From: Levi Zim <rsworktech@outlook.com>

kmalloc_nolock always fails for architectures that lack cmpxchg16b.
For example, this causes bpf_task_storage_get with flag
BPF_LOCAL_STORAGE_GET_F_CREATE to fails on riscv64 6.19 kernel.

Fix it by enabling use_kmalloc_nolock only when HAVE_CMPXCHG_DOUBLE

Fixes: f484f4a3e058 ("bpf: Replace bpf memory allocator with kmalloc_nolock() in local storage")
Cc: stable@vger.kernel.org
Signed-off-by: Levi Zim <rsworktech@outlook.com>
---
I find that bpf_task_storage_get with flag BPF_LOCAL_STORAGE_GET_F_CREATE
always fails for me on 6.19 kernel on riscv64 and bisected it.

In f484f4a3e058 ("bpf: Replace bpf memory allocator with kmalloc_nolock()
in local storage"), bpf memory allocator is replaced with kmalloc_nolock.
This approach is problematic for architectures that lack CMPXCHG_DOUBLE
because kmalloc_nolock always fails in this case:

In function kmalloc_nolock (kmalloc_nolock_noprof): 

	if (!(s->flags & __CMPXCHG_DOUBLE) && !kmem_cache_debug(s))
		/*
		 * kmalloc_nolock() is not supported on architectures that
		 * don't implement cmpxchg16b, but debug caches don't use
		 * per-cpu slab and per-cpu partial slabs. They rely on
		 * kmem_cache_node->list_lock, so kmalloc_nolock() can
		 * attempt to allocate from debug caches by
		 * spin_trylock_irqsave(&n->list_lock, ...)
		 */
		return NULL;

Fix it by enabling use_kmalloc_nolock only when HAVE_CMPXCHG_DOUBLE.

Note for stable: this only needs to be picked into v6.19 if the patch
makes it into 7.0.
---
 include/linux/bpf_local_storage.h | 2 ++
 kernel/bpf/bpf_cgrp_storage.c     | 2 +-
 kernel/bpf/bpf_local_storage.c    | 3 ++-
 kernel/bpf/bpf_task_storage.c     | 2 +-
 4 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/include/linux/bpf_local_storage.h b/include/linux/bpf_local_storage.h
index 8157e8da61d40..a7ae5dde15bcb 100644
--- a/include/linux/bpf_local_storage.h
+++ b/include/linux/bpf_local_storage.h
@@ -19,6 +19,8 @@
 
 #define BPF_LOCAL_STORAGE_CACHE_SIZE	16
 
+static const bool KMALLOC_NOLOCK_SUPPORTED = IS_ENABLED(CONFIG_HAVE_CMPXCHG_DOUBLE);
+
 struct bpf_local_storage_map_bucket {
 	struct hlist_head list;
 	rqspinlock_t lock;
diff --git a/kernel/bpf/bpf_cgrp_storage.c b/kernel/bpf/bpf_cgrp_storage.c
index c2a2ead1f466d..09c557e426968 100644
--- a/kernel/bpf/bpf_cgrp_storage.c
+++ b/kernel/bpf/bpf_cgrp_storage.c
@@ -114,7 +114,7 @@ static int notsupp_get_next_key(struct bpf_map *map, void *key, void *next_key)
 
 static struct bpf_map *cgroup_storage_map_alloc(union bpf_attr *attr)
 {
-	return bpf_local_storage_map_alloc(attr, &cgroup_cache, true);
+	return bpf_local_storage_map_alloc(attr, &cgroup_cache, KMALLOC_NOLOCK_SUPPORTED);
 }
 
 static void cgroup_storage_map_free(struct bpf_map *map)
diff --git a/kernel/bpf/bpf_local_storage.c b/kernel/bpf/bpf_local_storage.c
index 9c96a4477f81a..8e4b0fe6d12af 100644
--- a/kernel/bpf/bpf_local_storage.c
+++ b/kernel/bpf/bpf_local_storage.c
@@ -894,7 +894,8 @@ bpf_local_storage_map_alloc(union bpf_attr *attr,
 	 * preemptible context. Thus, enforce all storages to use
 	 * kmalloc_nolock() when CONFIG_PREEMPT_RT is enabled.
 	 */
-	smap->use_kmalloc_nolock = IS_ENABLED(CONFIG_PREEMPT_RT) ? true : use_kmalloc_nolock;
+	smap->use_kmalloc_nolock = IS_ENABLED(CONFIG_PREEMPT_RT) &&
+		KMALLOC_NOLOCK_SUPPORTED ? true : use_kmalloc_nolock;
 
 	smap->cache_idx = bpf_local_storage_cache_idx_get(cache);
 	return &smap->map;
diff --git a/kernel/bpf/bpf_task_storage.c b/kernel/bpf/bpf_task_storage.c
index 605506792b5b4..9f74fd1ef7f46 100644
--- a/kernel/bpf/bpf_task_storage.c
+++ b/kernel/bpf/bpf_task_storage.c
@@ -212,7 +212,7 @@ static int notsupp_get_next_key(struct bpf_map *map, void *key, void *next_key)
 
 static struct bpf_map *task_storage_map_alloc(union bpf_attr *attr)
 {
-	return bpf_local_storage_map_alloc(attr, &task_cache, true);
+	return bpf_local_storage_map_alloc(attr, &task_cache, KMALLOC_NOLOCK_SUPPORTED);
 }
 
 static void task_storage_map_free(struct bpf_map *map)

---
base-commit: e06e6b8001233241eb5b2e2791162f0585f50f4b
change-id: 20260314-bpf-kmalloc-nolock-60da80e613de

Best regards,
-- 
Levi Zim <rsworktech@outlook.com>

Re: [PATCH bpf] bpf: do not use kmalloc_nolock when !HAVE_CMPXCHG_DOUBLE

[bot+bpf-ci]

[-- Attachment #1: Type: text/plain, Size: 1254 bytes --]

> diff --git a/kernel/bpf/bpf_local_storage.c b/kernel/bpf/bpf_local_storage.c
> index 9c96a4477f81..8e4b0fe6d12a 100644
> --- a/kernel/bpf/bpf_local_storage.c
> +++ b/kernel/bpf/bpf_local_storage.c
> @@ -894,7 +894,8 @@ bpf_local_storage_map_alloc(union bpf_attr *attr,
>  	 * preemptible context. Thus, enforce all storages to use
>  	 * kmalloc_nolock() when CONFIG_PREEMPT_RT is enabled.
>  	 */
> -	smap->use_kmalloc_nolock = IS_ENABLED(CONFIG_PREEMPT_RT) ? true : use_kmalloc_nolock;
> +	smap->use_kmalloc_nolock = IS_ENABLED(CONFIG_PREEMPT_RT) &&
> +		KMALLOC_NOLOCK_SUPPORTED ? true : use_kmalloc_nolock;

Should the comment above be updated to reflect the new behavior?
It currently says "enforce all storages to use kmalloc_nolock()
when CONFIG_PREEMPT_RT is enabled", but the code now only enforces
kmalloc_nolock when both CONFIG_PREEMPT_RT and
CONFIG_HAVE_CMPXCHG_DOUBLE are enabled. On PREEMPT_RT without
HAVE_CMPXCHG_DOUBLE, the override is silently skipped and the
caller's value is used instead.


---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

CI run summary: https://github.com/kernel-patches/bpf/actions/runs/23086356774

Re: [PATCH bpf] bpf: do not use kmalloc_nolock when !HAVE_CMPXCHG_DOUBLE

[Levi Zim]

Hi,

On 3/14/26 7:09 PM, bot+bpf-ci@kernel.org wrote:
>> diff --git a/kernel/bpf/bpf_local_storage.c b/kernel/bpf/bpf_local_storage.c
>> index 9c96a4477f81..8e4b0fe6d12a 100644
>> --- a/kernel/bpf/bpf_local_storage.c
>> +++ b/kernel/bpf/bpf_local_storage.c
>> @@ -894,7 +894,8 @@ bpf_local_storage_map_alloc(union bpf_attr *attr,
>>  	 * preemptible context. Thus, enforce all storages to use
>>  	 * kmalloc_nolock() when CONFIG_PREEMPT_RT is enabled.
>>  	 */
>> -	smap->use_kmalloc_nolock = IS_ENABLED(CONFIG_PREEMPT_RT) ? true : use_kmalloc_nolock;
>> +	smap->use_kmalloc_nolock = IS_ENABLED(CONFIG_PREEMPT_RT) &&
>> +		KMALLOC_NOLOCK_SUPPORTED ? true : use_kmalloc_nolock;
> 
> Should the comment above be updated to reflect the new behavior?
> It currently says "enforce all storages to use kmalloc_nolock()
> when CONFIG_PREEMPT_RT is enabled", but the code now only enforces
> kmalloc_nolock when both CONFIG_PREEMPT_RT and
> CONFIG_HAVE_CMPXCHG_DOUBLE are enabled. On PREEMPT_RT without
> HAVE_CMPXCHG_DOUBLE, the override is silently skipped and the
> caller's value is used instead.

I didn't give it too much thought initially. So the kmalloc_nolock() usage in PREEMPT_RT
context is necessary for correctness. I will remove "&& KMALLOC_NOLOCK_SUPPORTED" in v2
and add a comment to explain the current limitation when PREEMPT_RT is combined with
an architecture that doesn't HAVE_CMPXCHG_DOUBLE.

Best regards,
Levi

> 
> ---
> AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
> See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
> 
> CI run summary: https://github.com/kernel-patches/bpf/actions/runs/23086356774

Re: [PATCH bpf] bpf: do not use kmalloc_nolock when !HAVE_CMPXCHG_DOUBLE

[kernel test robot]

Hi Levi,

kernel test robot noticed the following build warnings:

[auto build test WARNING on e06e6b8001233241eb5b2e2791162f0585f50f4b]

url:    https://github.com/intel-lab-lkp/linux/commits/Levi-Zim-via-B4-Relay/bpf-do-not-use-kmalloc_nolock-when-HAVE_CMPXCHG_DOUBLE/20260314-195705
base:   e06e6b8001233241eb5b2e2791162f0585f50f4b
patch link:    https://lore.kernel.org/r/20260314-bpf-kmalloc-nolock-v1-1-24abf3f75a9f%40outlook.com
patch subject: [PATCH bpf] bpf: do not use kmalloc_nolock when !HAVE_CMPXCHG_DOUBLE
config: x86_64-rhel-9.4 (https://download.01.org/0day-ci/archive/20260314/202603141857.sLJ0vJoa-lkp@intel.com/config)
compiler: gcc-14 (Debian 14.2.0-19) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260314/202603141857.sLJ0vJoa-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202603141857.sLJ0vJoa-lkp@intel.com/

All warnings (new ones prefixed by >>):

   In file included from include/net/bpf_sk_storage.h:15,
                    from net/core/sock.c:141:
>> include/linux/bpf_local_storage.h:22:19: warning: 'KMALLOC_NOLOCK_SUPPORTED' defined but not used [-Wunused-const-variable=]
      22 | static const bool KMALLOC_NOLOCK_SUPPORTED = IS_ENABLED(CONFIG_HAVE_CMPXCHG_DOUBLE);
         |                   ^~~~~~~~~~~~~~~~~~~~~~~~


vim +/KMALLOC_NOLOCK_SUPPORTED +22 include/linux/bpf_local_storage.h

    21	
  > 22	static const bool KMALLOC_NOLOCK_SUPPORTED = IS_ENABLED(CONFIG_HAVE_CMPXCHG_DOUBLE);
    23	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki