On 2026/1/23 00:02, [email protected] wrote:
>> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
>> index 5936f8e2996f..c5863487ee73 100644
>> --- a/include/linux/bpf.h
>> +++ b/include/linux/bpf.h
>> @@ -3935,9 +3935,16 @@ static inline int bpf_map_check_op_flags(struct
>> bpf_map *map, u64 flags, u64 all
>> if ((u32)flags & ~allowed_flags)
>> return -EINVAL;
>>
>> + /* BPF_NOEXIST and BPF_EXIST are mutually exclusive. */
>> + if ((flags & (BPF_NOEXIST | BPF_EXIST)) == (BPF_NOEXIST | BPF_EXIST))
>> + return -EINVAL;
>> +
>> if ((flags & BPF_F_LOCK) && !btf_record_has_field(map->record,
>> BPF_SPIN_LOCK))
>> return -EINVAL;
>>
>> + if ((flags & BPF_F_LOCK) && btf_record_has_field(map->record,
>> ~BPF_SPIN_LOCK))
>> + return -EOPNOTSUPP;
>> +
>> if (!(flags & BPF_F_CPU) && flags >> 32)
>> return -EINVAL;
>
> [ ... ]
>
>> diff --git a/kernel/bpf/bpf_local_storage.c b/kernel/bpf/bpf_local_storage.c
>> index e2fe6c32822b..80b50091cbbf 100644
>> --- a/kernel/bpf/bpf_local_storage.c
>> +++ b/kernel/bpf/bpf_local_storage.c
>> @@ -493,13 +493,6 @@ bpf_local_storage_update(void *owner, struct
>> bpf_local_storage_map *smap,
>> unsigned long flags;
>> int err;
>>
>> - /* BPF_EXIST and BPF_NOEXIST cannot be both set */
>> - if (unlikely((map_flags & ~BPF_F_LOCK) > BPF_EXIST) ||
>> - /* BPF_F_LOCK can only be used in a value with spin_lock */
>> - unlikely((map_flags & BPF_F_LOCK) &&
>> - !btf_record_has_field(smap->map.record, BPF_SPIN_LOCK)))
>> - return ERR_PTR(-EINVAL);
> ^^^^
>
> This validation was removed from bpf_local_storage_update() but it appears
> that no call to bpf_map_check_op_flags() was added in the local storage map
> update functions. For array and hash maps, calls to bpf_map_check_op_flags()
> were added in array_map_update_elem() and htab_map_update_elem(). However,
> the local storage update functions like bpf_cgrp_storage_update_elem(),
> bpf_fd_inode_storage_update_elem(), and bpf_fd_sk_storage_update_elem() pass
> map_flags directly to bpf_local_storage_update() without calling
> bpf_map_check_op_flags() first.
>
> The syscall path does call bpf_map_check_op_flags() in map_update_elem() with
> allowed_flags set to ~0, which catches the NOEXIST/EXIST mutual exclusion and
> BPF_F_LOCK validation. But with allowed_flags=~0, the check for unknown flags
> ((u32)flags & ~allowed_flags) always passes zero, meaning unknown flags like
> 8 or 16 would not be rejected.
>
Correct.
Such flags should indeed be rejected in the local storage update path.
Thanks,
Leon
> The old check (map_flags & ~BPF_F_LOCK) > BPF_EXIST rejected any unknown flag
> bits. Is there a missing call to bpf_map_check_op_flags() in the local storage
> update functions that should specify the allowed flags?
>
>> -
>> if (gfp_flags == GFP_KERNEL && (map_flags & ~BPF_F_LOCK) != BPF_NOEXIST)
>> return ERR_PTR(-EINVAL);
>
> [ ... ]
>
>
> ---
> 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/21254704407
