On 14/09/2016 20:51, Alexei Starovoitov wrote:
> On Wed, Sep 14, 2016 at 09:23:56AM +0200, Mickaël Salaün wrote:
>> This new arraymap looks like a set and brings new properties:
>> * strong typing of entries: the eBPF functions get the array type of
>>   elements instead of CONST_PTR_TO_MAP (e.g.
>>   CONST_PTR_TO_LANDLOCK_HANDLE_FS);
>> * force sequential filling (i.e. replace or append-only update), which
>>   allow quick browsing of all entries.
>>
>> This strong typing is useful to statically check if the content of a map
>> can be passed to an eBPF function. For example, Landlock use it to store
>> and manage kernel objects (e.g. struct file) instead of dealing with
>> userland raw data. This improve efficiency and ensure that an eBPF
>> program can only call functions with the right high-level arguments.
>>
>> The enum bpf_map_handle_type list low-level types (e.g.
>> BPF_MAP_HANDLE_TYPE_LANDLOCK_FS_FD) which are identified when
>> updating a map entry (handle). This handle types are used to infer a
>> high-level arraymap type which are listed in enum bpf_map_array_type
>> (e.g. BPF_MAP_ARRAY_TYPE_LANDLOCK_FS).
>>
>> For now, this new arraymap is only used by Landlock LSM (cf. next
>> commits) but it could be useful for other needs.
>>
>> Changes since v2:
>> * add a RLIMIT_NOFILE-based limit to the maximum number of arraymap
>>   handle entries (suggested by Andy Lutomirski)
>> * remove useless checks
>>
>> Changes since v1:
>> * arraymap of handles replace custom checker groups
>> * simpler userland API
>>
>> Signed-off-by: Mickaël Salaün <mic@digikod.net>
>> Cc: Alexei Starovoitov <ast@kernel.org>
>> Cc: Andy Lutomirski <luto@amacapital.net>
>> Cc: Daniel Borkmann <daniel@iogearbox.net>
>> Cc: David S. Miller <davem@davemloft.net>
>> Cc: Kees Cook <keescook@chromium.org>
>> Link: https://lkml.kernel.org/r/CALCETrWwTiz3kZTkEgOW24-DvhQq6LftwEXh77FD2G5o71yD7g@mail.gmail.com
>> ---
>>  include/linux/bpf.h      |  14 ++++
>>  include/uapi/linux/bpf.h |  18 +++++
>>  kernel/bpf/arraymap.c    | 203 +++++++++++++++++++++++++++++++++++++++++++++++
>>  kernel/bpf/verifier.c    |  12 ++-
>>  4 files changed, 246 insertions(+), 1 deletion(-)
>>
>> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
>> index fa9a988400d9..eae4ce4542c1 100644
>> --- a/include/linux/bpf.h
>> +++ b/include/linux/bpf.h
>> @@ -13,6 +13,10 @@
>>  #include <linux/percpu.h>
>>  #include <linux/err.h>
>>  
>> +#ifdef CONFIG_SECURITY_LANDLOCK
>> +#include <linux/fs.h> /* struct file */
>> +#endif /* CONFIG_SECURITY_LANDLOCK */
>> +
>>  struct perf_event;
>>  struct bpf_map;
>>  
>> @@ -38,6 +42,7 @@ struct bpf_map_ops {
>>  struct bpf_map {
>>  	atomic_t refcnt;
>>  	enum bpf_map_type map_type;
>> +	enum bpf_map_array_type map_array_type;
>>  	u32 key_size;
>>  	u32 value_size;
>>  	u32 max_entries;
>> @@ -187,6 +192,9 @@ struct bpf_array {
>>  	 */
>>  	enum bpf_prog_type owner_prog_type;
>>  	bool owner_jited;
>> +#ifdef CONFIG_SECURITY_LANDLOCK
>> +	u32 n_entries;	/* number of entries in a handle array */
>> +#endif /* CONFIG_SECURITY_LANDLOCK */
>>  	union {
>>  		char value[0] __aligned(8);
>>  		void *ptrs[0] __aligned(8);
>> @@ -194,6 +202,12 @@ struct bpf_array {
>>  	};
>>  };
>>  
>> +#ifdef CONFIG_SECURITY_LANDLOCK
>> +struct map_landlock_handle {
>> +	u32 type; /* enum bpf_map_handle_type */
>> +};
>> +#endif /* CONFIG_SECURITY_LANDLOCK */
>> +
>>  #define MAX_TAIL_CALL_CNT 32
>>  
>>  struct bpf_event_entry {
>> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
>> index 7cd36166f9b7..b68de57f7ab8 100644
>> --- a/include/uapi/linux/bpf.h
>> +++ b/include/uapi/linux/bpf.h
>> @@ -87,6 +87,15 @@ enum bpf_map_type {
>>  	BPF_MAP_TYPE_PERCPU_ARRAY,
>>  	BPF_MAP_TYPE_STACK_TRACE,P_TYPE_CGROUP_ARRAY
>>  	BPF_MAP_TYPE_CGROUP_ARRAY,
>> +	BPF_MAP_TYPE_LANDLOCK_ARRAY,
>> +};
>> +
>> +enum bpf_map_array_type {
>> +	BPF_MAP_ARRAY_TYPE_UNSPEC,
>> +};
>> +
>> +enum bpf_map_handle_type {
>> +	BPF_MAP_HANDLE_TYPE_UNSPEC,
>>  };
> 
> missing something. why it has to be special to have it's own
> fd array implementation?
> Please take a look how BPF_MAP_TYPE_PERF_EVENT_ARRAY, 
> BPF_MAP_TYPE_CGROUP_ARRAY and BPF_MAP_TYPE_PROG_ARRAY are done.
> The all store objects into array map that user space passes via FD.
> I think the same model should apply here.

The idea is to have multiple way for userland to describe a resource
(e.g. an open file descriptor, a path or a glob pattern). The kernel
representation could then be a "struct path *" or dedicated types (e.g.
custom glob).

Another interesting point (that could replace
check_map_func_compatibility()) is that BPF_MAP_TYPE_LANDLOCK_ARRAY
translate to dedicated (abstract) types (instead of CONST_PTR_TO_MAP)
thanks to bpf_reg_type_from_map(). This is useful to abstract userland
(map) interface with kernel object(s) dealing with that type.

A third point is that BPF_MAP_TYPE_LANDLOCK_ARRAY is a kind of set. It
is optimized to quickly walk through all the elements in a sequential way.