Re: [PATCH bpf-next v2] bpftool: Dump map id instead of value for map_of_maps types

[Yonghong Song <yhs@meta.com>]

On 4/24/23 9:10 PM, Xueming Feng wrote:
>> On 4/24/23 2:09 AM, Xueming Feng wrote:
>>> When using `bpftool map dump` in plain format, it is usually
>>> more convenient to show the inner map id instead of raw value.
>>> Changing this behavior would help with quick debugging with
>>> `bpftool`, without disrupting scripted behavior. Since user
>>> could dump the inner map with id, and need to convert value.
>>>
>>> Signed-off-by: Xueming Feng <kuro@kuroa.me>
>>> ---
>>> Changes in v2:
>>>     - Fix commit message grammar.
>>> 	- Change `print_uint` to only print to stdout, make `arg` const, and rename
>>> 	  `n` to `arg_size`.
>>>     - Make `print_uint` able to take any size of argument up to `unsigned long`,
>>> 		and print it as unsigned decimal.
>>>
>>> Thanks for the review and suggestions! I have changed my patch accordingly.
>>> There is a possibility that `arg_size` is larger than `unsigned long`,
>>> but previous review suggested that it should be up to the caller function to
>>> set `arg_size` correctly. So I didn't add check for that, should I?
>>>
>>>    tools/bpf/bpftool/main.c | 15 +++++++++++++++
>>>    tools/bpf/bpftool/main.h |  1 +
>>>    tools/bpf/bpftool/map.c  |  9 +++++++--
>>>    3 files changed, 23 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/tools/bpf/bpftool/main.c b/tools/bpf/bpftool/main.c
>>> index 08d0ac543c67..810c0dc10ecb 100644
>>> --- a/tools/bpf/bpftool/main.c
>>> +++ b/tools/bpf/bpftool/main.c
>>> @@ -251,6 +251,21 @@ int detect_common_prefix(const char *arg, ...)
>>>    	return 0;
>>>    }
>>>    
>>> +void print_uint(const void *arg, unsigned int arg_size)
>>> +{
>>> +	const unsigned char *data = arg;
>>> +	unsigned long val = 0ul;
>>> +
>>> +	#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
>>> +		memcpy(&val, data, arg_size);
>>> +	#else
>>> +		memcpy((unsigned char *)&val + sizeof(val) - arg_size,
>>> +		       data, arg_size);
>>> +	#endif
>>> +
>>> +	fprintf(stdout, "%lu", val);
>>> +}
>>> +
>>>    void fprint_hex(FILE *f, void *arg, unsigned int n, const char *sep)
>>>    {
>>>    	unsigned char *data = arg;
>>> diff --git a/tools/bpf/bpftool/main.h b/tools/bpf/bpftool/main.h
>>> index 0ef373cef4c7..0de671423431 100644
>>> --- a/tools/bpf/bpftool/main.h
>>> +++ b/tools/bpf/bpftool/main.h
>>> @@ -90,6 +90,7 @@ void __printf(1, 2) p_info(const char *fmt, ...);
>>>    
>>>    bool is_prefix(const char *pfx, const char *str);
>>>    int detect_common_prefix(const char *arg, ...);
>>> +void print_uint(const void *arg, unsigned int arg_size);
>>>    void fprint_hex(FILE *f, void *arg, unsigned int n, const char *sep);
>>>    void usage(void) __noreturn;
>>>    
>>> diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
>>> index aaeb8939e137..f5be4c0564cf 100644
>>> --- a/tools/bpf/bpftool/map.c
>>> +++ b/tools/bpf/bpftool/map.c
>>> @@ -259,8 +259,13 @@ static void print_entry_plain(struct bpf_map_info *info, unsigned char *key,
>>>    		}
>>>    
>>>    		if (info->value_size) {
>>> -			printf("value:%c", break_names ? '\n' : ' ');
>>> -			fprint_hex(stdout, value, info->value_size, " ");
>>> +			if (map_is_map_of_maps(info->type)) {
>>> +				printf("id:%c", break_names ? '\n' : ' ');
>> 1> +				print_uint(value, info->value_size);
> 
> On Mon, 24 Apr 2023 18:07:27 -0700, Yonghong Song wrote:
>> For all map_in_map types, the inner map value size is 32bit int which
>> represents a fd (for map creation) and a id (for map info), e.g., in
>> show_prog_maps() in prog.c. So maybe we can simplify the code as below:
>> 	printf("id: %u", *(unsigned int *)value);
> 
> That is true, maybe the "id" could also be changed to "map_id" to follow the
> convention. Do you think that `print_uint` could be useful in the future?
> If that is the case, should I keep using it here as an example usage, and to
> avoid dead code? Or should I just remove it?

Maybe, "inner_map_id" is a better choice. For array of maps, some array 
element value could be 0, implying "inner_map_id 0", but I think it is
okay, people should know a real inner_map_id (or any map_id) should 
never be 0.

Function "print_uint" is not needed any more. Please remove it.

Please add the command line to dump map values triggering the above 
change, also the actual dumps with and without this patch.

> 
>>> +			} else {
>>> +				printf("value:%c", break_names ? '\n' : ' ');
>>> +				fprint_hex(stdout, value, info->value_size, " ");
>>> +			}
>>>    		}
>>>    
>>>    		printf("\n");