Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Xueming Feng <kuro@kuroa.me>
To:     yhs@meta.com
Cc:     andrii@kernel.org, ast@kernel.org, bpf@vger.kernel.org,
        daniel@iogearbox.net, haoluo@google.com, john.fastabend@gmail.com,
        jolsa@kernel.org, kpsingh@kernel.org, kuro@kuroa.me,
        linux-kernel@vger.kernel.org, martin.lau@linux.dev,
        quentin@isovalent.com, sdf@google.com, song@kernel.org, yhs@fb.com
Subject: Re: [PATCH bpf-next v2] bpftool: Dump map id instead of value for map_of_maps types
Date:   Tue, 25 Apr 2023 12:10:50 +0800
Message-Id: <20230425041050.59727-1-kuro@kuroa.me>
In-Reply-To: <8b893118-6687-1d2b-d838-1a0c6ff7d886@meta.com>
References: <8b893118-6687-1d2b-d838-1a0c6ff7d886@meta.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

> 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?

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