Re: [PATCH v7] mem: telemetry support for memseg and element information

["Burakov, Anatoly" <anatoly.burakov@intel.com>]

On 10/25/2022 2:02 PM, Amit Prakash Shukla wrote:
> Changes adds telemetry support to display memory occupancy
> in memseg and the information of the elements allocated from
> a memseg based on arguments provided by user. This patch
> adds following endpoints:
> 
> 1. /eal/memseg_lists
> The command displays the memseg list from which the memory
> has been allocated.
> Example:
> --> /eal/memseg_lists
> {"/eal/memseg_lists": [0, 1]}
> 
> 2. /eal/memseg_list_info,<memseg-list-id>
> The command outputs the memsegs, from which the memory is
> allocated, for the memseg_list given as input.
> Example:
> --> /eal/memseg_list_info,0
> {"/eal/memseg_list_info": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, \
>   12, 13, 14, 15]}
> 
> 3. /eal/memseg_info,<memseg-list-id>,<memseg-id>
> The command outputs the memseg information based on the
> memseg-list and the memseg-id given as input.
> Example:
> --> /eal/memseg_info,0,10
> {"/eal/memseg_info": {"Memseg_list_index": 0, \
> "Memseg_index": 10, "Start_addr": "0x101600000", \
> "End_addr": "0x101800000", "Size": 2097152, "Hugepage_size": 2097152, \
> "Socket_id": 0, "flags": 0}}
> 
> --> /eal/memseg_info,0,15
> {"/eal/memseg_info": {"Memseg_list_index": 0, "Memseg_index": 15, \
> "Start_addr": "0x102000000", "End_addr": "0x102200000", \
> "Size": 2097152, "Hugepage_size": 2097152, "Socket_id": 0, "flags": 0}}
> 
> 4. /eal/mem_element_list,<heap-id>,<memseg-list-id>,<memseg-id>
> The command outputs number of elements in a memseg based
> on the heap-id, memseg-list-id and memseg-id given as input.
> Example:
> --> /eal/mem_element_list,0,0,10
> {"/eal/mem_element_list": {"Element_count": 52}}
> 
> --> /eal/mem_element_list,0,1,15
> {"/eal/mem_element_list": {"Element_count": 52}}
> 
> 5. /eal/mem_element_info,<heap-id>,<memseg-list-id>,<memseg-id>,  \
>     <elem-start-id>,<elem-end-id>
> The command outputs element information like element start
> address, end address, to which memseg it belongs, element
> state, element size. User can give a range of elements to be
> printed.
> Example:
> --> /eal/mem_element_info,0,0,15,1,2
> {"/eal/mem_element_info": {"element_1": {"msl_id": 0, "ms_id": 15, \
> "memseg_start_addr": "0x102000000", "memseg_end_addr": "0x102200000", \
> "element_start_addr": "0x102000b00", "element_end_addr": "0x102003380", \
> "element_size": 10368, "element_state": "Busy"}, "element_2": \
> {"msl_id": 0, "ms_id": 15, "memseg_start_addr": "0x102000000", \
> "memseg_end_addr": "0x102200000", "element_start_addr": "0x102003380", \
> "element_end_addr": "0x102005c00", "element_size": 10368, \
> "element_state": "Busy"}, "Element_count": 2}}
> 
> Signed-off-by: Amit Prakash Shukla <amitprakashs@marvell.com>
> ---



> +	ms_start_addr = ms->addr_64;
> +	ms_end_addr = (uint64_t)RTE_PTR_ADD(ms_start_addr, ms->len);
> +	ms_size = ms->len;
> +	hugepage_size = ms->hugepage_sz;
> +	ms_socket_id = ms->socket_id;
> +	ms_flags = ms->flags;
> +
> +	rte_mcfg_mem_read_unlock();
> +
> +	rte_tel_data_start_dict(d);
> +	rte_tel_data_add_dict_int(d, "Memseg_list_index", ms_list_idx);
> +	rte_tel_data_add_dict_int(d, "Memseg_index", ms_idx);
> +	snprintf(addr, ADDR_STR, "0x%"PRIx64, ms_start_addr);
> +	rte_tel_data_add_dict_string(d, "Start_addr", addr);
> +	snprintf(addr, ADDR_STR, "0x%"PRIx64, ms_end_addr);
> +	rte_tel_data_add_dict_string(d, "End_addr", addr);
> +	rte_tel_data_add_dict_u64(d, "Size", ms_size);
> +	rte_tel_data_add_dict_u64(d, "Hugepage_size", hugepage_size);

Seems like rte_tel_data_add_dict_u64 is deprecated now.

I would also suggest outputting IOVA address for memsegs as well (unless 
it's set to RTE_BAD_IOVA, in which case it should say something 
meaningful too).

Otherwise (and with above changes), LGTM

Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
-- 
Thanks,
Anatoly