Merge branch 'bugfix/sorting_and_list_of_lib_in_idf_size_v3.1' into 'release/v3.1'

tools: Fix sorting and list of libs in idf_size.py

See merge request idf/esp-idf!4285
This commit is contained in:
Angus Gratton 2019-03-05 11:03:51 +08:00
commit d6c52407fc

View file

@ -6,7 +6,7 @@
# Includes information which is not shown in "xtensa-esp32-elf-size",
# or easy to parse from "xtensa-esp32-elf-objdump" or raw map files.
#
# Copyright 2017 Espressif Systems (Shanghai) PTE LTD
# Copyright 2017-2018 Espressif Systems (Shanghai) PTE LTD
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
@ -20,6 +20,9 @@
# See the License for the specific language governing permissions and
# limitations under the License.
#
from __future__ import print_function
from __future__ import unicode_literals
from builtins import dict
import argparse, sys, subprocess, re
import os.path
import pprint
@ -48,12 +51,6 @@ def load_map_data(map_file):
sections = load_sections(map_file)
return memory_config, sections
def output_section_for_address(memory_config, address):
for m in memory_config.values():
if m["origin"] <= address and m["origin"] + m["length"] > address:
return m["name"]
return None
def load_memory_config(map_file):
""" Memory Configuration section is the total size of each output section """
result = {}
@ -183,7 +180,7 @@ def main():
print("Per-file contributions to ELF file:")
print_detailed_sizes(sections, "file", "Object File")
if args.archive_details:
print "Symbols within the archive:", args.archive_details, "(Not all symbols may be reported)"
print("Symbols within the archive:", args.archive_details, "(Not all symbols may be reported)")
print_archive_symbols(sections, args.archive_details)
def print_summary(memory_config, sections):
@ -199,7 +196,7 @@ def print_summary(memory_config, sections):
used_data = get_size(".dram0.data")
used_bss = get_size(".dram0.bss")
used_dram = used_data + used_bss
used_iram = sum( get_size(s) for s in sections.keys() if s.startswith(".iram0") )
used_iram = sum( get_size(s) for s in sections if s.startswith(".iram0") )
flash_code = get_size(".flash.text")
flash_rodata = get_size(".flash.rodata")
total_size = used_data + used_iram + flash_code + flash_rodata
@ -230,7 +227,7 @@ def print_detailed_sizes(sections, key, header):
"Total")
print("%24s %10s %6s %6s %10s %8s %7s" % headings)
result = {}
for k in (sizes.keys()):
for k in sizes:
v = sizes[k]
result[k] = {}
result[k]["data"] = v.get(".dram0.data", 0)
@ -243,7 +240,11 @@ def print_detailed_sizes(sections, key, header):
def return_total_size(elem):
val = elem[1]
return val["total"]
for k,v in sorted(result.items(), key=return_total_size, reverse=True):
def return_header(elem):
return elem[0]
s = sorted(list(result.items()), key=return_header)
# do a secondary sort in order to have consistent order (for diff-ing the output)
for k,v in sorted(s, key=return_total_size, reverse=True):
if ":" in k: # print subheadings for key of format archive:file
sh,k = k.split(":")
print("%24s %10d %6d %6d %10d %8d %7d" % (k[:24],
@ -269,12 +270,14 @@ def print_archive_symbols(sections, archive):
s["sym_name"] = re.sub("(.text.|.literal.|.data.|.bss.|.rodata.)", "", s["sym_name"]);
result[section_name][s["sym_name"]] = result[section_name].get(s["sym_name"], 0) + s["size"]
for t in interested_sections:
print "\nSymbols from section:", t
print("\nSymbols from section:", t)
section_total = 0
for key,val in sorted(result[t].items(), key=lambda (k,v): v, reverse=True):
print("%s(%d)"% (key.replace(t + ".", ""), val)),
s = sorted(list(result[t].items()), key=lambda k_v: k_v[0])
# do a secondary sort in order to have consistent order (for diff-ing the output)
for key,val in sorted(s, key=lambda k_v: k_v[1], reverse=True):
print(("%s(%d)"% (key.replace(t + ".", ""), val)), end=' ')
section_total += val
print "\nSection total:",section_total
print("\nSection total:",section_total)
if __name__ == "__main__":
main()