michaelh/zephyr
1
0
Fork 0
Code Releases Activity
zephyr/scripts/build/gen_relocate_app.py
Grzegorz Swiderski 460b6ef122 code_relocation: Add NOKEEP option 
When using the code and data relocation feature, every relocated symbol
would be marked with `KEEP()` in the generated linker script. Therefore,
if any input files contained unused code, then it wouldn't be discarded
by the linker, even when invoked with `--gc-sections`.

This can cause unexpected bloat, or other link-time issues stemming from
some symbols being discarded and others not.

On the other hand, this behavior has been present since the feature's
introduction, so it should remain default for the users who rely on it.

This patch introduces support for `zephyr_code_relocate(... NOKEEP)`.
This will suppress the generation of `KEEP()` statements for all symbols
in a particular library or set of files.

Much like `NOCOPY`, the `NOKEEP` flag is passed to `gen_relocate_app.py`
in string form. The script is now equipped to handle multiple such flags
when passed from CMake as a semicolon-separated list, like so:

   "SRAM2:NOCOPY;NOKEEP:/path/to/file1.c;/path/to/file2.c"

Documentation and tests are updated here as well.

Signed-off-by: Grzegorz Swiderski <grzegorz.swiderski@nordicsemi.no>
2024-01-15 13:20:17 +01:00

621 lines
22 KiB
Python
Raw Permalink Blame History

#!/usr/bin/env python3
#
# Copyright (c) 2018 Intel Corporation.
#
# SPDX-License-Identifier: Apache-2.0
#
"""
This script will relocate .text, .rodata, .data and .bss sections from required files
and places it in the required memory region. This memory region and file
are given to this python script in the form of a string.
Example of such a string would be::
SRAM2:COPY:/home/xyz/zephyr/samples/hello_world/src/main.c,\
SRAM1:COPY:/home/xyz/zephyr/samples/hello_world/src/main2.c, \
FLASH2:NOCOPY:/home/xyz/zephyr/samples/hello_world/src/main3.c
One can also specify the program header for a given memory region:
SRAM2\\ :phdr0:COPY:/home/xyz/zephyr/samples/hello_world/src/main.c
To invoke this script::
python3 gen_relocate_app.py -i input_string -o generated_linker -c generated_code
Configuration that needs to be sent to the python script.
- If the memory is like SRAM1/SRAM2/CCD/AON then place full object in
the sections
- If the memory type is appended with _DATA / _TEXT/ _RODATA/ _BSS only the
selected memory is placed in the required memory region. Others are
ignored.
- COPY/NOCOPY defines whether the script should generate the relocation code in
code_relocation.c or not
- NOKEEP will suppress the default behavior of marking every relocated symbol
with KEEP() in the generated linker script.
Multiple regions can be appended together like SRAM2_DATA_BSS
this will place data and bss inside SRAM2.
"""
import sys
import argparse
import os
import glob
import warnings
from collections import defaultdict
from enum import Enum
from pathlib import Path
from typing import NamedTuple
from typing import NewType
from typing import Tuple
from elftools.elf.elffile import ELFFile
from elftools.elf.sections import SymbolTableSection
MemoryRegion = NewType('MemoryRegion', str)
class SectionKind(Enum):
TEXT = "text"
RODATA = "rodata"
DATA = "data"
BSS = "bss"
LITERAL = "literal"
def __str__(self):
return self.name
@classmethod
def for_section_named(cls, name: str):
"""
Return the kind of section that includes a section with the given name.
>>> SectionKind.for_section_with_name(".rodata.str1.4")
<SectionKind.RODATA: 'rodata'>
>>> SectionKind.for_section_with_name(".device_deps")
None
"""
if ".text." in name:
return cls.TEXT
elif ".rodata." in name:
return cls.RODATA
elif ".data." in name:
return cls.DATA
elif ".bss." in name:
return cls.BSS
elif ".literal." in name:
return cls.LITERAL
else:
return None
class OutputSection(NamedTuple):
obj_file_name: str
section_name: str
keep: bool = True
PRINT_TEMPLATE = """
KEEP(*{obj_file_name}({section_name}))
"""
PRINT_TEMPLATE_NOKEEP = """
*{obj_file_name}({section_name})
"""
SECTION_LOAD_MEMORY_SEQ = """
__{0}_{1}_rom_start = LOADADDR(.{0}_{1}_reloc);
"""
LOAD_ADDRESS_LOCATION_FLASH = """
#ifdef CONFIG_XIP
GROUP_DATA_LINK_IN({0}, ROMABLE_REGION)
#else
GROUP_DATA_LINK_IN({0}, {0})
#endif
"""
LOAD_ADDRESS_LOCATION_FLASH_NOCOPY = """
GROUP_LINK_IN({0})
"""
LOAD_ADDRESS_LOCATION_BSS = "GROUP_LINK_IN({0})"
MPU_RO_REGION_START = """
_{0}_mpu_ro_region_start = ORIGIN({1});
"""
MPU_RO_REGION_END = """
_{0}_mpu_ro_region_end = .;
"""
# generic section creation format
LINKER_SECTION_SEQ = """
/* Linker section for memory region {2} for {3} section */
SECTION_PROLOGUE(.{0}_{1}_reloc,,)
{{
. = ALIGN(4);
{4}
. = ALIGN(4);
}} {5}
__{0}_{1}_reloc_end = .;
__{0}_{1}_reloc_start = ADDR(.{0}_{1}_reloc);
__{0}_{1}_reloc_size = __{0}_{1}_reloc_end - __{0}_{1}_reloc_start;
"""
LINKER_SECTION_SEQ_MPU = """
/* Linker section for memory region {2} for {3} section */
SECTION_PROLOGUE(.{0}_{1}_reloc,,)
{{
__{0}_{1}_reloc_start = .;
{4}
#if {6}
. = ALIGN({6});
#else
MPU_ALIGN(__{0}_{1}_reloc_size);
#endif
__{0}_{1}_reloc_end = .;
}} {5}
__{0}_{1}_reloc_size = __{0}_{1}_reloc_end - __{0}_{1}_reloc_start;
"""
SOURCE_CODE_INCLUDES = """
/* Auto generated code. Do not modify.*/
#include <zephyr/kernel.h>
#include <zephyr/linker/linker-defs.h>
#include <zephyr/kernel_structs.h>
#include <kernel_internal.h>
"""
EXTERN_LINKER_VAR_DECLARATION = """
extern char __{0}_{1}_reloc_start[];
extern char __{0}_{1}_rom_start[];
extern char __{0}_{1}_reloc_size[];
"""
DATA_COPY_FUNCTION = """
void data_copy_xip_relocation(void)
{{
{0}
}}
"""
BSS_ZEROING_FUNCTION = """
void bss_zeroing_relocation(void)
{{
{0}
}}
"""
MEMCPY_TEMPLATE = """
z_early_memcpy(&__{0}_{1}_reloc_start, &__{0}_{1}_rom_start,
(size_t) &__{0}_{1}_reloc_size);
"""
MEMSET_TEMPLATE = """
z_early_memset(&__{0}_bss_reloc_start, 0,
(size_t) &__{0}_bss_reloc_size);
"""
def region_is_default_ram(region_name: str) -> bool:
"""
Test whether a memory region with the given name is the system's default
RAM region or not.
This is used to determine whether some items need to be omitted from
custom regions and instead be placed in the default. In particular, mutable
data placed in the default RAM section is ignored and is allowed to be
handled normally by the linker because it is placed in that region anyway.
"""
return region_name == args.default_ram_region
def find_sections(filename: str) -> 'dict[SectionKind, list[OutputSection]]':
"""
Locate relocatable sections in the given object file.
The output value maps categories of sections to the list of actual sections
located in the object file that fit in that category.
"""
obj_file_path = Path(filename)
with open(obj_file_path, 'rb') as obj_file_desc:
full_lib = ELFFile(obj_file_desc)
if not full_lib:
sys.exit("Error parsing file: " + filename)
sections = [x for x in full_lib.iter_sections()]
out = defaultdict(list)
for section in sections:
section_kind = SectionKind.for_section_named(section.name)
if section_kind is None:
continue
out[section_kind].append(
OutputSection(obj_file_path.name, section.name)
)
# Common variables will be placed in the .bss section
# only after linking in the final executable. This "if" finds
# common symbols and warns the user of the problem.
# The solution to which is simply assigning a 0 to
# bss variable and it will go to the required place.
if isinstance(section, SymbolTableSection):
def is_common_symbol(s):
return s.entry["st_shndx"] == "SHN_COMMON"
for symbol in filter(is_common_symbol, section.iter_symbols()):
warnings.warn("Common variable found. Move "+
symbol.name + " to bss by assigning it to 0/NULL")
return out
def assign_to_correct_mem_region(
memory_region: str,
full_list_of_sections: 'dict[SectionKind, list[OutputSection]]'
) -> 'dict[MemoryRegion, dict[SectionKind, list[OutputSection]]]':
"""
Generate a mapping of memory region to collection of output sections to be
placed in each region.
"""
use_section_kinds, memory_region = section_kinds_from_memory_region(memory_region)
memory_region, _, align_size = memory_region.partition('_')
if align_size:
mpu_align[memory_region] = int(align_size)
keep_sections = '|NOKEEP' not in memory_region
memory_region = memory_region.replace('|NOKEEP', '')
output_sections = {}
for used_kind in use_section_kinds:
# Pass through section kinds that go into this memory region
output_sections[used_kind] = [
section._replace(keep=keep_sections)
for section in full_list_of_sections[used_kind]
]
return {MemoryRegion(memory_region): output_sections}
def section_kinds_from_memory_region(memory_region: str) -> 'Tuple[set[SectionKind], str]':
"""
Get the section kinds requested by the given memory region name.
Region names can be like RAM_RODATA_TEXT or just RAM; a section kind may
follow the region name. If no kinds are specified all are assumed.
In addition to the parsed kinds, the input region minus specifiers for those
kinds is returned.
>>> section_kinds_from_memory_region('SRAM2_TEXT')
({<SectionKind.TEXT: 'text'>}, 'SRAM2')
"""
out = set()
for kind in SectionKind:
specifier = f"_{kind}"
if specifier in memory_region:
out.add(kind)
memory_region = memory_region.replace(specifier, "")
if not out:
# No listed kinds implies all of the kinds
out = set(SectionKind)
return (out, memory_region)
def print_linker_sections(list_sections: 'list[OutputSection]'):
out = ''
for section in sorted(list_sections):
template = PRINT_TEMPLATE if section.keep else PRINT_TEMPLATE_NOKEEP
out += template.format(obj_file_name=section.obj_file_name,
section_name=section.section_name)
return out
def add_phdr(memory_type, phdrs):
return f'{memory_type} {phdrs[memory_type] if memory_type in phdrs else ""}'
def string_create_helper(
kind: SectionKind,
memory_type,
full_list_of_sections: 'dict[SectionKind, list[OutputSection]]',
load_address_in_flash,
is_copy,
phdrs
):
linker_string = ''
if load_address_in_flash:
if is_copy:
load_address_string = LOAD_ADDRESS_LOCATION_FLASH.format(add_phdr(memory_type, phdrs))
else:
load_address_string = LOAD_ADDRESS_LOCATION_FLASH_NOCOPY.format(add_phdr(memory_type, phdrs))
else:
load_address_string = LOAD_ADDRESS_LOCATION_BSS.format(add_phdr(memory_type, phdrs))
if full_list_of_sections[kind]:
# Create a complete list of funcs/ variables that goes in for this
# memory type
tmp = print_linker_sections(full_list_of_sections[kind])
if region_is_default_ram(memory_type) and kind in (SectionKind.DATA, SectionKind.BSS):
linker_string += tmp
else:
if not region_is_default_ram(memory_type) and kind is SectionKind.RODATA:
align_size = 0
if memory_type in mpu_align:
align_size = mpu_align[memory_type]
linker_string += LINKER_SECTION_SEQ_MPU.format(memory_type.lower(), kind.value, memory_type.upper(),
kind, tmp, load_address_string, align_size)
else:
if region_is_default_ram(memory_type) and kind in (SectionKind.TEXT, SectionKind.LITERAL):
align_size = 0
linker_string += LINKER_SECTION_SEQ_MPU.format(memory_type.lower(), kind.value, memory_type.upper(),
kind, tmp, load_address_string, align_size)
else:
linker_string += LINKER_SECTION_SEQ.format(memory_type.lower(), kind.value, memory_type.upper(),
kind, tmp, load_address_string)
if load_address_in_flash:
linker_string += SECTION_LOAD_MEMORY_SEQ.format(memory_type.lower(), kind.value, memory_type.upper(),
kind)
return linker_string
def generate_linker_script(linker_file, sram_data_linker_file, sram_bss_linker_file,
complete_list_of_sections, phdrs):
gen_string = ''
gen_string_sram_data = ''
gen_string_sram_bss = ''
for memory_type, full_list_of_sections in \
sorted(complete_list_of_sections.items()):
is_copy = bool("|COPY" in memory_type)
memory_type = memory_type.split("|", 1)[0]
if region_is_default_ram(memory_type) and is_copy:
gen_string += MPU_RO_REGION_START.format(memory_type.lower(), memory_type.upper())
gen_string += string_create_helper(SectionKind.LITERAL, memory_type, full_list_of_sections, 1, is_copy, phdrs)
gen_string += string_create_helper(SectionKind.TEXT, memory_type, full_list_of_sections, 1, is_copy, phdrs)
gen_string += string_create_helper(SectionKind.RODATA, memory_type, full_list_of_sections, 1, is_copy, phdrs)
if region_is_default_ram(memory_type) and is_copy:
gen_string += MPU_RO_REGION_END.format(memory_type.lower())
if region_is_default_ram(memory_type):
gen_string_sram_data += string_create_helper(SectionKind.DATA, memory_type, full_list_of_sections, 1, 1, phdrs)
gen_string_sram_bss += string_create_helper(SectionKind.BSS, memory_type, full_list_of_sections, 0, 1, phdrs)
else:
gen_string += string_create_helper(SectionKind.DATA, memory_type, full_list_of_sections, 1, 1, phdrs)
gen_string += string_create_helper(SectionKind.BSS, memory_type, full_list_of_sections, 0, 1, phdrs)
# finally writing to the linker file
with open(linker_file, "w") as file_desc:
file_desc.write(gen_string)
with open(sram_data_linker_file, "w") as file_desc:
file_desc.write(gen_string_sram_data)
with open(sram_bss_linker_file, "w") as file_desc:
file_desc.write(gen_string_sram_bss)
def generate_memcpy_code(memory_type, full_list_of_sections, code_generation):
generate_sections, memory_type = section_kinds_from_memory_region(memory_type)
# Non-BSS sections get copied to the destination memory, except data in
# main memory which gets copied automatically.
for kind in (SectionKind.TEXT, SectionKind.RODATA, SectionKind.DATA):
if region_is_default_ram(memory_type) and kind is SectionKind.DATA:
continue
if kind in generate_sections and full_list_of_sections[kind]:
code_generation["copy_code"] += MEMCPY_TEMPLATE.format(memory_type.lower(), kind.value)
code_generation["extern"] += EXTERN_LINKER_VAR_DECLARATION.format(
memory_type.lower(), kind.value)
# BSS sections in main memory are automatically zeroed; others need to have
# zeroing code generated.
if (SectionKind.BSS in generate_sections
and full_list_of_sections[SectionKind.BSS]
and not region_is_default_ram(memory_type)
):
code_generation["zero_code"] += MEMSET_TEMPLATE.format(memory_type.lower())
code_generation["extern"] += EXTERN_LINKER_VAR_DECLARATION.format(
memory_type.lower(), SectionKind.BSS.value)
return code_generation
def dump_header_file(header_file, code_generation):
code_string = ''
# create a dummy void function if there is no code to generate for
# bss/data/text regions
code_string += code_generation["extern"]
if code_generation["copy_code"]:
code_string += DATA_COPY_FUNCTION.format(code_generation["copy_code"])
else:
code_string += DATA_COPY_FUNCTION.format("return;")
if code_generation["zero_code"]:
code_string += BSS_ZEROING_FUNCTION.format(code_generation["zero_code"])
else:
code_string += BSS_ZEROING_FUNCTION.format("return;")
with open(header_file, "w") as header_file_desc:
header_file_desc.write(SOURCE_CODE_INCLUDES)
header_file_desc.write(code_string)
def parse_args():
global args
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter, allow_abbrev=False)
parser.add_argument("-d", "--directory", required=True,
help="obj file's directory")
parser.add_argument("-i", "--input_rel_dict", required=True, type=argparse.FileType('r'),
help="input file with dict src:memory type(sram2 or ccm or aon etc)")
parser.add_argument("-o", "--output", required=False, help="Output ld file")
parser.add_argument("-s", "--output_sram_data", required=False,
help="Output sram data ld file")
parser.add_argument("-b", "--output_sram_bss", required=False,
help="Output sram bss ld file")
parser.add_argument("-c", "--output_code", required=False,
help="Output relocation code header file")
parser.add_argument("-R", "--default_ram_region", default='SRAM',
help="Name of default RAM memory region for system")
parser.add_argument("-v", "--verbose", action="count", default=0,
help="Verbose Output")
args = parser.parse_args()
# return the absolute path for the object file.
def get_obj_filename(searchpath, filename):
# get the object file name which is almost always pended with .obj
obj_filename = filename.split("/")[-1] + ".obj"
for dirpath, _, files in os.walk(searchpath):
for filename1 in files:
if filename1 == obj_filename:
if filename.split("/")[-2] in dirpath.split("/")[-1]:
fullname = os.path.join(dirpath, filename1)
return fullname
# Extracts all possible components for the input string:
# <mem_region>[\ :program_header]:<flag_1>[;<flag_2>...]:<file_1>[;<file_2>...]
# Returns a 4-tuple with them: (mem_region, program_header, flags, files)
# If no `program_header` is defined, returns an empty string
def parse_input_string(line):
# Be careful when splitting by : to avoid breaking absolute paths on Windows
mem_region, rest = line.split(':', 1)
phdr = ''
if mem_region.endswith(' '):
mem_region = mem_region.rstrip()
phdr, rest = rest.split(':', 1)
# Split lists by semicolons, in part to support generator expressions
flag_list, file_list = (lst.split(';') for lst in rest.split(':', 1))
return mem_region, phdr, flag_list, file_list
# Create a dict with key as memory type and files as a list of values.
# Also, return another dict with program headers for memory regions
def create_dict_wrt_mem():
# need to support wild card *
rel_dict = dict()
phdrs = dict()
input_rel_dict = args.input_rel_dict.read()
if input_rel_dict == '':
sys.exit("Disable CONFIG_CODE_DATA_RELOCATION if no file needs relocation")
for line in input_rel_dict.split('|'):
if ':' not in line:
continue
mem_region, phdr, flag_list, file_list = parse_input_string(line)
# Handle any program header
if phdr != '':
phdrs[mem_region] = f':{phdr}'
file_name_list = []
# Use glob matching on each file in the list
for file_glob in file_list:
glob_results = glob.glob(file_glob)
if not glob_results:
warnings.warn("File: "+file_glob+" Not found")
continue
elif len(glob_results) > 1:
warnings.warn("Regex in file lists is deprecated, please use file(GLOB) instead")
file_name_list.extend(glob_results)
if len(file_name_list) == 0:
continue
if mem_region == '':
continue
if args.verbose:
print("Memory region ", mem_region, " Selected for files:", file_name_list)
mem_region = "|".join((mem_region, *flag_list))
if mem_region in rel_dict:
rel_dict[mem_region].extend(file_name_list)
else:
rel_dict[mem_region] = file_name_list
return rel_dict, phdrs
def main():
global mpu_align
mpu_align = {}
parse_args()
searchpath = args.directory
linker_file = args.output
sram_data_linker_file = args.output_sram_data
sram_bss_linker_file = args.output_sram_bss
rel_dict, phdrs = create_dict_wrt_mem()
complete_list_of_sections: 'dict[MemoryRegion, dict[SectionKind, list[OutputSection]]]' \
= defaultdict(lambda: defaultdict(list))
# Create/or truncate file contents if it already exists
# raw = open(linker_file, "w")
# for each memory_type, create text/rodata/data/bss sections for all obj files
for memory_type, files in rel_dict.items():
full_list_of_sections: 'dict[SectionKind, list[OutputSection]]' = defaultdict(list)
for filename in files:
obj_filename = get_obj_filename(searchpath, filename)
# the obj file wasn't found. Probably not compiled.
if not obj_filename:
continue
file_sections = find_sections(obj_filename)
# Merge sections from file into collection of sections for all files
for category, sections in file_sections.items():
full_list_of_sections[category].extend(sections)
# cleanup and attach the sections to the memory type after cleanup.
sections_by_category = assign_to_correct_mem_region(memory_type, full_list_of_sections)
for (region, section_category_map) in sections_by_category.items():
for (category, sections) in section_category_map.items():
complete_list_of_sections[region][category].extend(sections)
generate_linker_script(linker_file, sram_data_linker_file,
sram_bss_linker_file, complete_list_of_sections, phdrs)
code_generation = {"copy_code": '', "zero_code": '', "extern": ''}
for mem_type, list_of_sections in sorted(complete_list_of_sections.items()):
if "|COPY" in mem_type:
mem_type = mem_type.split("|", 1)[0]
code_generation = generate_memcpy_code(mem_type,
list_of_sections, code_generation)
dump_header_file(args.output_code, code_generation)
if __name__ == '__main__':
main()
View git blame Copy permalink