pw_bloat#
Utilities for generating binary size reports
Stable C++ GN
pw_bloat provides tools and helpers around using
Bloaty McBloatface including generating
size report cards for output binaries through Pigweed’s GN build
system.
Bloat report cards allow tracking the memory usage of a system over time as code changes are made and provide a breakdown of which parts of the code have the largest size impact.
pw bloat CLI command#
pw_bloat includes a plugin for the Pigweed command line capable of running
size reports on ELF binaries.
Note
The bloat CLI plugin is still experimental and only supports a small subset
of pw_bloat’s capabilities.
Basic usage#
Running a size report on a single executable#
By default, pw bloat assumes that
memoryregions symbols are defined in
binaries, and uses them to automatically generate a Bloaty config file.
$ pw bloat out/docs/obj/pw_result/size_report/bin/ladder_and_then.elf
▒█████▄ █▓ ▄███▒ ▒█ ▒█ ░▓████▒ ░▓████▒ ▒▓████▄
▒█░ █░ ░█▒ ██▒ ▀█▒ ▒█░ █ ▒█ ▒█ ▀ ▒█ ▀ ▒█ ▀█▌
▒█▄▄▄█░ ░█▒ █▓░ ▄▄░ ▒█░ █ ▒█ ▒███ ▒███ ░█ █▌
▒█▀ ░█░ ▓█ █▓ ░█░ █ ▒█ ▒█ ▄ ▒█ ▄ ░█ ▄█▌
▒█ ░█░ ░▓███▀ ▒█▓▀▓█░ ░▓████▒ ░▓████▒ ▒▓████▀
+----------------------+---------+
| memoryregions | sizes |
+======================+=========+
|FLASH |1,048,064|
|RAM | 196,608|
|VECTOR_TABLE | 512|
+======================+=========+
|Total |1,245,184|
+----------------------+---------+
Running a size report diff#
$ pw bloat out/docs/obj/pw_metric/size_report/bin/one_metric.elf \
--diff out/docs/obj/pw_metric/size_report/bin/base.elf \
-d symbols
▒█████▄ █▓ ▄███▒ ▒█ ▒█ ░▓████▒ ░▓████▒ ▒▓████▄
▒█░ █░ ░█▒ ██▒ ▀█▒ ▒█░ █ ▒█ ▒█ ▀ ▒█ ▀ ▒█ ▀█▌
▒█▄▄▄█░ ░█▒ █▓░ ▄▄░ ▒█░ █ ▒█ ▒███ ▒███ ░█ █▌
▒█▀ ░█░ ▓█ █▓ ░█░ █ ▒█ ▒█ ▄ ▒█ ▄ ░█ ▄█▌
▒█ ░█░ ░▓███▀ ▒█▓▀▓█░ ░▓████▒ ░▓████▒ ▒▓████▀
+-----------------------------------------------------------------------------------+
| |
+-----------------------------------------------------------------------------------+
| diff| memoryregions | symbols | sizes|
+=====+======================+===============================================+======+
| |FLASH | | -4|
| | |[section .FLASH.unused_space] | -408|
| | |main | +60|
| | |__sf_fake_stdout | +4|
| | |pw_boot_PreStaticMemoryInit | -2|
| | |_isatty | -2|
| NEW| |_GLOBAL__sub_I_group_foo | +84|
| NEW| |pw::metric::Group::~Group() | +34|
| NEW| |pw::intrusive_list_impl::List::insert_after() | +32|
| NEW| |pw::metric::Metric::Increment() | +32|
| NEW| |__cxa_atexit | +28|
| NEW| |pw::metric::Metric::Metric() | +28|
| NEW| |pw::metric::Metric::as_int() | +28|
| NEW| |pw::intrusive_list_impl::List::Item::unlist() | +20|
| NEW| |pw::metric::Group::Group() | +18|
| NEW| |pw::intrusive_list_impl::List::Item::previous()| +14|
| NEW| |pw::metric::TypedMetric<>::~TypedMetric() | +14|
| NEW| |__aeabi_atexit | +12|
+-----+----------------------+-----------------------------------------------+------+
| |RAM | | 0|
| | |[section .stack] | -32|
| NEW| |group_foo | +16|
| NEW| |metric_x | +12|
| NEW| |[section .static_init_ram] | +4|
+=====+======================+===============================================+======+
|Total| | | -4|
+-----+----------------------+-----------------------------------------------+------+
Specifying a custom Bloaty config file#
If the linker script for a target does not define memory regions, a custom
Bloaty config can be provided using the -c / --custom-config option.
$ pw bloat out/pw_strict_host_clang_debug/obj/pw_status/test/status_test -c targets/host/linux.bloaty
▒█████▄ █▓ ▄███▒ ▒█ ▒█ ░▓████▒ ░▓████▒ ▒▓████▄
▒█░ █░ ░█▒ ██▒ ▀█▒ ▒█░ █ ▒█ ▒█ ▀ ▒█ ▀ ▒█ ▀█▌
▒█▄▄▄█░ ░█▒ █▓░ ▄▄░ ▒█░ █ ▒█ ▒███ ▒███ ░█ █▌
▒█▀ ░█░ ▓█ █▓ ░█░ █ ▒█ ▒█ ▄ ▒█ ▄ ░█ ▄█▌
▒█ ░█░ ░▓███▀ ▒█▓▀▓█░ ░▓████▒ ░▓████▒ ▒▓████▀
+------------+---------------------+-------+
| segments | sections | sizes |
+============+=====================+=======+
|LOAD #3 [RX]| |138,176|
| |.text |137,524|
| |.plt | 608|
| |.init | 24|
| |.fini | 20|
+------------+---------------------+-------+
|LOAD #2 [R] | | 87,816|
| |.rela.dyn | 32,664|
| |.rodata | 23,176|
| |.eh_frame | 23,152|
| |.eh_frame_hdr | 4,236|
| |.gcc_except_table | 1,140|
| |.dynsym | 1,008|
| |.rela.plt | 888|
| |[ELF Program Headers]| 616|
| |.dynstr | 556|
| |.gnu.version_r | 116|
| |.gnu.version | 84|
| |[ELF Header] | 64|
| |.note.ABI-tag | 32|
| |.gnu.hash | 28|
| |.interp | 28|
| |.note.gnu.build-id | 28|
+------------+---------------------+-------+
|LOAD #5 [RW]| | 20,216|
| |.bss | 19,824|
| |.got.plt | 328|
| |.data | 64|
+------------+---------------------+-------+
|LOAD #4 [RW]| | 15,664|
| |.data.rel.ro | 12,240|
| |.relro_padding | 2,872|
| |.dynamic | 464|
| |.got | 56|
| |.fini_array | 16|
| |.init_array | 16|
+============+=====================+=======+
|Total | |261,872|
+------------+---------------------+-------+
Defining size reports in GN#
Diff size reports#
Size reports can be defined using the GN template pw_size_diff. The template
requires at least two executable targets on which to perform a size diff. The
base for the size diff can be specified either globally through the top-level
base argument, or individually per-binary within the binaries list.
Arguments#
base: Optional default base target for all listed binaries.source_filter: Optional global regex to filter labels in the diff output.data_sources: Optional global list of datasources from bloaty config filebinaries: List of binaries to size diff. Each binary specifies a target, a label for the diff, and optionally a base target, source filter, and data sources that override the global ones (if specified).
import("$dir_pw_bloat/bloat.gni")
executable("empty_base") {
sources = [ "empty_main.cc" ]
}
executable("hello_world_printf") {
sources = [ "hello_printf.cc" ]
}
executable("hello_world_iostream") {
sources = [ "hello_iostream.cc" ]
}
pw_size_diff("my_size_report") {
base = ":empty_base"
data_sources = "symbols,segments"
binaries = [
{
target = ":hello_world_printf"
label = "Hello world using printf"
},
{
target = ":hello_world_iostream"
label = "Hello world using iostream"
data_sources = "symbols"
},
]
}
A sample pw_size_diff reStructuredText size report table can be found
within module docs. For example, see the Size report
section of the pw_checksum module for more detail.
Single binary size reports#
Size reports can also be defined using pw_size_report, which provides
a size report for a single binary. The template requires a target binary.
Arguments#
target: Binary target to run size report on.data_sources: Optional list of data sources to organize outputs.source_filter: Optional regex to filter labels in the output.json_key_prefix: Optional prefix for key names in json size report.full_json_summary: Optional boolean to print json size report by labellevel hierarchy. Defaults to only use top-level label in size report.
ignore_unused_labels: Optional boolean to remove labels that have size ofzero in json size report.
import("$dir_pw_bloat/bloat.gni")
executable("hello_world_iostream") {
sources = [ "hello_iostream.cc" ]
}
pw_size_report("hello_world_iostream_size_report") {
target = ":hello_iostream"
data_sources = "segments,symbols"
source_filter = "pw::hello"
json_key_prefix = "hello_world_iostream"
full_json_summary = true
ignore_unused_labels = true
}
Example of the generated ASCII table for a single binary:
┌─────────────┬──────────────────────────────────────────────────┬──────┐
│segment_names│ symbols │ sizes│
├═════════════┼══════════════════════════════════════════════════┼══════┤
│FLASH │ │12,072│
│ │pw::kvs::KeyValueStore::InitializeMetadata() │ 684│
│ │pw::kvs::KeyValueStore::Init() │ 456│
│ │pw::kvs::internal::EntryCache::Find() │ 444│
│ │pw::kvs::FakeFlashMemory::Write() │ 240│
│ │pw::kvs::internal::Entry::VerifyChecksumInFlash() │ 228│
│ │pw::kvs::KeyValueStore::GarbageCollectSector() │ 220│
│ │pw::kvs::KeyValueStore::RemoveDeletedKeyEntries() │ 220│
│ │pw::kvs::KeyValueStore::AppendEntry() │ 204│
│ │pw::kvs::KeyValueStore::Get() │ 194│
│ │pw::kvs::internal::Entry::Read() │ 188│
│ │pw::kvs::ChecksumAlgorithm::Finish() │ 26│
│ │pw::kvs::internal::Entry::ReadKey() │ 26│
│ │pw::kvs::internal::Sectors::BaseAddress() │ 24│
│ │pw::kvs::ChecksumAlgorithm::Update() │ 20│
│ │pw::kvs::FlashTestPartition() │ 8│
│ │pw::kvs::FakeFlashMemory::Disable() │ 6│
│ │pw::kvs::FakeFlashMemory::Enable() │ 6│
│ │pw::kvs::FlashMemory::SelfTest() │ 6│
│ │pw::kvs::FlashPartition::Init() │ 6│
│ │pw::kvs::FlashPartition::sector_size_bytes() │ 6│
│ │pw::kvs::FakeFlashMemory::IsEnabled() │ 4│
├─────────────┼──────────────────────────────────────────────────┼──────┤
│RAM │ │ 1,424│
│ │test_kvs │ 992│
│ │pw::kvs::(anonymous namespace)::test_flash │ 384│
│ │pw::kvs::(anonymous namespace)::test_partition │ 24│
│ │pw::kvs::FakeFlashMemory::no_errors_ │ 12│
│ │borrowable_kvs │ 8│
│ │kvs_entry_count │ 4│
├═════════════┼══════════════════════════════════════════════════┼══════┤
│Total │ │13,496│
└─────────────┴──────────────────────────────────────────────────┴──────┘
Size reports are typically included in reStructuredText, as described in
Documentation integration. Size reports may also be printed in the build
output if desired. To enable this in the GN build
(pigweed/pw_bloat/bloat.gni), set the pw_bloat_SHOW_SIZE_REPORTS
build arg to true.
Collecting size report data#
Each pw_size_report target outputs a JSON file containing the sizes of all
top-level labels in the binary. (By default, this represents “segments”, i.e.
ELF program headers.) If full_json_summary is set to true, sizes for all
label levels are reported (i.e. default labels would show size of each symbol
per segment). If a build produces multiple images, it may be useful to collect
all of their sizes into a single file to provide a snapshot of sizes at some
point in time — for example, to display per-commit size deltas through CI.
The pw_size_report_aggregation template is provided to collect multiple size
reports’ data into a single JSON file.
Arguments#
deps: List ofpw_size_reporttargets whose data to collect.output: Path to the output JSON file.
import("$dir_pw_bloat/bloat.gni")
pw_size_report_aggregation("image_sizes") {
deps = [
":app_image_size_report",
":bootloader_image_size_report",
]
output = "$root_gen_dir/artifacts/image_sizes.json"
}
Documentation integration#
Bloat reports are easy to add to documentation files. All pw_size_diff
and pw_size_report targets output a file containing a tabular report card.
This file can be imported directly into a reStructuredText file using the
include directive.
For example, the simple_bloat_loop and simple_bloat_function size
reports under //pw_bloat/examples are imported into this file as follows:
Simple bloat loop example
^^^^^^^^^^^^^^^^^^^^^^^^^
.. include:: examples/simple_bloat_loop
Simple bloat function example
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
.. include:: examples/simple_bloat_function
Resulting in this output:
Simple bloat loop example#
Label |
Segment |
Delta |
||
|---|---|---|---|---|
Simple bloat loop |
FLASH
|
+32 |
Simple bloat function example#
Label |
Segment |
Delta |
||
|---|---|---|---|---|
Simple bloat function |
FLASH
|
+16 |
Additional Bloaty data sources#
Bloaty McBloatface by itself cannot help
answer some questions which embedded developers frequently face such as
understanding how much space is left. To address this, Pigweed provides Python
tooling (pw_bloat.bloaty_config) to generate bloaty configuration files
based on the final ELF files through small tweaks in the linker scripts to
expose extra information.
See the sections below on how to enable the additional data sections through modifications in your linker script(s).
As an example to generate the helper configuration which enables additional data
sources for example.elf if you’ve updated your linker script(s) accordingly,
simply run
python -m pw_bloaty.bloaty_config example.elf > example.bloaty. The
example.bloaty can then be used with bloaty using the -c flag, for
example
bloaty -c example.bloaty example.elf --domain vm -d memoryregions,utilization
which may return something like:
84.2% 1023Ki FLASH
94.2% 963Ki Free space
5.8% 59.6Ki Used space
15.8% 192Ki RAM
100.0% 192Ki Used space
0.0% 512 VECTOR_TABLE
96.9% 496 Free space
3.1% 16 Used space
0.0% 0 Not resident in memory
NAN% 0 Used space
utilization data source#
The most common question many embedded developers face when using bloaty is
how much space you are using and how much space is left. To correctly answer
this, section sizes must be used in order to correctly account for section
alignment requirements.
The generated utilization data source will work with any ELF file, where
Used Space is reported for the sum of virtual memory size of all sections.
Padding captures the amount of memory that is utilized to enfore alignment
requirements. Tracking Padding size can help monitor application growth
for changes that are too small to force realignment.
In order for Free Space to be reported, your linker scripts must include
properly aligned sections which span the unused remaining space for the relevant
memory region with the unused_space string anywhere in their name. This
typically means creating a trailing section which is pinned to span to the end
of the memory region.
For example imagine this partial example GNU LD linker script:
MEMORY
{
FLASH(rx) : \
ORIGIN = PW_BOOT_FLASH_BEGIN, \
LENGTH = PW_BOOT_FLASH_SIZE
RAM(rwx) : \
ORIGIN = PW_BOOT_RAM_BEGIN, \
LENGTH = PW_BOOT_RAM_SIZE
}
SECTIONS
{
/* Main executable code. */
.code : ALIGN(4)
{
/* Application code. */
*(.text)
*(.text*)
KEEP(*(.init))
KEEP(*(.fini))
. = ALIGN(4);
/* Constants.*/
*(.rodata)
*(.rodata*)
} >FLASH
/* Explicitly initialized global and static data. (.data)*/
.static_init_ram : ALIGN(4)
{
*(.data)
*(.data*)
. = ALIGN(4);
} >RAM AT> FLASH
/* Zero initialized global/static data. (.bss) */
.zero_init_ram (NOLOAD) : ALIGN(4)
{
*(.bss)
*(.bss*)
*(COMMON)
. = ALIGN(4);
} >RAM
}
Could be modified as follows to enable Free Space reporting:
MEMORY
{
FLASH(rx) : ORIGIN = PW_BOOT_FLASH_BEGIN, LENGTH = PW_BOOT_FLASH_SIZE
RAM(rwx) : ORIGIN = PW_BOOT_RAM_BEGIN, LENGTH = PW_BOOT_RAM_SIZE
/* Each memory region above has an associated .*.unused_space section that
* overlays the unused space at the end of the memory segment. These
* segments are used by pw_bloat.bloaty_config to create the utilization
* data source for bloaty size reports.
*
* These sections MUST be located immediately after the last section that is
* placed in the respective memory region or lld will issue a warning like:
*
* warning: ignoring memory region assignment for non-allocatable section
* '.VECTOR_TABLE.unused_space'
*
* If this warning occurs, it's also likely that LLD will have created quite
* large padded regions in the ELF file due to bad cursor operations. This
* can cause ELF files to balloon from hundreds of kilobytes to hundreds of
* megabytes.
*
* Attempting to add sections to the memory region AFTER the unused_space
* section will cause the region to overflow.
*/
}
SECTIONS
{
/* Main executable code. */
.code : ALIGN(4)
{
/* Application code. */
*(.text)
*(.text*)
KEEP(*(.init))
KEEP(*(.fini))
. = ALIGN(4);
/* Constants.*/
*(.rodata)
*(.rodata*)
} >FLASH
/* Explicitly initialized global and static data. (.data)*/
.static_init_ram : ALIGN(4)
{
*(.data)
*(.data*)
. = ALIGN(4);
} >RAM AT> FLASH
/* Defines a section representing the unused space in the FLASH segment.
* This MUST be the last section assigned to the FLASH region.
*/
PW_BLOAT_UNUSED_SPACE(FLASH)
/* Zero initialized global/static data. (.bss). */
.zero_init_ram (NOLOAD) : ALIGN(4)
{
*(.bss)
*(.bss*)
*(COMMON)
. = ALIGN(4);
} >RAM
/* Defines a section representing the unused space in the RAM segment. This
* MUST be the last section assigned to the RAM region.
*/
PW_BLOAT_UNUSED_SPACE(RAM)
}
The preprocessor macro PW_BLOAT_UNUSED_SPACE is defined in
pw_bloat/bloat_macros.ld. To use these macros include this file in your
pw_linker_script as follows:
pw_linker_script("my_linker_script") {
includes = [ "$dir_pw_bloat/bloat_macros.ld" ]
linker_script = "my_project_linker_script.ld"
}
Note that linker scripts are not natively supported by GN and can’t be provided
through deps, the bloat_macros.ld must be passed in the includes
list.
memoryregions data source#
Understanding how symbols, sections, and other data sources can be attributed
back to the memory regions defined in your linker script is another common
problem area. Unfortunately the ELF format does not include the original memory
regions, meaning bloaty can not do this today by itself. In addition, it’s
relatively common that there are multiple memory regions which alias to the same
memory but through different buses which could make attribution difficult.
Instead of taking the less portable and brittle approach to parse *.map
files, pw_bloat.bloaty_config consumes symbols which are defined in the
linker script with a special format to extract this information from the ELF
file: pw_bloat_config_memory_region_NAME_{start,end}{_N,}.
These symbols are defined by the preprocessor macros PW_BLOAT_MEMORY_REGION
and PW_BLOAT_MEMORY_REGION_MAP with the right address and size for the
regions. To use these macros include the pw_bloat/bloat_macros.ld in your
pw_linker_script as follows:
pw_linker_script("my_linker_script") {
includes = [ "$dir_pw_bloat/bloat_macros.ld" ]
linker_script = "my_project_linker_script.ld"
}
These symbols are then used to determine how to map segments to these memory regions. Note that segments must be used in order to account for inter-section padding which are not attributed against any sections.
As an example, if you have a single view in the single memory region named
FLASH, then you should include the following macro in your linker script to
generate the symbols needed for the that region:
PW_BLOAT_MEMORY_REGION(FLASH)
As another example, if you have two aliased memory regions (DCTM and
ITCM) into the same effective memory named you’d like to call RAM, then
you should produce the following four symbols in your linker script:
PW_BLOAT_MEMORY_REGION_MAP(RAM, ITCM)
PW_BLOAT_MEMORY_REGION_MAP(RAM, DTCM)