Filipe Cavalcanti
1ae7efd388
tools/espressif: add fill-flash-size to build if MERGEBIN selected
2024-12-24 10:49:39 +08:00
wangmingrong1
b3813bd450
gcov: Solve the problem of incorrect report generation caused by different compilation and test environments
...
Due to the makefile, the relative paths of the files are different, and
it is impossible to use existing tools such as gcov and lcov to specify
a base compilation path for them.
The solution for Linux is to copy a copy of the source code from
the compilation environment to the test environment,
and there is a place that is consistent with the absolute
path of the compilation environment.
This solution is to pass in the "-b" parameter to directly
modify the info file generated by lcov. The search path will be
executed in the next step of genhtml
Signed-off-by: wangmingrong1 <wangmingrong1@xiaomi.com>
2024-12-23 14:53:58 +08:00
wangmingrong1
e26e8bda0e
gcov: gcov.py, gcov_convert.py scripts merged
...
Build Documentation / build-html (push) Has been cancelled
Signed-off-by: wangmingrong1 <wangmingrong1@xiaomi.com>
2024-12-21 20:49:37 +08:00
wangmingrong1
7d6b2e4804
gcov/script: gcov.sh is implemented using Python
...
Signed-off-by: wangmingrong1 <wangmingrong1@xiaomi.com>
2024-12-21 20:49:37 +08:00
yaojingwei
bbd6931d61
nuttx/audio: add fake audio driver.
...
Build Documentation / build-html (push) Waiting to run
The fake audio driver aims to easily simulate the functions of audio driver through reading and writing files.
With it, you can conveniently customize and define various capture or playback audio drivers for debugging or automated testing.
Signed-off-by: yaojingwei <yaojingwei@xiaomi.com>
2024-12-20 11:47:08 -03:00
wangmingrong1
391aa9b5f3
gcov: Enhanced gcov script
...
1. Add "-s" to specify the path to collect gcno files
2. Add "-a" to specify the path to collect gcda files.
3. Add "-x" for copy-only operation
Signed-off-by: wangmingrong1 <wangmingrong1@xiaomi.com>
2024-12-18 20:24:14 +08:00
chao an
33d717d3a0
tools/parsetrace.py: fix typo lenght -> length
...
Signed-off-by: chao an <anchao@lixiang.com>
2024-12-17 20:48:07 +08:00
Matteo Golin
e96fb6b118
arch/bcm2711, boards/raspberry-pi-4b: Add preliminary support for the BCM2711 and Raspberry Pi 4B.
...
Includes a Mini-UART NSH console for the 4B, tested on the 4GB RAM
model. Part of an I2C driver which can only read, boiler-plate for a SPI
driver, and a GPIO driver with limited pins. Some tools are present for
automatically fetching the boot files and creating the `config.txt` file
based on selected Kconfig options.
2024-12-17 16:43:17 +08:00
Alin Jerpelea
f65a17da37
tools/macar-qcs.sh: migrate license to ASF
...
Xiaomi has signed the SGA and we can migrate the license to ASF
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2024-12-16 14:14:55 +08:00
YAMAMOTO Takashi
566e1c3a7d
tools/ci/platforms/darwin.sh: bump bloaty version
...
hopefully this can fix build failure on macos-13. (see below)
i tested "cibuild.sh -i" with macOS 15.2, x86-64, Xcode 16.1.
```
[470/495] Linking CXX executable bloaty
FAILED: bloaty
: && /Applications/Xcode_15.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/c++ -W -Wall -Wno-sign-compare -fdiagnostics-color=always -O2 -g1 -isysroot /Applications/Xcode_15.2.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX14.2.sdk -mmacosx-version-min=13.7 -Wl,-search_paths_first -Wl,-headerpad_max_install_names CMakeFiles/bloaty.dir/src/main.cc.o -o bloaty liblibbloaty.a third_party/protobuf/cmake/libprotoc.a third_party/re2/libre2.a third_party/capstone/libcapstone.a /Applications/Xcode_15.2.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX14.2.sdk/usr/lib/libz.tbd third_party/abseil-cpp/absl/strings/libabsl_strings.a third_party/protobuf/cmake/libprotobuf.a /Applications/Xcode_15.2.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX14.2.sdk/usr/lib/libz.tbd third_party/abseil-cpp/absl/strings/libabsl_strings_internal.a third_party/abseil-cpp/absl/base/libabsl_base.a third_party/abseil-cpp/absl/base/libabsl_spinlock_wait.a third_party/abseil-cpp/absl/numeric/libabsl_int128.a third_party/abseil-cpp/absl/base/libabsl_throw_delegate.a third_party/abseil-cpp/absl/types/libabsl_bad_optional_access.a third_party/abseil-cpp/absl/base/libabsl_raw_logging_internal.a third_party/abseil-cpp/absl/base/libabsl_log_severity.a && :
ld: Undefined symbols:
absl::debugging_internal::Demangle(char const*, char*, int), referenced from:
bloaty::ItaniumDemangle(std::__1::basic_string_view<char, std::__1::char_traits<char>>, bloaty::DataSource) in liblibbloaty.a[2](bloaty.cc.o)
clang: error: linker command failed with exit code 1 (use -v to see invocation)
```
2024-12-13 23:28:02 +08:00
hujun5
98ce36cdc7
remove fl_crefs
...
Signed-off-by: hujun5 <hujun5@xiaomi.com>
2024-12-13 23:04:10 +08:00
YAMAMOTO Takashi
fce4f2b3ee
tools/ci/platforms/darwin.sh: update bloaty pin
...
This seems to fix build errors on my environment.
(macOS 15.1.1, x86-64, xcode 16.1)
```
/Users/yamamoto/git/nuttx/tools/bloaty-src/third_party/abseil-cpp/absl/numeric/int128.cc:45:5: error: use of undeclared identifier 'ABSL_INTERNAL_ASSUME'
45 | ABSL_INTERNAL_ASSUME(hi != 0);
| ^
/Users/yamamoto/git/nuttx/tools/bloaty-src/third_party/abseil-cpp/absl/numeric/int128.cc:49:3: error: use of undeclared identifier 'ABSL_INTERNAL_ASSUME'
49 | ABSL_INTERNAL_ASSUME(low != 0);
| ^
/Users/yamamoto/git/nuttx/tools/bloaty-src/third_party/abseil-cpp/absl/numeric/int128.cc:141:9: error: redefinition of 'operator/'
141 | uint128 operator/(uint128 lhs, uint128 rhs) {
| ^
/usr/local/Cellar/abseil/20240722.0/include/absl/numeric/int128.h:1040:16: note: previous definition is here
1040 | inline uint128 operator/(uint128 lhs, uint128 rhs) {
| ^
/Users/yamamoto/git/nuttx/tools/bloaty-src/third_party/abseil-cpp/absl/numeric/int128.cc:152:9: error: redefinition of 'operator%'
152 | uint128 operator%(uint128 lhs, uint128 rhs) {
| ^
```
2024-12-13 18:23:29 +08:00
YAMAMOTO Takashi
ac3aa0185c
tools/ci/platforms/darwin.sh: Don't use PIP_USER if venv is activated
...
On my environment PIP_USER doesn't work well.
```
error: externally-managed-environment
```
(macOS 15.1.1, x86-64, xcode 16.1)
Also, I prefer to use virtualenv anyway.
2024-12-13 18:23:29 +08:00
edbf2e21e7
CI: Move rv-virt:python from risc-v-06 to risc-v-07
2024-12-12 23:38:35 +08:00
simbit18
1dcada4917
[ci] macOS platforms: added LLVM toolchain for Arm
...
added LLVM toolchain for Arm on Apple Intel
release-17.0.1
added LLVM toolchain for Arm on Apple Silicon
release-19.1.1
ì
2024-12-11 14:09:45 -03:00
zhangyuan29
fb4a246fcc
tools/ci: change sem_init_6_1 to support skip result
...
After SEM_VALUE_MAX change to INT_MAX, need to support
skip result.
Signed-off-by: zhangyuan29 <zhangyuan29@xiaomi.com>
2024-12-10 22:00:42 +08:00
wangmingrong1
909bf2dbb4
kasan/globals: fix compile error
...
../../../mm/kasan/global.c:58:44: error: type of 'g_global_region' does not match original declaration [-Werror=lto-type-mismatch]
58 | extern const struct kasan_global_region_s *g_global_region[];
| ^
kasan_globals.tmp:3:21: note: 'g_global_region' was previously declared here
3 | const unsigned long g_global_region[] = {
| ^
lto1: all warnings being treated as errors
Signed-off-by: wangmingrong1 <wangmingrong1@xiaomi.com>
2024-12-05 20:37:47 +08:00
wangmingrong1
cc88063646
debug symbol level: Use config instead
...
Signed-off-by: wangmingrong1 <wangmingrong1@xiaomi.com>
2024-12-04 22:36:45 +08:00
xuxingliang
08139fb1db
nxgdb/thread: fix ps command
...
Fix ps command reports wrong running stack corruption.
To get running tcb registers, need to read from frame register not the tcb context.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-12-02 13:43:52 +08:00
xuxingliang
1d55f111f3
gdb: check if pid is valid
...
Add check to make sure pid is valid.
Remove stack check for interrupt context.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-12-02 13:43:52 +08:00
xuxingliang
1f2435641d
nuttxgdb: fix memdump busyloop when nodesize is 0
...
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-12-02 13:43:52 +08:00
xuxingliang
659e3f87a0
gdb/mmleak: add diagnose API to report details of the leak
...
Example of json dump:
{
"title": "Memory Leak Report",
"summary": "Total 105 blks, 22768 bytes leaked",
"result": "fail",
"command": "mm leak",
"data": [
{
"count": 1,
"pid": 0,
"size": 48,
"address": 1100447608,
"seqno": 0,
"alive": true,
"backtrace": []
},
{
"count": 1,
"pid": 4,
"size": 368,
"address": 1100549056,
"seqno": 1041,
"alive": false,
"backtrace": [
{
"address": 1076428472,
"function": "<mm_zalloc+8>",
"source": "mm_heap/mm_zalloc.c:45"
},
{
"address": 1076424018,
"function": "<mm_calloc+18>",
"source": "mm_heap/mm_calloc.c:57"
},
{
"address": 1076419750,
"function": "<calloc+22>",
"source": "umm_heap/umm_calloc.c:73"
},
....
]
},
}
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-12-02 13:43:52 +08:00
Zhe Weng
3ad53656b9
tools/gdb: Fix gdbinit to unload all nuttxgdb related modules
...
There are many modules like nuttxgdb.* that we need to unload
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
2024-12-02 13:43:52 +08:00
trns1997
7d2d38f899
install kconfiglib using apt repo
2024-12-02 13:43:29 +08:00
chao an
2961ecab9f
sim/tflm: add tflite-micro demo into ci build
...
$ cmake -B build -DBOARD_CONFIG=sim/tflm -GNinja
$ cmake --build build
$ ./build/nuttx
NuttShell (NSH) NuttX-10.4.0
nsh> tflm -h
Utility to use tflite micro on nuttx.
[ -C ] Compile tflite model into c++ codes.
[ -E ] Do once evaluation (for profiling).
[ -i <str> ] Readable model file path.
[ -o <str> ] Writable c++ file path.
[ -p <str> ] Prefix of compiled code.
[ -a <int> ] Arena size (mempool).
[ -h ] Print this message.
nsh> tflm -E -i /data/MobileNet-v3-Small.tflite -o /data/MbileNet-v3-Small.out
0 (id=0): size=602112, offset=0, first_used=0 last_used=1
1 (id=1): size=602112, offset=602112, first_used=1 last_used=2
2 (id=2): size=602112, offset=0, first_used=2 last_used=3
3 (id=3): size=607504, offset=802816, first_used=3 last_used=4
...
* (id=114): size=4096, offset=0, first_used=114 last_used=115
* (id=115): size=4000, offset=4096, first_used=115 last_used=115
0: 00000000000000000000000000...................................................... (588k)
1: 0000000000000000000000000011111111111111111111111111............................ (1176k)
2: 2222222222222222222222222211111111111111111111111111............................ (1176k)
3: 22222222222222222222222222........333333333333333333333333333................... (1182k)
...
"Event","Tag","Ticks"
0,SUB,0
1,MUL,0
2,PAD,0
3,CONV_2D,1
4,HARD_SWISH,0
5,PAD,0
...
110,MEAN,0
111,FULLY_CONNECTED,0
112,FULLY_CONNECTED,0
113,MUL,0
114,FULLY_CONNECTED,0
"Unique Tag","Total ticks across all events with that tag."
SUB, 0
MUL, 0
PAD, 0
CONV_2D, 3
...
"total number of ticks", 3
nxai done!
Signed-off-by: chao an <anchao@lixiang.com>
2024-11-29 18:34:10 +08:00
yangao1
968bb179fe
nuttxgdb/rpmsg.py:dump rpmsg_service message
...
(gdb)rpmsg_service
g_rpmsg_cb:
rpmsg_cb_s at ns_match ns_bind
------------- -------- -------
0xf4e00ac0 0x44708225 <rpmsg_rtc_server_ns_match> 0x4470874c <rpmsg_rtc_server_ns_bind>
0xf4e27b20 0x440c8fd6 <syslog_rpmsg_ns_match> 0x440c966a <syslog_rpmsg_ns_bind>
0xf4e27af0 0x0 0x0
g_rpmsg:
Endpoint at Name local Addr dest Addr cb ns_bound_cb ns_unbind_cb
----------- ---- ---------- --------- -- ----------- ------------
0xf2302ac0 NS 53 53 0x44699520 <rpmsg_virtio_ns_callback> 0x0 0x0
0xf3c0f920 rpmsg-ttysensor 1025 1028 0x44704457 <uart_rpmsg_ept_cb> 0x0 0x0
0xf301d058 rpmsg-sensor 1026 1029 0x440c1680 <sensor_rpmsg_ept_cb> 0x440c1871 <sensor_rpmsg_device_ns_bound> 0x44693a18 <rpmsg_destroy_ept>
0xf2302a6c rpmsg-ping 1027 1036 0x440a10d2 <rpmsg_ping_ept_cb> 0x0 0x0
List update: rpmsg dump contains none initialized list, need to take care of it in
NxList.
Signed-off-by: yangao1 <yangao1@xiaomi.com>
2024-11-28 20:40:04 +08:00
simbit18
00f831028f
[ci] platforms: added toolchains
...
add toolchains for macOS on Apple Intel
xtensa-esp32s2
xtensa-esp32s3
add toolchains for macOS on Apple Silicon
xtensa-esp32s2
xtensa-esp32s3
add install for ubuntu.sh
xtensa-esp32s2
xtensa-esp32s3
add install for linux.sh
xtensa-esp32s2
xtensa-esp32s3
2024-11-26 10:14:42 +08:00
yinshengkai
e69903c939
tools/gdb: Add a singleton class for macros
...
Usage:
macro = Macro("nuttx/nuttx")
print(macro.CONFIG_MM_BACKTRACE)
if macro.CONFIG_MM_BACKTRACE:
print("mm backtrace is enabled")
else:
print("mm backtrace is disabled")
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
2024-11-25 22:58:53 +08:00
yinshengkai
9eaec7ba49
tools/gdb: move the macros cache files to a temporary directory
...
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
2024-11-25 22:58:53 +08:00
yinshengkai
7aa2dc24cd
tools/gdb: fix checkpatch warning
...
Use Python raw strings to avoid escaping:
SyntaxWarning: invalid escape sequence '\?'
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
2024-11-25 22:58:53 +08:00
xuxingliang
8031c9c1de
gdb/mm: cache global variables to save time of memleak check
...
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:42 +08:00
xuxingliang
a0e9b82a79
tools/gdb: always use name to read register
...
For unknown reason, for some cases, GDB refuses to read register using descriptor. Let's stick to the register name.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:42 +08:00
xuxingliang
9881fbae28
tools/gdb: fix thread command not working
...
`thread` is a GDB prefix command. Use `define` can only change it to a
user prefix command. In this case, `thread 3` is unable to pass
the argument 3 to python.
Use python code to register command to overwrite this behavior. It may
not work with future GDB, but all is good for now.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:42 +08:00
xuxingliang
efb890abd3
gdb/diagnose: use unified message for command error
...
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:42 +08:00
xuxingliang
5b477aacca
gdb/diag: catch gdb.error in diagnositics
...
gdb.error could report if memory is corrupted. Save and report the exception and continue to next command.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:42 +08:00
xuxingliang
191dbcc4f1
gdb/debug: add debugpy to debug python code
...
Usage:
(gdb) debugpy
Waiting for connection at localhost:5678
In VSCode, add new configuration in launch.
{
"name": "Python Debugger: Remote Attach",
"type": "debugpy",
"request": "attach",
"connect": {
"host": "localhost",
"port": 5678
}
}
and launch.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:30 +08:00
xuxingliang
6119a5ea63
gdb: print trace back when some modules failed to load
...
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:19 +08:00
yangao1
400790d40b
tools/gdb/gdbinit.py:add hint
...
Signed-off-by: yangao1 <yangao1@xiaomi.com>
2024-11-25 13:37:19 +08:00
xuxingliang
db9b5c4cee
tools/gdb: register commands gracefully
...
Report any error and continue to try to register next command.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:19 +08:00
xuxingliang
65a93e972c
tools/gdb: guide user for correct usage of nuttxgdb
...
Should use gdbinit.py instead of package __init__.py. Report the correct usage if user try to source the package directly.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 13:37:19 +08:00
xuxingliang
358261af19
tools/gdb: avoid direct access to tcb['name']
...
It could be disabled at compile time. Use utils.get_task_name to handle it
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
xuxingliang
41c4eb08c0
gdb/mm: fix prev free judging and region search
...
1. The prev free flag should use field size instead of preceding to compare.
2. If not prev-free, then prevnode should be None.
3. Cast type to mm_freenode_s in order to access flink, blink when node is free.
4. Heap itself is also included in first region.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
xuxingliang
8667f911bc
gdb/mm: fix memdump when mempool backtrace is disabled
...
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
xuxingliang
ee9673f62e
gdb/memdump: add option to exclude nodes from specific PID
...
This is useful to exclude memory nodes for mempool.
E.g.
(gdb) mm dump --nob --top 5 --sort size --nop --no-pid -1
Pool CNT PID Size Overhead Seqno Address Backtrace
H 1 0 16777264 44 13 0x41a79010
H 2 160 1536048 44 1129827490 0x41d6aad0
H 1 45374 1536176 44 1129829080 0x45ef1010
H 3 141 408440 44 145817 0x43905bf0
H 7 158 131120 44 32955 0x4308d090
Total 14 blks, 23528696 bytes
(gdb) mm dump --nob --top 5 --sort size --nop
Pool CNT PID Size Overhead Seqno Address Backtrace
H 1 0 16777264 44 13 0x41a79010
* H 6 -1 1048640 44 116954 0x436a0fd0
* H 5 -1 1048640 44 146821 0x43b24fd0
* H 3 -1 1048640 44 66161 0x4335bfd0
* H 3 -1 1048640 44 156337 0x43d3ffd0
Total 18 blks, 16777264 bytes
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
xuxingliang
f849092ede
gdb/memdump: add option to parse memdump log from device
...
igned-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
xuxingliang
06a696076a
gdb/memdump: reorganize the memdump parameters
...
Make the dump_nodes etc more common to use.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
xuxingliang
6d2ebfdec6
mm: dump node overheadp during memdump
...
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
xuxingliang
797f7a043d
gdb/memdump: show prev and next node in address finding mode
...
This will make it easier to debug memory corruption when agacent node is corrupted.
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
xuxingliang
dd7203aff7
gdb/mm: move memleak to standalone file
...
Signed-off-by: xuxingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00
Xu Xingliang
412644fcf2
gdb: optimize memory commands
...
Tested on a complex project, results are promising.
Command Time cost(s) Time saving(s) Peformance Boost
Before After
memleak 39.494172 22.366415 17.127757 1.8
memdump 41.872441 26.458386 15.414055 1.6
memdump -a 0x1234 28.116294 1.114119 27.002175 25.2
memdump --no-backtrace N/A 1.114119
memmap 7.973809 6.836468 1.137341 1.2
Signed-off-by: Xu Xingliang <xuxingliang@xiaomi.com>
2024-11-25 00:42:34 +08:00