Target stm init gcc alignement #3074

jamike · 2016-10-19T13:01:48Z

Description

With TOOLCHAIN GCC, Zero initialization is done twice for some STM target (at startup and in __start)
To Align the startup with other toolchain, the zero bss in startup is removed and mbed_sdk_init is performing again some init already done in SystemInit as for other toolchain

Status

READY

Steps to test or reproduce

All tests relative to STM target on TOOLCHAIN GCC_ARM are passed

zero bss is done in libc init after call to _start.

zero bss is done in libc init after call _start.

bridadan · 2016-10-19T14:55:30Z

/morph test

bridadan · 2016-10-19T14:55:48Z

@mbed-bot: TEST

HOST_OSES=ALL
BUILD_TOOLCHAINS=ALL
TARGETS=ALL

mbed-bot · 2016-10-19T16:20:16Z

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 0

All builds and test passed!

mbed-bot · 2016-10-20T06:58:39Z

[Build 1053]
FAILURE: Something went wrong when building and testing.

LMESTM · 2016-10-20T07:26:48Z

Tested on my side and definitely solves issues with GCC_ARM (HAL internal timeout not properly working on specific test cases)

bridadan · 2016-10-20T14:16:07Z

Failure is from a different platform, not related to your changes 👍

Gotta love that diff! 😄

Test results look ok, @0xc0170 or @c1728p9 would you like to do a quick review? This is a bit low-level for me I think.

c1728p9 · 2016-10-20T14:18:36Z

...TARGET_STM/TARGET_STM32F0/TARGET_DISCO_F051R8/device/TOOLCHAIN_GCC_ARM/startup_stm32f051x8.S

 /* Call the application's entry point.*/
-  bl main
-  bl exit
+//  bl main


Any reason this code was left in?

@jamike Please respond. we should not leave any dead code

ok, I will push a update of the serie with a dedicated commit for STM32F0 specific changes

@jamike I checked the commit messages, cant find a relevant info, why are you keeping this dead code in this file (line 86 and 88) ?

sorry, I missed your points as there is dead code in all family. For sure , I see no reason to keep dead code. So I update the patch list and remove the dead code introduced by my patches.

0xc0170 · 2016-10-26T11:48:38Z

...ARGET_STM/TARGET_STM32F0/TARGET_NUCLEO_F070RB/device/TOOLCHAIN_GCC_ARM/startup_stm32f070xb.S


 /* Call the clock system intitialization function.*/
  bl  SystemInit
 /* Call static constructors */
-  bl __libc_init_array
+//  bl __libc_init_array


Also this. remove it, specify a reason why in the commit message.

ok, I will push a update of the serie with a dedicated commit for STM32F0 specific changes and do the same the specific changes in STM32F1

ok, I will push a update of the serie with a dedicated commit for STM32F0 specific changes and do the same the specific changes in STM32L1

zero bss is done in libc init after call to _start.

zero bss, C++ init and the call to main. Remove direct call to __libc_init_array and main not needed as _start is beeing called.

zero bss is done in libc init after call to _start.

zero bss, C++ init and the call to main. Remove direct call to __libc_init_array and main not needed as _start is beeing called.

zero bss is done in libc init after call to _start.

Zero bss is done after the call to _start.

system_init, stops all on going timer. gcc _start , perform zero initialized. => HAL_Init must be done again also in GCC toolchain

bridadan · 2016-10-27T16:00:04Z

/morph test

bridadan · 2016-10-27T16:05:49Z

@mbed-bot: TEST

HOST_OSES=ALL
BUILD_TOOLCHAINS=ALL
TARGETS=ALL

mbed-bot · 2016-10-27T18:02:59Z

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 959

All builds and test passed!

mbed-bot · 2016-10-27T18:38:28Z

[Build 1068]
SUCCESS: Building succeeded and tests were run! Be sure to check the test results

dinau · 2016-10-28T06:15:35Z

Just FYI and for my information.
_start, _mainCRTStartup and memory clear problem:
#1432

dinau

Release mbed OS 5.2.2 and mbed lib v129 Known issues in this release There is currently a DNS resolution failure in Thread mode with this release. This causes a failure in the mbed-os-example-client. This will be fixed in a subsequent release. This can be worked around by reverting to mbed-os-5.2.0 Ports for Upcoming Targets 3011: Add u-blox Sara-N target. ARMmbed#3011 3099: MAX32625 ARMmbed#3099 3151: Add support for FRDM-K82F ARMmbed#3151 3177: New mcu k22512 fixing pr 3136 ARMmbed#3177 Fixes and Changes 2990: [tools] Parallel building of tests ARMmbed#2990 3008: NUCLEO_F072RB: Fix wrong timer channel number on pwm PB_5 pin ARMmbed#3008 3013: STM32xx - Change how the ADC internal pins are checked before pinmap_ ARMmbed#3013 3023: digital_loop tests update for STM32 ARMmbed#3023 3041: [nRF5] - added implementation of API of serial port flow control configuration. ARMmbed#3041 3092: [tools + tests] Adding parallelized build option for iar and uvision exporters ARMmbed#3092 3084: [nrf5] fix in Digital I/O : a gpioe pin was uninitialized badly ARMmbed#3084 3009: TRNG enabled. TRNG APIs implemented. REV A/B/C/D flags removed. Warnings removed ARMmbed#3009 3139: Handle [NOT_SUPPORTED] exception in make.py ARMmbed#3139 3074: Target stm init gcc alignement ARMmbed#3074 3140: [tests] Replacing getchar with RawSerial getc in greentea-client ARMmbed#3140 3158: Added support for 6lowpan PAN ID filter to mbed mesh api configuration ARMmbed#3158 2988: Update of can_api.c fixing ARMmbed#2987 ARMmbed#2988 3175: Updating IAR definition for the NCS36510 for IAR EW v7.8 ARMmbed#3175 3170: [tests] Preventing test from printing before Greentea __sync ARMmbed#3170 3169: [Update of ARMmbed#3014] Usb updates ARMmbed#3169 3143: CFStore fix needed for the Cloud Client ARMmbed#3143 3135: lwip - Fix memory leak in k64f cyclic-buffer overflow ARMmbed#3135 3048: Make update.py test compile examples prior to updating mbed-os version. ARMmbed#3048 3162: lwip/nsapi - Clean up warnings in network code ARMmbed#3162 3161: nsapi - Add better heuristic for the default record of DNS queries ARMmbed#3161 3173: [Exporters] Add a device_name to microbit entry in targets.json ARMmbed#3173 3072: i2c_loop tests update for STM32 ARMmbed#3072 2958: Allowing mbed_app.json files to be discovered for tests. ARMmbed#2958 2969: [nRF52] - switch irq priorities of driver handlers to the lowest level ARMmbed#2969 3078: lwip: Allow several configuration macros to be set externally (bis) ARMmbed#3078 3165: Add address type checks to NanostackInterface ARMmbed#3165 3166: nsapi_dns: Provide 2 IPv6-hosted default servers ARMmbed#3166 3171: [tools] Fixing project.py -S printing problem ARMmbed#3171 3172: [Exporters] New export-build tests ARMmbed#3172 3184: ARMmbed#3183 Compiler warning in trng_api.c with K64F ARMmbed#3184 3185: Update tests to fix build failures. Also make the code similar to oth ARMmbed#3185 3104: [NuMaker] Support CAN and fix PWM CLK error ARMmbed#3104 3182: Exporter documentation ARMmbed#3182 3186: MultiTech mDot - add back SPI3 pins ARMmbed#3186 3187: [Export-Make] Use internal class variable for resolving templates in makefiles ARMmbed#3187 3195: [Exporters - Make-based] Quote the shell call in mkdir and rmdir ARMmbed#3195 3204: [Export build-test] Directory traversal error ARMmbed#3204 3189: [Exporters - Make-based] Force make exporter to search PATH for compilers ARMmbed#3189 3200: Using Popen for uVision and unifying the structure of the build function ARMmbed#3200 3075: nsapi - Add standardized return types for size and errors ARMmbed#3075 3221: u-blox odin w2 drivers update ARMmbed#3221

Ports for Upcoming Targets 3011: Add u-blox Sara-N target. ARMmbed/mbed-os#3011 3099: MAX32625 ARMmbed/mbed-os#3099 3151: Add support for FRDM-K82F ARMmbed/mbed-os#3151 3177: New mcu k22512 fixing pr 3136 ARMmbed/mbed-os#3177 Fixes and Changes 3008: NUCLEO_F072RB: Fix wrong timer channel number on pwm PB_5 pin ARMmbed/mbed-os#3008 3013: STM32xx - Change how the ADC internal pins are checked before pinmap_ ARMmbed/mbed-os#3013 3041: [nRF5] - added implementation of API of serial port flow control configuration. ARMmbed/mbed-os#3041 3084: [nrf5] fix in Digital I/O : a gpioe pin was uninitialized badly ARMmbed/mbed-os#3084 3009: TRNG enabled. TRNG APIs implemented. REV A/B/C/D flags removed. Warnings removed ARMmbed/mbed-os#3009 3074: Target stm init gcc alignement ARMmbed/mbed-os#3074 2988: Update of can_api.c fixing #2987 ARMmbed/mbed-os#2988 3173: [Exporters] Add a device_name to microbit entry in targets.json ARMmbed/mbed-os#3173 2969: [nRF52] - switch irq priorities of driver handlers to the lowest level ARMmbed/mbed-os#2969 3184: #3183 Compiler warning in trng_api.c with K64F ARMmbed/mbed-os#3184 3104: [NuMaker] Support CAN and fix PWM CLK error ARMmbed/mbed-os#3104 3186: MultiTech mDot - add back SPI3 pins ARMmbed/mbed-os#3186 3075: nsapi - Add standardized return types for size and errors ARMmbed/mbed-os#3075 3221: u-blox odin w2 drivers update ARMmbed/mbed-os#3221

jamike added 2 commits October 19, 2016 14:54

STM32F207: GCC_ARM remove zero bss, in startup.

759837a

zero bss is done in libc init after call to _start.

STM32F3xxx : GCC_ARM remove zero bss, in startup.

26f3352

zero bss is done in libc init after call _start.

bridadan added the needs: review label Oct 20, 2016

c1728p9 approved these changes Oct 20, 2016

View reviewed changes

0xc0170 reviewed Oct 26, 2016

View reviewed changes

0xc0170 added needs: work and removed needs: review labels Oct 26, 2016

jamike added 3 commits October 26, 2016 14:41

STM32F4xx: GCC_ARM remove zero bss, in startup.

ee22d0a

zero bss is done in libc init after call to _start.

STM32L4xx : GCC_ARM remove zero bss, in startup.

93a0b9c

zero bss is done in libc init after call to _start.

STM32F7xx : GCC_ARM remove zero bss, in startup.

a711ff3

zero bss is done in libc init after call to _start.

jamike force-pushed the TARGET_STM_INIT_GCC_ALIGNEMENT branch from 7f1012a to 2309a2e Compare October 26, 2016 13:09

jamike added 7 commits October 27, 2016 10:38

STM32L1xx : GCC_ARM a call to _start which performs

f3b1c0a

zero bss, C++ init and the call to main. Remove direct call to __libc_init_array and main not needed as _start is beeing called.

STM32L1xx : GCC_ARM remove zero bss, in startup.

479a1a2

zero bss is done in libc init after call to _start.

STM32L0xx : GCC_ARM remove zero bss, in startup.

af090e3

zero bss is done in libc init after call to _start.

STM32F0xx : GCC_ARM use a call to _start which performs

9f20c46

zero bss, C++ init and the call to main. Remove direct call to __libc_init_array and main not needed as _start is beeing called.

STM32F0xx: GCC_ARM remove zero bss, in startup.

702f64a

zero bss is done in libc init after call to _start.

STM32F1xx : GCC_ARM remove zero bss done at startup.

b225a5c

Zero bss is done after the call to _start.

INIT:GCC add call to HAL_Init

4a69072

system_init, stops all on going timer. gcc _start , perform zero initialized. => HAL_Init must be done again also in GCC toolchain

jamike force-pushed the TARGET_STM_INIT_GCC_ALIGNEMENT branch from 2309a2e to 4a69072 Compare October 27, 2016 08:53

bridadan added the ready for merge label Oct 27, 2016

bridadan removed the needs: work label Oct 27, 2016

0xc0170 added release-version: 5.2.2 and removed ready for merge labels Oct 28, 2016

0xc0170 merged commit 4a4d09f into ARMmbed:master Oct 28, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Target stm init gcc alignement #3074

Target stm init gcc alignement #3074

jamike commented Oct 19, 2016

bridadan commented Oct 19, 2016

bridadan commented Oct 19, 2016

mbed-bot commented Oct 19, 2016

mbed-bot commented Oct 20, 2016

LMESTM commented Oct 20, 2016

bridadan commented Oct 20, 2016

c1728p9 Oct 20, 2016

0xc0170 Oct 26, 2016 •

edited

Loading

jamike Oct 26, 2016

0xc0170 Oct 27, 2016

jamike Oct 27, 2016

0xc0170 Oct 26, 2016

jamike Oct 26, 2016

jamike Oct 26, 2016

bridadan commented Oct 27, 2016

bridadan commented Oct 27, 2016

mbed-bot commented Oct 27, 2016

mbed-bot commented Oct 27, 2016

dinau commented Oct 28, 2016 •

edited

Loading

Target stm init gcc alignement #3074

Target stm init gcc alignement #3074

Conversation

jamike commented Oct 19, 2016

Description

Status

Steps to test or reproduce

bridadan commented Oct 19, 2016

bridadan commented Oct 19, 2016

mbed-bot commented Oct 19, 2016

Output

mbed-bot commented Oct 20, 2016

LMESTM commented Oct 20, 2016

bridadan commented Oct 20, 2016

Choose a reason for hiding this comment

0xc0170 Oct 26, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bridadan commented Oct 27, 2016

bridadan commented Oct 27, 2016

mbed-bot commented Oct 27, 2016

Output

mbed-bot commented Oct 27, 2016

dinau commented Oct 28, 2016 • edited Loading

0xc0170 Oct 26, 2016 •

edited

Loading

dinau commented Oct 28, 2016 •

edited

Loading