lto1.exe: internal compiler error: in splice_child_die, at dwarf2out.c:4905

[Andrey B]

trying to migrate from 2014 Mentor/CodeSourcery G++ Lite to 5.4_2016q3 or 6_2017-q1 but both fail with same

lto1.exe: internal compiler error: in splice_child_die, at dwarf2out.c:4905
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
lto-wrapper.exe: fatal error: arm-none-eabi-g++ returned 1 exit status
compilation terminated.
c:/progra~1/gnu_tools/5.4_2016q3/bin/../lib/gcc/arm-none-eabi/5.4.1/../../../../arm-none-eabi/bin/ld.exe: error: lto-wrapper failed

I was able to isolate the issue to if -g is passed to linker or not: does not fail without -g, fails with -g or g3

compiling on windows 7 64 bit, issue is reproducible 100% of the time.

[Andrey B (arro239)]

cs-make all
Building target: rusefi.elf
Invoking: Cross ARM C++ Linker
arm-none-eabi-g++ -mcpu=cortex-m4 -mthumb -mfloat-abi=softfp -mfpu=fpv4-sp-d16 -O0 -fmessage-length=0 -ffunction-sections -fdata-sections -g -T "..\config\stm32f4ems\STM32F407xG_CCM.ld" -Xlinker --gc-sections -Wl,-Map,"rusefi.map" -Wl,-Map,"rusefi.map" -o "rusefi.elf" ./util/LocalVersionHolder.o ./util/cli_registry.o ./util/crc.o ./util/cyclic_buffer.o ./util/data_buffer.o ./util/datalogging.o ./util/efilib.o ./util/efilib2.o ./util/histogram.o ./util/listener_array.o ./util/loggingcentral.o ./util/rfiutil.o ./hw_layer/stm32f4/mpu_util.o ./hw_layer/stm32f4/stm32f4xx_hal_flash.o ./hw_layer/stm32f4/stm32f4xx_hal_flash_ex.o ./hw_layer/serial_over_usb/usbcfg.o ./hw_layer/serial_over_usb/usbconsole.o ./hw_layer/sensors/CJ125.o ./hw_layer/mass_storage/usb_msd.o ./hw_layer/mass_storage/usb_msd_cfg.o ./hw_layer/lcd/lcd_HD44780.o ./hw_layer/HIP9011.o ./hw_layer/accelerometer.o ./hw_layer/adc_inputs.o ./hw_layer/board_test.o ./hw_layer/can_hw.o ./hw_layer/digital_input_hw.o ./hw_layer/flash.o ./hw_layer/gpio_helper.o ./hw_layer/hardware.o ./hw_layer/io_pins.o ./hw_layer/joystick.o ./hw_layer/max31855.o ./hw_layer/mcp3208.o ./hw_layer/microsecond_timer.o ./hw_layer/mmc_card.o ./hw_layer/neo6m.o ./hw_layer/pin_repository.o ./hw_layer/pwm_generator.o ./hw_layer/rtc_helper.o ./hw_layer/servo.o ./hw_layer/stepper.o ./hw_layer/trigger_input.o ./hw_layer/vehicle_speed.o ./development/test/test.o ./development/test/testbmk.o ./development/hw_layer/poten.o ./development/engine_emulator.o ./development/engine_sniffer.o ./development/rfi_perftest.o ./development/sensor_chart.o ./development/trigger_emulator.o ./development/wave_analyzer.o ./controllers/trigger/decoders/trigger_bmw.o ./controllers/trigger/decoders/trigger_chrysler.o ./controllers/trigger/decoders/trigger_gm.o ./controllers/trigger/decoders/trigger_honda.o ./controllers/trigger/decoders/trigger_mazda.o ./controllers/trigger/decoders/trigger_mitsubishi.o ./controllers/trigger/decoders/trigger_nissan.o ./controllers/trigger/decoders/trigger_rover.o ./controllers/trigger/decoders/trigger_subaru.o ./controllers/trigger/decoders/trigger_toyota.o ./controllers/trigger/decoders/trigger_universal.o ./controllers/trigger/main_trigger_callback.o ./controllers/trigger/rpm_calculator.o ./controllers/trigger/spark_logic.o ./controllers/trigger/trigger_central.o ./controllers/trigger/trigger_decoder.o ./controllers/trigger/trigger_emulator_algo.o ./controllers/trigger/trigger_simulator.o ./controllers/trigger/trigger_structure.o ./controllers/system/SingleTimerExecutor.o ./controllers/system/efiGpio.o ./controllers/system/event_queue.o ./controllers/system/pwm_generator_logic.o ./controllers/system/signal_executor_sleep.o ./controllers/sensors/allsensors.o ./controllers/sensors/ego.o ./controllers/sensors/hip9011_lookup.o ./controllers/sensors/maf.o ./controllers/sensors/map.o ./controllers/sensors/thermistors.o ./controllers/sensors/tps.o ./controllers/sensors/voltage.o ./controllers/math/biquad.o ./controllers/math/engine_math.o ./controllers/math/pid.o ./controllers/math/speed_density.o ./controllers/core/EfiWave.o ./cont...

[Andrey B (arro239)]

Apprantly chibios people are aware of this as "there is a bug with gcc-arm-none-eabi-4_8-2014q1 which prevent to use -flto and -gddb together."

As soon as I've removed -flto things seems to work.

[Thomas Preud'homme (thomas-preudhomme)]

Hi there,

Could you post instructions to reproduce this issue? Which version of Chibios shows this problem and how to build it?

Best regards.

[Andrey B (arro239)]

I know it's not the best way to reproduce an issue but can you download my whole project? It embedds Chibios 2.6, I believe ChibiOS prior to 3.0 is affected. While yes this is an order version of ChibiOS I would assume gcc should not fail with internal error regardless?

Download & unzip http://rusefi.com/data/issue_1674525.zip
make

expected result: successful compilation & linking or human-readable error message

actual result:
lto1.exe: internal compiler error: in splice_child_die, at dwarf2out.c:4905
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
lto-wrapper.exe: fatal error: C:\Progra~1\GNU_Tools\5.4_2016q3\bin\arm-none-eabi-gcc.exe returned 1 exit status
compilation terminated.
c:/progra~1/gnu_tools/5.4_2016q3/bin/../lib/gcc/arm-none-eabi/5.4.1/../../../../arm-none-eabi/bin/ld.exe: error: lto-wrapper failed
collect2.exe: error: ld returned 1 exit status
make: *** [chibios/os/ports/GCC/ARMCMx/rules.mk:211: build/rusefi.elf] Error 1

[Thomas Preud'homme (thomas-preudhomme)]

Hi Andrey,

Perfect, I can reproduce it. Yes you're right, however old is ChibiOS in there does not matter, the compiler should just not crash.

Best regards.

[Thomas Preud'homme (thomas-preudhomme)]

This appears to be fixed in our 7-2017Q4 release. Can you confirm Andrey?

[Thomas Preud'homme (thomas-preudhomme)]

Note you'll need 2 more -Wno-* for the compilation to work with GCC 7.

[Andrey B (arro239)]

This now compiles just fine, I guess we can close this one!