STM32CubeIde inline assembler error

[PBobo.1]

Previously, I used this delay function in inline assembler in my projects (on Cortex M3, M4):

static inline void delay_m4(uint32_t cnt){
asm volatile(
"1: SUBS %0,#1" "\n\t"
" bne 1b"
:"=r" (cnt)
: "0" (cnt)
);
}

Now I'm trying to add this function to a project on a Cortex M0+ microcontroller (STM32G0XX).
The compiler gives me this error message:
Error: instruction not supported in Thumb16 mode -- `subs r3,#1'
I'm sure this is a compiler error because:
1) the documentation for the Cortex M0+ "PM0223" lists the subs instruction as available;
2) looking through the assembly listings of other Cortex M0+ projects, I see that
the subs instruction is used constantly.

STM32CubeIDE
Version: 1.16.1
Build: 22882_20240916_0822 (UTC)

 

[TDK]

Unified syntax is required for the "subs" instruction:

static inline void delay_m4(uint32_t cnt) {
	asm volatile(
			".syntax unified \n\t"
			"1: SUBS %0,#1 \n\t"
			" bne 1b \n\t"
			:"=r" (cnt)
			: "0" (cnt)
	);
}

Dig more if you want:

M0 inline assembly: subs r0, #32 not supported - NXP Community

Directives that Change the Instruction Type — TI Arm Clang Compiler Tools User's Guide