Re: [PATCH V3] powerpc/bug: Add ARCH_WARN_ASM and refactor _EMIT_BUG_ENTRY for Rust support

[Gary Guo]

On Thu Sep 10, 2026 at 11:08 AM BST, Mukesh Kumar Chaurasiya (IBM) wrote:
> The Rust kernel infrastructure generates inline asm for WARN() via
> ARCH_WARN_ASM(file, line, flags, size), expanding it through a C
> preprocessor pass (generated_arch_warn_asm.rs.S) to produce an
> arch-specific asm template string for use in Rust's core::arch macros.
>
> powerpc currently lacks ARCH_WARN_ASM and ARCH_WARN_REACHABLE, causing
> Rust builds to fail on powerpc with
> ```
> error: no rules expected `ARCH_WARN_ASM`
>    --> /home/linkmauve/dev/linux/wii/rust/kernel/generated_arch_warn_asm.rs:1:28
>     |
>   1 | ::kernel::concat_literals!(ARCH_WARN_ASM("{file}", "{line}", "{flags}", "{size}"))

I think we probably want to catch this earlier by have something like

#ifndef ARCH_WARN_ASM
#error "ARCH_WARM_ASM is not defined"
#endif

in generated_arch_warn_asm.rs.S.

>     |                            ^^^^^^^^^^^^^ no rules expected this token in macro call
>     |
>    ::: ../rust/kernel/lib.rs:279:1
>     |
> 279 | macro_rules! concat_literals {
>     | ---------------------------- when calling this macro
>     |
>     = note: while trying to match sequence start
>
> error: no rules expected `ARCH_WARN_REACHABLE`
>    --> /home/linkmauve/dev/linux/wii/rust/kernel/generated_arch_reachable_asm.rs:1:28
>     |
>   1 | ::kernel::concat_literals!(ARCH_WARN_REACHABLE)
>     |                            ^^^^^^^^^^^^^^^^^^^ no rules expected this token in macro call
>     |
>    ::: ../rust/kernel/lib.rs:279:1
>     |
> 279 | macro_rules! concat_literals {
>     | ---------------------------- when calling this macro
>     |
>     = note: while trying to match sequence start
>
> error: aborting due to 2 previous errors
> ```
>
> To add ARCH_WARN_ASM, _EMIT_BUG_ENTRY first needs to be refactored.
> The old definition was a bare macro with no parameters, relying on
> positional asm operand references (%0-%3), hardcoding the backward
> reference to local label 1b, and including .org/.previous directives
> inline. That made it impossible to compose as a plain string outside of
> an asm operand context, and left an invisible contract that callers must
> always emit their trap at label 1:.
>
> Refactor _EMIT_BUG_ENTRY to take explicit (label, file, line, flags)
> string arguments via string concatenation. This removes the dependency
> on asm operand numbering and makes the trap label an explicit argument,
> so the caller's intent is visible at the call site and a future caller
> using a different label cannot silently produce a wrong bug table entry.
>
> Move the .org and .previous directives out of _EMIT_BUG_ENTRY and into
> each call site, so BUG_ENTRY() can still pass sizeof(struct bug_entry)
> as an asm operand while ARCH_WARN_ASM can supply its own size string
> independently.
>
> Add ARCH_WARN_REACHABLE as an empty define, matching the arm64
> convention, indicating that no additional reachability annotation is
> needed after a WARN on powerpc.
>
> This brings powerpc into line with x86, arm64, s390, and riscv, all of
> which already define ARCH_WARN_ASM and ARCH_WARN_REACHABLE.
>
> Reported-by: FUJITA Tomonori <tomo@xxxxxxxxxxxx>
> Closes: https://lore.kernel.org/all/anG67Q6Y59kDqh-c@desktop
> Fixes: 73b741adb264 ("rust: Add PowerPC support")
> Signed-off-by: Mukesh Kumar Chaurasiya (IBM) <mkchauras@xxxxxxxxx>
> ---
> Changelog:
> V2 -> V3:
>  - Add label argument in _EMIT_BUG_ENTRY
> V2: https://lore.kernel.org/all/20260910071252.1950488-2-mkchauras@xxxxxxxxx
>
> V1 -> V2:
> - commit message now has error, fixes tag and closes tag
> V1: https://lore.kernel.org/all/20260819084825.969116-1-mkchauras@xxxxxxxxx
>
>  arch/powerpc/include/asm/bug.h | 36 +++++++++++++++++++---------------
>  1 file changed, 20 insertions(+), 16 deletions(-)
>
> diff --git a/arch/powerpc/include/asm/bug.h b/arch/powerpc/include/asm/bug.h
> index 0db48977c70c..df2183c35945 100644
> --- a/arch/powerpc/include/asm/bug.h
> +++ b/arch/powerpc/include/asm/bug.h
> @@ -32,34 +32,38 @@
>  #endif /* verbose */
>  
>  #else /* !__ASSEMBLER__ */
> -/* _EMIT_BUG_ENTRY expects args %0,%1,%2,%3 to be FILE, LINE, flags and
> -   sizeof(struct bug_entry), respectively */
>  #ifdef CONFIG_DEBUG_BUGVERBOSE
> -#define _EMIT_BUG_ENTRY				\
> -	".section __bug_table,\"aw\"\n"		\
> -	"2:	.4byte 1b - .\n"		\
> -	"	.4byte %0 - .\n"		\
> -	"	.short %1, %2\n"		\
> -	".org 2b+%3\n"				\
> -	".previous\n"
> +#define _EMIT_BUG_ENTRY(label, file, line, flags)	\
> +	".section __bug_table,\"aw\"\n"			\
> +	"2:	.4byte " label "b - .\n"		\

"b" is part of the label. "1b" itself is a label and "1" is just an integer.

If the code uses

    _EMIT_BUG_ENTRY(..)
    "1: ..."

then the correct label would be "1f".

Best,
Gary


> +	"	.4byte " file " - .\n"			\
> +	"	.short " line ", " flags "\n"
>  #else
> -#define _EMIT_BUG_ENTRY				\
> -	".section __bug_table,\"aw\"\n"		\
> -	"2:	.4byte 1b - .\n"		\
> -	"	.short %2\n"			\
> -	".org 2b+%3\n"				\
> -	".previous\n"
> +#define _EMIT_BUG_ENTRY(label, file, line, flags)	\
> +	".section __bug_table,\"aw\"\n"			\
> +	"2:	.4byte " label "b - .\n"		\
> +	"	.short " flags "\n"
>  #endif
>  
>  #define BUG_ENTRY(cond_str, insn, flags, ...)		\
>  	__asm__ __volatile__(				\
>  		"1:	" insn "\n"			\
> -		_EMIT_BUG_ENTRY				\
> +		_EMIT_BUG_ENTRY("1", "%0", "%1", "%2")	\
> +		".org 2b+%3\n"				\
> +		".previous\n"				\
>  		: : "i" (WARN_CONDITION_STR(cond_str) __FILE__), "i" (__LINE__),	\
>  		  "i" (flags),				\
>  		  "i" (sizeof(struct bug_entry)),	\
>  		  ##__VA_ARGS__)
>  
> +#define ARCH_WARN_ASM(file, line, flags, size)		\
> +		"1:	twi 31, 0, 0\n"			\
> +		_EMIT_BUG_ENTRY("1", file, line, flags)	\
> +		".org 2b+" size "\n"			\
> +		".previous\n"
> +
> +#define ARCH_WARN_REACHABLE
> +
>  /*
>   * BUG_ON() and WARN_ON() do their best to cooperate with compile-time
>   * optimisations. However depending on the complexity of the condition