Re: [PATCH] x86/boot: use __noreturn instead of directly __attribute__ definition

From: Ingo Molnar
Date: Tue Apr 07 2015 - 04:35:41 EST

* Alexander Kuleshov <kuleshovmail@xxxxxxxxx> wrote:

> arch/x86/boot/boot.h defines a couple functions as die and etc..., with
> 'noreturn' attribute. Let's use __noreturn macro instead of directly
> __attribute__ declaration from the <linux/compiler.h>.
> We no need to include <linux/compiler.h> to the arch/x86/boot/boot.h,
> because boot.h already includes "bitops.h" which already includes
> <linux/compiler.h>.
> Signed-off-by: Alexander Kuleshov <kuleshovmail@xxxxxxxxx>
> ---
> arch/x86/boot/boot.h | 7 +++----
> 1 file changed, 3 insertions(+), 4 deletions(-)
> diff --git a/arch/x86/boot/boot.h b/arch/x86/boot/boot.h
> index bd49ec6..3351528 100644
> --- a/arch/x86/boot/boot.h
> +++ b/arch/x86/boot/boot.h
> @@ -305,7 +305,7 @@ void console_init(void);
> void query_edd(void);
> /* header.S */
> -void __attribute__((noreturn)) die(void);
> +void __noreturn die(void);
> /* mca.c */
> int query_mca(void);
> @@ -314,11 +314,10 @@ int query_mca(void);
> int detect_memory(void);
> /* pm.c */
> -void __attribute__((noreturn)) go_to_protected_mode(void);
> +void __noreturn go_to_protected_mode(void);
> /* pmjump.S */
> -void __attribute__((noreturn))
> - protected_mode_jump(u32 entrypoint, u32 bootparams);
> +void __noreturn protected_mode_jump(u32 entrypoint, u32 bootparams);
> /* printf.c */
> int sprintf(char *buf, const char *fmt, ...);

Please don't bother producing and sending me such trivial patches
unless they:

- fix a real bug (in which case they are not trivial patches anymore)

- or are part of a larger (non-trivial!) series that does some real,
substantial work on this code that tries to:

- fix existing code
- speed up existing code
- or expand upon existing code with new code

The reason I'm not applying your patch is that trivial patches with no
substance following them up have more costs than benefits:

- they lead to pointless churn:

- they take up Git space for no good reason
- they slow down bisection of real changes
- they take up (valuable!) reviewer bandwidth
- they take up maintainer bandwidth

there's literally a million pointless cleanup patches that could be
done on the kernel, and we don't want to add a million commits to the
kernel tree.

This applies for this patch but also for other future patches you
might intend to send for code that I (co-)maintain.

My advice to you is to try to raise beyond newbie patches and write
something more substantial that helps Linux:

- take a look at the many bugs on and try to
analyze, reproduce or fix them

- go read kernel code, understand it and try to find real bugs.

- go test the latest kernels and find bugs in it. The fresher the
code, the more likely it is that it has bugs.

- go read kernel code and try to expand upon it

Fortunately it's not hard to contribute to the kernel: there's
literally an infinite amount of work to be done on the kernel, and I
welcome productive contributions - but churning out trivial patches
with no substantial patches following them up is not productive and in
fact they are harmful once you are not a totally fresh newbie kernel
developer anymore...


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at