linux-kernel - Re: [coreboot] checkpatch: Question regarding asmlinkage and storage class

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Sun, 19 Mar 2017 10:24:23 +0100
From:   Paul Menzel <paulepanter@...rs.sourceforge.net>
To:     Joe Perches <joe@...ches.com>, Andy Whitcroft <apw@...onical.com>
Cc:     linux-kernel@...r.kernel.org, coreboot@...eboot.org
Subject: Re: [coreboot] checkpatch: Question regarding asmlinkage and
 storage class

Dear Joe,


Am Sonntag, den 19.03.2017, 01:31 -0700 schrieb Joe Perches:
> On Sat, 2017-03-18 at 13:15 +0100, Paul Menzel wrote:
> > Dear checkpatch developers,
> > 
> > 
> > The coreboot project started using checkpatch.pl, and now some effort
> > is going into fixing issues pointed out by `checkpatch.pl`.
> > 
> > The file `src/arch/x86/acpi_s3.c` in coreboot contains the code below.
> > 
> > ```
> >    205	void (*acpi_do_wakeup)(uintptr_t vector, u32 backup_source, u32 backup_target,
> >    206		u32 backup_size) asmlinkage = (void *)WAKEUP_BASE;
> > ```
> > 
> > The warning is
> > 
> > > WARNING: storage class should be at the beginning of the declaration
> > 
> > which raised the question below [2].
> > 
> > > And I am waiting for someone to answer why checkpatch.pl claims
> > > asmlinkage as a storage-class in the first place.
> 
> []
> > In coreboot the macro is defined similarly to Linux.
> > 
> > ```
> > #define asmlinkage __attribute__((regparm(0)))
> > #define alwaysinline inline __attribute__((always_inline))
> > ```
> 
> Are they similar?
> 
> $ git grep -i "define.*ASMLINKAGE\b" include
> include/linux/linkage.h:#define CPP_ASMLINKAGE extern "C"
> include/linux/linkage.h:#define CPP_ASMLINKAGE
> include/linux/linkage.h:#define asmlinkage CPP_ASMLINKAGE

Yes, for x86 (with `CONFIG_X86_32`) they are.

```
$ git grep asmlinkage | grep regparm
arch/x86/include/asm/linkage.h:#def
ine asmlinkage CPP_ASMLINKAGE __attribute__((regparm(0)))
$ nl -ba arch/x86/include/asm/linkage.h | head -11
     1	#ifndef _ASM_X86_LINKAGE_H
     2	#define _ASM_X86_LINKAGE_H
     3	
     4	#include <linux/stringify.h>
     5	
     6	#undef notrace
     7	#define notrace __attribute__((no_instrument_function))
     8	
     9	#ifdef CONFIG_X86_32
    10	#define asmlinkage CPP_ASMLINKAGE __attribute__((regparm(0)))
    11	#endif /* CONFIG_X86_32 */
```

> I believe asmlinkage is defined just to avoid
> possible asm/c++ symbol decorations.
> 
> > In Linux, commit 9c0ca6f9 (update checkpatch.pl to version 0.10) seems
> > to have introduced the check. The commit message contains “asmlinkage
> > is also a storage type”.
> > 
> > Furthermore, `checkpatch.pl` doesn’t seem to warn about the code below.
> > 
> > ```
> > void __attribute__((weak)) mainboard_suspend_resume(void)
> > ```
> > 
> > This raises the question below.
> > 
> > > It appears coreboot proper mostly followed this placement for
> > > function attributes before. It would be nice if we were consistent,
> > > specially if checkpatch starts to complaint about these.
> > 
> > Is there another reason, besides not having that implemented?
> > 
> > I am looking forward to your answers.


Kind regards,

Paul


> > [1] https://review.coreboot.org/#/c/18865/1/src/arch/x86/acpi_s3.c@205
> > [2] https://review.coreboot.org/18865/
> > [3] https://review.coreboot.org/#/c/18865/1/src/arch/x86/acpi_s3.c@244
Download attachment "signature.asc" of type "application/pgp-signature" (196 bytes)