linux-kernel - Re: [PATCH 1/3] m68k/mac: Improve NMI handler

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

Date:   Tue, 3 Jan 2017 15:31:34 +1100 (AEDT)
From:   Finn Thain <fthain@...egraphics.com.au>
To:     Geert Uytterhoeven <geert@...ux-m68k.org>
cc:     Linux/m68k <linux-m68k@...r.kernel.org>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/3] m68k/mac: Improve NMI handler


On Mon, 2 Jan 2017, Geert Uytterhoeven wrote:

> >
> > -/*
> > - * console_loglevel determines NMI handler function
> > - */
> > +extern void show_registers(struct pt_regs *);
> 
> Seems like we do have a declaration in ... <linux/kprobes.h>.
> 

Yes, and it would have to be moved outside of the #ifdef CONFIG_KPROBES 
portion before it could be used by m68k, openrisc, cris or mn10300, which 
all lack HAVE_KPROBES.

I can't see why linux/kprobes.h is more appropriate for this declaration 
than, say linux/sched.h.

Despite the checkpatch warning, placing the extern at the call site is 
very popular.

$ egrep -r "extern.*show_registers" arch/
arch/mips/kernel/unaligned.c:extern void show_registers(struct pt_regs *regs);
arch/openrisc/kernel/process.c: extern void show_registers(struct pt_regs *regs);
arch/cris/kernel/traps.c:extern void show_registers(struct pt_regs *regs);
arch/cris/mm/fault.c:extern void show_registers(struct pt_regs *regs);
arch/cris/arch-v32/mach-a3/arbiter.c:extern void show_registers(struct pt_regs *regs);
arch/cris/arch-v32/kernel/time.c:extern void show_registers(struct pt_regs *regs);
arch/cris/arch-v32/mach-fs/arbiter.c:extern void show_registers(struct pt_regs *regs);
arch/mn10300/include/asm/gdb-stub.h:extern void show_registers_only(struct pt_regs *regs);
arch/mn10300/include/asm/processor.h:extern void show_registers(struct pt_regs *regs);
arch/frv/include/asm/gdb-stub.h:extern void show_registers_only(struct pt_regs *regs);
$ 
$ egrep -r "extern.*show_registers" include/
include/linux/kprobes.h:extern void show_registers(struct pt_regs *regs);
$ 

The situation with show_regs(struct pt_regs *) is a bit better in that the 
declaration is not found at multiple call sites (as with cris) but still 
appears in multiple header files.

$ egrep -r "extern.*show_regs" arch/
arch/x86/include/asm/kdebug.h:extern void __show_regs(struct pt_regs *regs, int all);
arch/nios2/include/asm/ptrace.h:extern void show_regs(struct pt_regs *);
arch/arm/include/asm/bug.h:extern void __show_regs(struct pt_regs *);
arch/tile/include/asm/stack.h:extern void tile_show_regs(struct pt_regs *);
arch/arm64/include/asm/system_misc.h:extern void __show_regs(struct pt_regs *);
arch/c6x/include/asm/ptrace.h:extern void show_regs(struct pt_regs *);
arch/avr32/include/asm/processor.h:extern void show_regs_log_lvl(struct pt_regs *regs, const char *log_lvl);
arch/unicore32/kernel/setup.h:extern void __show_regs(struct pt_regs *);
arch/alpha/kernel/proto.h:extern void dik_show_regs(struct pt_regs *regs, unsigned long *r9_15);
$ 
$ egrep -r "extern.*show_regs" include/
include/linux/sched.h:extern void show_regs(struct pt_regs *);
$ 

I guess we could put a show_registers() declaration in 
arch/m68k/include/asm/ptrace.h, as that's where struct pt_regs definition 
comes from (actually uapi/asm/ptrace.h). Both nios2 and c6x do this for 
show_regs().

Or we could just leave the patch as it is, because thus far m68k has no 
need for any declaration except a sole call site in macints.c.

--