linux-kernel - Re: [PATCH v9 3.2 2/9] uprobes: handle breakpoint and signal step exception.

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:	Wed, 18 Jan 2012 16:17:49 +0530
From:	Srikar Dronamraju <srikar@...ux.vnet.ibm.com>
To:	Mike Frysinger <vapier@...too.org>
Cc:	Anton Arapov <anton@...hat.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Oleg Nesterov <oleg@...hat.com>, Ingo Molnar <mingo@...e.hu>,
	Andrew Morton <akpm@...ux-foundation.org>,
	LKML <linux-kernel@...r.kernel.org>,
	Linux-mm <linux-mm@...ck.org>, Andi Kleen <andi@...stfloor.org>,
	Christoph Hellwig <hch@...radead.org>,
	Steven Rostedt <rostedt@...dmis.org>,
	Roland McGrath <roland@...k.frob.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	Masami Hiramatsu <masami.hiramatsu.pt@...achi.com>,
	Arnaldo Carvalho de Melo <acme@...radead.org>,
	Ananth N Mavinakayanahalli <ananth@...ibm.com>,
	Stephen Rothwell <sfr@...b.auug.org.au>
Subject: Re: [PATCH v9 3.2 2/9] uprobes: handle breakpoint and signal step
 exception.

> On Wednesday 18 January 2012 04:02:32 Srikar Dronamraju wrote:
> > >   Can we use existing SET_IP() instead of set_instruction_pointer() ?
> > 
> > Oleg had already commented about this in one his uprobes reviews.
> > 
> > The GET_IP/SET_IP available in include/asm-generic/ptrace.h doesnt work
> > on all archs. Atleast it doesnt work on powerpc when I tried it.
> 
> so migrate the arches you need over to it.

One question that could be asked is why arent we using instruction_pointer
instead of GET_IP since instruction_pointer is being defined in 25
places and with references in 120 places.

> 
> > Also most archs define instruction_pointer(). So I thought (rather Peter
> > Zijlstra suggested the name set_instruction_pointer())
> > set_instruction_pointer was a better bet than SET_IP. I
> 
> asm-generic/ptrace.h already has instruction_pointer_set()
> 
> > Also I dont see any usage for SET_IP/GET_IP.
> 
> i think you mean "users" here ?  the usage should be fairly obvious.  both 
> macros are used by asm-generic/ptrace.h internally, but (currently) rarely 
> defined by arches themselves (by design).  the funcs that are based on these 
> GET/SET helpers though do get used in many places.
> 
> simply grep arch/*/include/asm/ptrace.h

here are the stats

$ grep -r -w GET_IP * | wc -l 
5
$ grep -r -w SET_IP * | wc -l 
3
$ grep -r -w instruction_pointer * | wc -l 
120
$ grep -r -w instruction_pointer_set * | wc -l
3

The only place I saw GET_IP was used was to define SET_IP
The only place I saw SET_IP was used was to define
instruction_pointer_set.
The only place  I saw instruction_pointer_set being used is drivers/misc/kgdbts.c 

instruction_pointer was defined in close to 25 places.

> 
> > May be we should have something like this in
> > include/asm-generic/ptrace.h
> > 
> > #ifdef instruction_pointer
> > #define GET_IP(regs)		(instruction_pointer(regs))
> > #define set_instruction_pointer(regs, val) (instruction_pointer(regs) =
> > (val))
> > #define SET_IP(regs, val)	(set_instruction_pointer(regs,val))
> > #endif
> > 
> 
> what you propose here won't work on all arches which is the whole point of 
> {G,S}ET_IP in the first place.  i proposed a similar idea before and was shot 
> down for exactly that reason.  look at ia64 for an obvious example.

Sorry, I didnt quite understand this.
Was it that people objected to instruction_pointer or 
Is it that instruction_pointer and GET_IP will work differently on few
architectures or 
Is it people had an objection to defining instruction_pointer.

So let me rephrase here. Initially we used set_ip. But Peter suggested
that the name be changed to set_instruction_pointer so that it goes with 
instruction_pointer. I also felt that set_instruction_pointer was
better.  However I am okay with any other name including
SET_IP/instruction_pointer_set. I have no issues in moving the
set_instruction_pointer to arch/*/ptrace.h files it it helps (including
include/asm-generic/ptrace.h).

But I think  we should either have GET_IP or instruction_pointer.
Similarly either SET_IP/set_instruction_pointer{_set}. Since
instruction_pointer is more widely used, I would side by the
instruction_pointer.

> 
> > or should we do away with GET_IP/SET_IP esp since there are no many
> > users?
> 
> no, the point is to migrate to asm-generic/ptrace.h, not away from it.

I think the rational for having asm-generic/ptrace.h was to have define
a way to get the instruction_pointer such that the each archs dont have
to define their own definition unless and untill its necessary.

If yes, then why did we choose the names GET_IP/SET_IP instead of
instruction_pointer and the like.

-- 
Thanks and Regards
Srikar

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/