linux-kernel - Re: [PATCH 3/7] ptrace: move SIGTRAP on exec(2) logic to ptrace

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]

Message-ID: <20110623085814.GG30101@htj.dyndns.org>
Date:	Thu, 23 Jun 2011 10:58:14 +0200
From:	Tejun Heo <tj@...nel.org>
To:	Oleg Nesterov <oleg@...hat.com>
Cc:	linux-kernel@...r.kernel.org, akpm@...ux-foundation.org,
	torvalds@...ux-foundation.org, hch@...radead.org
Subject: Re: [PATCH 3/7] ptrace: move SIGTRAP on exec(2) logic to
 ptrace_event()

Hello,

On Tue, Jun 21, 2011 at 10:40:01PM +0200, Oleg Nesterov wrote:
> > The intention is to concentrate ptrace specific logic in
> > ptrace_event().  We'll have more of them, mostly dependent on
> > PT_SEIZED and I don't think it's a good idea to scatter them across
> > the kernel.  They're of no interest outside of ptrace after all.  I
> > think it's better to have them collected in one place than scattered
> > around.
> 
> This was one of the reasons for tracehooks ;)

Yeah, sure.  The problem with tracehook is not that it's a
fundamentally wrong thing to do but it's way over done than actually
necessary and full of requirements which are of no interest to inside
the upstream kernel.

> OK, we can move this helper to ptrace.h although I do not think this
> makes sense. As for "scattered around", imho the code which calculates
> trace in do_fork() falls into the same category.
> 
> I still can't understand why ptrace_event() should check EVENT_EXEC.
> This is the special case, it should be handled specially. And while
> I think this is not that important, this is not friendly to do_fork,
> compiler has to generate the code to check event.

The generated code should be the same.  I didn't check all the cases
but the few I check didn't really change.

> But OK, I applied 1-5 and 7. This is minor, and we can reconsider this
> later. I mean, right now I think I'll send the cleanup later, and you
> will have to explain your nack ;)

Yeah, yeah, it's a minor point one way or the other.  I was (and still
is) thinking about folding most of ptrace-specific event logics into
ptrace_event() and this was just a first step as I think adding more
condition checks in fs/exec.c doesn't make much sense - ie. I might
end up adding PT_SEZIED there too.

Thanks.

-- 
tejun
--
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/