linux-mips
[Top] [All Lists]

Re: [PATCH 9/10] signal: do not use save_static_function() anymore

To: "Atsushi Nemoto" <anemo@mba.ocn.ne.jp>
Subject: Re: [PATCH 9/10] signal: do not use save_static_function() anymore
From: "Franck Bui-Huu" <vagabon.xyz@gmail.com>
Date: Thu, 8 Feb 2007 21:05:03 +0100
Cc: ralf@linux-mips.org, linux-mips@linux-mips.org
Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=O7er7CwmzqEvt3Ht1aRStkfp7pHHoWmA8LiHXA1poMJ7Z0GzvatUQvPDkS4SUG77G1sx/pqAhNk4FjtQa2//lCE3jMtHqrBSe5LilMtq9Qtf1UKcM+QGntDnbaW1mTTAohOroBDWu8jPwVu4LElEKzZDpIv7Begcs/G3oDuX3zg=
In-reply-to: <20070209.013507.52129192.anemo@mba.ocn.ne.jp>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <cda58cb80702080053m6f22dc15td3b8c447e2abbda1@mail.gmail.com> <20070208.223637.108120499.anemo@mba.ocn.ne.jp> <cda58cb80702080739y18d31a34gc184a0cc96c86fb0@mail.gmail.com> <20070209.013507.52129192.anemo@mba.ocn.ne.jp>
Sender: linux-mips-bounce@linux-mips.org
On 2/8/07, Atsushi Nemoto <anemo@mba.ocn.ne.jp> wrote:
On Thu, 8 Feb 2007 16:39:42 +0100, "Franck Bui-Huu" <vagabon.xyz@gmail.com> 
wrote:
> But the points are:
>
>       - get rid of saving static registers in setup_sigcontext()
>       - get rid of restoring static registers in restore_sigcontext()
>       - free space in the signal frame

I'm afraid of ABI compatibility.  Someone might try to handle SIGSEGV
and dump all registers to debug the program without debugger...


Yes that's the main issue with this change. We could make it
configurable with an option which would depend on CONFIG_EMBEDDED or
something. Therefore someone can turn on the optimization if he really
wants it on his platform. But we would still lose the extra space gain
in the signal frame.

Note: I think that such programs can have trouble with current code
anyway... What would happen if the sig handler is run when returning
from a syscall ? In this case wouldn't sig context contain almost
garbage ?

--
              Franck

<Prev in Thread] Current Thread [Next in Thread>