From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from palinux.external.hp.com ([192.25.206.14]:47531 "EHLO mail.parisc-linux.org") by vger.kernel.org with ESMTP id S932110AbWJ1Wrk (ORCPT ); Sat, 28 Oct 2006 18:47:40 -0400 Date: Sat, 28 Oct 2006 16:47:38 -0600 From: Matthew Wilcox Subject: Re: Generic compat_sys_rt_sigqueueinfo Message-ID: <20061028224738.GT5591@parisc-linux.org> References: <20061028223730.GC3243@athena.road.mcmartin.ca> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20061028223730.GC3243@athena.road.mcmartin.ca> Sender: linux-arch-owner@vger.kernel.org To: Kyle McMartin Cc: akpm@osdl.org, linux-arch@vger.kernel.org List-ID: On Sat, Oct 28, 2006 at 06:37:30PM -0400, Kyle McMartin wrote: > Every 32-on-64 arch (except parisc) was implementing effectively the > same sys32_rt_sigqueueinfo. Add a generic implementation taken from > x86_64 and add it to kernel/compat.c, which required the addition of > copy_siginfo_from_user32 on mips and powerpc. sparc64 had a sufficient > implementation for this purpose, so crib it for both those platforms. Why can't we have a compat_copy_siginfo_from_user() (and compat_copy_siginfo_to_user() for that matter) defined in kernel/compat.c for all architectures? The user32 convention really grates for some reason ;-)