From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnd Bergmann Subject: Re: [PATCH v4] mmc: Add ioctl to let userspace apps send ACMDs Date: Mon, 18 Apr 2011 16:39:48 +0200 Message-ID: <201104181639.48730.arnd@arndb.de> References: <201104130100.39810.arnd@arndb.de> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-15" Content-Transfer-Encoding: 7bit Return-path: Received: from moutng.kundenserver.de ([212.227.126.187]:55372 "EHLO moutng.kundenserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753258Ab1DROkA (ORCPT ); Mon, 18 Apr 2011 10:40:00 -0400 In-Reply-To: Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: John Calixto Cc: =?utf-8?q?Micha=C5=82_Miros=C5=82aw?= , linux-mmc@vger.kernel.org, Chris Ball , Andrei Warkentin On Wednesday 13 April 2011, John Calixto wrote: > Gah! OK, so much for the v5 I just sent then. What do you think about > the compat_ioctl that I sent in v2? It means having the extra 32-bit > compat structure, but at least all the compat overhead is conditional > upon CONFIG_COMPAT. If you're not using CONFIG_COMPAT, you don't get > any compat cruft. > A single __u64 is sufficient, because the user space will do the correct conversion from pointer to 64-bit integer then. The only conversion you need to worry about is the actual pointer to the main structure, which needs the compat_ptr() magic. Arnd