Return-path: Received: from mail-qw0-f46.google.com ([209.85.216.46]:53488 "EHLO mail-qw0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753428Ab1FJLRY (ORCPT ); Fri, 10 Jun 2011 07:17:24 -0400 Received: by qwk3 with SMTP id 3so1164133qwk.19 for ; Fri, 10 Jun 2011 04:17:24 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <4DF1E973.1030709@broadcom.com> References: <1307630701-9170-1-git-send-email-rvossen@broadcom.com> <1307630701-9170-13-git-send-email-rvossen@broadcom.com> <4DF1E973.1030709@broadcom.com> From: Jonas Gorski Date: Fri, 10 Jun 2011 13:17:04 +0200 Message-ID: (sfid-20110610_131727_885667_47B19A27) Subject: Re: [PATCH 12/21] staging: brcm80211: use __BIG_ENDIAN macro in dma.c To: Arend van Spriel Cc: Roland Vossen , "gregkh@suse.de" , "devel@linuxdriverproject.org" , "linux-wireless@vger.kernel.org" Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: Hi Arend, On 10 June 2011 11:52, Arend van Spriel wrote: > Hi Jonas, > > You are right as with previous mips related patches. Consider this a first > step which does not change functionality but aims to use the proper linux > provided macros instead of our own incarnations. Actually as far as I can tell this changes (non-)functionality: With the broken big endiness check the driver probably just doesn't work properly on MIPS BE systems, because some endianess conversion is missing*. With the endianess check fixed, the special handling for the Broadcom MIPS get's enabled, potentially crashing the system (or changing random registers), thus making it *more* broken for non BMIPS systems. Therefore I'd rather consider this a regression. Especially since (AFAIK) the BMIPS systems that can take advantage of this aren't supported in Linux (yet). If it were just making things less efficient, I wouldn't be complaining (at least not that much ;-). If you can say that this is a code path that isn't taken anyway, then my point is moot (for now). I don't know the driver enough to tell that. Jonas *assuming brcm80211 depends on this on BE systems and it isn't just an optimization