From: Jean-Christophe PLAGNIOL-VILLARD Date: Wed, 8 Jul 2009 00:07:10 +0200 Subject: [U-Boot] [PATCH v4 3/6] ARM: Add macros.h to be used in assembler file. In-Reply-To: <20090707212807.D0D2F832E416@gemini.denx.de> References: <4a62b27ae616d894b9302d1b73a3e4a71bbdfbf1.1244918138.git.lilja.magnus@gmail.com> <6fd22327241accc2efcada8380923c8e9c523790.1244918138.git.lilja.magnus@gmail.com> <20090707193316.EF032832E416@gemini.denx.de> <20090707202432.GK30172@game.jcrosoft.org> <20090707212807.D0D2F832E416@gemini.denx.de> Message-ID: <20090707220710.GF30395@game.jcrosoft.org> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de On 23:28 Tue 07 Jul , Wolfgang Denk wrote: > Dear Jean-Christophe PLAGNIOL-VILLARD, > > In message <20090707202432.GK30172@game.jcrosoft.org> you wrote: > > > > > > + * These macros provide a convenient way to write 8, 16 and 32 bit data > > > > + * to any address. > > > > > > I think this comment might be wrong. It should probably read "... to > > > any _memory_ address". Using these macros to access device registers > > > and the like is probably nore recommended? > > you can access cpu register without problem with these macros > > Is it guaranteed that none of the existing ARM architectures and all > the plethora of their specific implementations of the zillions of > SoCs do not require any form of memory barrier instructions when > accessing device registers? Even when these are external peripherals > on some external bus? > > Sorry for such a stupid question about things that are eventually > obvious to you, but I don't know ARM processors that well... you may need it in specific case as cache, tlb, co-processor or smp otherwise you do not need it so normaly no-need for register access Best Regards, J.