From: baruch@tkos.co.il (Baruch Siach) Date: Mon, 15 Mar 2010 08:22:22 +0200 Subject: [PATCH 1/2] mx1: fix build when DEBUG_LL is enabled In-Reply-To: <20100311105637.GE16376@n2100.arm.linux.org.uk> References: <20100311103940.GA19320@pengutronix.de> <20100311105637.GE16376@n2100.arm.linux.org.uk> Message-ID: <20100315062222.GA27515@jasper.tkos.co.il> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hi Russel, Uwe, On Thu, Mar 11, 2010 at 10:56:37AM +0000, Russell King - ARM Linux wrote: > On Thu, Mar 11, 2010 at 11:39:41AM +0100, Uwe Kleine-K?nig wrote: > > Hello, > > > > On Tue, Mar 09, 2010 at 12:12:04PM +0200, Baruch Siach wrote: > > > Commit "mx1: prefix SOC specific defines with MX1_ and deprecate old names" > > > redefined the IO_ADDRESS macro in a way that is incompatible with > > > debug-macro.S. Reintroduce AVIC_IO_ADDRESS, and use it in debug-macro.S. > > This feels wrong. Can you please instead fix IMX_IO_ADDRESS? > > > > Something like: > > > > #ifdef __ASSEMBLY__ > > #define IOREG(x) (x) > > #else > > #define IOREG(x) ((void __force __iomem *)(x)) > > #endif > > > > #define IMX_IO_ADDRESS(addr, module) IOREG(\ > > ... > > > > could work. Don't know if the cast to unsigned long works in assembler. > > Maybe it's unneeded? Don't remember. > > This is what I put in place for OMAP: > > #ifdef __ASSEMBLER__ > #define IOMEM(x) (x) > #else > #define IOMEM(x) ((void __force __iomem *)(x)) > #endif > > and then define all your device virtual addresses as normal hex integers > wrapped in IOMEM(). > > You can then use those symbols in assembly and C without any harm. OK. I'll try this approach and send an updated patch shortly. baruch -- ~. .~ Tk Open Systems =}------------------------------------------------ooO--U--Ooo------------{= - baruch at tkos.co.il - tel: +972.2.679.5364, http://www.tkos.co.il -