From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Luis R. Rodriguez" Subject: Re: linux-next: build failure after merge of the tip tree Date: Tue, 28 Jul 2015 18:34:19 +0200 Message-ID: <20150728163419.GZ30479@wotan.suse.de> References: <20150728153303.2ba27786@canb.auug.org.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from mx2.suse.de ([195.135.220.15]:36480 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751124AbbG1Qeb (ORCPT ); Tue, 28 Jul 2015 12:34:31 -0400 Content-Disposition: inline In-Reply-To: <20150728153303.2ba27786@canb.auug.org.au> Sender: linux-next-owner@vger.kernel.org List-ID: To: Stephen Rothwell , akpm@linux-foundation.org, benh@kernel.crashing.org Cc: Thomas Gleixner , Ingo Molnar , "H. Peter Anvin" , Peter Zijlstra , linux-next@vger.kernel.org, linux-kernel@vger.kernel.org On Tue, Jul 28, 2015 at 03:33:03PM +1000, Stephen Rothwell wrote: > Hi all, > > After merging the tip tree, today's linux-next build (powerpc > allyesconfig) failed like this: > > drivers/video/fbdev/aty/atyfb_base.c: In function 'atyfb_setup_generic': > drivers/video/fbdev/aty/atyfb_base.c:3447:2: error: implicit declaration of function 'ioremap_uc' [-Werror=implicit-function-declaration] > par->ati_regbase = ioremap_uc(info->fix.mmio_start, 0x1000); > ^ > drivers/video/fbdev/aty/atyfb_base.c:3447:19: warning: assignment makes pointer from integer without a cast > par->ati_regbase = ioremap_uc(info->fix.mmio_start, 0x1000); > ^ > > Caused by commits > > 3cc2dac5be3f ("drivers/video/fbdev/atyfb: Replace MTRR UC hole with strong UC") > 8c7ea50c010b ("x86/mm, asm-generic: Add IOMMU ioremap_uc() variant default") > > The latter defines ioremap_uc() for x86 and those architectures that > use asm-generic/io.h - which is not all of them :-( . The former commit > then uses ioremap_uc(). > > I have reverted commit 3cc2dac5be3f (and 7d89a3cb159a that follows it) > for today. This should be fixed by: http://git.kernel.org/tip/8c7ea50c010b2f1e006ad37c43f98202a31de2cb The way it was defined was to return NULL if an arch does not have it, *but* if the asm-generic io.h header is not included on some archs it will still fail, which leaves us no option but to then poke and define its implementaiton for archs which opt-out of asm-generic io.h Benh, in this case I believe its OK to to just map it to ioremap(), let me know what you think. diff --git a/arch/powerpc/include/asm/io.h b/arch/powerpc/include/asm/io.h index a8d2ef30d473..91db9153cd44 100644 --- a/arch/powerpc/include/asm/io.h +++ b/arch/powerpc/include/asm/io.h @@ -721,6 +721,7 @@ extern void __iomem *ioremap_prot(phys_addr_t address, unsigned long size, unsigned long flags); extern void __iomem *ioremap_wc(phys_addr_t address, unsigned long size); #define ioremap_nocache(addr, size) ioremap((addr), (size)) +#define ioremap_uc(addr, size) ioremap((addr), (size)) extern void iounmap(volatile void __iomem *addr); Luis