Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1764877AbXLUCar (ORCPT ); Thu, 20 Dec 2007 21:30:47 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755116AbXLUCaR (ORCPT ); Thu, 20 Dec 2007 21:30:17 -0500 Received: from ozlabs.org ([203.10.76.45]:35983 "EHLO ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753161AbXLUCaM (ORCPT ); Thu, 20 Dec 2007 21:30:12 -0500 To: CC: Andrew Morton , , , , , , From: Benjamin Herrenschmidt Date: Fri, 21 Dec 2007 13:30:07 +1100 Subject: [PATCH 1/2] DMA buffer alignment annotations Message-Id: <20071221023010.C56B7DDDF3@ozlabs.org> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4563 Lines: 117 This patch based on some earlier work by Roland Dreier introduces a pair of annotations that can be used to enforce alignment of objects that can be DMA'ed into, and to enforce that an DMA'able object within a structure isn't sharing a cache line with some other object. Such sharing of a data structure between DMA and non-DMA objects isn't a recommended practice, but it does happen and in some case might even make sense, so we now have a way to make it work propertly. The current patch only enables such alignment for some PowerPC platforms that do not have coherent caches. Other platforms such as ARM, MIPS, etc... can define ARCH_MIN_DMA_ALIGNMENT if they want to benefit from this, I don't know them well enough to do it myself. The initial issue I'm fixing (in a second patch) by using these is the SCSI sense buffer which is currently part of the scsi command structure and can be DMA'ed to. On non-coherent platforms, this causes various corruptions as this cache line is shared with various other fields of the scsi_cmnd data structure. Signed-off-by: Benjamin Herrenschmidt --- Documentation/DMA-mapping.txt | 32 ++++++++++++++++++++++++++++++++ include/asm-generic/page.h | 10 ++++++++++ include/asm-powerpc/page.h | 8 ++++++++ 3 files changed, 50 insertions(+) --- linux-merge.orig/include/asm-generic/page.h 2007-07-27 13:44:45.000000000 +1000 +++ linux-merge/include/asm-generic/page.h 2007-12-21 13:07:28.000000000 +1100 @@ -20,6 +20,16 @@ static __inline__ __attribute_const__ in return order; } +#ifndef ARCH_MIN_DMA_ALIGNMENT +#define __dma_aligned +#define __dma_buffer +#else +#define __dma_aligned __attribute__((aligned(ARCH_MIN_DMA_ALIGNMENT))) +#define __dma_buffer __dma_buffer_line(__LINE__) +#define __dma_buffer_line(line) __dma_aligned;\ + char __dma_pad_##line[0] __dma_aligned +#endif + #endif /* __ASSEMBLY__ */ #endif /* __KERNEL__ */ Index: linux-merge/include/asm-powerpc/page.h =================================================================== --- linux-merge.orig/include/asm-powerpc/page.h 2007-09-28 11:42:10.000000000 +1000 +++ linux-merge/include/asm-powerpc/page.h 2007-12-21 13:15:02.000000000 +1100 @@ -77,6 +77,14 @@ #define VM_DATA_DEFAULT_FLAGS64 (VM_READ | VM_WRITE | \ VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC) +/* + * On non cache coherent platforms, we enforce cache aligned DMA + * buffers inside of structures + */ +#ifdef CONFIG_NOT_COHERENT_CACHE +#define ARCH_MIN_DMA_ALIGNMENT L1_CACHE_BYTES +#endif + #ifdef __powerpc64__ #include #else Index: linux-merge/Documentation/DMA-mapping.txt =================================================================== --- linux-merge.orig/Documentation/DMA-mapping.txt 2007-12-21 13:17:14.000000000 +1100 +++ linux-merge/Documentation/DMA-mapping.txt 2007-12-21 13:20:00.000000000 +1100 @@ -75,6 +75,38 @@ What about block I/O and networking buff networking subsystems make sure that the buffers they use are valid for you to DMA from/to. +Note that on non-cache-coherent architectures, having a DMA buffer +that shares a cache line with other data can lead to memory +corruption. + +The __dma_buffer macro exists to allow safe DMA buffers to be declared +easily and portably as part of larger structures without causing bloat +on cache-coherent architectures. To get this macro, architectures have +to define ARCH_MIN_DMA_ALIGNMENT to the requested alignment value in +their asm/page.h before including asm-generic/page.h + +Of course these structures must be contained in memory that can be +used for DMA as described above. + +To use __dma_buffer, just declare a struct like: + + struct mydevice { + int field1; + char buffer[BUFFER_SIZE] __dma_buffer; + int field2; + }; + +If this is used in code like: + + struct mydevice *dev; + dev = kmalloc(sizeof *dev, GFP_KERNEL); + +then dev->buffer will be safe for DMA on all architectures. On a +cache-coherent architecture the members of dev will be aligned exactly +as they would have been without __dma_buffer; on a non-cache-coherent +architecture buffer and field2 will be aligned so that buffer does not +share a cache line with any other data. + DMA addressing limitations Does your device have any DMA addressing limitations? For example, is -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/