From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:46633) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1RLV3y-0008Jv-Nn for qemu-devel@nongnu.org; Wed, 02 Nov 2011 03:16:43 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1RLV3x-0001z5-G0 for qemu-devel@nongnu.org; Wed, 02 Nov 2011 03:16:42 -0400 Received: from mx1.redhat.com ([209.132.183.28]:2012) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1RLV3x-0001yw-7M for qemu-devel@nongnu.org; Wed, 02 Nov 2011 03:16:41 -0400 Date: Wed, 2 Nov 2011 09:17:37 +0200 From: "Michael S. Tsirkin" Message-ID: <20111102071737.GB5613@redhat.com> References: <1320041218-30487-1-git-send-email-david@gibson.dropbear.id.au> <1320041218-30487-4-git-send-email-david@gibson.dropbear.id.au> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1320041218-30487-4-git-send-email-david@gibson.dropbear.id.au> Subject: Re: [Qemu-devel] [PATCH 03/14] Add stub functions for PCI device models to do PCI DMA List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: David Gibson Cc: joerg.roedel@amd.com, agraf@suse.de, qemu-devel@nongnu.org, avi@redhat.com, eduard.munteanu@linux360.ro, rth@twiddle.net On Mon, Oct 31, 2011 at 05:06:47PM +1100, David Gibson wrote: > This patch adds functions to pci.[ch] to perform PCI DMA operations. > At present, these are just stubs which perform directly cpu physical > memory accesses. Stubs are included which are analogous to > cpu_physical_memory_{read,write}(), the stX_phys() and ldX_phys() > functions and cpu_physical_memory_{map,unmap}(). > > In addition, a wrapper around qemu_sglist_init() is provided, which > also takes a PCIDevice *. It's assumed that _init() is the only > sglist function which will need wrapping, the idea being that once we > have IOMMU support whatever IOMMU context handle the wrapper derives > from the PCI device will be stored within the sglist structure for > later use. > > Using these stubs, however, distinguishes PCI device DMA transactions from > other accesses to physical memory, which will allow PCI IOMMU support to > be added in one place, rather than updating every PCI driver at that time. > > That is, it allows us to update individual PCI drivers to support an IOMMU > without having yet determined the details of how the IOMMU emulation will > operate. This will let us remove the most bitrot-sensitive part of an > IOMMU patch in advance. > > Signed-off-by: David Gibson > Signed-off-by: Alexey Kardashevskiy > --- > hw/pci.h | 67 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > 1 files changed, 67 insertions(+), 0 deletions(-) > > diff --git a/hw/pci.h b/hw/pci.h > index 86a81c8..c449a90 100644 > --- a/hw/pci.h > +++ b/hw/pci.h > @@ -6,6 +6,7 @@ > > #include "qdev.h" > #include "memory.h" > +#include "dma.h" > > /* PCI includes legacy ISA access. */ > #include "isa.h" > @@ -487,4 +488,70 @@ static inline uint32_t pci_config_size(const PCIDevice *d) > return pci_is_express(d) ? PCIE_CONFIG_SPACE_SIZE : PCI_CONFIG_SPACE_SIZE; > } > > +/* DMA access functions */ > +static inline int pci_dma_rw(PCIDevice *dev, dma_addr_t addr, > + void *buf, dma_addr_t len, DMADirection dir) > +{ > + cpu_physical_memory_rw(addr, buf, len, dir == DMA_DIRECTION_FROM_DEVICE); > + return 0; > +} > + > +static inline int pci_dma_read(PCIDevice *dev, dma_addr_t addr, > + void *buf, dma_addr_t len) > +{ > + return pci_dma_rw(dev, addr, buf, len, DMA_DIRECTION_TO_DEVICE); > +} > + > +static inline int pci_dma_write(PCIDevice *dev, dma_addr_t addr, > + const void *buf, dma_addr_t len) > +{ > + return pci_dma_rw(dev, addr, (void *) buf, len, DMA_DIRECTION_FROM_DEVICE); > +} > + > +#define PCI_DMA_DEFINE_LDST(_l, _s, _bits) \ > + static inline uint##_bits##_t ld##_l##_pci_dma(PCIDevice *dev, \ > + dma_addr_t addr) \ > + { \ > + return ld##_l##_phys(addr); \ > + } \ > + static inline void st##_s##_pci_dma(PCIDevice *dev, \ > + dma_addr_t addr, uint##_bits##_t val) \ > + { \ > + st##_s##_phys(addr, val); \ > + } > + I'd much rather name APIs pci_dma_XXX. why use a suffix and not a prefix? > +PCI_DMA_DEFINE_LDST(ub, b, 8); > +PCI_DMA_DEFINE_LDST(uw_le, w_le, 16) > +PCI_DMA_DEFINE_LDST(l_le, l_le, 32); > +PCI_DMA_DEFINE_LDST(q_le, q_le, 64); > +PCI_DMA_DEFINE_LDST(uw_be, w_be, 16) > +PCI_DMA_DEFINE_LDST(l_be, l_be, 32); > +PCI_DMA_DEFINE_LDST(q_be, q_be, 64); > + > +#undef PCI_DMA_DEFINE_LDST > + > +static inline void *pci_dma_map(PCIDevice *dev, dma_addr_t addr, > + dma_addr_t *plen, DMADirection dir) > +{ > + target_phys_addr_t len = *plen; > + void *buf; > + > + buf = cpu_physical_memory_map(addr, &len, dir == DMA_DIRECTION_FROM_DEVICE); > + *plen = len; > + return buf; > +} > + > +static inline void pci_dma_unmap(PCIDevice *dev, void *buffer, dma_addr_t len, > + DMADirection dir, dma_addr_t access_len) > +{ > + cpu_physical_memory_unmap(buffer, len, dir == DMA_DIRECTION_FROM_DEVICE, > + access_len); > +} > + > +static inline void pci_dma_sglist_init(QEMUSGList *qsg, PCIDevice *dev, > + int alloc_hint) > +{ > + qemu_sglist_init(qsg, alloc_hint); > +} > + > #endif > -- > 1.7.7