| Message ID | 1492117462-19886-1-git-send-email-sstabellini@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Apr 13, 2017 at 02:04:21PM -0700, Stefano Stabellini wrote: > The following commit: > > commit 815dd18788fe0d41899f51b91d0560279cf16b0d > Author: Bart Van Assche <bart.vanassche@sandisk.com> > Date: Fri Jan 20 13:04:04 2017 -0800 > > treewide: Consolidate get_dma_ops() implementations > > rearranges get_dma_ops in a way that xen_dma_ops are not returned when > running on Xen anymore, dev->dma_ops is returned instead (see > arch/arm/include/asm/dma-mapping.h:get_arch_dma_ops and > include/linux/dma-mapping.h:get_dma_ops). > > Fix the problem by storing dev->dma_ops in dev_archdata, and setting > dev->dma_ops to xen_dma_ops. This way, xen_dma_ops is returned naturally > by get_dma_ops. The Xen code can retrieve the original dev->dma_ops from > dev_archdata when needed. It also allows us to remove __generic_dma_ops > from common headers. > > Signed-off-by: Stefano Stabellini <sstabellini@kernel.org> > Tested-by: Julien Grall <julien.grall@arm.com> > Suggested-by: Catalin Marinas <catalin.marinas@arm.com> > CC: linux@armlinux.org.uk > CC: catalin.marinas@arm.com > CC: will.deacon@arm.com > CC: boris.ostrovsky@oracle.com > CC: jgross@suse.com > CC: Julien Grall <julien.grall@arm.com> Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Hello Russell, Can I have your acked-by on the following fix (original patch is 1492117462-19886-1-git-send-email-sstabellini@kernel.org)? Thanks, Stefano On Tue, 18 Apr 2017, Catalin Marinas wrote: > On Thu, Apr 13, 2017 at 02:04:21PM -0700, Stefano Stabellini wrote: > > The following commit: > > > > commit 815dd18788fe0d41899f51b91d0560279cf16b0d > > Author: Bart Van Assche <bart.vanassche@sandisk.com> > > Date: Fri Jan 20 13:04:04 2017 -0800 > > > > treewide: Consolidate get_dma_ops() implementations > > > > rearranges get_dma_ops in a way that xen_dma_ops are not returned when > > running on Xen anymore, dev->dma_ops is returned instead (see > > arch/arm/include/asm/dma-mapping.h:get_arch_dma_ops and > > include/linux/dma-mapping.h:get_dma_ops). > > > > Fix the problem by storing dev->dma_ops in dev_archdata, and setting > > dev->dma_ops to xen_dma_ops. This way, xen_dma_ops is returned naturally > > by get_dma_ops. The Xen code can retrieve the original dev->dma_ops from > > dev_archdata when needed. It also allows us to remove __generic_dma_ops > > from common headers. > > > > Signed-off-by: Stefano Stabellini <sstabellini@kernel.org> > > Tested-by: Julien Grall <julien.grall@arm.com> > > Suggested-by: Catalin Marinas <catalin.marinas@arm.com> > > CC: linux@armlinux.org.uk > > CC: catalin.marinas@arm.com > > CC: will.deacon@arm.com > > CC: boris.ostrovsky@oracle.com > > CC: jgross@suse.com > > CC: Julien Grall <julien.grall@arm.com> > > Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> > > > > diff --git a/arch/arm/include/asm/device.h b/arch/arm/include/asm/device.h > > index 220ba20..36ec9c8 100644 > > --- a/arch/arm/include/asm/device.h > > +++ b/arch/arm/include/asm/device.h > > @@ -16,6 +16,9 @@ struct dev_archdata { > > #ifdef CONFIG_ARM_DMA_USE_IOMMU > > struct dma_iommu_mapping *mapping; > > #endif > > +#ifdef CONFIG_XEN > > + const struct dma_map_ops *dev_dma_ops; > > +#endif > > bool dma_coherent; > > }; > > > > diff --git a/arch/arm/include/asm/dma-mapping.h b/arch/arm/include/asm/dma-mapping.h > > index 7166569..680d3f3 100644 > > --- a/arch/arm/include/asm/dma-mapping.h > > +++ b/arch/arm/include/asm/dma-mapping.h > > @@ -16,19 +16,9 @@ > > extern const struct dma_map_ops arm_dma_ops; > > extern const struct dma_map_ops arm_coherent_dma_ops; > > > > -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > > -{ > > - if (dev && dev->dma_ops) > > - return dev->dma_ops; > > - return &arm_dma_ops; > > -} > > - > > static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > > { > > - if (xen_initial_domain()) > > - return xen_dma_ops; > > - else > > - return __generic_dma_ops(NULL); > > + return &arm_dma_ops; > > } > > > > #define HAVE_ARCH_DMA_SUPPORTED 1 > > diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c > > index 63eabb0..a614459 100644 > > --- a/arch/arm/mm/dma-mapping.c > > +++ b/arch/arm/mm/dma-mapping.c > > @@ -2396,6 +2396,13 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > > dma_ops = arm_get_dma_map_ops(coherent); > > > > set_dma_ops(dev, dma_ops); > > + > > +#ifdef CONFIG_XEN > > + if (xen_initial_domain()) { > > + dev->archdata.dev_dma_ops = dev->dma_ops; > > + dev->dma_ops = xen_dma_ops; > > + } > > +#endif > > } > > > > void arch_teardown_dma_ops(struct device *dev) > > diff --git a/arch/arm64/include/asm/device.h b/arch/arm64/include/asm/device.h > > index 73d5bab..5a5fa47 100644 > > --- a/arch/arm64/include/asm/device.h > > +++ b/arch/arm64/include/asm/device.h > > @@ -20,6 +20,9 @@ struct dev_archdata { > > #ifdef CONFIG_IOMMU_API > > void *iommu; /* private IOMMU data */ > > #endif > > +#ifdef CONFIG_XEN > > + const struct dma_map_ops *dev_dma_ops; > > +#endif > > bool dma_coherent; > > }; > > > > diff --git a/arch/arm64/include/asm/dma-mapping.h b/arch/arm64/include/asm/dma-mapping.h > > index 505756c..5392dbe 100644 > > --- a/arch/arm64/include/asm/dma-mapping.h > > +++ b/arch/arm64/include/asm/dma-mapping.h > > @@ -27,11 +27,8 @@ > > #define DMA_ERROR_CODE (~(dma_addr_t)0) > > extern const struct dma_map_ops dummy_dma_ops; > > > > -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > > +static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > > { > > - if (dev && dev->dma_ops) > > - return dev->dma_ops; > > - > > /* > > * We expect no ISA devices, and all other DMA masters are expected to > > * have someone call arch_setup_dma_ops at device creation time. > > @@ -39,14 +36,6 @@ static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > > return &dummy_dma_ops; > > } > > > > -static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > > -{ > > - if (xen_initial_domain()) > > - return xen_dma_ops; > > - else > > - return __generic_dma_ops(NULL); > > -} > > - > > void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > > const struct iommu_ops *iommu, bool coherent); > > #define arch_setup_dma_ops arch_setup_dma_ops > > diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c > > index 81cdb2e..7f8b37e 100644 > > --- a/arch/arm64/mm/dma-mapping.c > > +++ b/arch/arm64/mm/dma-mapping.c > > @@ -977,4 +977,11 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > > > > dev->archdata.dma_coherent = coherent; > > __iommu_setup_dma_ops(dev, dma_base, size, iommu); > > + > > +#ifdef CONFIG_XEN > > + if (xen_initial_domain()) { > > + dev->archdata.dev_dma_ops = dev->dma_ops; > > + dev->dma_ops = xen_dma_ops; > > + } > > +#endif > > } > > diff --git a/include/xen/arm/page-coherent.h b/include/xen/arm/page-coherent.h > > index 95ce6ac..b0a2bfc 100644 > > --- a/include/xen/arm/page-coherent.h > > +++ b/include/xen/arm/page-coherent.h > > @@ -2,8 +2,16 @@ > > #define _ASM_ARM_XEN_PAGE_COHERENT_H > > > > #include <asm/page.h> > > +#include <asm/dma-mapping.h> > > #include <linux/dma-mapping.h> > > > > +static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > > +{ > > + if (dev && dev->archdata.dev_dma_ops) > > + return dev->archdata.dev_dma_ops; > > + return get_arch_dma_ops(NULL); > > +} > > + > > void __xen_dma_map_page(struct device *hwdev, struct page *page, > > dma_addr_t dev_addr, unsigned long offset, size_t size, > > enum dma_data_direction dir, unsigned long attrs);
On 19/04/17 19:25, Stefano Stabellini wrote: > Hello Russell, > > Can I have your acked-by on the following fix (original patch is > 1492117462-19886-1-git-send-email-sstabellini@kernel.org)? Stefano, through which tree should this go? ARM or Xen or other? Juergen > > Thanks, > > Stefano > > > On Tue, 18 Apr 2017, Catalin Marinas wrote: >> On Thu, Apr 13, 2017 at 02:04:21PM -0700, Stefano Stabellini wrote: >>> The following commit: >>> >>> commit 815dd18788fe0d41899f51b91d0560279cf16b0d >>> Author: Bart Van Assche <bart.vanassche@sandisk.com> >>> Date: Fri Jan 20 13:04:04 2017 -0800 >>> >>> treewide: Consolidate get_dma_ops() implementations >>> >>> rearranges get_dma_ops in a way that xen_dma_ops are not returned when >>> running on Xen anymore, dev->dma_ops is returned instead (see >>> arch/arm/include/asm/dma-mapping.h:get_arch_dma_ops and >>> include/linux/dma-mapping.h:get_dma_ops). >>> >>> Fix the problem by storing dev->dma_ops in dev_archdata, and setting >>> dev->dma_ops to xen_dma_ops. This way, xen_dma_ops is returned naturally >>> by get_dma_ops. The Xen code can retrieve the original dev->dma_ops from >>> dev_archdata when needed. It also allows us to remove __generic_dma_ops >>> from common headers. >>> >>> Signed-off-by: Stefano Stabellini <sstabellini@kernel.org> >>> Tested-by: Julien Grall <julien.grall@arm.com> >>> Suggested-by: Catalin Marinas <catalin.marinas@arm.com> >>> CC: linux@armlinux.org.uk >>> CC: catalin.marinas@arm.com >>> CC: will.deacon@arm.com >>> CC: boris.ostrovsky@oracle.com >>> CC: jgross@suse.com >>> CC: Julien Grall <julien.grall@arm.com> >> >> Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> >> >> >>> diff --git a/arch/arm/include/asm/device.h b/arch/arm/include/asm/device.h >>> index 220ba20..36ec9c8 100644 >>> --- a/arch/arm/include/asm/device.h >>> +++ b/arch/arm/include/asm/device.h >>> @@ -16,6 +16,9 @@ struct dev_archdata { >>> #ifdef CONFIG_ARM_DMA_USE_IOMMU >>> struct dma_iommu_mapping *mapping; >>> #endif >>> +#ifdef CONFIG_XEN >>> + const struct dma_map_ops *dev_dma_ops; >>> +#endif >>> bool dma_coherent; >>> }; >>> >>> diff --git a/arch/arm/include/asm/dma-mapping.h b/arch/arm/include/asm/dma-mapping.h >>> index 7166569..680d3f3 100644 >>> --- a/arch/arm/include/asm/dma-mapping.h >>> +++ b/arch/arm/include/asm/dma-mapping.h >>> @@ -16,19 +16,9 @@ >>> extern const struct dma_map_ops arm_dma_ops; >>> extern const struct dma_map_ops arm_coherent_dma_ops; >>> >>> -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) >>> -{ >>> - if (dev && dev->dma_ops) >>> - return dev->dma_ops; >>> - return &arm_dma_ops; >>> -} >>> - >>> static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) >>> { >>> - if (xen_initial_domain()) >>> - return xen_dma_ops; >>> - else >>> - return __generic_dma_ops(NULL); >>> + return &arm_dma_ops; >>> } >>> >>> #define HAVE_ARCH_DMA_SUPPORTED 1 >>> diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c >>> index 63eabb0..a614459 100644 >>> --- a/arch/arm/mm/dma-mapping.c >>> +++ b/arch/arm/mm/dma-mapping.c >>> @@ -2396,6 +2396,13 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, >>> dma_ops = arm_get_dma_map_ops(coherent); >>> >>> set_dma_ops(dev, dma_ops); >>> + >>> +#ifdef CONFIG_XEN >>> + if (xen_initial_domain()) { >>> + dev->archdata.dev_dma_ops = dev->dma_ops; >>> + dev->dma_ops = xen_dma_ops; >>> + } >>> +#endif >>> } >>> >>> void arch_teardown_dma_ops(struct device *dev) >>> diff --git a/arch/arm64/include/asm/device.h b/arch/arm64/include/asm/device.h >>> index 73d5bab..5a5fa47 100644 >>> --- a/arch/arm64/include/asm/device.h >>> +++ b/arch/arm64/include/asm/device.h >>> @@ -20,6 +20,9 @@ struct dev_archdata { >>> #ifdef CONFIG_IOMMU_API >>> void *iommu; /* private IOMMU data */ >>> #endif >>> +#ifdef CONFIG_XEN >>> + const struct dma_map_ops *dev_dma_ops; >>> +#endif >>> bool dma_coherent; >>> }; >>> >>> diff --git a/arch/arm64/include/asm/dma-mapping.h b/arch/arm64/include/asm/dma-mapping.h >>> index 505756c..5392dbe 100644 >>> --- a/arch/arm64/include/asm/dma-mapping.h >>> +++ b/arch/arm64/include/asm/dma-mapping.h >>> @@ -27,11 +27,8 @@ >>> #define DMA_ERROR_CODE (~(dma_addr_t)0) >>> extern const struct dma_map_ops dummy_dma_ops; >>> >>> -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) >>> +static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) >>> { >>> - if (dev && dev->dma_ops) >>> - return dev->dma_ops; >>> - >>> /* >>> * We expect no ISA devices, and all other DMA masters are expected to >>> * have someone call arch_setup_dma_ops at device creation time. >>> @@ -39,14 +36,6 @@ static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) >>> return &dummy_dma_ops; >>> } >>> >>> -static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) >>> -{ >>> - if (xen_initial_domain()) >>> - return xen_dma_ops; >>> - else >>> - return __generic_dma_ops(NULL); >>> -} >>> - >>> void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, >>> const struct iommu_ops *iommu, bool coherent); >>> #define arch_setup_dma_ops arch_setup_dma_ops >>> diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c >>> index 81cdb2e..7f8b37e 100644 >>> --- a/arch/arm64/mm/dma-mapping.c >>> +++ b/arch/arm64/mm/dma-mapping.c >>> @@ -977,4 +977,11 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, >>> >>> dev->archdata.dma_coherent = coherent; >>> __iommu_setup_dma_ops(dev, dma_base, size, iommu); >>> + >>> +#ifdef CONFIG_XEN >>> + if (xen_initial_domain()) { >>> + dev->archdata.dev_dma_ops = dev->dma_ops; >>> + dev->dma_ops = xen_dma_ops; >>> + } >>> +#endif >>> } >>> diff --git a/include/xen/arm/page-coherent.h b/include/xen/arm/page-coherent.h >>> index 95ce6ac..b0a2bfc 100644 >>> --- a/include/xen/arm/page-coherent.h >>> +++ b/include/xen/arm/page-coherent.h >>> @@ -2,8 +2,16 @@ >>> #define _ASM_ARM_XEN_PAGE_COHERENT_H >>> >>> #include <asm/page.h> >>> +#include <asm/dma-mapping.h> >>> #include <linux/dma-mapping.h> >>> >>> +static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) >>> +{ >>> + if (dev && dev->archdata.dev_dma_ops) >>> + return dev->archdata.dev_dma_ops; >>> + return get_arch_dma_ops(NULL); >>> +} >>> + >>> void __xen_dma_map_page(struct device *hwdev, struct page *page, >>> dma_addr_t dev_addr, unsigned long offset, size_t size, >>> enum dma_data_direction dir, unsigned long attrs); >
On Wed, 19 Apr 2017, Juergen Gross wrote: > On 19/04/17 19:25, Stefano Stabellini wrote: > > Hello Russell, > > > > Can I have your acked-by on the following fix (original patch is > > 1492117462-19886-1-git-send-email-sstabellini@kernel.org)? > > Stefano, through which tree should this go? ARM or Xen or other? I think it could go via the Xen tree, but we still need an ack from the ARM maintainer for changes to code under arch/arm. > Juergen > > > > > Thanks, > > > > Stefano > > > > > > On Tue, 18 Apr 2017, Catalin Marinas wrote: > >> On Thu, Apr 13, 2017 at 02:04:21PM -0700, Stefano Stabellini wrote: > >>> The following commit: > >>> > >>> commit 815dd18788fe0d41899f51b91d0560279cf16b0d > >>> Author: Bart Van Assche <bart.vanassche@sandisk.com> > >>> Date: Fri Jan 20 13:04:04 2017 -0800 > >>> > >>> treewide: Consolidate get_dma_ops() implementations > >>> > >>> rearranges get_dma_ops in a way that xen_dma_ops are not returned when > >>> running on Xen anymore, dev->dma_ops is returned instead (see > >>> arch/arm/include/asm/dma-mapping.h:get_arch_dma_ops and > >>> include/linux/dma-mapping.h:get_dma_ops). > >>> > >>> Fix the problem by storing dev->dma_ops in dev_archdata, and setting > >>> dev->dma_ops to xen_dma_ops. This way, xen_dma_ops is returned naturally > >>> by get_dma_ops. The Xen code can retrieve the original dev->dma_ops from > >>> dev_archdata when needed. It also allows us to remove __generic_dma_ops > >>> from common headers. > >>> > >>> Signed-off-by: Stefano Stabellini <sstabellini@kernel.org> > >>> Tested-by: Julien Grall <julien.grall@arm.com> > >>> Suggested-by: Catalin Marinas <catalin.marinas@arm.com> > >>> CC: linux@armlinux.org.uk > >>> CC: catalin.marinas@arm.com > >>> CC: will.deacon@arm.com > >>> CC: boris.ostrovsky@oracle.com > >>> CC: jgross@suse.com > >>> CC: Julien Grall <julien.grall@arm.com> > >> > >> Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> > >> > >> > >>> diff --git a/arch/arm/include/asm/device.h b/arch/arm/include/asm/device.h > >>> index 220ba20..36ec9c8 100644 > >>> --- a/arch/arm/include/asm/device.h > >>> +++ b/arch/arm/include/asm/device.h > >>> @@ -16,6 +16,9 @@ struct dev_archdata { > >>> #ifdef CONFIG_ARM_DMA_USE_IOMMU > >>> struct dma_iommu_mapping *mapping; > >>> #endif > >>> +#ifdef CONFIG_XEN > >>> + const struct dma_map_ops *dev_dma_ops; > >>> +#endif > >>> bool dma_coherent; > >>> }; > >>> > >>> diff --git a/arch/arm/include/asm/dma-mapping.h b/arch/arm/include/asm/dma-mapping.h > >>> index 7166569..680d3f3 100644 > >>> --- a/arch/arm/include/asm/dma-mapping.h > >>> +++ b/arch/arm/include/asm/dma-mapping.h > >>> @@ -16,19 +16,9 @@ > >>> extern const struct dma_map_ops arm_dma_ops; > >>> extern const struct dma_map_ops arm_coherent_dma_ops; > >>> > >>> -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > >>> -{ > >>> - if (dev && dev->dma_ops) > >>> - return dev->dma_ops; > >>> - return &arm_dma_ops; > >>> -} > >>> - > >>> static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > >>> { > >>> - if (xen_initial_domain()) > >>> - return xen_dma_ops; > >>> - else > >>> - return __generic_dma_ops(NULL); > >>> + return &arm_dma_ops; > >>> } > >>> > >>> #define HAVE_ARCH_DMA_SUPPORTED 1 > >>> diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c > >>> index 63eabb0..a614459 100644 > >>> --- a/arch/arm/mm/dma-mapping.c > >>> +++ b/arch/arm/mm/dma-mapping.c > >>> @@ -2396,6 +2396,13 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > >>> dma_ops = arm_get_dma_map_ops(coherent); > >>> > >>> set_dma_ops(dev, dma_ops); > >>> + > >>> +#ifdef CONFIG_XEN > >>> + if (xen_initial_domain()) { > >>> + dev->archdata.dev_dma_ops = dev->dma_ops; > >>> + dev->dma_ops = xen_dma_ops; > >>> + } > >>> +#endif > >>> } > >>> > >>> void arch_teardown_dma_ops(struct device *dev) > >>> diff --git a/arch/arm64/include/asm/device.h b/arch/arm64/include/asm/device.h > >>> index 73d5bab..5a5fa47 100644 > >>> --- a/arch/arm64/include/asm/device.h > >>> +++ b/arch/arm64/include/asm/device.h > >>> @@ -20,6 +20,9 @@ struct dev_archdata { > >>> #ifdef CONFIG_IOMMU_API > >>> void *iommu; /* private IOMMU data */ > >>> #endif > >>> +#ifdef CONFIG_XEN > >>> + const struct dma_map_ops *dev_dma_ops; > >>> +#endif > >>> bool dma_coherent; > >>> }; > >>> > >>> diff --git a/arch/arm64/include/asm/dma-mapping.h b/arch/arm64/include/asm/dma-mapping.h > >>> index 505756c..5392dbe 100644 > >>> --- a/arch/arm64/include/asm/dma-mapping.h > >>> +++ b/arch/arm64/include/asm/dma-mapping.h > >>> @@ -27,11 +27,8 @@ > >>> #define DMA_ERROR_CODE (~(dma_addr_t)0) > >>> extern const struct dma_map_ops dummy_dma_ops; > >>> > >>> -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > >>> +static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > >>> { > >>> - if (dev && dev->dma_ops) > >>> - return dev->dma_ops; > >>> - > >>> /* > >>> * We expect no ISA devices, and all other DMA masters are expected to > >>> * have someone call arch_setup_dma_ops at device creation time. > >>> @@ -39,14 +36,6 @@ static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > >>> return &dummy_dma_ops; > >>> } > >>> > >>> -static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > >>> -{ > >>> - if (xen_initial_domain()) > >>> - return xen_dma_ops; > >>> - else > >>> - return __generic_dma_ops(NULL); > >>> -} > >>> - > >>> void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > >>> const struct iommu_ops *iommu, bool coherent); > >>> #define arch_setup_dma_ops arch_setup_dma_ops > >>> diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c > >>> index 81cdb2e..7f8b37e 100644 > >>> --- a/arch/arm64/mm/dma-mapping.c > >>> +++ b/arch/arm64/mm/dma-mapping.c > >>> @@ -977,4 +977,11 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > >>> > >>> dev->archdata.dma_coherent = coherent; > >>> __iommu_setup_dma_ops(dev, dma_base, size, iommu); > >>> + > >>> +#ifdef CONFIG_XEN > >>> + if (xen_initial_domain()) { > >>> + dev->archdata.dev_dma_ops = dev->dma_ops; > >>> + dev->dma_ops = xen_dma_ops; > >>> + } > >>> +#endif > >>> } > >>> diff --git a/include/xen/arm/page-coherent.h b/include/xen/arm/page-coherent.h > >>> index 95ce6ac..b0a2bfc 100644 > >>> --- a/include/xen/arm/page-coherent.h > >>> +++ b/include/xen/arm/page-coherent.h > >>> @@ -2,8 +2,16 @@ > >>> #define _ASM_ARM_XEN_PAGE_COHERENT_H > >>> > >>> #include <asm/page.h> > >>> +#include <asm/dma-mapping.h> > >>> #include <linux/dma-mapping.h> > >>> > >>> +static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > >>> +{ > >>> + if (dev && dev->archdata.dev_dma_ops) > >>> + return dev->archdata.dev_dma_ops; > >>> + return get_arch_dma_ops(NULL); > >>> +} > >>> + > >>> void __xen_dma_map_page(struct device *hwdev, struct page *page, > >>> dma_addr_t dev_addr, unsigned long offset, size_t size, > >>> enum dma_data_direction dir, unsigned long attrs); > > >
Hi Russell, Given the outstanding regression we need to fix as soon as possible, I'll queue these patches on the xentip tree for 4.12. Please shout if you disagree. Thank you. - Stefano On Wed, 19 Apr 2017, Stefano Stabellini wrote: > Hello Russell, > > Can I have your acked-by on the following fix (original patch is > 1492117462-19886-1-git-send-email-sstabellini@kernel.org)? > > Thanks, > > Stefano > > > On Tue, 18 Apr 2017, Catalin Marinas wrote: > > On Thu, Apr 13, 2017 at 02:04:21PM -0700, Stefano Stabellini wrote: > > > The following commit: > > > > > > commit 815dd18788fe0d41899f51b91d0560279cf16b0d > > > Author: Bart Van Assche <bart.vanassche@sandisk.com> > > > Date: Fri Jan 20 13:04:04 2017 -0800 > > > > > > treewide: Consolidate get_dma_ops() implementations > > > > > > rearranges get_dma_ops in a way that xen_dma_ops are not returned when > > > running on Xen anymore, dev->dma_ops is returned instead (see > > > arch/arm/include/asm/dma-mapping.h:get_arch_dma_ops and > > > include/linux/dma-mapping.h:get_dma_ops). > > > > > > Fix the problem by storing dev->dma_ops in dev_archdata, and setting > > > dev->dma_ops to xen_dma_ops. This way, xen_dma_ops is returned naturally > > > by get_dma_ops. The Xen code can retrieve the original dev->dma_ops from > > > dev_archdata when needed. It also allows us to remove __generic_dma_ops > > > from common headers. > > > > > > Signed-off-by: Stefano Stabellini <sstabellini@kernel.org> > > > Tested-by: Julien Grall <julien.grall@arm.com> > > > Suggested-by: Catalin Marinas <catalin.marinas@arm.com> > > > CC: linux@armlinux.org.uk > > > CC: catalin.marinas@arm.com > > > CC: will.deacon@arm.com > > > CC: boris.ostrovsky@oracle.com > > > CC: jgross@suse.com > > > CC: Julien Grall <julien.grall@arm.com> > > > > Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> > > > > > > > diff --git a/arch/arm/include/asm/device.h b/arch/arm/include/asm/device.h > > > index 220ba20..36ec9c8 100644 > > > --- a/arch/arm/include/asm/device.h > > > +++ b/arch/arm/include/asm/device.h > > > @@ -16,6 +16,9 @@ struct dev_archdata { > > > #ifdef CONFIG_ARM_DMA_USE_IOMMU > > > struct dma_iommu_mapping *mapping; > > > #endif > > > +#ifdef CONFIG_XEN > > > + const struct dma_map_ops *dev_dma_ops; > > > +#endif > > > bool dma_coherent; > > > }; > > > > > > diff --git a/arch/arm/include/asm/dma-mapping.h b/arch/arm/include/asm/dma-mapping.h > > > index 7166569..680d3f3 100644 > > > --- a/arch/arm/include/asm/dma-mapping.h > > > +++ b/arch/arm/include/asm/dma-mapping.h > > > @@ -16,19 +16,9 @@ > > > extern const struct dma_map_ops arm_dma_ops; > > > extern const struct dma_map_ops arm_coherent_dma_ops; > > > > > > -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > > > -{ > > > - if (dev && dev->dma_ops) > > > - return dev->dma_ops; > > > - return &arm_dma_ops; > > > -} > > > - > > > static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > > > { > > > - if (xen_initial_domain()) > > > - return xen_dma_ops; > > > - else > > > - return __generic_dma_ops(NULL); > > > + return &arm_dma_ops; > > > } > > > > > > #define HAVE_ARCH_DMA_SUPPORTED 1 > > > diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c > > > index 63eabb0..a614459 100644 > > > --- a/arch/arm/mm/dma-mapping.c > > > +++ b/arch/arm/mm/dma-mapping.c > > > @@ -2396,6 +2396,13 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > > > dma_ops = arm_get_dma_map_ops(coherent); > > > > > > set_dma_ops(dev, dma_ops); > > > + > > > +#ifdef CONFIG_XEN > > > + if (xen_initial_domain()) { > > > + dev->archdata.dev_dma_ops = dev->dma_ops; > > > + dev->dma_ops = xen_dma_ops; > > > + } > > > +#endif > > > } > > > > > > void arch_teardown_dma_ops(struct device *dev) > > > diff --git a/arch/arm64/include/asm/device.h b/arch/arm64/include/asm/device.h > > > index 73d5bab..5a5fa47 100644 > > > --- a/arch/arm64/include/asm/device.h > > > +++ b/arch/arm64/include/asm/device.h > > > @@ -20,6 +20,9 @@ struct dev_archdata { > > > #ifdef CONFIG_IOMMU_API > > > void *iommu; /* private IOMMU data */ > > > #endif > > > +#ifdef CONFIG_XEN > > > + const struct dma_map_ops *dev_dma_ops; > > > +#endif > > > bool dma_coherent; > > > }; > > > > > > diff --git a/arch/arm64/include/asm/dma-mapping.h b/arch/arm64/include/asm/dma-mapping.h > > > index 505756c..5392dbe 100644 > > > --- a/arch/arm64/include/asm/dma-mapping.h > > > +++ b/arch/arm64/include/asm/dma-mapping.h > > > @@ -27,11 +27,8 @@ > > > #define DMA_ERROR_CODE (~(dma_addr_t)0) > > > extern const struct dma_map_ops dummy_dma_ops; > > > > > > -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > > > +static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > > > { > > > - if (dev && dev->dma_ops) > > > - return dev->dma_ops; > > > - > > > /* > > > * We expect no ISA devices, and all other DMA masters are expected to > > > * have someone call arch_setup_dma_ops at device creation time. > > > @@ -39,14 +36,6 @@ static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > > > return &dummy_dma_ops; > > > } > > > > > > -static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) > > > -{ > > > - if (xen_initial_domain()) > > > - return xen_dma_ops; > > > - else > > > - return __generic_dma_ops(NULL); > > > -} > > > - > > > void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > > > const struct iommu_ops *iommu, bool coherent); > > > #define arch_setup_dma_ops arch_setup_dma_ops > > > diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c > > > index 81cdb2e..7f8b37e 100644 > > > --- a/arch/arm64/mm/dma-mapping.c > > > +++ b/arch/arm64/mm/dma-mapping.c > > > @@ -977,4 +977,11 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, > > > > > > dev->archdata.dma_coherent = coherent; > > > __iommu_setup_dma_ops(dev, dma_base, size, iommu); > > > + > > > +#ifdef CONFIG_XEN > > > + if (xen_initial_domain()) { > > > + dev->archdata.dev_dma_ops = dev->dma_ops; > > > + dev->dma_ops = xen_dma_ops; > > > + } > > > +#endif > > > } > > > diff --git a/include/xen/arm/page-coherent.h b/include/xen/arm/page-coherent.h > > > index 95ce6ac..b0a2bfc 100644 > > > --- a/include/xen/arm/page-coherent.h > > > +++ b/include/xen/arm/page-coherent.h > > > @@ -2,8 +2,16 @@ > > > #define _ASM_ARM_XEN_PAGE_COHERENT_H > > > > > > #include <asm/page.h> > > > +#include <asm/dma-mapping.h> > > > #include <linux/dma-mapping.h> > > > > > > +static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) > > > +{ > > > + if (dev && dev->archdata.dev_dma_ops) > > > + return dev->archdata.dev_dma_ops; > > > + return get_arch_dma_ops(NULL); > > > +} > > > + > > > void __xen_dma_map_page(struct device *hwdev, struct page *page, > > > dma_addr_t dev_addr, unsigned long offset, size_t size, > > > enum dma_data_direction dir, unsigned long attrs); >
Hi Stefano, On 24/04/17 20:16, Stefano Stabellini wrote: > Given the outstanding regression we need to fix as soon as possible, > I'll queue these patches on the xentip tree for 4.12. It looks like there is another rc for 4.11. I am wondering whether you could try to send a pull request to Linus so it can be fixed in 4.11? Cheers,
On Tue, 25 Apr 2017, Julien Grall wrote: > Hi Stefano, > > On 24/04/17 20:16, Stefano Stabellini wrote: > > Given the outstanding regression we need to fix as soon as possible, > > I'll queue these patches on the xentip tree for 4.12. > > It looks like there is another rc for 4.11. I am wondering whether you could > try to send a pull request to Linus so it can be fixed in 4.11? No, especially without input from Russell.
On Tue, Apr 25, 2017 at 10:22:00AM -0700, Stefano Stabellini wrote: > On Tue, 25 Apr 2017, Julien Grall wrote: > > On 24/04/17 20:16, Stefano Stabellini wrote: > > > Given the outstanding regression we need to fix as soon as possible, > > > I'll queue these patches on the xentip tree for 4.12. > > > > It looks like there is another rc for 4.11. I am wondering whether you could > > try to send a pull request to Linus so it can be fixed in 4.11? > > No, especially without input from Russell. You could add a "Fixes" tag to the first patch so that it gets back-ported to 4.11.x once it hits mainline (the second patch is just cosmetic): Fixes: 815dd18788fe ("treewide: Consolidate get_dma_ops() implementations")
On Wed, 26 Apr 2017, Catalin Marinas wrote: > On Tue, Apr 25, 2017 at 10:22:00AM -0700, Stefano Stabellini wrote: > > On Tue, 25 Apr 2017, Julien Grall wrote: > > > On 24/04/17 20:16, Stefano Stabellini wrote: > > > > Given the outstanding regression we need to fix as soon as possible, > > > > I'll queue these patches on the xentip tree for 4.12. > > > > > > It looks like there is another rc for 4.11. I am wondering whether you could > > > try to send a pull request to Linus so it can be fixed in 4.11? > > > > No, especially without input from Russell. > > You could add a "Fixes" tag to the first patch so that it gets > back-ported to 4.11.x once it hits mainline (the second patch is just > cosmetic): > > Fixes: 815dd18788fe ("treewide: Consolidate get_dma_ops() implementations") I did add: Cc: <stable@vger.kernel.org> [4.11+] I thought that would be enough to trigger a backport?
On Wed, Apr 26, 2017 at 10:00:30AM -0700, Stefano Stabellini wrote: > On Wed, 26 Apr 2017, Catalin Marinas wrote: > > On Tue, Apr 25, 2017 at 10:22:00AM -0700, Stefano Stabellini wrote: > > > On Tue, 25 Apr 2017, Julien Grall wrote: > > > > On 24/04/17 20:16, Stefano Stabellini wrote: > > > > > Given the outstanding regression we need to fix as soon as possible, > > > > > I'll queue these patches on the xentip tree for 4.12. > > > > > > > > It looks like there is another rc for 4.11. I am wondering whether you could > > > > try to send a pull request to Linus so it can be fixed in 4.11? > > > > > > No, especially without input from Russell. > > > > You could add a "Fixes" tag to the first patch so that it gets > > back-ported to 4.11.x once it hits mainline (the second patch is just > > cosmetic): > > > > Fixes: 815dd18788fe ("treewide: Consolidate get_dma_ops() implementations") > > I did add: > > Cc: <stable@vger.kernel.org> [4.11+] > > I thought that would be enough to trigger a backport? Either would do. I didn't see it in the patch you posted, that's why I mentioned it.
On Wed, 26 Apr 2017, Catalin Marinas wrote: > On Wed, Apr 26, 2017 at 10:00:30AM -0700, Stefano Stabellini wrote: > > On Wed, 26 Apr 2017, Catalin Marinas wrote: > > > On Tue, Apr 25, 2017 at 10:22:00AM -0700, Stefano Stabellini wrote: > > > > On Tue, 25 Apr 2017, Julien Grall wrote: > > > > > On 24/04/17 20:16, Stefano Stabellini wrote: > > > > > > Given the outstanding regression we need to fix as soon as possible, > > > > > > I'll queue these patches on the xentip tree for 4.12. > > > > > > > > > > It looks like there is another rc for 4.11. I am wondering whether you could > > > > > try to send a pull request to Linus so it can be fixed in 4.11? > > > > > > > > No, especially without input from Russell. > > > > > > You could add a "Fixes" tag to the first patch so that it gets > > > back-ported to 4.11.x once it hits mainline (the second patch is just > > > cosmetic): > > > > > > Fixes: 815dd18788fe ("treewide: Consolidate get_dma_ops() implementations") > > > > I did add: > > > > Cc: <stable@vger.kernel.org> [4.11+] > > > > I thought that would be enough to trigger a backport? > > Either would do. I didn't see it in the patch you posted, that's why I > mentioned it. Thanks, I appreciated it!
diff --git a/arch/arm/include/asm/device.h b/arch/arm/include/asm/device.h index 220ba20..36ec9c8 100644 --- a/arch/arm/include/asm/device.h +++ b/arch/arm/include/asm/device.h @@ -16,6 +16,9 @@ struct dev_archdata { #ifdef CONFIG_ARM_DMA_USE_IOMMU struct dma_iommu_mapping *mapping; #endif +#ifdef CONFIG_XEN + const struct dma_map_ops *dev_dma_ops; +#endif bool dma_coherent; }; diff --git a/arch/arm/include/asm/dma-mapping.h b/arch/arm/include/asm/dma-mapping.h index 7166569..680d3f3 100644 --- a/arch/arm/include/asm/dma-mapping.h +++ b/arch/arm/include/asm/dma-mapping.h @@ -16,19 +16,9 @@ extern const struct dma_map_ops arm_dma_ops; extern const struct dma_map_ops arm_coherent_dma_ops; -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) -{ - if (dev && dev->dma_ops) - return dev->dma_ops; - return &arm_dma_ops; -} - static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) { - if (xen_initial_domain()) - return xen_dma_ops; - else - return __generic_dma_ops(NULL); + return &arm_dma_ops; } #define HAVE_ARCH_DMA_SUPPORTED 1 diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c index 63eabb0..a614459 100644 --- a/arch/arm/mm/dma-mapping.c +++ b/arch/arm/mm/dma-mapping.c @@ -2396,6 +2396,13 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, dma_ops = arm_get_dma_map_ops(coherent); set_dma_ops(dev, dma_ops); + +#ifdef CONFIG_XEN + if (xen_initial_domain()) { + dev->archdata.dev_dma_ops = dev->dma_ops; + dev->dma_ops = xen_dma_ops; + } +#endif } void arch_teardown_dma_ops(struct device *dev) diff --git a/arch/arm64/include/asm/device.h b/arch/arm64/include/asm/device.h index 73d5bab..5a5fa47 100644 --- a/arch/arm64/include/asm/device.h +++ b/arch/arm64/include/asm/device.h @@ -20,6 +20,9 @@ struct dev_archdata { #ifdef CONFIG_IOMMU_API void *iommu; /* private IOMMU data */ #endif +#ifdef CONFIG_XEN + const struct dma_map_ops *dev_dma_ops; +#endif bool dma_coherent; }; diff --git a/arch/arm64/include/asm/dma-mapping.h b/arch/arm64/include/asm/dma-mapping.h index 505756c..5392dbe 100644 --- a/arch/arm64/include/asm/dma-mapping.h +++ b/arch/arm64/include/asm/dma-mapping.h @@ -27,11 +27,8 @@ #define DMA_ERROR_CODE (~(dma_addr_t)0) extern const struct dma_map_ops dummy_dma_ops; -static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) +static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) { - if (dev && dev->dma_ops) - return dev->dma_ops; - /* * We expect no ISA devices, and all other DMA masters are expected to * have someone call arch_setup_dma_ops at device creation time. @@ -39,14 +36,6 @@ static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) return &dummy_dma_ops; } -static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus) -{ - if (xen_initial_domain()) - return xen_dma_ops; - else - return __generic_dma_ops(NULL); -} - void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, const struct iommu_ops *iommu, bool coherent); #define arch_setup_dma_ops arch_setup_dma_ops diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c index 81cdb2e..7f8b37e 100644 --- a/arch/arm64/mm/dma-mapping.c +++ b/arch/arm64/mm/dma-mapping.c @@ -977,4 +977,11 @@ void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, dev->archdata.dma_coherent = coherent; __iommu_setup_dma_ops(dev, dma_base, size, iommu); + +#ifdef CONFIG_XEN + if (xen_initial_domain()) { + dev->archdata.dev_dma_ops = dev->dma_ops; + dev->dma_ops = xen_dma_ops; + } +#endif } diff --git a/include/xen/arm/page-coherent.h b/include/xen/arm/page-coherent.h index 95ce6ac..b0a2bfc 100644 --- a/include/xen/arm/page-coherent.h +++ b/include/xen/arm/page-coherent.h @@ -2,8 +2,16 @@ #define _ASM_ARM_XEN_PAGE_COHERENT_H #include <asm/page.h> +#include <asm/dma-mapping.h> #include <linux/dma-mapping.h> +static inline const struct dma_map_ops *__generic_dma_ops(struct device *dev) +{ + if (dev && dev->archdata.dev_dma_ops) + return dev->archdata.dev_dma_ops; + return get_arch_dma_ops(NULL); +} + void __xen_dma_map_page(struct device *hwdev, struct page *page, dma_addr_t dev_addr, unsigned long offset, size_t size, enum dma_data_direction dir, unsigned long attrs);
The following commit: commit 815dd18788fe0d41899f51b91d0560279cf16b0d Author: Bart Van Assche <bart.vanassche@sandisk.com> Date: Fri Jan 20 13:04:04 2017 -0800 treewide: Consolidate get_dma_ops() implementations rearranges get_dma_ops in a way that xen_dma_ops are not returned when running on Xen anymore, dev->dma_ops is returned instead (see arch/arm/include/asm/dma-mapping.h:get_arch_dma_ops and include/linux/dma-mapping.h:get_dma_ops). Fix the problem by storing dev->dma_ops in dev_archdata, and setting dev->dma_ops to xen_dma_ops. This way, xen_dma_ops is returned naturally by get_dma_ops. The Xen code can retrieve the original dev->dma_ops from dev_archdata when needed. It also allows us to remove __generic_dma_ops from common headers. Signed-off-by: Stefano Stabellini <sstabellini@kernel.org> Tested-by: Julien Grall <julien.grall@arm.com> Suggested-by: Catalin Marinas <catalin.marinas@arm.com> CC: linux@armlinux.org.uk CC: catalin.marinas@arm.com CC: will.deacon@arm.com CC: boris.ostrovsky@oracle.com CC: jgross@suse.com CC: Julien Grall <julien.grall@arm.com> --- arch/arm/include/asm/device.h | 3 +++ arch/arm/include/asm/dma-mapping.h | 12 +----------- arch/arm/mm/dma-mapping.c | 7 +++++++ arch/arm64/include/asm/device.h | 3 +++ arch/arm64/include/asm/dma-mapping.h | 13 +------------ arch/arm64/mm/dma-mapping.c | 7 +++++++ include/xen/arm/page-coherent.h | 8 ++++++++ 7 files changed, 30 insertions(+), 23 deletions(-)