Re: dma coherent memory user-space maps
Sorry, but travel for meeting. I'll try to get to it as quick as I can.
Re: dma coherent memory user-space maps
On 10/8/19 2:34 PM, Thomas Hellström (VMware) wrote: Hi, Christoph, Following our previous discussion I wonder if something along the lines of the following could work / be acceptible typedef unsigned long dma_pfn_t /* Opaque pfn type. Arch dependent. This could if needed be a struct with a pointer and an offset */ /* Similar to vmf_insert_mixed() */ vm_fault_t dma_vmf_insert_mixed(struct device *dev, struct vm_area_struct *vma, unsigned long addr, dma_pfn_t dpfn, unsigned long attrs); /* Similar to vmf_insert_pfn_pmd() */ vm_fault_t dma_vmf_insert_pfn_pmd(struct device *dev, struct vm_area_struct *vma, unsigned long addr, dma_pfn_t dpfn, unsigned long attrs); /* Like vmap, but takes struct dma_pfns. */ extern void *dma_vmap(struct device *dev, dma_pfn_t dpfns[], unsigned int count, unsigned long flags, unsigned long attrs); /* Obtain struct dma_pfn pointers from a dma coherent allocation */ int dma_get_dpfns(struct device *dev, void *cpu_addr, dma_addr_t dma_addr, pgoff_t offset, pgoff_t num, dma_pfn_t dpfns[]); I figure, for most if not all architectures we could use an ordinary pfn as dma_pfn_t, but the dma layer would still have control over how those pfns are obtained and how they are used in the kernel's mapping APIs. If so, I could start looking at this, time permitting, for the cases where the pfn can be obtained from the kernel address or from arch_dma_coherent_to_pfn(), and also the needed work to have a tailored vmap_pfn(). Thanks, /Thomas Ping? Thanks, Thomas
dma coherent memory user-space maps
Hi, Christoph, Following our previous discussion I wonder if something along the lines of the following could work / be acceptible typedef unsigned long dma_pfn_t /* Opaque pfn type. Arch dependent. This could if needed be a struct with a pointer and an offset */ /* Similar to vmf_insert_mixed() */ vm_fault_t dma_vmf_insert_mixed(struct device *dev, struct vm_area_struct *vma, unsigned long addr, dma_pfn_t dpfn, unsigned long attrs); /* Similar to vmf_insert_pfn_pmd() */ vm_fault_t dma_vmf_insert_pfn_pmd(struct device *dev, struct vm_area_struct *vma, unsigned long addr, dma_pfn_t dpfn, unsigned long attrs); /* Like vmap, but takes struct dma_pfns. */ extern void *dma_vmap(struct device *dev, dma_pfn_t dpfns[], unsigned int count, unsigned long flags, unsigned long attrs); /* Obtain struct dma_pfn pointers from a dma coherent allocation */ int dma_get_dpfns(struct device *dev, void *cpu_addr, dma_addr_t dma_addr, pgoff_t offset, pgoff_t num, dma_pfn_t dpfns[]); I figure, for most if not all architectures we could use an ordinary pfn as dma_pfn_t, but the dma layer would still have control over how those pfns are obtained and how they are used in the kernel's mapping APIs. If so, I could start looking at this, time permitting, for the cases where the pfn can be obtained from the kernel address or from arch_dma_coherent_to_pfn(), and also the needed work to have a tailored vmap_pfn(). Thanks, /Thomas