Hello, On Monday, September 26, 2011 3:30 PM Mauro Carvalho Chehab wrote: > Em 26-09-2011 08:21, Marek Szyprowski escreveu: > > Hello, > > > > On Monday, September 26, 2011 1:14 PM Mauro Carvalho Chehab wrote: > > > >>> Scott Jiang (1): > >>> vb2: add vb2_get_unmapped_area in vb2 core > >> > >>> diff --git a/include/media/videobuf2-core.h b/include/media/videobuf2-core.h > >>> index ea55c08..977410b 100644 > >>> --- a/include/media/videobuf2-core.h > >>> +++ b/include/media/videobuf2-core.h > >>> @@ -309,6 +309,13 @@ int vb2_streamon(struct vb2_queue *q, enum v4l2_buf_type type); > >>> int vb2_streamoff(struct vb2_queue *q, enum v4l2_buf_type type); > >>> > >>> int vb2_mmap(struct vb2_queue *q, struct vm_area_struct *vma); > >>> +#ifndef CONFIG_MMU > >>> +unsigned long vb2_get_unmapped_area(struct vb2_queue *q, > >>> + unsigned long addr, > >>> + unsigned long len, > >>> + unsigned long pgoff, > >>> + unsigned long flags); > >>> +#endif > >>> unsigned int vb2_poll(struct vb2_queue *q, struct file *file, poll_table *wait); > >>> size_t vb2_read(struct vb2_queue *q, char __user *data, size_t count, > >>> loff_t *ppos, int nonblock); > >> > >> This sounds me like a hack, as it is passing the problem of working with a non-mmu > >> capable hardware to the driver, inserting architecture-dependent bits on them. > >> > >> The proper way to do it is to provide a vb2 core support to handle the non-mmu case > >> inside it. > > > > This is exactly what this patch does - it provides generic vb2 implementation for > > fops->get_unmapped_area callback which any vb2 ready driver can use. This operation > > is used only on NON-MMU systems. Please check drivers/media/video/v4l2-dev.c file and > > the implementation of get_unmapped_area there. Similar code is used by uvc driver. > > At least there, there is a: > #ifdef CONFIG_MMU > #define v4l2_get_unmapped_area NULL > #else > ... > #endif > > block, so, in thesis, a driver can be written to support both cases without inserting > #ifdefs inside it. videobuf2 functions are not meant to be used as direct callbacks in fops so defining it as NULL makes no sense at all. > Ideally, I would prefer if all those iommu-specific calls would be inside the core. > A driver should not need to do anything special in order to support a different > (sub)architecture. It is not about IOMMU at all. Some (embedded) systems don't have MMU at all. Drivers on such systems cannot do regular mmap operation. Instead it is emulated with get_unmapped_area fops callback and some (u)libC magic. This patch just provides vb2_get_unmapped_area function. Drivers have to call it from their respective my_driver_get_unmapped_area() function provided in its fops. Implementing it makes sense only on NO-MMU systems. I really see no other way of dealing with this. Best regards -- Marek Szyprowski Samsung Poland R&D Center -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
