with this patch, we can allocate physically continuous or non-continuous memory and also it creates scatterlist for iommu support so allocated memory region can be mapped to iommu page table using scatterlist.
Signed-off-by: Inki Dae inki.dae@samsung.com Signed-off-by: Kyungmin Park kyungmin.park@samsung.com --- drivers/gpu/drm/exynos/exynos_drm_buf.c | 148 +++++++++++++-- drivers/gpu/drm/exynos/exynos_drm_buf.h | 4 +- drivers/gpu/drm/exynos/exynos_drm_fbdev.c | 4 +- drivers/gpu/drm/exynos/exynos_drm_gem.c | 284 ++++++++++++++++++++++++++--- drivers/gpu/drm/exynos/exynos_drm_gem.h | 11 +- include/drm/exynos_drm.h | 6 + 6 files changed, 407 insertions(+), 50 deletions(-)
diff --git a/drivers/gpu/drm/exynos/exynos_drm_buf.c b/drivers/gpu/drm/exynos/exynos_drm_buf.c index 3cf785c..554f674 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_buf.c +++ b/drivers/gpu/drm/exynos/exynos_drm_buf.c @@ -25,45 +25,151 @@
#include "drmP.h" #include "drm.h" +#include "exynos_drm.h"
#include "exynos_drm_drv.h" #include "exynos_drm_gem.h" #include "exynos_drm_buf.h"
static int lowlevel_buffer_allocate(struct drm_device *dev, - struct exynos_drm_gem_buf *buffer) + unsigned int flags, struct exynos_drm_gem_buf *buf) { + int ret = 0; + DRM_DEBUG_KMS("%s\n", __FILE__);
- buffer->kvaddr = dma_alloc_writecombine(dev->dev, buffer->size, - &buffer->dma_addr, GFP_KERNEL); - if (!buffer->kvaddr) { - DRM_ERROR("failed to allocate buffer.\n"); - return -ENOMEM; + /* + * allocate only physically continuous memory and + * non-continuous memory would be allocated by exynos + * gem framework. + */ + if (!(flags & EXYNOS_BO_NONCONTIG)) { + dma_addr_t start_addr, end_addr; + unsigned int npages, page_size, i = 0; + struct scatterlist *sgl; + + if (buf->dma_addr) { + DRM_DEBUG_KMS("already allocated.\n"); + return -EINVAL; + } + + /* + * according to desired size, it sets page size + * for performance with using iommu so physically + * continuous memory could be mapped to iommu in + * multiple page size. + */ + if (buf->size >= SZ_1M) { + npages = (buf->size >> SECTION_SHIFT) + 1; + page_size = SECTION_SIZE; + } else if (buf->size >= SZ_64K) { + npages = (buf->size >> 16) + 1; + page_size = SZ_64K; + } else { + npages = (buf->size >> PAGE_SHIFT) + 1; + page_size = PAGE_SIZE; + } + + buf->sgt = kzalloc(sizeof(struct sg_table), GFP_KERNEL); + if (!buf->sgt) { + DRM_ERROR("failed to allocate sg table.\n"); + return -ENOMEM; + } + + ret = sg_alloc_table(buf->sgt, npages, GFP_KERNEL); + if (ret < 0) { + DRM_ERROR("failed to initialize sg table.\n"); + kfree(buf->sgt); + buf->sgt = NULL; + return -ENOMEM; + } + + buf->kvaddr = dma_alloc_writecombine(dev->dev, buf->size, + &buf->dma_addr, GFP_KERNEL); + if (!buf->kvaddr) { + DRM_ERROR("failed to allocate buffer.\n"); + ret = -ENOMEM; + goto err1; + } + + start_addr = buf->dma_addr; + end_addr = buf->dma_addr + buf->size; + + buf->pages = kzalloc(sizeof(struct page) * npages, GFP_KERNEL); + if (!buf->pages) { + DRM_ERROR("failed to allocate pages.\n"); + ret = -ENOMEM; + goto err2; + } + + sgl = buf->sgt->sgl; + + while (i < npages) { + buf->pages[i] = phys_to_page(start_addr); + sg_set_page(sgl, buf->pages[i], page_size, 0); + sg_dma_address(sgl) = start_addr; + start_addr += page_size; + if (end_addr - start_addr < page_size) + break; + sgl = sg_next(sgl); + i++; + } + + buf->pages[i] = phys_to_page(start_addr); + + sgl = sg_next(sgl); + sg_set_page(sgl, buf->pages[i+1], end_addr - start_addr, 0); }
DRM_DEBUG_KMS("vaddr(0x%lx), dma_addr(0x%lx), size(0x%lx)\n", - (unsigned long)buffer->kvaddr, - (unsigned long)buffer->dma_addr, - buffer->size); - - return 0; + (unsigned long)buf->kvaddr, + (unsigned long)buf->dma_addr, + buf->size); + + return ret; +err2: + dma_free_writecombine(dev->dev, buf->size, buf->kvaddr, + (dma_addr_t)buf->dma_addr); + buf->dma_addr = (dma_addr_t)NULL; +err1: + sg_free_table(buf->sgt); + kfree(buf->sgt); + buf->sgt = NULL; + + return ret; }
static void lowlevel_buffer_deallocate(struct drm_device *dev, - struct exynos_drm_gem_buf *buffer) + unsigned int flags, struct exynos_drm_gem_buf *buf) { DRM_DEBUG_KMS("%s.\n", __FILE__);
- if (buffer->dma_addr && buffer->size) - dma_free_writecombine(dev->dev, buffer->size, buffer->kvaddr, - (dma_addr_t)buffer->dma_addr); - else - DRM_DEBUG_KMS("buffer data are invalid.\n"); + /* + * release only physically continuous memory and + * non-continuous memory would be released by exynos + * gem framework. + */ + if (!(flags & EXYNOS_BO_NONCONTIG)) { + sg_free_table(buf->sgt); + + kfree(buf->sgt); + buf->sgt = NULL; + + kfree(buf->pages); + buf->pages = NULL; + + if (buf->dma_addr && buf->size) + dma_free_writecombine(dev->dev, buf->size, buf->kvaddr, + (dma_addr_t)buf->dma_addr); + else + DRM_DEBUG_KMS("buffer data are invalid.\n"); + + buf->dma_addr = (dma_addr_t)NULL; + } }
struct exynos_drm_gem_buf *exynos_drm_buf_create(struct drm_device *dev, - unsigned int size) + unsigned int flags, unsigned int size) { struct exynos_drm_gem_buf *buffer;
@@ -82,7 +188,7 @@ struct exynos_drm_gem_buf *exynos_drm_buf_create(struct drm_device *dev, * allocate memory region with size and set the memory information * to vaddr and dma_addr of a buffer object. */ - if (lowlevel_buffer_allocate(dev, buffer) < 0) { + if (lowlevel_buffer_allocate(dev, flags, buffer) < 0) { kfree(buffer); return NULL; } @@ -91,7 +197,7 @@ struct exynos_drm_gem_buf *exynos_drm_buf_create(struct drm_device *dev, }
void exynos_drm_buf_destroy(struct drm_device *dev, - struct exynos_drm_gem_buf *buffer) + unsigned int flags, struct exynos_drm_gem_buf *buffer) { DRM_DEBUG_KMS("%s.\n", __FILE__);
@@ -100,7 +206,7 @@ void exynos_drm_buf_destroy(struct drm_device *dev, return; }
- lowlevel_buffer_deallocate(dev, buffer); + lowlevel_buffer_deallocate(dev, flags, buffer);
kfree(buffer); buffer = NULL; diff --git a/drivers/gpu/drm/exynos/exynos_drm_buf.h b/drivers/gpu/drm/exynos/exynos_drm_buf.h index c913f2b..3422267 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_buf.h +++ b/drivers/gpu/drm/exynos/exynos_drm_buf.h @@ -28,10 +28,10 @@
/* allocate physical memory. */ struct exynos_drm_gem_buf *exynos_drm_buf_create(struct drm_device *dev, - unsigned int size); + unsigned int flags, unsigned int size);
/* remove allocated physical memory. */ void exynos_drm_buf_destroy(struct drm_device *dev, - struct exynos_drm_gem_buf *buffer); + unsigned int flags, struct exynos_drm_gem_buf *buffer);
#endif diff --git a/drivers/gpu/drm/exynos/exynos_drm_fbdev.c b/drivers/gpu/drm/exynos/exynos_drm_fbdev.c index 3508700..4e86d15 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fbdev.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fbdev.c @@ -151,7 +151,9 @@ static int exynos_drm_fbdev_create(struct drm_fb_helper *helper, }
size = mode_cmd.pitches[0] * mode_cmd.height; - exynos_gem_obj = exynos_drm_gem_create(dev, size); + + /* 0 means to allocate physically continuous memory */ + exynos_gem_obj = exynos_drm_gem_create(dev, 0, size); if (IS_ERR(exynos_gem_obj)) { ret = PTR_ERR(exynos_gem_obj); goto out; diff --git a/drivers/gpu/drm/exynos/exynos_drm_gem.c b/drivers/gpu/drm/exynos/exynos_drm_gem.c index 025abb3..36d081d 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_gem.c +++ b/drivers/gpu/drm/exynos/exynos_drm_gem.c @@ -26,6 +26,7 @@ #include "drmP.h" #include "drm.h"
+#include <linux/shmem_fs.h> #include <drm/exynos_drm.h>
#include "exynos_drm_drv.h" @@ -55,6 +56,181 @@ static unsigned int convert_to_vm_err_msg(int msg) return out_msg; }
+static unsigned int mask_gem_flags(unsigned int flags) +{ + return flags &= EXYNOS_BO_NONCONTIG; +} + +static struct page **exynos_gem_get_pages(struct drm_gem_object *obj, + gfp_t gfpmask) +{ + struct inode *inode; + struct address_space *mapping; + struct page *p, **pages; + int i, npages; + + /* This is the shared memory object that backs the GEM resource */ + inode = obj->filp->f_path.dentry->d_inode; + mapping = inode->i_mapping; + + npages = obj->size >> PAGE_SHIFT; + + pages = drm_malloc_ab(npages, sizeof(struct page *)); + if (pages == NULL) + return ERR_PTR(-ENOMEM); + + gfpmask |= mapping_gfp_mask(mapping); + + for (i = 0; i < npages; i++) { + p = shmem_read_mapping_page_gfp(mapping, i, gfpmask); + if (IS_ERR(p)) + goto fail; + pages[i] = p; + } + + return pages; + +fail: + while (i--) + page_cache_release(pages[i]); + + drm_free_large(pages); + return ERR_PTR(PTR_ERR(p)); +} + +static void exynos_gem_put_pages(struct drm_gem_object *obj, + struct page **pages, + bool dirty, bool accessed) +{ + int i, npages; + + npages = obj->size >> PAGE_SHIFT; + + for (i = 0; i < npages; i++) { + if (dirty) + set_page_dirty(pages[i]); + + if (accessed) + mark_page_accessed(pages[i]); + + /* Undo the reference we took when populating the table */ + page_cache_release(pages[i]); + } + + drm_free_large(pages); +} + +static int exynos_drm_gem_map_pages(struct drm_gem_object *obj, + struct vm_area_struct *vma, + unsigned long f_vaddr, + pgoff_t page_offset) +{ + struct exynos_drm_gem_obj *exynos_gem_obj = to_exynos_gem_obj(obj); + struct exynos_drm_gem_buf *buf = exynos_gem_obj->buffer; + unsigned long pfn; + + if (exynos_gem_obj->flags & EXYNOS_BO_NONCONTIG) { + unsigned long usize = buf->size; + + if (!buf->pages) + return -EINTR; + + while (usize > 0) { + pfn = page_to_pfn(buf->pages[page_offset++]); + vm_insert_mixed(vma, f_vaddr, pfn); + f_vaddr += PAGE_SIZE; + usize -= PAGE_SIZE; + } + + return 0; + } + + pfn = (buf->dma_addr >> PAGE_SHIFT) + page_offset; + + return vm_insert_mixed(vma, f_vaddr, pfn); +} + +static int exynos_drm_gem_get_pages(struct drm_gem_object *obj) +{ + struct exynos_drm_gem_obj *exynos_gem_obj = to_exynos_gem_obj(obj); + struct exynos_drm_gem_buf *buf = exynos_gem_obj->buffer; + struct scatterlist *sgl; + struct page **pages; + unsigned int npages, i = 0; + int ret; + + if (buf->pages) { + DRM_DEBUG_KMS("already allocated.\n"); + return -EINVAL; + } + + pages = exynos_gem_get_pages(obj, GFP_KERNEL); + if (IS_ERR(pages)) { + DRM_ERROR("failed to get pages.\n"); + return PTR_ERR(pages); + } + + npages = obj->size >> PAGE_SHIFT; + + buf->sgt = kzalloc(sizeof(struct sg_table), GFP_KERNEL); + if (!buf->sgt) { + DRM_ERROR("failed to allocate sg table.\n"); + ret = -ENOMEM; + goto err; + } + + ret = sg_alloc_table(buf->sgt, npages, GFP_KERNEL); + if (ret < 0) { + DRM_ERROR("failed to initialize sg table.\n"); + ret = -EFAULT; + goto err1; + } + + sgl = buf->sgt->sgl; + + /* set all pages to sg list. */ + while (i < npages) { + sg_set_page(sgl, pages[i], PAGE_SIZE, 0); + sg_dma_address(sgl) = page_to_phys(pages[i]); + i++; + sgl = sg_next(sgl); + } + + /* add some codes for UNCACHED type here. TODO */ + + buf->pages = pages; + return ret; +err1: + kfree(buf->sgt); + buf->sgt = NULL; +err: + exynos_gem_put_pages(obj, pages, true, false); + return ret; + +} + +static void exynos_drm_gem_put_pages(struct drm_gem_object *obj) +{ + struct exynos_drm_gem_obj *exynos_gem_obj = to_exynos_gem_obj(obj); + struct exynos_drm_gem_buf *buf = exynos_gem_obj->buffer; + + /* + * if buffer typs is EXYNOS_BO_NONCONTIG then release all pages + * allocated at gem fault handler. + */ + if ((exynos_gem_obj->flags & EXYNOS_BO_NONCONTIG) && + buf->pages) { + sg_free_table(buf->sgt); + kfree(buf->sgt); + buf->sgt = NULL; + + exynos_gem_put_pages(obj, buf->pages, true, false); + buf->pages = NULL; + } + + /* add some codes for UNCACHED type here. TODO */ +} + static int exynos_drm_gem_handle_create(struct drm_gem_object *obj, struct drm_file *file_priv, unsigned int *handle) @@ -90,7 +266,10 @@ void exynos_drm_gem_destroy(struct exynos_drm_gem_obj *exynos_gem_obj)
DRM_DEBUG_KMS("handle count = %d\n", atomic_read(&obj->handle_count));
- exynos_drm_buf_destroy(obj->dev, exynos_gem_obj->buffer); + exynos_drm_gem_put_pages(obj); + + exynos_drm_buf_destroy(obj->dev, exynos_gem_obj->flags, + exynos_gem_obj->buffer);
if (obj->map_list.map) drm_gem_free_mmap_offset(obj); @@ -114,6 +293,7 @@ static struct exynos_drm_gem_obj *exynos_drm_gem_init(struct drm_device *dev, return NULL; }
+ exynos_gem_obj->size = size; obj = &exynos_gem_obj->base;
ret = drm_gem_object_init(dev, obj, size); @@ -129,27 +309,49 @@ static struct exynos_drm_gem_obj *exynos_drm_gem_init(struct drm_device *dev, }
struct exynos_drm_gem_obj *exynos_drm_gem_create(struct drm_device *dev, - unsigned long size) + unsigned int flags, + unsigned long size) { struct exynos_drm_gem_buf *buffer; struct exynos_drm_gem_obj *exynos_gem_obj; + int ret;
size = roundup(size, PAGE_SIZE); DRM_DEBUG_KMS("%s: size = 0x%lx\n", __FILE__, size);
- buffer = exynos_drm_buf_create(dev, size); + buffer = exynos_drm_buf_create(dev, flags, size); if (!buffer) return ERR_PTR(-ENOMEM);
exynos_gem_obj = exynos_drm_gem_init(dev, size); if (!exynos_gem_obj) { - exynos_drm_buf_destroy(dev, buffer); + exynos_drm_buf_destroy(dev, flags, buffer); return ERR_PTR(-ENOMEM); }
exynos_gem_obj->buffer = buffer;
+ /* + * allocate all pages as desired size if user wants to allocate + * physically non-continuous memory. + */ + if (flags & EXYNOS_BO_NONCONTIG) { + ret = exynos_drm_gem_get_pages(&exynos_gem_obj->base); + if (ret < 0) + goto err; + } + + /* set memory type and cache attribute from user side. */ + exynos_gem_obj->flags = mask_gem_flags(flags); + + return exynos_gem_obj; + +err: + exynos_drm_buf_destroy(dev, flags, buffer); + drm_gem_object_release(&exynos_gem_obj->base); + kfree(exynos_gem_obj); + return ERR_PTR(ret); }
int exynos_drm_gem_create_ioctl(struct drm_device *dev, void *data, @@ -161,7 +363,7 @@ int exynos_drm_gem_create_ioctl(struct drm_device *dev, void *data,
DRM_DEBUG_KMS("%s\n", __FILE__);
- exynos_gem_obj = exynos_drm_gem_create(dev, args->size); + exynos_gem_obj = exynos_drm_gem_create(dev, args->flags, args->size); if (IS_ERR(exynos_gem_obj)) return PTR_ERR(exynos_gem_obj);
@@ -200,7 +402,8 @@ static int exynos_drm_gem_mmap_buffer(struct file *filp, struct drm_gem_object *obj = filp->private_data; struct exynos_drm_gem_obj *exynos_gem_obj = to_exynos_gem_obj(obj); struct exynos_drm_gem_buf *buffer; - unsigned long pfn, vm_size; + unsigned long pfn, vm_size, usize, uaddr = vma->vm_start; + int ret;
DRM_DEBUG_KMS("%s\n", __FILE__);
@@ -210,7 +413,8 @@ static int exynos_drm_gem_mmap_buffer(struct file *filp, vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); vma->vm_file = filp;
- vm_size = vma->vm_end - vma->vm_start; + vm_size = usize = vma->vm_end - vma->vm_start; + /* * a buffer contains information to physically continuous memory * allocated by user request or at framebuffer creation. @@ -221,18 +425,37 @@ static int exynos_drm_gem_mmap_buffer(struct file *filp, if (vm_size > buffer->size) return -EINVAL;
- /* - * get page frame number to physical memory to be mapped - * to user space. - */ - pfn = ((unsigned long)exynos_gem_obj->buffer->dma_addr) >> PAGE_SHIFT; - - DRM_DEBUG_KMS("pfn = 0x%lx\n", pfn); - - if (remap_pfn_range(vma, vma->vm_start, pfn, vm_size, - vma->vm_page_prot)) { - DRM_ERROR("failed to remap pfn range.\n"); - return -EAGAIN; + if (exynos_gem_obj->flags & EXYNOS_BO_NONCONTIG) { + int i = 0; + + if (!buffer->pages) + return -EINVAL; + + do { + ret = vm_insert_page(vma, uaddr, buffer->pages[i++]); + if (ret) { + DRM_ERROR("failed to remap user space.\n"); + return ret; + } + + uaddr += PAGE_SIZE; + usize -= PAGE_SIZE; + } while (usize > 0); + } else { + /* + * get page frame number to physical memory to be mapped + * to user space. + */ + pfn = ((unsigned long)exynos_gem_obj->buffer->dma_addr) >> + PAGE_SHIFT; + + DRM_DEBUG_KMS("pfn = 0x%lx\n", pfn); + + if (remap_pfn_range(vma, vma->vm_start, pfn, vm_size, + vma->vm_page_prot)) { + DRM_ERROR("failed to remap pfn range.\n"); + return -EAGAIN; + } }
return 0; @@ -312,9 +535,9 @@ int exynos_drm_gem_dumb_create(struct drm_file *file_priv, */
args->pitch = args->width * args->bpp >> 3; - args->size = args->pitch * args->height; + args->size = PAGE_ALIGN(args->pitch * args->height);
- exynos_gem_obj = exynos_drm_gem_create(dev, args->size); + exynos_gem_obj = exynos_drm_gem_create(dev, args->flags, args->size); if (IS_ERR(exynos_gem_obj)) return PTR_ERR(exynos_gem_obj);
@@ -398,20 +621,31 @@ int exynos_drm_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf) struct drm_gem_object *obj = vma->vm_private_data; struct exynos_drm_gem_obj *exynos_gem_obj = to_exynos_gem_obj(obj); struct drm_device *dev = obj->dev; - unsigned long pfn; + unsigned long f_vaddr; pgoff_t page_offset; int ret;
page_offset = ((unsigned long)vmf->virtual_address - vma->vm_start) >> PAGE_SHIFT; + f_vaddr = (unsigned long)vmf->virtual_address;
mutex_lock(&dev->struct_mutex);
- pfn = (((unsigned long)exynos_gem_obj->buffer->dma_addr) >> - PAGE_SHIFT) + page_offset; + /* + * allocate all pages as desired size if user wants to allocate + * physically non-continuous memory. + */ + if (exynos_gem_obj->flags & EXYNOS_BO_NONCONTIG) { + ret = exynos_drm_gem_get_pages(obj); + if (ret < 0) + goto err; + }
- ret = vm_insert_mixed(vma, (unsigned long)vmf->virtual_address, pfn); + ret = exynos_drm_gem_map_pages(obj, vma, f_vaddr, page_offset); + if (ret < 0) + DRM_ERROR("failed to map pages.\n");
+err: mutex_unlock(&dev->struct_mutex);
return convert_to_vm_err_msg(ret); diff --git a/drivers/gpu/drm/exynos/exynos_drm_gem.h b/drivers/gpu/drm/exynos/exynos_drm_gem.h index 67cdc91..096267d 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_gem.h +++ b/drivers/gpu/drm/exynos/exynos_drm_gem.h @@ -36,11 +36,15 @@ * @dma_addr: bus address(accessed by dma) to allocated memory region. * - this address could be physical address without IOMMU and * device address with IOMMU. + * @sgt: sg table to transfer page data. + * @pages: contain all pages to allocated memory region. * @size: size of allocated memory region. */ struct exynos_drm_gem_buf { void __iomem *kvaddr; dma_addr_t dma_addr; + struct sg_table *sgt; + struct page **pages; unsigned long size; };
@@ -55,6 +59,8 @@ struct exynos_drm_gem_buf { * by user request or at framebuffer creation. * continuous memory region allocated by user request * or at framebuffer creation. + * @size: total memory size to physically non-continuous memory region. + * @flags: indicate memory type to allocated buffer and cache attruibute. * * P.S. this object would be transfered to user as kms_bo.handle so * user can access the buffer through kms_bo.handle. @@ -62,6 +68,8 @@ struct exynos_drm_gem_buf { struct exynos_drm_gem_obj { struct drm_gem_object base; struct exynos_drm_gem_buf *buffer; + unsigned long size; + unsigned int flags; };
/* destroy a buffer with gem object */ @@ -69,7 +77,8 @@ void exynos_drm_gem_destroy(struct exynos_drm_gem_obj *exynos_gem_obj);
/* create a new buffer with gem object */ struct exynos_drm_gem_obj *exynos_drm_gem_create(struct drm_device *dev, - unsigned long size); + unsigned int flags, + unsigned long size);
/* * request gem object creation and buffer allocation as the size diff --git a/include/drm/exynos_drm.h b/include/drm/exynos_drm.h index aff2313..81c9cb7 100644 --- a/include/drm/exynos_drm.h +++ b/include/drm/exynos_drm.h @@ -79,6 +79,12 @@ struct drm_exynos_plane_set_zpos { __s32 zpos; };
+/* memory type definitions. */ +enum e_drm_exynos_gem_mem_type { + /* Physically Non-Continuous memory. */ + EXYNOS_BO_NONCONTIG = 1 << 0 +}; + #define DRM_EXYNOS_GEM_CREATE 0x00 #define DRM_EXYNOS_GEM_MAP_OFFSET 0x01 #define DRM_EXYNOS_GEM_MMAP 0x02