diff options
Diffstat (limited to 'drivers/gpu/drm/emgd/emgd/gmm/gmm.c')
| -rw-r--r-- | drivers/gpu/drm/emgd/emgd/gmm/gmm.c | 806 |
1 files changed, 0 insertions, 806 deletions
diff --git a/drivers/gpu/drm/emgd/emgd/gmm/gmm.c b/drivers/gpu/drm/emgd/emgd/gmm/gmm.c deleted file mode 100644 index 587aa5f0b0fd..000000000000 --- a/drivers/gpu/drm/emgd/emgd/gmm/gmm.c +++ /dev/null @@ -1,806 +0,0 @@ -/* -*- pse-c -*- - *----------------------------------------------------------------------------- - * Filename: gmm.c - * $Revision: 1.44 $ - *----------------------------------------------------------------------------- - * Copyright © 2002-2010, Intel Corporation. - * - * This program is free software; you can redistribute it and/or modify it - * under the terms and conditions of the GNU General Public License, - * version 2, as published by the Free Software Foundation. - * - * This program is distributed in the hope it will be useful, but WITHOUT - * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or - * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for - * more details. - * - * You should have received a copy of the GNU General Public License along with - * this program; if not, write to the Free Software Foundation, Inc., - * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. - * - *----------------------------------------------------------------------------- - * Description: - * Very basic video memory managment functions required by HAL. - *----------------------------------------------------------------------------- - */ - -#define MODULE_NAME hal.gmm - -#include <igd_debug.h> -#include <drmP.h> -#include <memlist.h> -#include <io.h> -#include <memory.h> - -#include <asm/agp.h> - -#define AGP_PHYS_MEMORY 2 /* Physical contigous memory */ - - -gmm_context_t gmm_context; - -gmm_chunk_t *gmm_get_chunk(igd_context_t *context, unsigned long offset); -static int gmm_flush_cache(void); -static int gmm_alloc_linear_surface(unsigned long *offset, - unsigned long pixel_format, - unsigned int *width, - unsigned int *height, - unsigned int *pitch, - unsigned long *size, - unsigned long type, - unsigned long flags, - unsigned long phys); - -static int gmm_alloc_chunk_space(gmm_context_t *gmm_context, - unsigned long *offset, - unsigned long size, - unsigned long phys, - unsigned long flags); - -static int gmm_get_page_list(unsigned long offset, - unsigned long **pages, - unsigned long *page_cnt); - -gmm_mem_buffer_t *emgd_alloc_pages(unsigned long num_pages, int type); -void emgd_free_pages(gmm_mem_buffer_t *mem); -void emgd_gtt_remove(igd_context_t *context, gmm_mem_buffer_t *mem, - unsigned long offset); -void emgd_gtt_insert(igd_context_t *context, gmm_mem_buffer_t *mem, - unsigned long offset); - - -static void gmm_free(unsigned long offset) -{ - gmm_chunk_t *chunk; - - EMGD_DEBUG("Enter gmm_free(0x%lx)", offset); - - /* Walk the chunk list */ - chunk = gmm_context.head_chunk; - while (chunk) { - if (chunk->offset == offset) { - if (chunk->used == 0) { - EMGD_DEBUG("WARNING: The chunk 0x%lx is already freed", offset); - } else { - EMGD_DEBUG("Freeing the chunk 0x%lx", offset); - } - - /* - * What to do if the ref count is > 0? Unmapping is - * probably the right thing since nothing should try - * to use this. If something does, it should probably - * fail. - */ - if (chunk->ref_cnt > 0) { - EMGD_DEBUG("WARNING: The chunk 0x%lx is mapped", offset); - chunk->ref_cnt = 0; - vunmap(chunk->addr); - chunk->addr = NULL; - } - /* Free the array of page address, if applicable: */ - if (chunk->page_addresses != NULL) { - EMGD_DEBUG("About to free chunk->page_addresses = 0x%p", - chunk->page_addresses); - OS_FREE(chunk->page_addresses); - chunk->page_addresses = NULL; - } - - chunk->used = 0; /* mark as free */ - return; - } - chunk = chunk->next; - } - - EMGD_ERROR("gmm_free() did not find the chunk 0x%lx to free", offset); - return; -} - -static int gmm_alloc_region(unsigned long *offset, - unsigned long *size, - unsigned int type, - unsigned long flags) -{ - int ret; - unsigned long aligned_size; - unsigned long phys = 0; - - EMGD_TRACE_ENTER; - EMGD_DEBUG("Parameters: size=%lu, type=%d, flags=0x%lx", *size, type, flags); - - *offset = 0; - - switch(type) { - case IGD_GMM_REGION_TYPE_OVLREG: - flags |= IGD_GMM_REGION_ALIGN_MMAP; - phys = 1; - break; - case IGD_GMM_REGION_TYPE_OVLREG64: - flags |= IGD_GMM_REGION_ALIGN_64K; - phys = 1; - break; - case IGD_GMM_REGION_TYPE_HWSTATUS: - flags |= IGD_GMM_REGION_ALIGN_MMAP; - *size = 4096; - phys = 1; - break; - case IGD_GMM_REGION_TYPE_DMA: - flags |= IGD_GMM_REGION_ALIGN_MMAP; - break; - case IGD_GMM_REGION_TYPE_PERSISTENT: - flags |= IGD_GMM_REGION_ALIGN_MMAP; - break; - case IGD_GMM_REGION_TYPE_BPL: - flags |= IGD_GMM_REGION_ALIGN_MMAP; - phys = 1; - break; - case IGD_GMM_REGION_TYPE_CONTEXT: - flags |= IGD_GMM_REGION_ALIGN_CONTEXT | IGD_GMM_REGION_ALIGN_MMAP; - *size = 4096; - phys = 1; - break; - default: - EMGD_ERROR_EXIT("Invalid Region type requested: 0x%8.8x", type); - return -IGD_INVAL; - } - - aligned_size = (*size + 4095) & ~4095; - EMGD_DEBUG("aligned_size=%lu", aligned_size); - - do { - ret = gmm_alloc_chunk_space(&gmm_context, offset, aligned_size, phys, - flags); - } while ((ret == -IGD_ERROR_NOMEM) && gmm_flush_cache()); - - EMGD_DEBUG("EXIT Returning %d", ret); - return ret; -} - -static int gmm_get_num_surface(unsigned long *count) -{ - gmm_chunk_t *chunk; - - EMGD_TRACE_ENTER; - - /* Walk the chunk list */ - chunk = gmm_context.head_chunk; - *count = 0; - while (chunk) { - (*count)++; - chunk = chunk->next; - } - - EMGD_TRACE_EXIT; - return 0; -} - -static int gmm_get_surface_list(unsigned long allocated_size, - unsigned long *list_size, - igd_surface_list_t **surface_list) -{ - gmm_chunk_t *chunk; - igd_surface_list_t *tmp_list; - - EMGD_TRACE_ENTER; - gmm_get_num_surface(list_size); - - if (*list_size > 0){ - *surface_list = vmalloc(*list_size * sizeof(igd_surface_list_t)); - - /* Walk the chunk list */ - chunk = gmm_context.head_chunk; - tmp_list = *surface_list; - - while (chunk){ - tmp_list->offset = chunk->offset; - tmp_list->size = chunk->size; - - chunk = chunk->next; - tmp_list++; - } - } - EMGD_TRACE_EXIT; - return 0; -} - -static int gmm_alloc_surface(unsigned long *offset, - unsigned long pixel_format, - unsigned int *width, - unsigned int *height, - unsigned int *pitch, - unsigned long *size, - unsigned int type, - unsigned long *flags) -{ - int ret; - unsigned long phys; - - EMGD_TRACE_ENTER; - EMGD_DEBUG("Parameters: pixel_format=0x%08lx,", pixel_format); - EMGD_DEBUG(" width=%u, height=%u", *width, *height); - EMGD_DEBUG(" pitch=%u, type=%d, flags=0x%08lx", *pitch, type, *flags); - - *offset = 0; - *size = 0; - if (! (*flags & IGD_MIN_PITCH)) { - *pitch = 0; - } - - if (*flags & IGD_SURFACE_CURSOR) { - phys = 1; - } else { - phys = 0; - } - - ret = gmm_alloc_linear_surface(offset, pixel_format, width, height, pitch, - size, type, *flags, phys); - - EMGD_DEBUG("EXIT Returning %d", ret); - return ret; -} - - -/* - * Given an offset, find the chunk and return the physical address. - */ -static int gmm_virt_to_phys(unsigned long offset, - unsigned long *physical) -{ - gmm_chunk_t *chunk; - - EMGD_TRACE_ENTER; - EMGD_DEBUG("Looking for offset=0x%lx", offset); - - /* Walk the chunk list */ - chunk = gmm_context.head_chunk; - while (chunk) { - if (chunk->offset == offset) { - *physical = chunk->gtt_mem->physical; - EMGD_DEBUG("Physical address = 0x%08lx", *physical); - EMGD_TRACE_EXIT; - return 0; - } - chunk = chunk->next; - } - - /* offset not found */ - EMGD_ERROR_EXIT("Did not find offset (0x%lx); returning %d", - offset, -IGD_ERROR_NOMEM); - return -IGD_ERROR_NOMEM; -} - - -static int gmm_flush_cache(void) -{ - EMGD_DEBUG("Enter gmm_flush_cache(), which is stubbed"); - return 0; -} - -static void gmm_save(igd_context_t *context, void **state) -{ - EMGD_DEBUG("Enter gmm_save(), which is stubbed"); - return; -} - -static void gmm_restore(igd_context_t *context, void *state) -{ - EMGD_DEBUG("Enter gmm_restore(), which is stubbed"); - return; -} - -/* - * Create a virtual address mapping for a block of video memory. - */ -static void *gmm_map(unsigned long offset) -{ - gmm_chunk_t *chunk; - struct page **page_map; - int i; - void *addr = NULL; - unsigned long num_pages; - unsigned long size; - - EMGD_TRACE_ENTER; - EMGD_DEBUG("Parameter: offset=0x%lx", offset); - - chunk = gmm_get_chunk(gmm_context.context, offset); - - if (chunk == NULL) { - printk(KERN_ERR"[EMGD] gmm_map: Failed to find chunk: 0x%lx\n", offset); - return NULL; - } - - /* - * Check if this as been mapped already and return that map instead - * of remapping it. - */ - chunk->ref_cnt++; - if (chunk->addr) { - EMGD_DEBUG("This chunk is already mapped!"); - return chunk->addr; - } - - /* - * Read the physical addresses of the allocation from the GTT - * and convert that to a page list. - */ - - size = PAGE_ALIGN(chunk->size); - num_pages = size / PAGE_SIZE; - - if (num_pages != chunk->gtt_mem->page_count) { - printk(KERN_ERR"[EMGD] gmm_map: Num pages don't match %lu vs. %lu\n", - num_pages, (unsigned long)chunk->gtt_mem->page_count); - return NULL; - } - - /* - * chunk->memory is an agp_memory struct - * This has an elements struct page **pages and page_count - */ - page_map = vmalloc(num_pages * sizeof(struct page *)); - if (page_map == NULL) { - printk(KERN_ERR"[EMGD] gmm_map: vmalloc failed.\n"); - return NULL; - } - - for (i = 0; i < num_pages; i++) { - page_map[i] = chunk->gtt_mem->pages[i]; - } - - addr = vmap(page_map, num_pages, VM_MAP, PAGE_AGP); - vfree(page_map); - chunk->addr = addr; - - EMGD_DEBUG("Mapped address = 0x%p", addr); - EMGD_TRACE_EXIT; - - return addr; -} - - -static void gmm_unmap(void *addr) -{ - gmm_chunk_t *chunk; - - EMGD_TRACE_ENTER; - EMGD_DEBUG("Parameter: addr=0x%p", addr); - - /* Look up the chunk that was mapped to this address */ - chunk = gmm_context.head_chunk; - while (chunk) { - if (chunk->addr == addr) { - EMGD_DEBUG("The chunk with addr=0x%p has the offset = 0x%08lx", addr, - chunk->offset); - chunk->ref_cnt--; - if (chunk->ref_cnt == 0) { - EMGD_DEBUG("About to call vunmap(0x%p)", addr); - vunmap(addr); - chunk->addr = NULL; - } - return; - } - chunk = chunk->next; - } - - EMGD_TRACE_EXIT; -} - -int gmm_init(igd_context_t *context, - unsigned long scratch_mem, - unsigned long max_fb_mem) -{ - EMGD_TRACE_ENTER; - EMGD_DEBUG("Parameters: scratch_mem=0x%lx, max_fb_mem=%lu", - scratch_mem, max_fb_mem); - - context->dispatch.gmm_alloc_surface = gmm_alloc_surface; - context->dispatch.gmm_alloc_region = gmm_alloc_region; - context->dispatch.gmm_virt_to_phys = gmm_virt_to_phys; - context->dispatch.gmm_free = gmm_free; - context->dispatch.gmm_memstat = NULL; - context->dispatch.gmm_alloc_cached = NULL; - context->dispatch.gmm_free_cached = NULL; - context->dispatch.gmm_alloc_cached_region = NULL; - context->dispatch.gmm_free_cached_region = NULL; - context->dispatch.gmm_flush_cache = gmm_flush_cache; - context->dispatch.gmm_alloc_reservation = NULL; - context->dispatch.gmm_alloc_heap = NULL; - context->dispatch.gmm_alloc_heap_block = NULL; - context->dispatch.gmm_free_heap_block = NULL; - context->dispatch.gmm_get_heap_from_block = NULL; - context->dispatch.gmm_get_pvtheap_size = NULL; - context->dispatch.gmm_get_cache_mem = NULL; - context->dispatch.gmm_alloc_persistent_region = NULL; - context->dispatch.gmm_free_persistent_region = NULL; - context->dispatch.gmm_map = gmm_map; - context->dispatch.gmm_unmap = gmm_unmap; - context->dispatch.gmm_get_page_list = gmm_get_page_list; - context->dispatch.gmm_get_num_surface = gmm_get_num_surface; - context->dispatch.gmm_get_surface_list = gmm_get_surface_list; - - context->mod_dispatch.gmm_save = gmm_save; - context->mod_dispatch.gmm_restore = gmm_restore; - - gmm_context.context = context; - gmm_context.head_chunk = NULL; - gmm_context.tail_chunk = NULL; - - /* Reserve memory for framebuffer ??? */ - - EMGD_DEBUG("EXIT Returning %d", 0); - return 0; -} - - -void gmm_shutdown(igd_context_t *context) -{ - gmm_chunk_t *chunk, *del; - struct drm_device *dev; - - EMGD_TRACE_ENTER; - - dev = (struct drm_device *)context->drm_dev; - - /* Walk the chunk list */ - chunk = gmm_context.head_chunk; - while (chunk) { - EMGD_DEBUG("process chunk at 0x%lx", chunk->offset); - if (chunk->used == 1) { - EMGD_ERROR("Chunk at 0x%lx not properly freed", chunk->offset); - } - - if (chunk->addr != NULL) { - vunmap(chunk->addr); - } - - if (chunk->bound) { - emgd_gtt_remove(context, chunk->gtt_mem, chunk->offset); - } - emgd_free_pages(chunk->gtt_mem); - - /* Free the array of page address, if applicable: */ - if (chunk->page_addresses != NULL) { - EMGD_DEBUG("About to free chunk->page_addresses = 0x%p", - chunk->page_addresses); - OS_FREE(chunk->page_addresses); - } - - /* Free the chunk */ - del = chunk; - chunk = chunk->next; - OS_FREE(del); - } - - EMGD_TRACE_EXIT; - return; -} - -gmm_chunk_t *gmm_get_chunk(igd_context_t *context, unsigned long offset) -{ - gmm_chunk_t *chunk; - - chunk = gmm_context.head_chunk; - while (chunk) { - if (chunk->offset == offset) { - return chunk; - } - chunk = chunk->next; - } - - printk(KERN_ERR "[EMGD] gmm_get_chunk: Failed to find chunk 0x%lx\n", - offset); - return NULL; -} - - - -static int gmm_alloc_linear_surface(unsigned long *offset, - unsigned long pixel_format, - unsigned int *width, - unsigned int *height, - unsigned int *pitch, - unsigned long *size, - unsigned long type, - unsigned long flags, - unsigned long phys) -{ - int ret; - unsigned long align; - unsigned long min_pitch; - unsigned long test = 1; - - EMGD_TRACE_ENTER; - EMGD_DEBUG("Parameters: pixel_format=0x%08lx,", pixel_format); - EMGD_DEBUG(" width=%u, height=%u", *width, *height); - EMGD_DEBUG(" pitch=%u, size=%lu, type=%lu", *pitch, *size, type); - EMGD_DEBUG(" flags=0x%08lx; phys=%lu", flags, phys); - - /* Validate surface */ - if (! *width) { - *width = 1; - } - - if (! *height) { - *height = 1; - } - - /* Set the minimum surface pitch */ - min_pitch = (IGD_PF_DEPTH(pixel_format) * *width) >> 3; - if (min_pitch < *pitch) { - min_pitch = *pitch; - } - - /* Make sure pitch is power of two */ - while (test < min_pitch) { - test <<= 1; - } - min_pitch = test; - - /* - * Size should be based on pixel format and pitch, not just pitch. - * For YUV surfaces, it is smaller than RGB surfaces. - */ - switch (IGD_PF_TYPE(pixel_format)) { - case PF_TYPE_YUV_PLANAR: - *size = min_pitch * (*height + (*height>>1)); - break; - case PF_TYPE_YUV_PACKED: - /* FIXME: What should this really be? */ - *size = min_pitch * *height; - break; - default: - *size = min_pitch * *height; - break; - } - - *pitch = min_pitch; - - /* Page align size */ - align = (*size + 4095) & ~4095; - - /* - * Flags provide information on the type of surface being requested - * 0x04 = cursor surface - * 0x08 = overlay surface - * 0x10 = display surface - * 0x40 = video surface - */ - do { - ret = gmm_alloc_chunk_space(&gmm_context, offset, *size, phys, flags); - } while ((ret == -IGD_ERROR_NOMEM) && gmm_flush_cache()); - - EMGD_DEBUG("EXIT Returning %d", ret); - return ret; -} - - - -/* - * Maintain a very simple linear linked list of memory allocations. Try - * to re-use freed blocks. No error checking is done and alignment is - * hard codeded. - */ - -static int gmm_alloc_chunk_space(gmm_context_t *gmm_context, - unsigned long *offset, - unsigned long size, - unsigned long phys, - unsigned long flags) -{ - gmm_chunk_t *chunk; - struct drm_device *dev; - - EMGD_TRACE_ENTER; - EMGD_DEBUG("Parameters: size=%lu; phys=%lu", size, phys); - EMGD_DEBUG(" flags=0x%08lx", flags); - - /* Check for a free chunk of sufficent size */ - chunk = gmm_context->head_chunk; - while (chunk) { - if ((chunk->used == 0) && (chunk->size >= size) && - (chunk->type == (phys ? AGP_PHYS_MEMORY : AGP_NORMAL_MEMORY))) { - chunk->used = 1; - *offset = chunk->offset; - EMGD_DEBUG("Re-using old chunk with offset=0x%lx", chunk->offset); - EMGD_DEBUG("EXIT Returning %d", 0); - return 0; - } - chunk = chunk->next; - } - - /* Allocate a new chunk */ - chunk = (gmm_chunk_t *)OS_ALLOC(sizeof(gmm_chunk_t)); - if (!chunk) { - printk(KERN_ALERT "[EMGD] Cannot allocate gmm_chunk_t"); - EMGD_ERROR_EXIT("Returning %d", -IGD_ERROR_NOMEM); - return -IGD_ERROR_NOMEM; - } - OS_MEMSET(chunk, 0, sizeof(gmm_chunk_t)); - - /* - * First allocate the memory from the gart driver. If this failes, - * don't bother allocating a new chunk. - */ - dev = (struct drm_device *)gmm_context->context->drm_dev; - chunk->size = size; - chunk->pages = (size + PAGE_SIZE - 1) / PAGE_SIZE; - /* - * If we need phyical contiguous memory, then we need to - * set the type to AGP_PHYS_MEMORY, otherwise use AGP_NORMAL_MEMORY - */ - if (phys) { - chunk->type = AGP_PHYS_MEMORY; - EMGD_DEBUG("Allocate AGP_PHYS; size = 0x%08lx", chunk->size); - } else { - chunk->type = AGP_NORMAL_MEMORY; - EMGD_DEBUG("Allocate AGP_NORMAL; size = 0x%08lx", chunk->size); - } - - if (dev == NULL) { - OS_FREE(chunk); - EMGD_ERROR_EXIT("drm device is NULL; Returning %d", -IGD_ERROR_NOMEM); - return -IGD_ERROR_NOMEM; - } - - - /* Allocate memory from the AGPGART */ - chunk->gtt_mem = emgd_alloc_pages(chunk->pages, chunk->type); - if (!chunk->gtt_mem) { - OS_FREE(chunk); - printk(KERN_ALERT "[EMGD] Failed to allocated AGP memory.\n"); - EMGD_DEBUG("gmm_alloc_chunk_space() returning %d", -IGD_ERROR_NOMEM); - return -IGD_ERROR_NOMEM; - } - - - chunk->used = 1; - chunk->ref_cnt = 0; - chunk->page_addresses = NULL; - - /* - * Get the next available offset that works for this allocation. - * Currently this just uses the next linear offset available. No - * attempt is made to keep track of or utilize gaps introduced - * because of alignments. - * - * Eventually, this should mainting different "heaps" of offsets - * for different types of allocations. For example, display vs. - * video. - * - * See igd_gmm.h for the different surface types supported. Below - * are the ones of interest - * - * #define IGD_SURFACE_RENDER 0x00000001 - * #define IGD_SURFACE_CURSOR 0x00000004 - * #define IGD_SURFACE_OVERLAY 0x00000008 - * #define IGD_SURFACE_DISPLAY 0x00000010 - * #define IGD_SURFACE_VIDEO 0x00000040 - * #define IGD_SURFACE_VIDEO_ENCODE 0x00000080 - * #define IGD_SURFACE_SYS_MEM 0x00008000 - * #define IGD_SURFACE_PHYS_PTR 0x00010000 - * - */ - if (gmm_context->tail_chunk == NULL) { - chunk->offset = 0; - } else { - chunk->offset = gmm_context->tail_chunk->offset + - gmm_context->tail_chunk->size; - } - EMGD_DEBUG("- Before alignment: offset=0x%lx", chunk->offset); - - /* - * Alignment varies depending on the type of surface being allocated. - */ - if (flags & IGD_SURFACE_DISPLAY) { - /* 256KB aligned */ - chunk->offset = (chunk->offset + 0x3ffff) & ~0x3ffff; - } else { - /* 4KB aligned */ - chunk->offset = (chunk->offset + 0x0fff) & ~0x0fff; - } - - EMGD_DEBUG("- After alignment: offset=0x%lx", chunk->offset); - chunk->next = NULL; - - if (gmm_context->head_chunk == NULL) { - gmm_context->head_chunk = chunk; - } else { - gmm_context->tail_chunk->next = chunk; - } - gmm_context->tail_chunk = chunk; - - /* Bind the gart memory to the offset */ - /* - * This updates the GTT table with the actual allocated pages - * so the display hardware can access the memory. - * - * TODO: Add ability to use other MMU's depending on the - * type of memory requested. - */ - emgd_gtt_insert(gmm_context->context, chunk->gtt_mem, chunk->offset); - chunk->bound = 1; - - /* - * Physical is only meaningfull for single page or contiguous pages. - * It represents the physical address of the first allocated page. - */ - if (chunk->gtt_mem->physical == 0x0) { - chunk->gtt_mem->physical = page_to_phys(chunk->gtt_mem->pages[0]); - } - - *offset = chunk->offset; - - EMGD_DEBUG("Allocated chunk @ 0x%lx (0x%lx)", chunk->offset, - (unsigned long)chunk->gtt_mem->physical); - EMGD_TRACE_EXIT; - return 0; -} - - -static int gmm_get_page_list(unsigned long offset, - unsigned long **pages, - unsigned long *page_cnt) -{ - gmm_chunk_t *chunk; - int i; - - EMGD_TRACE_ENTER; - EMGD_DEBUG("Parameters: offset=0x%08lx", offset); - EMGD_DEBUG(" pages=0x%p, *pages=0x%p", pages, *pages); - chunk = gmm_get_chunk(gmm_context.context, offset); - - if (chunk == NULL) { - printk(KERN_ERR"[EMGD] gmm_get_page_list: Failed to find chunk: " - "0x%lx\n", offset); - return -IGD_ERROR_NOMEM; - } - - *page_cnt = chunk->gtt_mem->page_count; - - /* Allocate an array of page addresses: */ - if (chunk->page_addresses == NULL) { - chunk->page_addresses = OS_ALLOC(sizeof(unsigned long) * *page_cnt); - EMGD_DEBUG("Allocated chunk->page_addresses = 0x%p", - chunk->page_addresses); - if (chunk->page_addresses == NULL) { - printk(KERN_ERR "[EMGD] gmm_get_page_list: failed to allocate the " - "array of page addresses for chunk: 0x%lx\n", offset); - return -IGD_ERROR_NOMEM; - } - } else { - EMGD_DEBUG("Re-using chunk->page_addresses = 0x%p", - chunk->page_addresses); - } - - /* Populate the array with the starting addresses of the pages: */ - for (i = 0; i < *page_cnt; i++) { - chunk->page_addresses[i] = page_to_phys(chunk->gtt_mem->pages[i]); - } - - *pages = chunk->page_addresses; - - EMGD_DEBUG("*pages=0x%p", *pages); - EMGD_DEBUG("page_count=%lu", *page_cnt); - EMGD_TRACE_EXIT; - return 0; -} |