Linaro-mm-sig

linaro-mm-sig@lists.linaro.org

10 participants
2527 discussions

[RFC v2] implicit drm synchronization wip with dma-buf

by Maarten Lankhorst

Well, V2 time! I was hinted to look at ttm_execbuf_util, and it does indeed contain some nice code. My goal was to extend dma-buf in a generic way now, some elements from v1 are remaining, most notably a dma-buf used for syncing. However it is expected now that dma-buf syncing will work in a very specific way now, slightly more strict than in v1. Instead of each buffer having their own dma-buf I put in 1 per command submission. This submission hasn't been run-time tested yet, but I expect the api to go something like this. Intended to be used like this: list_init(&head); add_list_tail(&validate1->entry, &head); add_list_tail(&validate2->entry, &head); add_list_tail(&validate3->entry, &head); r = dmabufmgr_eu_reserve_buffers(&head); if (r) return r; // add waits on cpu or gpu list_for_each_entry(validate, ...) { if (!validate->sync_buf) continue; // Check attachments to see if we already imported sync_buf // somewhere, if not attach to it. // waiting until cur_seq - dmabuf->sync_val >= 0 on either cpu // or as command submitted to gpu // sync_buf itself is a dma-buf, so it should be trivial // TODO: sync_buf should NEVER be validated, add is_sync_buf to dma_buf? // If this step fails: dmabufmgr_eu_backoff_reservation // else: // dmabufmgr_eu_fence_buffer_objects(our_own_sync_buf, // hwchannel * max(minhwalign, 4), ++counter[hwchannel]); // XXX: Do we still require a minimum alignment? I set up 16 for nouveau, // but this is no longer needed in this design since it only matters // for writes for which nouveau would already control the offset. } // Some time after execbuffer was executed, doesn't have to be right away but before // getting in the danger of our own counter wrapping around: // grab dmabufmgr.lru_lock, and cleanup by unreffing sync_buf when // sync_buf == ownbuf, sync_ofs == ownofs, and sync_val == saved_counter // In the meantime someone else or even us might have reserved this dma_buf // again, which is why all those checks are needed before unreffing. diff --git a/drivers/base/Makefile b/drivers/base/Makefile index 5aa2d70..86e7598 100644 --- a/drivers/base/Makefile +++ b/drivers/base/Makefile @@ -10,7 +10,7 @@ obj-$(CONFIG_CMA) += dma-contiguous.o obj-y += power/ obj-$(CONFIG_HAS_DMA) += dma-mapping.o obj-$(CONFIG_HAVE_GENERIC_DMA_COHERENT) += dma-coherent.o -obj-$(CONFIG_DMA_SHARED_BUFFER) += dma-buf.o +obj-$(CONFIG_DMA_SHARED_BUFFER) += dma-buf.o dma-buf-mgr.o dma-buf-mgr-eu.o obj-$(CONFIG_ISA) += isa.o obj-$(CONFIG_FW_LOADER) += firmware_class.o obj-$(CONFIG_NUMA) += node.o diff --git a/drivers/base/dma-buf-mgr-eu.c b/drivers/base/dma-buf-mgr-eu.c new file mode 100644 index 0000000..27ebc68 --- /dev/null +++ b/drivers/base/dma-buf-mgr-eu.c @@ -0,0 +1,170 @@ +/* + * Copyright (C) 2012 Canonical Ltd + * + * Based on ttm_bo.c which bears the following copyright notice, + * but is dual licensed: + * + * Copyright (c) 2006-2009 VMware, Inc., Palo Alto, CA., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include <linux/dma-buf-mgr.h> +#include <linux/sched.h> +#include <linux/export.h> + +static void dmabufmgr_eu_backoff_reservation_locked(struct list_head *list) +{ + struct dmabufmgr_validate *entry; + + list_for_each_entry(entry, list, head) { + struct dma_buf *bo = entry->bo; + if (!entry->reserved) + continue; + + entry->reserved = false; + atomic_set(&bo->reserved, 0); + wake_up_all(&bo->event_queue); + if (entry->sync_buf) + dma_buf_put(entry->sync_buf); + entry->sync_buf = NULL; + } +} + +static int +dmabufmgr_eu_wait_unreserved_locked(struct list_head *list, + struct dma_buf *bo) +{ + int ret; + + spin_unlock(&dmabufmgr.lru_lock); + ret = dmabufmgr_bo_wait_unreserved(bo, true); + spin_lock(&dmabufmgr.lru_lock); + if (unlikely(ret != 0)) + dmabufmgr_eu_backoff_reservation_locked(list); + return ret; +} + +void +dmabufmgr_eu_backoff_reservation(struct list_head *list) +{ + struct dmabufmgr_validate *entry; + + if (list_empty(list)) + return; + + entry = list_first_entry(list, struct dmabufmgr_validate, head); + spin_lock(&dmabufmgr.lru_lock); + dmabufmgr_eu_backoff_reservation_locked(list); + spin_unlock(&dmabufmgr.lru_lock); +} +EXPORT_SYMBOL_GPL(dmabufmgr_eu_backoff_reservation); + +int +dmabufmgr_eu_reserve_buffers(struct list_head *list) +{ + struct dmabufmgr_validate *entry; + int ret; + u32 val_seq; + + if (list_empty(list)) + return 0; + + list_for_each_entry(entry, list, head) { + entry->reserved = false; + entry->sync_buf = NULL; + } + +retry: + spin_lock(&dmabufmgr.lru_lock); + val_seq = dmabufmgr.counter++; + + list_for_each_entry(entry, list, head) { + struct dma_buf *bo = entry->bo; + +retry_this_bo: + ret = dmabufmgr_bo_reserve_locked(bo, true, true, true, val_seq); + switch (ret) { + case 0: + break; + case -EBUSY: + ret = dmabufmgr_eu_wait_unreserved_locked(list, bo); + if (unlikely(ret != 0)) { + spin_unlock(&dmabufmgr.lru_lock); + return ret; + } + goto retry_this_bo; + case -EAGAIN: + dmabufmgr_eu_backoff_reservation_locked(list); + spin_unlock(&dmabufmgr.lru_lock); + ret = dmabufmgr_bo_wait_unreserved(bo, true); + if (unlikely(ret != 0)) + return ret; + goto retry; + default: + dmabufmgr_eu_backoff_reservation_locked(list); + spin_unlock(&dmabufmgr.lru_lock); + return ret; + } + + entry->reserved = true; + if (bo->sync_buf) + get_dma_buf(bo->sync_buf); + entry->sync_buf = bo->sync_buf; + entry->sync_ofs = bo->sync_ofs; + entry->sync_val = bo->sync_val; + } + spin_unlock(&dmabufmgr.lru_lock); + + return 0; +} +EXPORT_SYMBOL_GPL(dmabufmgr_eu_reserve_buffers); + +void +dmabufmgr_eu_fence_buffer_objects(struct dma_buf *sync_buf, u32 ofs, u32 seq, struct list_head *list) +{ + struct dmabufmgr_validate *entry; + struct dma_buf *bo; + + if (list_empty(list) || WARN_ON(!sync_buf)) + return; + + spin_lock(&dmabufmgr.lru_lock); + + list_for_each_entry(entry, list, head) { + bo = entry->bo; + dmabufmgr_bo_unreserve_locked(bo); + entry->reserved = false; + if (entry->sync_buf) + dma_buf_put(entry->sync_buf); + entry->sync_buf = NULL; + + get_dma_buf(sync_buf); + bo->sync_buf = sync_buf; + bo->sync_ofs = ofs; + bo->sync_val = seq; + } + + spin_unlock(&dmabufmgr.lru_lock); +} +EXPORT_SYMBOL_GPL(dmabufmgr_eu_fence_buffer_objects); diff --git a/drivers/base/dma-buf-mgr.c b/drivers/base/dma-buf-mgr.c new file mode 100644 index 0000000..14756ff --- /dev/null +++ b/drivers/base/dma-buf-mgr.c @@ -0,0 +1,149 @@ +/* + * Copyright (C) 2012 Canonical Ltd + * + * Based on ttm_bo.c which bears the following copyright notice, + * but is dual licensed: + * + * Copyright (c) 2006-2009 VMware, Inc., Palo Alto, CA., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ +/* + * Authors: Thomas Hellstrom <thellstrom-at-vmware-dot-com> + */ + + +#include <linux/fs.h> +#include <linux/slab.h> +#include <linux/dma-buf-mgr.h> +#include <linux/anon_inodes.h> +#include <linux/export.h> +#include <linux/sched.h> +#include <linux/list.h> + +/* Based on ttm_bo.c with vm_lock and fence_lock removed + * lru_lock takes care of fence_lock as well + */ +struct dmabufmgr dmabufmgr = { + .lru_lock = __SPIN_LOCK_UNLOCKED(dmabufmgr.lru_lock), + .counter = 1, +}; + +int +dmabufmgr_bo_reserve_locked(struct dma_buf *bo, + bool interruptible, bool no_wait, + bool use_sequence, u32 sequence) +{ + int ret; + + while (unlikely(atomic_cmpxchg(&bo->reserved, 0, 1) != 0)) { + /** + * Deadlock avoidance for multi-bo reserving. + */ + if (use_sequence && bo->seq_valid) { + /** + * We've already reserved this one. + */ + if (unlikely(sequence == bo->val_seq)) + return -EDEADLK; + /** + * Already reserved by a thread that will not back + * off for us. We need to back off. + */ + if (unlikely(sequence - bo->val_seq < (1 << 31))) + return -EAGAIN; + } + + if (no_wait) + return -EBUSY; + + spin_unlock(&dmabufmgr.lru_lock); + ret = dmabufmgr_bo_wait_unreserved(bo, interruptible); + spin_lock(&dmabufmgr.lru_lock); + + if (unlikely(ret)) + return ret; + } + + if (use_sequence) { + /** + * Wake up waiters that may need to recheck for deadlock, + * if we decreased the sequence number. + */ + if (unlikely((bo->val_seq - sequence < (1 << 31)) + || !bo->seq_valid)) + wake_up_all(&bo->event_queue); + + bo->val_seq = sequence; + bo->seq_valid = true; + } else { + bo->seq_valid = false; + } + + return 0; +} +EXPORT_SYMBOL_GPL(dmabufmgr_bo_reserve_locked); + +int +dmabufmgr_bo_reserve(struct dma_buf *bo, + bool interruptible, bool no_wait, + bool use_sequence, u32 sequence) +{ + int ret; + + spin_lock(&dmabufmgr.lru_lock); + ret = dmabufmgr_bo_reserve_locked(bo, interruptible, no_wait, + use_sequence, sequence); + spin_unlock(&dmabufmgr.lru_lock); + + return ret; +} +EXPORT_SYMBOL_GPL(dmabufmgr_bo_reserve); + +int +dmabufmgr_bo_wait_unreserved(struct dma_buf *bo, bool interruptible) +{ + if (interruptible) { + return wait_event_interruptible(bo->event_queue, + atomic_read(&bo->reserved) == 0); + } else { + wait_event(bo->event_queue, atomic_read(&bo->reserved) == 0); + return 0; + } +} +EXPORT_SYMBOL_GPL(dmabufmgr_bo_wait_unreserved); + +void dmabufmgr_bo_unreserve_locked(struct dma_buf *bo) +{ + atomic_set(&bo->reserved, 0); + wake_up_all(&bo->event_queue); +} +EXPORT_SYMBOL_GPL(dmabufmgr_bo_unreserve_locked); + +void dmabufmgr_bo_unreserve(struct dma_buf *bo) +{ + spin_lock(&dmabufmgr.lru_lock); + dmabufmgr_bo_unreserve_locked(bo); + spin_unlock(&dmabufmgr.lru_lock); +} +EXPORT_SYMBOL_GPL(dmabufmgr_bo_unreserve); diff --git a/drivers/base/dma-buf.c b/drivers/base/dma-buf.c index 24e88fe..01c4f71 100644 --- a/drivers/base/dma-buf.c +++ b/drivers/base/dma-buf.c @@ -40,6 +40,9 @@ static int dma_buf_release(struct inode *inode, struct file *file) dmabuf = file->private_data; dmabuf->ops->release(dmabuf); + BUG_ON(waitqueue_active(&dmabuf->event_queue)); + if (dmabuf->sync_buf) + dma_buf_put(dmabuf->sync_buf); kfree(dmabuf); return 0; } @@ -119,6 +122,7 @@ struct dma_buf *dma_buf_export(void *priv, const struct dma_buf_ops *ops, mutex_init(&dmabuf->lock); INIT_LIST_HEAD(&dmabuf->attachments); + init_waitqueue_head(&dmabuf->event_queue); return dmabuf; } diff --git a/include/linux/dma-buf-mgr.h b/include/linux/dma-buf-mgr.h new file mode 100644 index 0000000..b26462e --- /dev/null +++ b/include/linux/dma-buf-mgr.h @@ -0,0 +1,84 @@ +/* + * Header file for dma buffer sharing framework. + * + * Copyright(C) 2011 Linaro Limited. All rights reserved. + * Author: Sumit Semwal <sumit.semwal(a)ti.com> + * + * Many thanks to linaro-mm-sig list, and specially + * Arnd Bergmann <arnd(a)arndb.de>, Rob Clark <rob(a)ti.com> and + * Daniel Vetter <daniel(a)ffwll.ch> for their support in creation and + * refining of this idea. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License version 2 as published by + * the Free Software Foundation. + * + * This program is distributed in the hope that 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, see <http://www.gnu.org/licenses/>. + */ +#ifndef __DMA_BUF_MGR_H__ +#define __DMA_BUF_MGR_H__ + +#include <linux/dma-buf.h> +#include <linux/list.h> + +/** Size of each hwcontext in synchronization dma-buf */ +#define DMABUFMGR_HWCONTEXT_SYNC_ALIGN 16 + +struct dmabufmgr { + spinlock_t lru_lock; + + u32 counter; +}; +extern struct dmabufmgr dmabufmgr; + +extern int +dmabufmgr_bo_reserve_locked(struct dma_buf *bo, + bool interruptible, bool no_wait, + bool use_sequence, u32 sequence); + +extern int +dmabufmgr_bo_reserve(struct dma_buf *bo, + bool interruptible, bool no_wait, + bool use_sequence, u32 sequence); + +extern void +dmabufmgr_bo_unreserve_locked(struct dma_buf *bo); + +extern void +dmabufmgr_bo_unreserve(struct dma_buf *bo); + +extern int +dmabufmgr_bo_wait_unreserved(struct dma_buf *bo, bool interruptible); + +/* execbuf util support for reservations + * matches ttm_execbuf_util + */ +struct dmabufmgr_validate { + struct list_head head; + struct dma_buf *bo; + bool reserved; + + /* If non-null, check for attachments */ + struct dma_buf *sync_buf; + u32 sync_ofs, sync_val; +}; + +/** reserve a linked list of struct dmabufmgr_validate entries */ +extern int +dmabufmgr_eu_reserve_buffers(struct list_head *list); + +/** Undo reservation */ +extern void +dmabufmgr_eu_backoff_reservation(struct list_head *list); + +/** Commit reservation */ +extern void +dmabufmgr_eu_fence_buffer_objects(struct dma_buf *sync_buf, u32 ofs, u32 val, struct list_head *list); + +#endif /* __DMA_BUF_MGR_H__ */ diff --git a/include/linux/dma-buf.h b/include/linux/dma-buf.h index eb48f38..b2ab395 100644 --- a/include/linux/dma-buf.h +++ b/include/linux/dma-buf.h @@ -113,6 +113,8 @@ struct dma_buf_ops { * @attachments: list of dma_buf_attachment that denotes all devices attached. * @ops: dma_buf_ops associated with this buffer object. * @priv: exporter specific private data for this buffer object. + * @bufmgr_entry: used by dmabufmgr + * @bufdev: used by dmabufmgr */ struct dma_buf { size_t size; @@ -122,6 +124,24 @@ struct dma_buf { /* mutex to serialize list manipulation and attach/detach */ struct mutex lock; void *priv; + + /** dmabufmgr members */ + wait_queue_head_t event_queue; + + /** + * dmabufmgr members protected by the dmabufmgr::lru_lock. + */ + u32 val_seq; + bool seq_valid; + + struct dma_buf *sync_buf; + u32 sync_ofs, sync_val; + + /** + * dmabufmgr members protected by the dmabufmgr::lru_lock + * only when written to. + */ + atomic_t reserved; }; /**

12 years, 5 months

[PATCHv4 0/2] ARM: replace custom consistent dma region with vmalloc

by Marek Szyprowski

Hello! This quick update on the patchset which replaces custom consistent dma regions usage in dma-mapping framework in favour of generic vmalloc areas created on demand for each allocation. The main purpose for this patchset is to remove 2MiB limit of dma coherent/writecombine allocations. In this version arch-independent VM_DMA flag has been replaced with ARM-specific VM_ARM_DMA_CONSISTENT flag. This patch is based on vanilla v3.5-rc4 release. Best regards Marek Szyprowski Samsung Poland R&D Center Changelog: v4: - replaced arch-independent VM_DMA flag with ARM-specific VM_ARM_DMA_CONSISTENT flag v3: http://thread.gmane.org/gmane.linux.kernel.mm/80028/ - rebased onto v3.4-rc2: added support for IOMMU-aware implementation of dma-mapping calls, unified with CMA coherent dma pool - implemented changes requested by Minchan Kim: added more checks for vmarea->flags & VM_DMA, renamed some variables, removed obsole locks, squashed find_vm_area() exporting patch into the main redesign patch v2: http://thread.gmane.org/gmane.linux.kernel.mm/78563 - added support for atomic allocations (served from preallocated pool) - minor cleanup here and there - rebased onto v3.4-rc7 v1: http://thread.gmane.org/gmane.linux.kernel.mm/76703 - initial version Patch summary: Marek Szyprowski (2): mm: vmalloc: use const void * for caller argument ARM: dma-mapping: remove custom consistent dma region Documentation/kernel-parameters.txt | 2 +- arch/arm/include/asm/dma-mapping.h | 2 +- arch/arm/mm/dma-mapping.c | 505 +++++++++++++---------------------- arch/arm/mm/mm.h | 3 + include/linux/vmalloc.h | 9 +- mm/vmalloc.c | 28 ++- 6 files changed, 207 insertions(+), 342 deletions(-) -- 1.7.1.569.g6f426

12 years, 5 months

[PATCH] common: dma-mapping: add support for generic dma_mmap_* calls

by Marek Szyprowski

Commit 9adc5374 ('common: dma-mapping: introduce mmap method') added a generic method for implementing mmap user call to dma_map_ops structure. This patch converts ARM and PowerPC architectures (the only providers of dma_mmap_coherent/dma_mmap_writecombine calls) to use this generic dma_map_ops based call and adds a generic cross architecture definition for dma_mmap_attrs, dma_mmap_coherent, dma_mmap_writecombine functions. The generic mmap virt_to_page-based fallback implementation is provided for architectures which don't provide their own implementation for mmap method. Signed-off-by: Marek Szyprowski <m.szyprowski(a)samsung.com> Reviewed-by: Kyungmin Park <kyungmin.park(a)samsung.com> --- Hello, This patch is a continuation of my works on dma-mapping cleanup and unification. Previous works (commit 58bca4a8fa ('Merge branch 'for-linus' of git://git.linaro.org/people/mszyprowski/linux-dma-mapping') has been merged to v3.4-rc2. Now I've focuses on providing implementation for all architectures so the drivers and some cross-architecture common helpers (like for example videobuf2) can start using this new api. I'm not 100% sure if the PowerPC changes are correct. The cases of dma_iommu_ops and vio_dma_mapping_ops are a bit suspicious for me, but I have no way to test and check if my changes works for that hardware. Best regards Marek Szyprowski Samsung Poland R&D Center --- arch/arm/include/asm/dma-mapping.h | 19 --------------- arch/powerpc/include/asm/dma-mapping.h | 8 +++--- arch/powerpc/kernel/dma-iommu.c | 1 + arch/powerpc/kernel/dma-swiotlb.c | 1 + arch/powerpc/kernel/dma.c | 36 +++++++++++++++------------- arch/powerpc/kernel/vio.c | 1 + drivers/base/dma-mapping.c | 31 +++++++++++++++++++++++++ include/asm-generic/dma-coherent.h | 1 + include/asm-generic/dma-mapping-common.h | 37 ++++++++++++++++++++++++++++++ 9 files changed, 95 insertions(+), 40 deletions(-) diff --git a/arch/arm/include/asm/dma-mapping.h b/arch/arm/include/asm/dma-mapping.h index bbef15d..8645088 100644 --- a/arch/arm/include/asm/dma-mapping.h +++ b/arch/arm/include/asm/dma-mapping.h @@ -186,17 +186,6 @@ extern int arm_dma_mmap(struct device *dev, struct vm_area_struct *vma, void *cpu_addr, dma_addr_t dma_addr, size_t size, struct dma_attrs *attrs); -#define dma_mmap_coherent(d, v, c, h, s) dma_mmap_attrs(d, v, c, h, s, NULL) - -static inline int dma_mmap_attrs(struct device *dev, struct vm_area_struct *vma, - void *cpu_addr, dma_addr_t dma_addr, - size_t size, struct dma_attrs *attrs) -{ - struct dma_map_ops *ops = get_dma_ops(dev); - BUG_ON(!ops); - return ops->mmap(dev, vma, cpu_addr, dma_addr, size, attrs); -} - static inline void *dma_alloc_writecombine(struct device *dev, size_t size, dma_addr_t *dma_handle, gfp_t flag) { @@ -213,14 +202,6 @@ static inline void dma_free_writecombine(struct device *dev, size_t size, return dma_free_attrs(dev, size, cpu_addr, dma_handle, &attrs); } -static inline int dma_mmap_writecombine(struct device *dev, struct vm_area_struct *vma, - void *cpu_addr, dma_addr_t dma_addr, size_t size) -{ - DEFINE_DMA_ATTRS(attrs); - dma_set_attr(DMA_ATTR_WRITE_COMBINE, &attrs); - return dma_mmap_attrs(dev, vma, cpu_addr, dma_addr, size, &attrs); -} - /* * This can be called during boot to increase the size of the consistent * DMA region above it's default value of 2MB. It must be called before the diff --git a/arch/powerpc/include/asm/dma-mapping.h b/arch/powerpc/include/asm/dma-mapping.h index 62678e3..7816087 100644 --- a/arch/powerpc/include/asm/dma-mapping.h +++ b/arch/powerpc/include/asm/dma-mapping.h @@ -27,7 +27,10 @@ extern void *dma_direct_alloc_coherent(struct device *dev, size_t size, extern void dma_direct_free_coherent(struct device *dev, size_t size, void *vaddr, dma_addr_t dma_handle, struct dma_attrs *attrs); - +extern int dma_direct_mmap_coherent(struct device *dev, + struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t handle, + size_t size, struct dma_attrs *attrs); #ifdef CONFIG_NOT_COHERENT_CACHE /* @@ -207,11 +210,8 @@ static inline phys_addr_t dma_to_phys(struct device *dev, dma_addr_t daddr) #define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f) #define dma_free_noncoherent(d, s, v, h) dma_free_coherent(d, s, v, h) -extern int dma_mmap_coherent(struct device *, struct vm_area_struct *, - void *, dma_addr_t, size_t); #define ARCH_HAS_DMA_MMAP_COHERENT - static inline void dma_cache_sync(struct device *dev, void *vaddr, size_t size, enum dma_data_direction direction) { diff --git a/arch/powerpc/kernel/dma-iommu.c b/arch/powerpc/kernel/dma-iommu.c index bcfdcd2..2d7bb8c 100644 --- a/arch/powerpc/kernel/dma-iommu.c +++ b/arch/powerpc/kernel/dma-iommu.c @@ -109,6 +109,7 @@ static u64 dma_iommu_get_required_mask(struct device *dev) struct dma_map_ops dma_iommu_ops = { .alloc = dma_iommu_alloc_coherent, .free = dma_iommu_free_coherent, + .mmap = dma_direct_mmap_coherent, .map_sg = dma_iommu_map_sg, .unmap_sg = dma_iommu_unmap_sg, .dma_supported = dma_iommu_dma_supported, diff --git a/arch/powerpc/kernel/dma-swiotlb.c b/arch/powerpc/kernel/dma-swiotlb.c index 4ab88da..4694365 100644 --- a/arch/powerpc/kernel/dma-swiotlb.c +++ b/arch/powerpc/kernel/dma-swiotlb.c @@ -49,6 +49,7 @@ static u64 swiotlb_powerpc_get_required(struct device *dev) struct dma_map_ops swiotlb_dma_ops = { .alloc = dma_direct_alloc_coherent, .free = dma_direct_free_coherent, + .mmap = dma_direct_mmap_coherent, .map_sg = swiotlb_map_sg_attrs, .unmap_sg = swiotlb_unmap_sg_attrs, .dma_supported = swiotlb_dma_supported, diff --git a/arch/powerpc/kernel/dma.c b/arch/powerpc/kernel/dma.c index b1ec983..062bf20 100644 --- a/arch/powerpc/kernel/dma.c +++ b/arch/powerpc/kernel/dma.c @@ -65,6 +65,24 @@ void dma_direct_free_coherent(struct device *dev, size_t size, #endif } +int dma_direct_mmap_coherent(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t handle, size_t size, + struct dma_attrs *attrs) +{ + unsigned long pfn; + +#ifdef CONFIG_NOT_COHERENT_CACHE + vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); + pfn = __dma_get_coherent_pfn((unsigned long)cpu_addr); +#else + pfn = page_to_pfn(virt_to_page(cpu_addr)); +#endif + return remap_pfn_range(vma, vma->vm_start, + pfn + vma->vm_pgoff, + vma->vm_end - vma->vm_start, + vma->vm_page_prot); +} + static int dma_direct_map_sg(struct device *dev, struct scatterlist *sgl, int nents, enum dma_data_direction direction, struct dma_attrs *attrs) @@ -154,6 +172,7 @@ static inline void dma_direct_sync_single(struct device *dev, struct dma_map_ops dma_direct_ops = { .alloc = dma_direct_alloc_coherent, .free = dma_direct_free_coherent, + .mmap = dma_direct_mmap_coherent, .map_sg = dma_direct_map_sg, .unmap_sg = dma_direct_unmap_sg, .dma_supported = dma_direct_dma_supported, @@ -211,20 +230,3 @@ static int __init dma_init(void) } fs_initcall(dma_init); -int dma_mmap_coherent(struct device *dev, struct vm_area_struct *vma, - void *cpu_addr, dma_addr_t handle, size_t size) -{ - unsigned long pfn; - -#ifdef CONFIG_NOT_COHERENT_CACHE - vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); - pfn = __dma_get_coherent_pfn((unsigned long)cpu_addr); -#else - pfn = page_to_pfn(virt_to_page(cpu_addr)); -#endif - return remap_pfn_range(vma, vma->vm_start, - pfn + vma->vm_pgoff, - vma->vm_end - vma->vm_start, - vma->vm_page_prot); -} -EXPORT_SYMBOL_GPL(dma_mmap_coherent); diff --git a/arch/powerpc/kernel/vio.c b/arch/powerpc/kernel/vio.c index cb87301..dda3d9a 100644 --- a/arch/powerpc/kernel/vio.c +++ b/arch/powerpc/kernel/vio.c @@ -613,6 +613,7 @@ static u64 vio_dma_get_required_mask(struct device *dev) struct dma_map_ops vio_dma_mapping_ops = { .alloc = vio_dma_iommu_alloc_coherent, .free = vio_dma_iommu_free_coherent, + .mmap = dma_direct_mmap_coherent, .map_sg = vio_dma_iommu_map_sg, .unmap_sg = vio_dma_iommu_unmap_sg, .map_page = vio_dma_iommu_map_page, diff --git a/drivers/base/dma-mapping.c b/drivers/base/dma-mapping.c index 6f3676f..db5db02 100644 --- a/drivers/base/dma-mapping.c +++ b/drivers/base/dma-mapping.c @@ -10,6 +10,7 @@ #include <linux/dma-mapping.h> #include <linux/export.h> #include <linux/gfp.h> +#include <asm-generic/dma-coherent.h> /* * Managed DMA API @@ -218,3 +219,33 @@ void dmam_release_declared_memory(struct device *dev) EXPORT_SYMBOL(dmam_release_declared_memory); #endif + +/* + * Create userspace mapping for the DMA-coherent memory. + */ +int dma_common_mmap(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t dma_addr, size_t size) +{ + int ret = -ENXIO; +#ifdef CONFIG_MMU + unsigned long user_count = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; + unsigned long count = PAGE_ALIGN(size) >> PAGE_SHIFT; + unsigned long pfn = page_to_pfn(virt_to_page(cpu_addr)); + unsigned long off = vma->vm_pgoff; + + vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); + + if (dma_mmap_from_coherent(dev, vma, cpu_addr, size, &ret)) + return ret; + + if (off < count && user_count <= (count - off)) { + ret = remap_pfn_range(vma, vma->vm_start, + pfn + off, + user_count << PAGE_SHIFT, + vma->vm_page_prot); + } +#endif /* CONFIG_MMU */ + + return ret; +} +EXPORT_SYMBOL(dma_common_mmap); diff --git a/include/asm-generic/dma-coherent.h b/include/asm-generic/dma-coherent.h index abfb268..2be8a2d 100644 --- a/include/asm-generic/dma-coherent.h +++ b/include/asm-generic/dma-coherent.h @@ -29,6 +29,7 @@ dma_mark_declared_memory_occupied(struct device *dev, #else #define dma_alloc_from_coherent(dev, size, handle, ret) (0) #define dma_release_from_coherent(dev, order, vaddr) (0) +#define dma_mmap_from_coherent(dev, vma, vaddr, order, ret) (0) #endif #endif diff --git a/include/asm-generic/dma-mapping-common.h b/include/asm-generic/dma-mapping-common.h index 2e248d8..9073aeb 100644 --- a/include/asm-generic/dma-mapping-common.h +++ b/include/asm-generic/dma-mapping-common.h @@ -176,4 +176,41 @@ dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg, #define dma_map_sg(d, s, n, r) dma_map_sg_attrs(d, s, n, r, NULL) #define dma_unmap_sg(d, s, n, r) dma_unmap_sg_attrs(d, s, n, r, NULL) +extern int dma_common_mmap(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t dma_addr, size_t size); + +/** + * dma_mmap_attrs - map a coherent DMA allocation into user space + * @dev: valid struct device pointer, or NULL for ISA and EISA-like devices + * @vma: vm_area_struct describing requested user mapping + * @cpu_addr: kernel CPU-view address returned from dma_alloc_attrs + * @handle: device-view address returned from dma_alloc_attrs + * @size: size of memory originally requested in dma_alloc_attrs + * @attrs: attributes of mapping properties requested in dma_alloc_attrs + * + * Map a coherent DMA buffer previously allocated by dma_alloc_attrs + * into user space. The coherent DMA buffer must not be freed by the + * driver until the user space mapping has been released. + */ +static inline int +dma_mmap_attrs(struct device *dev, struct vm_area_struct *vma, void *cpu_addr, + dma_addr_t dma_addr, size_t size, struct dma_attrs *attrs) +{ + struct dma_map_ops *ops = get_dma_ops(dev); + BUG_ON(!ops); + if (ops->mmap) + return ops->mmap(dev, vma, cpu_addr, dma_addr, size, attrs); + return dma_common_mmap(dev, vma, cpu_addr, dma_addr, size); +} + +#define dma_mmap_coherent(d, v, c, h, s) dma_mmap_attrs(d, v, c, h, s, NULL) + +static inline int dma_mmap_writecombine(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t dma_addr, size_t size) +{ + DEFINE_DMA_ATTRS(attrs); + dma_set_attr(DMA_ATTR_WRITE_COMBINE, &attrs); + return dma_mmap_attrs(dev, vma, cpu_addr, dma_addr, size, &attrs); +} + #endif -- 1.7.1.569.g6f426

12 years, 5 months

[PATCH 0/2] Add return value checks in the drm-dmabuf

by Subash Patel

From: Subash Patel <subash.rp(a)samsung.com> This patch series is split and re-send of my original patch, after request from Inki Dae. Below two patches add the required error checks in drm dmabuf when the system fails to allocate pages for the scatter-gather. This is very rare situation, and occurs when the system is under memory pressure. Scatter-gather asks for memory using sg_kmalloc() and this can return no memory. Without below return value checks, the code will crash the system due to invalid pointer access. Subash Patel (2): DRM: Exynos: return NULL if exynos_pages_to_sg fails DRM: Exynos: check for null in return value of dma_buf_map_attachment() drivers/gpu/drm/exynos/exynos_drm_dmabuf.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) -- 1.7.9.5

12 years, 6 months

implicit drm synchronization wip with dma-buf

by Maarten Lankhorst

Hey, Due to inertia, I thought I would take a shot at implicit synchronization as well. I have just barely enough to make it work for nouveau to synchronize with itself now using the cpu. Hopefully the general idea is correct but I feel the implementation wrong. There are 2 ways to get deadlocks if no proper care is taken to avoid it, the first being 2 tasks taking each device's lock in a different order, the second 2 devices waiting on completion of each other before starting own work. The easiest way to avoid this is to introduce a global dma_buf_submit_mutex so in cases where synchronization is needed. This way only 1 submission involving dma-buffer synchronisation can be made simultaneously. This will make it impossible to deadlock because even if you take dma mutex->dev a mutex->dev b mutex, and swap a and b, the dmabuf mutex would prevent this from being done at the same time. That leaves the real problem of the synchronization itself. I felt that because the code involved was already sharing dma-buf's, the easiest way to implement it would be.. another dma-buf. Some hardware might have specific requirements on them, so I haven't pinned down the exact details yet. It's a bit of intermingling between drm and dma-buf namespace since it is an early wip, any comments are welcome though. This is what I used so far: #define DRM_PRIME_FENCE_MAX 2 struct drm_prime_fence { struct dma_buf *sync_buf; uint64_t offset; uint32_t value; enum { /* Nop is to allow preparations in case dma_buf * is different for release, so the call will * never fail at that point. */ DRM_PRIME_FENCE_NOP = 0, DRM_PRIME_FENCE_WAIT_EQ, // DRM_PRIME_FENCE_WAIT_GE, /* block while ((int)(cur - expected) < 0); */ DRM_PRIME_FENCE_SET } op; }; and added to struct dma_buf_ops: /* drm_prime_fence_ is written by function to indicate what is needed * to acquire this buffer, up to DRM_PRIME_FENCE_MAX buffers are allowed * sync_acquire returns a negative value on error, otherwise * amounts of fence ops that need to be executed. * * Release is not allowed to fail and merely returns number of * fence ops that needs to be executed after command stream is done. * Abort occurs when there's a failure between acquire and release, * for example because dma-buf's from multiple devices are involved * and the other one failed to acquire. */ int (*sync_acquire)(struct dma_buf *, struct drm_prime_fence fence[2], unsigned long align, unsigned long release_write); int (*sync_release)(struct dma_buf * struct drm_prime_fence fence[2]); void (*sync_abort)(struct dma_buf *); I'm not completely sure about this part yet, align can be seen as minimum alignment requirement, ideally I would negotiate those earlier but I haven't found the correct place yet, maybe on attach? nouveau writes a 16 byte stamp as part of it's semaphore ops (4 bytes programmable, 4 bytes pad, 8 bytes timestamp) which is why I need to communicate those requirements somehow. Not all nouveau cards wold support DRM_PRIME_FENCE_WAIT_GE either. I think there is a great power in making the sync object itself just another dma-buf that can be written to and/or read. Especially since all graphics card have some way to write an arbitrary 4-byte value to an arbitrary location (even the oldest intel cards have a blitter! :D). I'm hoping for more input into making the api better for other users too, which is why I'm posting this as early as I had something working (for some definition of working). Thoughts? ~Maarten

12 years, 6 months

[PATCH v6 0/4] Add CMA heap for ION memory manager

by Benjamin Gaignard

The goal of those patches is to allow ION clients (drivers or userland applications) to use Contiguous Memory Allocator (CMA). To get more info about CMA: http://lists.linaro.org/pipermail/linaro-mm-sig/2012-February/001328.html patches version 6: - add private field in ion_platform_heap to pass the device linked with CMA. - rework CMA heap to use private field. - prepare CMA heap for incoming dma_common_get_sgtable function http://lists.linaro.org/pipermail/linaro-mm-sig/2012-June/002109.html - simplify ion-ux500 driver. patches version 5: - port patches on android kernel 3.4 where ION use dmabuf - add ion_cma_heap_map_dma and ion_cma_heap_unmap_dma functions patches version 4: - add ION_HEAP_TYPE_DMA heap type in ion_heap_type enum. - CMA heap is now a "native" ION heap. - add ion_heap_create_full function to keep backward compatibilty. - clean up included files in CMA heap - ux500-ion is using ion_heap_create_full instead of ion_heap_create patches version 3: - add a private field in ion_heap structure instead of expose ion_device structure to all heaps - ion_cma_heap is no more a platform driver - ion_cma_heap use ion_heap private field to store the device pointer and make the link with reserved CMA regions - provide ux500-ion driver and configuration file for snowball board to give an example of how use CMA heaps patches version 2: - fix comments done by Andy Green Benjamin Gaignard (4): fix ion_platform_data definition add private field in ion_heap and ion_platform_heap structure add CMA heap add test/example driver for ux500 platform arch/arm/mach-ux500/board-mop500.c | 64 +++++++++++++ drivers/gpu/ion/Kconfig | 5 + drivers/gpu/ion/Makefile | 5 +- drivers/gpu/ion/ion_cma_heap.c | 179 ++++++++++++++++++++++++++++++++++++ drivers/gpu/ion/ion_heap.c | 11 +++ drivers/gpu/ion/ion_priv.h | 8 ++ drivers/gpu/ion/ux500/Makefile | 1 + drivers/gpu/ion/ux500/ux500_ion.c | 134 +++++++++++++++++++++++++++ include/linux/ion.h | 7 +- 9 files changed, 412 insertions(+), 2 deletions(-) create mode 100644 drivers/gpu/ion/ion_cma_heap.c create mode 100644 drivers/gpu/ion/ux500/Makefile create mode 100644 drivers/gpu/ion/ux500/ux500_ion.c -- 1.7.10

12 years, 6 months

[PATCH 2/2] DRM: Exynos: check for null in return value of dma_buf_map_attachment()

by Subash Patel

From: Subash Patel <subash.rp(a)samsung.com> dma_buf_map_attachment() can return NULL and valid sg as return value. Hence the check for the returned scatter-gather must be using the inline function IS_ERR_OR_NULL() in place of IS_ERR() Change-Id: I33218480e220f6a26a1e726b336bf533a95363de Signed-off-by: Subash Patel <subash.ramaswamy(a)linaro.org> CC: dri-devel(a)lists.freedesktop.org CC: linux-samsung-soc(a)vger.kernel.org CC: linaro-mm-sig(a)lists.linaro.org CC: inki.dae(a)samsung.com CC: airlied(a)redhat.com CC: olofj(a)chromium.org --- drivers/gpu/drm/exynos/exynos_drm_dmabuf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c b/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c index 52cf761..c908a29 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c +++ b/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c @@ -245,7 +245,7 @@ struct drm_gem_object *exynos_dmabuf_prime_import(struct drm_device *drm_dev, sgt = dma_buf_map_attachment(attach, DMA_BIDIRECTIONAL); - if (IS_ERR(sgt)) { + if (IS_ERR_OR_NULL(sgt)) { ret = PTR_ERR(sgt); goto err_buf_detach; } -- 1.7.9.5

12 years, 6 months

[PATCH 1/2] DRM: Exynos: return NULL if exynos_pages_to_sg fails

by Subash Patel

From: Subash Patel <subash.rp(a)samsung.com> exynos_pages_to_sg() internally calls sg_kmalloc() which can return no pages when the system is under high memory crunch. One such instance is chromeos-install in the chromeos. This patch adds check for the return value of the function in subject to return NULL on failure. Change-Id: I541ed30491a926ebe72738225041c9f2d88007bc Signed-off-by: Subash Patel <subash.ramaswamy(a)linaro.org> CC: dri-devel(a)lists.freedesktop.org CC: linux-samsung-soc(a)vger.kernel.org CC: linaro-mm-sig(a)lists.linaro.org CC: inki.dae(a)samsung.com CC: airlied(a)redhat.com CC: olofj(a)chromium.org --- drivers/gpu/drm/exynos/exynos_drm_dmabuf.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c b/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c index 97325c1..52cf761 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c +++ b/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c @@ -87,6 +87,10 @@ static struct sg_table * npages = buf->size / buf->page_size; sgt = exynos_pages_to_sg(buf->pages, npages, buf->page_size); + if (!sgt) { + DRM_DEBUG_PRIME("exynos_pages_to_sg returned NULL!\n"); + goto err_unlock; + } nents = dma_map_sg(attach->dev, sgt->sgl, sgt->nents, dir); DRM_DEBUG_PRIME("npages = %d buffer size = 0x%lx page_size = 0x%lx\n", -- 1.7.9.5

12 years, 6 months

[PATCH] DRM: exynos: return NULL if exynos_pages_to_sg fails

by Subash Patel

From: Subash Patel <subash.rp(a)samsung.com> exynos_pages_to_sg() internally calls sg_kmalloc() which can return no pages when the system is under high memory crunch. One such instance is chromeos-install in the chromeos. This patch adds check for the return value of the function in subject to return NULL on failure. Signed-off-by: Subash Patel <subash.ramaswamy(a)linaro.org> CC: dri-devel(a)lists.freedesktop.org CC: linux-samsung-soc(a)vger.kernel.org CC: linaro-mm-sig(a)lists.linaro.org CC: inki.dae(a)samsung.com CC: airlied(a)redhat.com CC: olofj(a)chromium.org --- drivers/gpu/drm/exynos/exynos_drm_dmabuf.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c b/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c index 97325c1..c908a29 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c +++ b/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c @@ -87,6 +87,10 @@ static struct sg_table * npages = buf->size / buf->page_size; sgt = exynos_pages_to_sg(buf->pages, npages, buf->page_size); + if (!sgt) { + DRM_DEBUG_PRIME("exynos_pages_to_sg returned NULL!\n"); + goto err_unlock; + } nents = dma_map_sg(attach->dev, sgt->sgl, sgt->nents, dir); DRM_DEBUG_PRIME("npages = %d buffer size = 0x%lx page_size = 0x%lx\n", @@ -241,7 +245,7 @@ struct drm_gem_object *exynos_dmabuf_prime_import(struct drm_device *drm_dev, sgt = dma_buf_map_attachment(attach, DMA_BIDIRECTIONAL); - if (IS_ERR(sgt)) { + if (IS_ERR_OR_NULL(sgt)) { ret = PTR_ERR(sgt); goto err_buf_detach; } -- 1.7.9.5

12 years, 6 months

[GIT PULL] DMA-mapping fixes for v3.5-rc5

by Marek Szyprowski

Hi Linus, I would like to ask for pulling another minor fixup for ARM dma-mapping redesign and extensions merged in v3.5-rc1. The following changes since commit 6b16351acbd415e66ba16bf7d473ece1574cf0bc: Linux 3.5-rc4 (2012-06-24 12:53:04 -0700) with the top-most commit 593f47355467b9ef44293698817e2bdb347e2d11 ARM: dma-mapping: fix buffer chunk allocation order are available in the git repository at: git://git.linaro.org/people/mszyprowski/linux-dma-mapping.git fixes-for-linus Marek Szyprowski (1): ARM: dma-mapping: fix buffer chunk allocation order arch/arm/mm/dma-mapping.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) Thanks! Best regards Marek Szyprowski Samsung Poland R&D Center

12 years, 6 months

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

Linaro-mm-sig