Linaro-mm-sig January 2013

linaro-mm-sig@lists.linaro.org

30 participants
13 discussions

[PATCH] mm: dmapool: use provided gfp flags for all dma_alloc_coherent() calls

by Marek Szyprowski

dmapool always calls dma_alloc_coherent() with GFP_ATOMIC flag, regardless the flags provided by the caller. This causes excessive pruning of emergency memory pools without any good reason. This patch changes the code to correctly use gfp flags provided by the dmapool caller. This should solve the dmapool usage on ARM architecture, where GFP_ATOMIC DMA allocations can be served only from the special, very limited memory pool. Reported-by: Soren Moch <smoch(a)web.de> Reported-by: Thomas Petazzoni <thomas.petazzoni(a)free-electrons.com> Signed-off-by: Marek Szyprowski <m.szyprowski(a)samsung.com> --- mm/dmapool.c | 27 +++++++-------------------- 1 file changed, 7 insertions(+), 20 deletions(-) diff --git a/mm/dmapool.c b/mm/dmapool.c index c5ab33b..86de9b2 100644 --- a/mm/dmapool.c +++ b/mm/dmapool.c @@ -62,8 +62,6 @@ struct dma_page { /* cacheable header for 'allocation' bytes */ unsigned int offset; }; -#define POOL_TIMEOUT_JIFFIES ((100 /* msec */ * HZ) / 1000) - static DEFINE_MUTEX(pools_lock); static ssize_t @@ -227,7 +225,6 @@ static struct dma_page *pool_alloc_page(struct dma_pool *pool, gfp_t mem_flags) memset(page->vaddr, POOL_POISON_FREED, pool->allocation); #endif pool_initialise_page(pool, page); - list_add(&page->page_list, &pool->page_list); page->in_use = 0; page->offset = 0; } else { @@ -315,30 +312,21 @@ void *dma_pool_alloc(struct dma_pool *pool, gfp_t mem_flags, might_sleep_if(mem_flags & __GFP_WAIT); spin_lock_irqsave(&pool->lock, flags); - restart: list_for_each_entry(page, &pool->page_list, page_list) { if (page->offset < pool->allocation) goto ready; } - page = pool_alloc_page(pool, GFP_ATOMIC); - if (!page) { - if (mem_flags & __GFP_WAIT) { - DECLARE_WAITQUEUE(wait, current); - __set_current_state(TASK_UNINTERRUPTIBLE); - __add_wait_queue(&pool->waitq, &wait); - spin_unlock_irqrestore(&pool->lock, flags); + /* pool_alloc_page() might sleep, so temporarily drop &pool->lock */ + spin_unlock_irqrestore(&pool->lock, flags); - schedule_timeout(POOL_TIMEOUT_JIFFIES); + page = pool_alloc_page(pool, mem_flags); + if (!page) + return NULL; - spin_lock_irqsave(&pool->lock, flags); - __remove_wait_queue(&pool->waitq, &wait); - goto restart; - } - retval = NULL; - goto done; - } + spin_lock_irqsave(&pool->lock, flags); + list_add(&page->page_list, &pool->page_list); ready: page->in_use++; offset = page->offset; @@ -348,7 +336,6 @@ void *dma_pool_alloc(struct dma_pool *pool, gfp_t mem_flags, #ifdef DMAPOOL_DEBUG memset(retval, POOL_POISON_ALLOCATED, pool->size); #endif - done: spin_unlock_irqrestore(&pool->lock, flags); return retval; } -- 1.7.9.5

11 years, 7 months

[PATCH 0/2] ARM: dma-mapping: add highmem support for coherent allocation

by Marek Szyprowski

Hello, This is the last missing piece to let us efficiently use large DMA buffers on systems with lots of memory, which have support for himem enabled. The first patch adds support for CMA regions placed in high memory zones, the second one also enables allocations of individual pages from high memory zone for IOMMU-mapped devices. Those two changes let us to significantly save low memory for other tasks. Best regards Marek Szyprowski Samsung Poland R&D Center Patch summary: Marek Szyprowski (2): ARM: dma-mapping: add support for CMA regions placed in highmem zone ARM: dma-mapping: use himem for DMA buffers for IOMMU-mapped devices arch/arm/mm/dma-mapping.c | 70 +++++++++++++++++++++++++++++++++------------ 1 file changed, 52 insertions(+), 18 deletions(-) -- 1.7.9.5

11 years, 10 months

[PATCH 0/7] cross-device reservation for dma-buf support

by Maarten Lankhorst

So I'm resending the patch series for reservations. This is identical to my git tree at http://cgit.freedesktop.org/~mlankhorst/linux/ Some changes have been made since my last version. Most notably is the use of mutexes now instead of creating my own lock primitive, that would end up being duplicate anyway. The git tree also has a version of i915 and radeon working together like that. It's probably a bit hacky, but it works on my macbook pro 8.2. :) I haven't had any reply on the mutex extensions when I sent them out separately, so I'm including it in the series. The idea is that for lockdep purposes, the seqno is tied to a locking a class. This locking class it not exclusive, but as can be seen from the last patch in the series, it catches all violations we care about.

11 years, 10 months

CDF discussions at FOSDEM

by Laurent Pinchart

Hi everybody, Would anyone be interested in meeting at the FOSDEM to discuss the Common Display Framework ? There will be a CDF meeting at the ELC at the end of February, the FOSDEM would be a good venue for European developers. -- Regards, Laurent Pinchart

11 years, 10 months

[RFC] ARM: dma-mapping: Return 0 if no ->set_dma_mask()

by Hiroshi Doyu

struct dma_map_ops iommu_ops doesn't have ->set_dma_mask, which causes crash when dma_set_mask() is called from some driver. Signed-off-by: Hiroshi Doyu <hdoyu(a)nvidia.com> --- arch/arm/include/asm/dma-mapping.h | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/arch/arm/include/asm/dma-mapping.h b/arch/arm/include/asm/dma-mapping.h index a58e0f5..95328bf 100644 --- a/arch/arm/include/asm/dma-mapping.h +++ b/arch/arm/include/asm/dma-mapping.h @@ -32,7 +32,11 @@ static inline void set_dma_ops(struct device *dev, struct dma_map_ops *ops) static inline int dma_set_mask(struct device *dev, u64 mask) { - return get_dma_ops(dev)->set_dma_mask(dev, mask); + struct dma_map_ops *ops = get_dma_ops(dev); + + if (ops->set_dma_mask) + return ops->set_dma_mask(dev, mask); + return 0; } #ifdef __arch_page_to_dma -- 1.7.9.5

11 years, 10 months

Common Display Framework BoF at ELC

by Jesse Barker

Hi all, I've just confirmed the schedule for ELC ( http://events.linuxfoundation.org/events/embedded-linux-conference/schedule), and we're on for Thursday at 4pm. I requested 2 slots to give us some flexibility (that's the "Part I & Part II" on the schedule). Whether you are attending the rest of ELC or not, if you care about CDF, please come by. cheers, Jesse

11 years, 11 months

[RFC] ARM: dma-mapping: Add DMA attribute to skip iommu mapping

by Abhinav Kochhar

Adding a new dma attribute which can be used by the platform drivers to avoid creating iommu mappings. In some cases the buffers are allocated by display controller driver using dma alloc apis but are not used for scanout. Though the buffers are allocated by display controller but are only used for sharing among different devices. With this attribute the platform drivers can choose not to create iommu mapping at the time of buffer allocation and only create the mapping when they access this buffer. Change-Id: I2178b3756170982d814e085ca62474d07b616a21 Signed-off-by: Abhinav Kochhar <abhinav.k(a)samsung.com> --- arch/arm/mm/dma-mapping.c | 8 +++++--- include/linux/dma-attrs.h | 1 + 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c index c0f0f43..e73003c 100644 --- a/arch/arm/mm/dma-mapping.c +++ b/arch/arm/mm/dma-mapping.c @@ -1279,9 +1279,11 @@ static void *arm_iommu_alloc_attrs(struct device *dev, size_t size, if (!pages) return NULL; - *handle = __iommu_create_mapping(dev, pages, size); - if (*handle == DMA_ERROR_CODE) - goto err_buffer; + if (!dma_get_attr(DMA_ATTR_NO_IOMMU_MAPPING, attrs)) { + *handle = __iommu_create_mapping(dev, pages, size); + if (*handle == DMA_ERROR_CODE) + goto err_buffer; + } if (dma_get_attr(DMA_ATTR_NO_KERNEL_MAPPING, attrs)) return pages; diff --git a/include/linux/dma-attrs.h b/include/linux/dma-attrs.h index c8e1831..1f04419 100644 --- a/include/linux/dma-attrs.h +++ b/include/linux/dma-attrs.h @@ -15,6 +15,7 @@ enum dma_attr { DMA_ATTR_WEAK_ORDERING, DMA_ATTR_WRITE_COMBINE, DMA_ATTR_NON_CONSISTENT, + DMA_ATTR_NO_IOMMU_MAPPING, DMA_ATTR_NO_KERNEL_MAPPING, DMA_ATTR_SKIP_CPU_SYNC, DMA_ATTR_FORCE_CONTIGUOUS, -- 1.7.8.6

11 years, 11 months

Attendance at ELC San Francisco

by Jesse Barker

Hi all, With all of the attention that the Common Display Framework is getting, I was wondering if it was worth having a BoF discussion at ELC next month in San Francisco. This will be only a couple of weeks after FOSDEM, but given the pace that things seem to be moving, that could be a great opportunity either to have a follow-on discussion, or simply to involve a slightly different cross-section of the community in a face-to-face discussion. If folks could let me know that they'll be at ELC and are interested in a BoF there, I'll look into getting it set up. cheers, Jesse

11 years, 11 months

[PATCH] gpu: ion: Add ion_share_dma_buf_kernel

by Johan Mossberg

ion_share_dma_buf_kernel enables you to share ion buffers via dma buf for kernel only use cases. Useful for example when a GPU driver using ion wants to share its output buffers with a 3d party display controller driver supporting dma buf. Signed-off-by: Johan Mossberg <johan.mossberg(a)stericsson.com> --- drivers/gpu/ion/ion.c | 18 ++++++++++++++++-- include/linux/ion.h | 8 ++++++++ 2 files changed, 24 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/ion/ion.c b/drivers/gpu/ion/ion.c index 748ff7d..31bb5e1 100644 --- a/drivers/gpu/ion/ion.c +++ b/drivers/gpu/ion/ion.c @@ -991,12 +991,12 @@ struct dma_buf_ops dma_buf_ops = { .kunmap = ion_dma_buf_kunmap, }; -int ion_share_dma_buf(struct ion_client *client, struct ion_handle *handle) +struct dma_buf *ion_share_dma_buf_kernel(struct ion_client *client, + struct ion_handle *handle) { struct ion_buffer *buffer; struct dma_buf *dmabuf; bool valid_handle; - int fd; mutex_lock(&client->lock); valid_handle = ion_handle_validate(client, handle); @@ -1013,6 +1013,20 @@ int ion_share_dma_buf(struct ion_client *client, struct ion_handle *handle) ion_buffer_put(buffer); return PTR_ERR(dmabuf); } + + return dmabuf; +} +EXPORT_SYMBOL(ion_share_dma_buf_kernel); + +int ion_share_dma_buf(struct ion_client *client, struct ion_handle *handle) +{ + struct dma_buf *dmabuf; + int fd; + + dmabuf = ion_share_dma_buf_kernel(client, handle); + if (IS_ERR(dmabuf)) + return PTR_ERR(dmabuf); + fd = dma_buf_fd(dmabuf, O_CLOEXEC); if (fd < 0) dma_buf_put(dmabuf); diff --git a/include/linux/ion.h b/include/linux/ion.h index 3ccc75a..cfb831d 100644 --- a/include/linux/ion.h +++ b/include/linux/ion.h @@ -210,6 +210,14 @@ void *ion_map_kernel(struct ion_client *client, struct ion_handle *handle); void ion_unmap_kernel(struct ion_client *client, struct ion_handle *handle); /** + * ion_share_dma_buf_kernel() - share buffer as dma-buf + * @client: the client + * @handle: the handle + */ +struct dma_buf *ion_share_dma_buf_kernel(struct ion_client *client, + struct ion_handle *buf); + +/** * ion_share_dma_buf() - given an ion client, create a dma-buf fd * @client: the client * @handle: the handle -- 1.8.0

11 years, 11 months

Re: [Linaro-mm-sig] [PATCHv2] gpu: ion: Add ion_share_dma_buf_kernel

by Johan Mossberg

Hi Rebecca, Is this patch accepted for ion mainline or do you have more comments/questions? /Johan Mossberg On 12/13/2012 10:24 AM, Johan Mossberg wrote: > ion_share_dma_buf_kernel enables you to share ion buffers via dma buf > for kernel only use cases. Useful for example when a GPU driver using > ion wants to share its output buffers with a 3d party display > controller driver supporting dma buf. > > Signed-off-by: Johan Mossberg <johan.mossberg(a)stericsson.com> > --- > drivers/gpu/ion/ion.c | 22 ++++++++++++++++++---- > include/linux/ion.h | 8 ++++++++ > 2 files changed, 26 insertions(+), 4 deletions(-) > > diff --git a/drivers/gpu/ion/ion.c b/drivers/gpu/ion/ion.c > index 3872095..e7b0d0b 100644 > --- a/drivers/gpu/ion/ion.c > +++ b/drivers/gpu/ion/ion.c > @@ -955,19 +955,19 @@ struct dma_buf_ops dma_buf_ops = { > .kunmap = ion_dma_buf_kunmap, > }; > > -int ion_share_dma_buf(struct ion_client *client, struct ion_handle *handle) > +struct dma_buf *ion_share_dma_buf_kernel(struct ion_client *client, > + struct ion_handle *handle) > { > struct ion_buffer *buffer; > struct dma_buf *dmabuf; > bool valid_handle; > - int fd; > > mutex_lock(&client->lock); > valid_handle = ion_handle_validate(client, handle); > mutex_unlock(&client->lock); > if (!valid_handle) { > WARN(1, "%s: invalid handle passed to share.\n", __func__); > - return -EINVAL; > + return ERR_PTR(-EINVAL); > } > > buffer = handle->buffer; > @@ -975,8 +975,22 @@ int ion_share_dma_buf(struct ion_client *client, struct ion_handle *handle) > dmabuf = dma_buf_export(buffer, &dma_buf_ops, buffer->size, O_RDWR); > if (IS_ERR(dmabuf)) { > ion_buffer_put(buffer); > - return PTR_ERR(dmabuf); > + return dmabuf; > } > + > + return dmabuf; > +} > +EXPORT_SYMBOL(ion_share_dma_buf_kernel); > + > +int ion_share_dma_buf(struct ion_client *client, struct ion_handle *handle) > +{ > + struct dma_buf *dmabuf; > + int fd; > + > + dmabuf = ion_share_dma_buf_kernel(client, handle); > + if (IS_ERR(dmabuf)) > + return PTR_ERR(dmabuf); > + > fd = dma_buf_fd(dmabuf, O_CLOEXEC); > if (fd < 0) > dma_buf_put(dmabuf); > diff --git a/include/linux/ion.h b/include/linux/ion.h > index a7d399c..8720e9b 100644 > --- a/include/linux/ion.h > +++ b/include/linux/ion.h > @@ -205,6 +205,14 @@ void *ion_map_kernel(struct ion_client *client, struct ion_handle *handle); > void ion_unmap_kernel(struct ion_client *client, struct ion_handle *handle); > > /** > + * ion_share_dma_buf_kernel() - share buffer as dma-buf > + * @client: the client > + * @handle: the handle > + */ > +struct dma_buf *ion_share_dma_buf_kernel(struct ion_client *client, > + struct ion_handle *buf); > + > +/** > * ion_share_dma_buf() - given an ion client, create a dma-buf fd > * @client: the client > * @handle: the handle > -- > 1.8.0 > > > _______________________________________________ > Linaro-mm-sig mailing list > Linaro-mm-sig(a)lists.linaro.org > http://lists.linaro.org/mailman/listinfo/linaro-mm-sig > >

11 years, 11 months

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

Linaro-mm-sig January 2013